1FFMPEG-FILTERS(1) FFMPEG-FILTERS(1)
2
6 ffmpeg-filters - FFmpeg filters
7
9 This document describes filters, sources, and sinks provided by the
10 libavfilter library.
11
13 Filtering in FFmpeg is enabled through the libavfilter library.
14 In libavfilter, a filter can have multiple inputs and multiple outputs.
16 To illustrate the sorts of things that are possible, we consider the
17 following filtergraph.
18 [main]
20 input --> split ---------------------> overlay --> output
21 | ^
22 |[tmp] [flip]|
23 +-----> crop --> vflip -------+
24 This filtergraph splits the input stream in two streams, then sends one
26 stream through the crop filter and the vflip filter, before merging it
27 back with the other stream by overlaying it on top. You can use the
28 following command to achieve this:
29 ffmpeg -i INPUT -vf "split [main][tmp]; [tmp] crop=iw:ih/2:0:0, vflip [flip]; [main][flip] overlay=0:H/2" OUTPUT
31 The result will be that the top half of the video is mirrored onto the
33 bottom half of the output video.
34 Filters in the same linear chain are separated by commas, and distinct
36 linear chains of filters are separated by semicolons. In our example,
37 crop,vflip are in one linear chain, split and overlay are separately in
38 another. The points where the linear chains join are labelled by names
39 enclosed in square brackets. In the example, the split filter generates
40 two outputs that are associated to the labels [main] and [tmp].
41 The stream sent to the second output of split, labelled as [tmp], is
43 processed through the crop filter, which crops away the lower half part
44 of the video, and then vertically flipped. The overlay filter takes in
45 input the first unchanged output of the split filter (which was
46 labelled as [main]), and overlay on its lower half the output generated
47 by the crop,vflip filterchain.
48 Some filters take in input a list of parameters: they are specified
50 after the filter name and an equal sign, and are separated from each
51 other by a colon.
52 There exist so-called source filters that do not have an audio/video
54 input, and sink filters that will not have audio/video output.
55
57 The graph2dot program included in the FFmpeg tools directory can be
58 used to parse a filtergraph description and issue a corresponding
59 textual representation in the dot language.
60 Invoke the command:
62 graph2dot -h
64 to see how to use graph2dot.
66 You can then pass the dot description to the dot program (from the
68 graphviz suite of programs) and obtain a graphical representation of
69 the filtergraph.
70 For example the sequence of commands:
72 echo <GRAPH_DESCRIPTION> | \
74 tools/graph2dot -o graph.tmp && \
75 dot -Tpng graph.tmp -o graph.png && \
76 display graph.png
77 can be used to create and display an image representing the graph
79 described by the GRAPH_DESCRIPTION string. Note that this string must
80 be a complete self-contained graph, with its inputs and outputs
81 explicitly defined. For example if your command line is of the form:
82 ffmpeg -i infile -vf scale=640:360 outfile
84 your GRAPH_DESCRIPTION string will need to be of the form:
86 nullsrc,scale=640:360,nullsink
88 you may also need to set the nullsrc parameters and add a format filter
90 in order to simulate a specific input file.
91
93 A filtergraph is a directed graph of connected filters. It can contain
94 cycles, and there can be multiple links between a pair of filters. Each
95 link has one input pad on one side connecting it to one filter from
96 which it takes its input, and one output pad on the other side
97 connecting it to one filter accepting its output.
98 Each filter in a filtergraph is an instance of a filter class
100 registered in the application, which defines the features and the
101 number of input and output pads of the filter.
102 A filter with no input pads is called a "source", and a filter with no
104 output pads is called a "sink".
105 Filtergraph syntax
107 A filtergraph has a textual representation, which is recognized by the
108 -filter/-vf/-af and -filter_complex options in ffmpeg and -vf/-af in
109 ffplay, and by the "avfilter_graph_parse_ptr()" function defined in
110 libavfilter/avfilter.h.
111 A filterchain consists of a sequence of connected filters, each one
113 connected to the previous one in the sequence. A filterchain is
114 represented by a list of ","-separated filter descriptions.
115 A filtergraph consists of a sequence of filterchains. A sequence of
117 filterchains is represented by a list of ";"-separated filterchain
118 descriptions.
119 A filter is represented by a string of the form:
121 [in_link_1]...[in_link_N]filter_name@id=arguments[out_link_1]...[out_link_M]
122 filter_name is the name of the filter class of which the described
124 filter is an instance of, and has to be the name of one of the filter
125 classes registered in the program optionally followed by "@id". The
126 name of the filter class is optionally followed by a string
127 "=arguments".
128 arguments is a string which contains the parameters used to initialize
130 the filter instance. It may have one of two forms:
131 • A ':'-separated list of key=value pairs.
133 • A ':'-separated list of value. In this case, the keys are assumed
135 to be the option names in the order they are declared. E.g. the
136 "fade" filter declares three options in this order -- type,
137 start_frame and nb_frames. Then the parameter list in:0:30 means
138 that the value in is assigned to the option type, 0 to start_frame
139 and 30 to nb_frames.
140 • A ':'-separated list of mixed direct value and long key=value
142 pairs. The direct value must precede the key=value pairs, and
143 follow the same constraints order of the previous point. The
144 following key=value pairs can be set in any preferred order.
145 If the option value itself is a list of items (e.g. the "format" filter
147 takes a list of pixel formats), the items in the list are usually
148 separated by |.
149 The list of arguments can be quoted using the character ' as initial
151 and ending mark, and the character \ for escaping the characters within
152 the quoted text; otherwise the argument string is considered terminated
153 when the next special character (belonging to the set []=;,) is
154 encountered.
155 The name and arguments of the filter are optionally preceded and
157 followed by a list of link labels. A link label allows one to name a
158 link and associate it to a filter output or input pad. The preceding
159 labels in_link_1 ... in_link_N, are associated to the filter input
160 pads, the following labels out_link_1 ... out_link_M, are associated to
161 the output pads.
162 When two link labels with the same name are found in the filtergraph, a
164 link between the corresponding input and output pad is created.
165 If an output pad is not labelled, it is linked by default to the first
167 unlabelled input pad of the next filter in the filterchain. For
168 example in the filterchain
169 nullsrc, split[L1], [L2]overlay, nullsink
171 the split filter instance has two output pads, and the overlay filter
173 instance two input pads. The first output pad of split is labelled
174 "L1", the first input pad of overlay is labelled "L2", and the second
175 output pad of split is linked to the second input pad of overlay, which
176 are both unlabelled.
177 In a filter description, if the input label of the first filter is not
179 specified, "in" is assumed; if the output label of the last filter is
180 not specified, "out" is assumed.
181 In a complete filterchain all the unlabelled filter input and output
183 pads must be connected. A filtergraph is considered valid if all the
184 filter input and output pads of all the filterchains are connected.
185 Libavfilter will automatically insert scale filters where format
187 conversion is required. It is possible to specify swscale flags for
188 those automatically inserted scalers by prepending "sws_flags=flags;"
189 to the filtergraph description.
190 Here is a BNF description of the filtergraph syntax:
192 <NAME> ::= sequence of alphanumeric characters and '_'
194 <FILTER_NAME> ::= <NAME>["@"<NAME>]
195 <LINKLABEL> ::= "[" <NAME> "]"
196 <LINKLABELS> ::= <LINKLABEL> [<LINKLABELS>]
197 <FILTER_ARGUMENTS> ::= sequence of chars (possibly quoted)
198 <FILTER> ::= [<LINKLABELS>] <FILTER_NAME> ["=" <FILTER_ARGUMENTS>] [<LINKLABELS>]
199 <FILTERCHAIN> ::= <FILTER> [,<FILTERCHAIN>]
200 <FILTERGRAPH> ::= [sws_flags=<flags>;] <FILTERCHAIN> [;<FILTERGRAPH>]
201 Notes on filtergraph escaping
203 Filtergraph description composition entails several levels of escaping.
204 See the "Quoting and escaping" section in the ffmpeg-utils(1) manual
205 for more information about the employed escaping procedure.
206 A first level escaping affects the content of each filter option value,
208 which may contain the special character ":" used to separate values, or
209 one of the escaping characters "\'".
210 A second level escaping affects the whole filter description, which may
212 contain the escaping characters "\'" or the special characters "[],;"
213 used by the filtergraph description.
214 Finally, when you specify a filtergraph on a shell commandline, you
216 need to perform a third level escaping for the shell special characters
217 contained within it.
218 For example, consider the following string to be embedded in the
220 drawtext filter description text value:
221 this is a 'string': may contain one, or more, special characters
223 This string contains the "'" special escaping character, and the ":"
225 special character, so it needs to be escaped in this way:
226 text=this is a \'string\'\: may contain one, or more, special characters
228 A second level of escaping is required when embedding the filter
230 description in a filtergraph description, in order to escape all the
231 filtergraph special characters. Thus the example above becomes:
232 drawtext=text=this is a \\\'string\\\'\\: may contain one\, or more\, special characters
234 (note that in addition to the "\'" escaping special characters, also
236 "," needs to be escaped).
237 Finally an additional level of escaping is needed when writing the
239 filtergraph description in a shell command, which depends on the
240 escaping rules of the adopted shell. For example, assuming that "\" is
241 special and needs to be escaped with another "\", the previous string
242 will finally result in:
243 -vf "drawtext=text=this is a \\\\\\'string\\\\\\'\\\\: may contain one\\, or more\\, special characters"
245
247 Some filters support a generic enable option. For the filters
248 supporting timeline editing, this option can be set to an expression
249 which is evaluated before sending a frame to the filter. If the
250 evaluation is non-zero, the filter will be enabled, otherwise the frame
251 will be sent unchanged to the next filter in the filtergraph.
252 The expression accepts the following values:
254 t timestamp expressed in seconds, NAN if the input timestamp is
256 unknown
257 n sequential number of the input frame, starting from 0
259 pos the position in the file of the input frame, NAN if unknown
261 w
263 h width and height of the input frame if video
264 Additionally, these filters support an enable command that can be used
266 to re-define the expression.
267 Like any other filtering option, the enable option follows the same
269 rules.
270 For example, to enable a blur filter (smartblur) from 10 seconds to 3
272 minutes, and a curves filter starting at 3 seconds:
273 smartblur = enable='between(t,10,3*60)',
275 curves = enable='gte(t,3)' : preset=cross_process
276 See "ffmpeg -filters" to view which filters have timeline support.
278
280 Some options can be changed during the operation of the filter using a
281 command. These options are marked 'T' on the output of ffmpeg -h
282 filter=<name of filter>. The name of the command is the name of the
283 option and the argument is the new value.
284
286 Some filters with several inputs support a common set of options.
287 These options can only be set by name, not with the short notation.
288 eof_action
290 The action to take when EOF is encountered on the secondary input;
291 it accepts one of the following values:
292 repeat
294 Repeat the last frame (the default).
295 endall
297 End both streams.
298 pass
300 Pass the main input through.
301 shortest
303 If set to 1, force the output to terminate when the shortest input
304 terminates. Default value is 0.
305 repeatlast
307 If set to 1, force the filter to extend the last frame of secondary
308 streams until the end of the primary stream. A value of 0 disables
309 this behavior. Default value is 1.
310
312 When you configure your FFmpeg build, you can disable any of the
313 existing filters using "--disable-filters". The configure output will
314 show the audio filters included in your build.
315 Below is a description of the currently available audio filters.
317 acompressor
319 A compressor is mainly used to reduce the dynamic range of a signal.
320 Especially modern music is mostly compressed at a high ratio to improve
321 the overall loudness. It's done to get the highest attention of a
322 listener, "fatten" the sound and bring more "power" to the track. If a
323 signal is compressed too much it may sound dull or "dead" afterwards or
324 it may start to "pump" (which could be a powerful effect but can also
325 destroy a track completely). The right compression is the key to reach
326 a professional sound and is the high art of mixing and mastering.
327 Because of its complex settings it may take a long time to get the
328 right feeling for this kind of effect.
329 Compression is done by detecting the volume above a chosen level
331 "threshold" and dividing it by the factor set with "ratio". So if you
332 set the threshold to -12dB and your signal reaches -6dB a ratio of 2:1
333 will result in a signal at -9dB. Because an exact manipulation of the
334 signal would cause distortion of the waveform the reduction can be
335 levelled over the time. This is done by setting "Attack" and "Release".
336 "attack" determines how long the signal has to rise above the threshold
337 before any reduction will occur and "release" sets the time the signal
338 has to fall below the threshold to reduce the reduction again. Shorter
339 signals than the chosen attack time will be left untouched. The
340 overall reduction of the signal can be made up afterwards with the
341 "makeup" setting. So compressing the peaks of a signal about 6dB and
342 raising the makeup to this level results in a signal twice as loud than
343 the source. To gain a softer entry in the compression the "knee"
344 flattens the hard edge at the threshold in the range of the chosen
345 decibels.
346 The filter accepts the following options:
348 level_in
350 Set input gain. Default is 1. Range is between 0.015625 and 64.
351 mode
353 Set mode of compressor operation. Can be "upward" or "downward".
354 Default is "downward".
355 threshold
357 If a signal of stream rises above this level it will affect the
358 gain reduction. By default it is 0.125. Range is between
359 0.00097563 and 1.
360 ratio
362 Set a ratio by which the signal is reduced. 1:2 means that if the
363 level rose 4dB above the threshold, it will be only 2dB above after
364 the reduction. Default is 2. Range is between 1 and 20.
365 attack
367 Amount of milliseconds the signal has to rise above the threshold
368 before gain reduction starts. Default is 20. Range is between 0.01
369 and 2000.
370 release
372 Amount of milliseconds the signal has to fall below the threshold
373 before reduction is decreased again. Default is 250. Range is
374 between 0.01 and 9000.
375 makeup
377 Set the amount by how much signal will be amplified after
378 processing. Default is 1. Range is from 1 to 64.
379 knee
381 Curve the sharp knee around the threshold to enter gain reduction
382 more softly. Default is 2.82843. Range is between 1 and 8.
383 link
385 Choose if the "average" level between all channels of input stream
386 or the louder("maximum") channel of input stream affects the
387 reduction. Default is "average".
388 detection
390 Should the exact signal be taken in case of "peak" or an RMS one in
391 case of "rms". Default is "rms" which is mostly smoother.
392 mix How much to use compressed signal in output. Default is 1. Range
394 is between 0 and 1.
395 Commands
397 This filter supports the all above options as commands.
399 acontrast
401 Simple audio dynamic range compression/expansion filter.
402 The filter accepts the following options:
404 contrast
406 Set contrast. Default is 33. Allowed range is between 0 and 100.
407 acopy
409 Copy the input audio source unchanged to the output. This is mainly
410 useful for testing purposes.
411 acrossfade
413 Apply cross fade from one input audio stream to another input audio
414 stream. The cross fade is applied for specified duration near the end
415 of first stream.
416 The filter accepts the following options:
418 nb_samples, ns
420 Specify the number of samples for which the cross fade effect has
421 to last. At the end of the cross fade effect the first input audio
422 will be completely silent. Default is 44100.
423 duration, d
425 Specify the duration of the cross fade effect. See the Time
426 duration section in the ffmpeg-utils(1) manual for the accepted
427 syntax. By default the duration is determined by nb_samples. If
428 set this option is used instead of nb_samples.
429 overlap, o
431 Should first stream end overlap with second stream start. Default
432 is enabled.
433 curve1
435 Set curve for cross fade transition for first stream.
436 curve2
438 Set curve for cross fade transition for second stream.
439 For description of available curve types see afade filter
441 description.
442 Examples
444 • Cross fade from one input to another:
446 ffmpeg -i first.flac -i second.flac -filter_complex acrossfade=d=10:c1=exp:c2=exp output.flac
448 • Cross fade from one input to another but without overlapping:
450 ffmpeg -i first.flac -i second.flac -filter_complex acrossfade=d=10:o=0:c1=exp:c2=exp output.flac
452 acrossover
454 Split audio stream into several bands.
455 This filter splits audio stream into two or more frequency ranges.
457 Summing all streams back will give flat output.
458 The filter accepts the following options:
460 split
462 Set split frequencies. Those must be positive and increasing.
463 order
465 Set filter order for each band split. This controls filter roll-off
466 or steepness of filter transfer function. Available values are:
467 2nd 12 dB per octave.
469 4th 24 dB per octave.
471 6th 36 dB per octave.
473 8th 48 dB per octave.
475 10th
477 60 dB per octave.
478 12th
480 72 dB per octave.
481 14th
483 84 dB per octave.
484 16th
486 96 dB per octave.
487 18th
489 108 dB per octave.
490 20th
492 120 dB per octave.
493 Default is 4th.
495 level
497 Set input gain level. Allowed range is from 0 to 1. Default value
498 is 1.
499 gains
501 Set output gain for each band. Default value is 1 for all bands.
502 precision
504 Set which precision to use when processing samples.
505 auto
507 Auto pick internal sample format depending on other filters.
508 float
510 Always use single-floating point precision sample format.
511 double
513 Always use double-floating point precision sample format.
514 Default value is "auto".
516 Examples
518 • Split input audio stream into two bands (low and high) with split
520 frequency of 1500 Hz, each band will be in separate stream:
521 ffmpeg -i in.flac -filter_complex 'acrossover=split=1500[LOW][HIGH]' -map '[LOW]' low.wav -map '[HIGH]' high.wav
523 • Same as above, but with higher filter order:
525 ffmpeg -i in.flac -filter_complex 'acrossover=split=1500:order=8th[LOW][HIGH]' -map '[LOW]' low.wav -map '[HIGH]' high.wav
527 • Same as above, but also with additional middle band (frequencies
529 between 1500 and 8000):
530 ffmpeg -i in.flac -filter_complex 'acrossover=split=1500 8000:order=8th[LOW][MID][HIGH]' -map '[LOW]' low.wav -map '[MID]' mid.wav -map '[HIGH]' high.wav
532 acrusher
534 Reduce audio bit resolution.
535 This filter is bit crusher with enhanced functionality. A bit crusher
537 is used to audibly reduce number of bits an audio signal is sampled
538 with. This doesn't change the bit depth at all, it just produces the
539 effect. Material reduced in bit depth sounds more harsh and "digital".
540 This filter is able to even round to continuous values instead of
541 discrete bit depths. Additionally it has a D/C offset which results in
542 different crushing of the lower and the upper half of the signal. An
543 Anti-Aliasing setting is able to produce "softer" crushing sounds.
544 Another feature of this filter is the logarithmic mode. This setting
546 switches from linear distances between bits to logarithmic ones. The
547 result is a much more "natural" sounding crusher which doesn't gate low
548 signals for example. The human ear has a logarithmic perception, so
549 this kind of crushing is much more pleasant. Logarithmic crushing is
550 also able to get anti-aliased.
551 The filter accepts the following options:
553 level_in
555 Set level in.
556 level_out
558 Set level out.
559 bits
561 Set bit reduction.
562 mix Set mixing amount.
564 mode
566 Can be linear: "lin" or logarithmic: "log".
567 dc Set DC.
569 aa Set anti-aliasing.
571 samples
573 Set sample reduction.
574 lfo Enable LFO. By default disabled.
576 lforange
578 Set LFO range.
579 lforate
581 Set LFO rate.
582 Commands
584 This filter supports the all above options as commands.
586 acue
588 Delay audio filtering until a given wallclock timestamp. See the cue
589 filter.
590 adeclick
592 Remove impulsive noise from input audio.
593 Samples detected as impulsive noise are replaced by interpolated
595 samples using autoregressive modelling.
596 window, w
598 Set window size, in milliseconds. Allowed range is from 10 to 100.
599 Default value is 55 milliseconds. This sets size of window which
600 will be processed at once.
601 overlap, o
603 Set window overlap, in percentage of window size. Allowed range is
604 from 50 to 95. Default value is 75 percent. Setting this to a very
605 high value increases impulsive noise removal but makes whole
606 process much slower.
607 arorder, a
609 Set autoregression order, in percentage of window size. Allowed
610 range is from 0 to 25. Default value is 2 percent. This option also
611 controls quality of interpolated samples using neighbour good
612 samples.
613 threshold, t
615 Set threshold value. Allowed range is from 1 to 100. Default value
616 is 2. This controls the strength of impulsive noise which is going
617 to be removed. The lower value, the more samples will be detected
618 as impulsive noise.
619 burst, b
621 Set burst fusion, in percentage of window size. Allowed range is 0
622 to 10. Default value is 2. If any two samples detected as noise
623 are spaced less than this value then any sample between those two
624 samples will be also detected as noise.
625 method, m
627 Set overlap method.
628 It accepts the following values:
630 add, a
632 Select overlap-add method. Even not interpolated samples are
633 slightly changed with this method.
634 save, s
636 Select overlap-save method. Not interpolated samples remain
637 unchanged.
638 Default value is "a".
640 adeclip
642 Remove clipped samples from input audio.
643 Samples detected as clipped are replaced by interpolated samples using
645 autoregressive modelling.
646 window, w
648 Set window size, in milliseconds. Allowed range is from 10 to 100.
649 Default value is 55 milliseconds. This sets size of window which
650 will be processed at once.
651 overlap, o
653 Set window overlap, in percentage of window size. Allowed range is
654 from 50 to 95. Default value is 75 percent.
655 arorder, a
657 Set autoregression order, in percentage of window size. Allowed
658 range is from 0 to 25. Default value is 8 percent. This option also
659 controls quality of interpolated samples using neighbour good
660 samples.
661 threshold, t
663 Set threshold value. Allowed range is from 1 to 100. Default value
664 is 10. Higher values make clip detection less aggressive.
665 hsize, n
667 Set size of histogram used to detect clips. Allowed range is from
668 100 to 9999. Default value is 1000. Higher values make clip
669 detection less aggressive.
670 method, m
672 Set overlap method.
673 It accepts the following values:
675 add, a
677 Select overlap-add method. Even not interpolated samples are
678 slightly changed with this method.
679 save, s
681 Select overlap-save method. Not interpolated samples remain
682 unchanged.
683 Default value is "a".
685 adecorrelate
687 Apply decorrelation to input audio stream.
688 The filter accepts the following options:
690 stages
692 Set decorrelation stages of filtering. Allowed range is from 1 to
693 16. Default value is 6.
694 seed
696 Set random seed used for setting delay in samples across channels.
697 adelay
699 Delay one or more audio channels.
700 Samples in delayed channel are filled with silence.
702 The filter accepts the following option:
704 delays
706 Set list of delays in milliseconds for each channel separated by
707 '|'. Unused delays will be silently ignored. If number of given
708 delays is smaller than number of channels all remaining channels
709 will not be delayed. If you want to delay exact number of samples,
710 append 'S' to number. If you want instead to delay in seconds,
711 append 's' to number.
712 all Use last set delay for all remaining channels. By default is
714 disabled. This option if enabled changes how option "delays" is
715 interpreted.
716 Examples
718 • Delay first channel by 1.5 seconds, the third channel by 0.5
720 seconds and leave the second channel (and any other channels that
721 may be present) unchanged.
722 adelay=1500|0|500
724 • Delay second channel by 500 samples, the third channel by 700
726 samples and leave the first channel (and any other channels that
727 may be present) unchanged.
728 adelay=0|500S|700S
730 • Delay all channels by same number of samples:
732 adelay=delays=64S:all=1
734 adenorm
736 Remedy denormals in audio by adding extremely low-level noise.
737 This filter shall be placed before any filter that can produce
739 denormals.
740 A description of the accepted parameters follows.
742 level
744 Set level of added noise in dB. Default is "-351". Allowed range
745 is from -451 to -90.
746 type
748 Set type of added noise.
749 dc Add DC signal.
751 ac Add AC signal.
753 square
755 Add square signal.
756 pulse
758 Add pulse signal.
759 Default is "dc".
761 Commands
763 This filter supports the all above options as commands.
765 aderivative, aintegral
767 Compute derivative/integral of audio stream.
768 Applying both filters one after another produces original audio.
770 adynamicequalizer
772 Apply dynamic equalization to input audio stream.
773 A description of the accepted options follows.
775 threshold
777 Set the detection threshold used to trigger equalization.
778 Threshold detection is using bandpass filter. Default value is 0.
779 Allowed range is from 0 to 100.
780 dfrequency
782 Set the detection frequency in Hz used for bandpass filter used to
783 trigger equalization. Default value is 1000 Hz. Allowed range is
784 between 2 and 1000000 Hz.
785 dqfactor
787 Set the detection resonance factor for bandpass filter used to
788 trigger equalization. Default value is 1. Allowed range is from
789 0.001 to 1000.
790 tfrequency
792 Set the target frequency of equalization filter. Default value is
793 1000 Hz. Allowed range is between 2 and 1000000 Hz.
794 tqfactor
796 Set the target resonance factor for target equalization filter.
797 Default value is 1. Allowed range is from 0.001 to 1000.
798 attack
800 Set the amount of milliseconds the signal from detection has to
801 rise above the detection threshold before equalization starts.
802 Default is 20. Allowed range is between 1 and 2000.
803 release
805 Set the amount of milliseconds the signal from detection has to
806 fall below the detection threshold before equalization ends.
807 Default is 200. Allowed range is between 1 and 2000.
808 knee
810 Curve the sharp knee around the detection threshold to calculate
811 equalization gain more softly. Default is 1. Allowed range is
812 between 0 and 8.
813 ratio
815 Set the ratio by which the equalization gain is raised. Default is
816 1. Allowed range is between 1 and 20.
817 makeup
819 Set the makeup offset in dB by which the equalization gain is
820 raised. Default is 0. Allowed range is between 0 and 30.
821 range
823 Set the max allowed cut/boost amount in dB. Default is 0. Allowed
824 range is from 0 to 200.
825 slew
827 Set the slew factor. Default is 1. Allowed range is from 1 to 200.
828 mode
830 Set the mode of filter operation, can be one of the following:
831 listen
833 Output only isolated bandpass signal.
834 cut Cut frequencies above detection threshold.
836 boost
838 Boost frequencies bellow detection threshold.
839 Default mode is cut.
841 tftype
843 Set the type of target filter, can be one of the following:
844 bell
846 lowshelf
847 highshelf
848 Default type is bell.
850 Commands
852 This filter supports the all above options as commands.
854 adynamicsmooth
856 Apply dynamic smoothing to input audio stream.
857 A description of the accepted options follows.
859 sensitivity
861 Set an amount of sensitivity to frequency fluctations. Default is
862 2. Allowed range is from 0 to 1e+06.
863 basefreq
865 Set a base frequency for smoothing. Default value is 22050.
866 Allowed range is from 2 to 1e+06.
867 Commands
869 This filter supports the all above options as commands.
871 aecho
873 Apply echoing to the input audio.
874 Echoes are reflected sound and can occur naturally amongst mountains
876 (and sometimes large buildings) when talking or shouting; digital echo
877 effects emulate this behaviour and are often used to help fill out the
878 sound of a single instrument or vocal. The time difference between the
879 original signal and the reflection is the "delay", and the loudness of
880 the reflected signal is the "decay". Multiple echoes can have
881 different delays and decays.
882 A description of the accepted parameters follows.
884 in_gain
886 Set input gain of reflected signal. Default is 0.6.
887 out_gain
889 Set output gain of reflected signal. Default is 0.3.
890 delays
892 Set list of time intervals in milliseconds between original signal
893 and reflections separated by '|'. Allowed range for each "delay" is
894 "(0 - 90000.0]". Default is 1000.
895 decays
897 Set list of loudness of reflected signals separated by '|'.
898 Allowed range for each "decay" is "(0 - 1.0]". Default is 0.5.
899 Examples
901 • Make it sound as if there are twice as many instruments as are
903 actually playing:
904 aecho=0.8:0.88:60:0.4
906 • If delay is very short, then it sounds like a (metallic) robot
908 playing music:
909 aecho=0.8:0.88:6:0.4
911 • A longer delay will sound like an open air concert in the
913 mountains:
914 aecho=0.8:0.9:1000:0.3
916 • Same as above but with one more mountain:
918 aecho=0.8:0.9:1000|1800:0.3|0.25
920 aemphasis
922 Audio emphasis filter creates or restores material directly taken from
923 LPs or emphased CDs with different filter curves. E.g. to store music
924 on vinyl the signal has to be altered by a filter first to even out the
925 disadvantages of this recording medium. Once the material is played
926 back the inverse filter has to be applied to restore the distortion of
927 the frequency response.
928 The filter accepts the following options:
930 level_in
932 Set input gain.
933 level_out
935 Set output gain.
936 mode
938 Set filter mode. For restoring material use "reproduction" mode,
939 otherwise use "production" mode. Default is "reproduction" mode.
940 type
942 Set filter type. Selects medium. Can be one of the following:
943 col select Columbia.
945 emi select EMI.
947 bsi select BSI (78RPM).
949 riaa
951 select RIAA.
952 cd select Compact Disc (CD).
954 50fm
956 select 50Xs (FM).
957 75fm
959 select 75Xs (FM).
960 50kf
962 select 50Xs (FM-KF).
963 75kf
965 select 75Xs (FM-KF).
966 Commands
968 This filter supports the all above options as commands.
970 aeval
972 Modify an audio signal according to the specified expressions.
973 This filter accepts one or more expressions (one for each channel),
975 which are evaluated and used to modify a corresponding audio signal.
976 It accepts the following parameters:
978 exprs
980 Set the '|'-separated expressions list for each separate channel.
981 If the number of input channels is greater than the number of
982 expressions, the last specified expression is used for the
983 remaining output channels.
984 channel_layout, c
986 Set output channel layout. If not specified, the channel layout is
987 specified by the number of expressions. If set to same, it will use
988 by default the same input channel layout.
989 Each expression in exprs can contain the following constants and
991 functions:
992 ch channel number of the current expression
994 n number of the evaluated sample, starting from 0
996 s sample rate
998 t time of the evaluated sample expressed in seconds
1000 nb_in_channels
1002 nb_out_channels
1003 input and output number of channels
1004 val(CH)
1006 the value of input channel with number CH
1007 Note: this filter is slow. For faster processing you should use a
1009 dedicated filter.
1010 Examples
1012 • Half volume:
1014 aeval=val(ch)/2:c=same
1016 • Invert phase of the second channel:
1018 aeval=val(0)|-val(1)
1020 aexciter
1022 An exciter is used to produce high sound that is not present in the
1023 original signal. This is done by creating harmonic distortions of the
1024 signal which are restricted in range and added to the original signal.
1025 An Exciter raises the upper end of an audio signal without simply
1026 raising the higher frequencies like an equalizer would do to create a
1027 more "crisp" or "brilliant" sound.
1028 The filter accepts the following options:
1030 level_in
1032 Set input level prior processing of signal. Allowed range is from
1033 0 to 64. Default value is 1.
1034 level_out
1036 Set output level after processing of signal. Allowed range is from
1037 0 to 64. Default value is 1.
1038 amount
1040 Set the amount of harmonics added to original signal. Allowed
1041 range is from 0 to 64. Default value is 1.
1042 drive
1044 Set the amount of newly created harmonics. Allowed range is from
1045 0.1 to 10. Default value is 8.5.
1046 blend
1048 Set the octave of newly created harmonics. Allowed range is from
1049 -10 to 10. Default value is 0.
1050 freq
1052 Set the lower frequency limit of producing harmonics in Hz.
1053 Allowed range is from 2000 to 12000 Hz. Default is 7500 Hz.
1054 ceil
1056 Set the upper frequency limit of producing harmonics. Allowed
1057 range is from 9999 to 20000 Hz. If value is lower than 10000 Hz no
1058 limit is applied.
1059 listen
1061 Mute the original signal and output only added harmonics. By
1062 default is disabled.
1063 Commands
1065 This filter supports the all above options as commands.
1067 afade
1069 Apply fade-in/out effect to input audio.
1070 A description of the accepted parameters follows.
1072 type, t
1074 Specify the effect type, can be either "in" for fade-in, or "out"
1075 for a fade-out effect. Default is "in".
1076 start_sample, ss
1078 Specify the number of the start sample for starting to apply the
1079 fade effect. Default is 0.
1080 nb_samples, ns
1082 Specify the number of samples for which the fade effect has to
1083 last. At the end of the fade-in effect the output audio will have
1084 the same volume as the input audio, at the end of the fade-out
1085 transition the output audio will be silence. Default is 44100.
1086 start_time, st
1088 Specify the start time of the fade effect. Default is 0. The value
1089 must be specified as a time duration; see the Time duration section
1090 in the ffmpeg-utils(1) manual for the accepted syntax. If set this
1091 option is used instead of start_sample.
1092 duration, d
1094 Specify the duration of the fade effect. See the Time duration
1095 section in the ffmpeg-utils(1) manual for the accepted syntax. At
1096 the end of the fade-in effect the output audio will have the same
1097 volume as the input audio, at the end of the fade-out transition
1098 the output audio will be silence. By default the duration is
1099 determined by nb_samples. If set this option is used instead of
1100 nb_samples.
1101 curve
1103 Set curve for fade transition.
1104 It accepts the following values:
1106 tri select triangular, linear slope (default)
1108 qsin
1110 select quarter of sine wave
1111 hsin
1113 select half of sine wave
1114 esin
1116 select exponential sine wave
1117 log select logarithmic
1119 ipar
1121 select inverted parabola
1122 qua select quadratic
1124 cub select cubic
1126 squ select square root
1128 cbr select cubic root
1130 par select parabola
1132 exp select exponential
1134 iqsin
1136 select inverted quarter of sine wave
1137 ihsin
1139 select inverted half of sine wave
1140 dese
1142 select double-exponential seat
1143 desi
1145 select double-exponential sigmoid
1146 losi
1148 select logistic sigmoid
1149 sinc
1151 select sine cardinal function
1152 isinc
1154 select inverted sine cardinal function
1155 nofade
1157 no fade applied
1158 Commands
1160 This filter supports the all above options as commands.
1162 Examples
1164 • Fade in first 15 seconds of audio:
1166 afade=t=in:ss=0:d=15
1168 • Fade out last 25 seconds of a 900 seconds audio:
1170 afade=t=out:st=875:d=25
1172 afftdn
1174 Denoise audio samples with FFT.
1175 A description of the accepted parameters follows.
1177 noise_reduction, nr
1179 Set the noise reduction in dB, allowed range is 0.01 to 97.
1180 Default value is 12 dB.
1181 noise_floor, nf
1183 Set the noise floor in dB, allowed range is -80 to -20. Default
1184 value is -50 dB.
1185 noise_type, nt
1187 Set the noise type.
1188 It accepts the following values:
1190 white, w
1192 Select white noise.
1193 vinyl, v
1195 Select vinyl noise.
1196 shellac, s
1198 Select shellac noise.
1199 custom, c
1201 Select custom noise, defined in "bn" option.
1202 Default value is white noise.
1204 band_noise, bn
1206 Set custom band noise profile for every one of 15 bands. Bands are
1207 separated by ' ' or '|'.
1208 residual_floor, rf
1210 Set the residual floor in dB, allowed range is -80 to -20. Default
1211 value is -38 dB.
1212 track_noise, tn
1214 Enable noise floor tracking. By default is disabled. With this
1215 enabled, noise floor is automatically adjusted.
1216 track_residual, tr
1218 Enable residual tracking. By default is disabled.
1219 output_mode, om
1221 Set the output mode.
1222 It accepts the following values:
1224 input, i
1226 Pass input unchanged.
1227 output, o
1229 Pass noise filtered out.
1230 noise, n
1232 Pass only noise.
1233 Default value is output.
1235 adaptivity, ad
1237 Set the adaptivity factor, used how fast to adapt gains adjustments
1238 per each frequency bin. Value 0 enables instant adaptation, while
1239 higher values react much slower. Allowed range is from 0 to 1.
1240 Default value is 0.5.
1241 floor_offset, fo
1243 Set the noise floor offset factor. This option is used to adjust
1244 offset applied to measured noise floor. It is only effective when
1245 noise floor tracking is enabled. Allowed range is from -2.0 to
1246 2.0. Default value is 1.0.
1247 noise_link, nl
1249 Set the noise link used for multichannel audio.
1250 It accepts the following values:
1252 none
1254 Use unchanged channel's noise floor.
1255 min Use measured min noise floor of all channels.
1257 max Use measured max noise floor of all channels.
1259 average
1261 Use measured average noise floor of all channels.
1262 Default value is min.
1264 band_multiplier, bm
1266 Set the band multiplier factor, used how much to spread bands
1267 across frequency bins. Allowed range is from 0.2 to 5. Default
1268 value is 1.25.
1269 sample_noise, sn
1271 Toggle capturing and measurement of noise profile from input audio.
1272 It accepts the following values:
1274 start, begin
1276 Start sample noise capture.
1277 stop, end
1279 Stop sample noise capture and measure new noise band profile.
1280 Default value is "none".
1282 gain_smooth, gs
1284 Set gain smooth spatial radius, used to smooth gains applied to
1285 each frequency bin. Useful to reduce random music noise artefacts.
1286 Higher values increases smoothing of gains. Allowed range is from
1287 0 to 50. Default value is 0.
1288 Commands
1290 This filter supports the some above mentioned options as commands.
1292 Examples
1294 • Reduce white noise by 10dB, and use previously measured noise floor
1296 of -40dB:
1297 afftdn=nr=10:nf=-40
1299 • Reduce white noise by 10dB, also set initial noise floor to -80dB
1301 and enable automatic tracking of noise floor so noise floor will
1302 gradually change during processing:
1303 afftdn=nr=10:nf=-80:tn=1
1305 • Reduce noise by 20dB, using noise floor of -40dB and using commands
1307 to take noise profile of first 0.4 seconds of input audio:
1308 asendcmd=0.0 afftdn sn start,asendcmd=0.4 afftdn sn stop,afftdn=nr=20:nf=-40
1310 afftfilt
1312 Apply arbitrary expressions to samples in frequency domain.
1313 real
1315 Set frequency domain real expression for each separate channel
1316 separated by '|'. Default is "re". If the number of input channels
1317 is greater than the number of expressions, the last specified
1318 expression is used for the remaining output channels.
1319 imag
1321 Set frequency domain imaginary expression for each separate channel
1322 separated by '|'. Default is "im".
1323 Each expression in real and imag can contain the following
1325 constants and functions:
1326 sr sample rate
1328 b current frequency bin number
1330 nb number of available bins
1332 ch channel number of the current expression
1334 chs number of channels
1336 pts current frame pts
1338 re current real part of frequency bin of current channel
1340 im current imaginary part of frequency bin of current channel
1342 real(b, ch)
1344 Return the value of real part of frequency bin at location
1345 (bin,channel)
1346 imag(b, ch)
1348 Return the value of imaginary part of frequency bin at location
1349 (bin,channel)
1350 win_size
1352 Set window size. Allowed range is from 16 to 131072. Default is
1353 4096
1354 win_func
1356 Set window function.
1357 It accepts the following values:
1359 rect
1361 bartlett
1362 hann, hanning
1363 hamming
1364 blackman
1365 welch
1366 flattop
1367 bharris
1368 bnuttall
1369 bhann
1370 sine
1371 nuttall
1372 lanczos
1373 gauss
1374 tukey
1375 dolph
1376 cauchy
1377 parzen
1378 poisson
1379 bohman
1380 Default is "hann".
1382 overlap
1384 Set window overlap. If set to 1, the recommended overlap for
1385 selected window function will be picked. Default is 0.75.
1386 Examples
1388 • Leave almost only low frequencies in audio:
1390 afftfilt="'real=re * (1-clip((b/nb)*b,0,1))':imag='im * (1-clip((b/nb)*b,0,1))'"
1392 • Apply robotize effect:
1394 afftfilt="real='hypot(re,im)*sin(0)':imag='hypot(re,im)*cos(0)':win_size=512:overlap=0.75"
1396 • Apply whisper effect:
1398 afftfilt="real='hypot(re,im)*cos((random(0)*2-1)*2*3.14)':imag='hypot(re,im)*sin((random(1)*2-1)*2*3.14)':win_size=128:overlap=0.8"
1400 afir
1402 Apply an arbitrary Finite Impulse Response filter.
1403 This filter is designed for applying long FIR filters, up to 60 seconds
1405 long.
1406 It can be used as component for digital crossover filters, room
1408 equalization, cross talk cancellation, wavefield synthesis,
1409 auralization, ambiophonics, ambisonics and spatialization.
1410 This filter uses the streams higher than first one as FIR coefficients.
1412 If the non-first stream holds a single channel, it will be used for all
1413 input channels in the first stream, otherwise the number of channels in
1414 the non-first stream must be same as the number of channels in the
1415 first stream.
1416 It accepts the following parameters:
1418 dry Set dry gain. This sets input gain.
1420 wet Set wet gain. This sets final output gain.
1422 length
1424 Set Impulse Response filter length. Default is 1, which means whole
1425 IR is processed.
1426 gtype
1428 Enable applying gain measured from power of IR.
1429 Set which approach to use for auto gain measurement.
1431 none
1433 Do not apply any gain.
1434 peak
1436 select peak gain, very conservative approach. This is default
1437 value.
1438 dc select DC gain, limited application.
1440 gn select gain to noise approach, this is most popular one.
1442 irgain
1444 Set gain to be applied to IR coefficients before filtering.
1445 Allowed range is 0 to 1. This gain is applied after any gain
1446 applied with gtype option.
1447 irfmt
1449 Set format of IR stream. Can be "mono" or "input". Default is
1450 "input".
1451 maxir
1453 Set max allowed Impulse Response filter duration in seconds.
1454 Default is 30 seconds. Allowed range is 0.1 to 60 seconds.
1455 response
1457 Show IR frequency response, magnitude(magenta), phase(green) and
1458 group delay(yellow) in additional video stream. By default it is
1459 disabled.
1460 channel
1462 Set for which IR channel to display frequency response. By default
1463 is first channel displayed. This option is used only when response
1464 is enabled.
1465 size
1467 Set video stream size. This option is used only when response is
1468 enabled.
1469 rate
1471 Set video stream frame rate. This option is used only when response
1472 is enabled.
1473 minp
1475 Set minimal partition size used for convolution. Default is 8192.
1476 Allowed range is from 1 to 32768. Lower values decreases latency
1477 at cost of higher CPU usage.
1478 maxp
1480 Set maximal partition size used for convolution. Default is 8192.
1481 Allowed range is from 8 to 32768. Lower values may increase CPU
1482 usage.
1483 nbirs
1485 Set number of input impulse responses streams which will be
1486 switchable at runtime. Allowed range is from 1 to 32. Default is
1487 1.
1488 ir Set IR stream which will be used for convolution, starting from 0,
1490 should always be lower than supplied value by "nbirs" option.
1491 Default is 0. This option can be changed at runtime via commands.
1492 precision
1494 Set which precision to use when processing samples.
1495 auto
1497 Auto pick internal sample format depending on other filters.
1498 float
1500 Always use single-floating point precision sample format.
1501 double
1503 Always use double-floating point precision sample format.
1504 Default value is auto.
1506 Examples
1508 • Apply reverb to stream using mono IR file as second input, complete
1510 command using ffmpeg:
1511 ffmpeg -i input.wav -i middle_tunnel_1way_mono.wav -lavfi afir output.wav
1513 aformat
1515 Set output format constraints for the input audio. The framework will
1516 negotiate the most appropriate format to minimize conversions.
1517 It accepts the following parameters:
1519 sample_fmts, f
1521 A '|'-separated list of requested sample formats.
1522 sample_rates, r
1524 A '|'-separated list of requested sample rates.
1525 channel_layouts, cl
1527 A '|'-separated list of requested channel layouts.
1528 See the Channel Layout section in the ffmpeg-utils(1) manual for
1530 the required syntax.
1531 If a parameter is omitted, all values are allowed.
1533 Force the output to either unsigned 8-bit or signed 16-bit stereo
1535 aformat=sample_fmts=u8|s16:channel_layouts=stereo
1537 afreqshift
1539 Apply frequency shift to input audio samples.
1540 The filter accepts the following options:
1542 shift
1544 Specify frequency shift. Allowed range is -INT_MAX to INT_MAX.
1545 Default value is 0.0.
1546 level
1548 Set output gain applied to final output. Allowed range is from 0.0
1549 to 1.0. Default value is 1.0.
1550 order
1552 Set filter order used for filtering. Allowed range is from 1 to 16.
1553 Default value is 8.
1554 Commands
1556 This filter supports the all above options as commands.
1558 afwtdn
1560 Reduce broadband noise from input samples using Wavelets.
1561 A description of the accepted options follows.
1563 sigma
1565 Set the noise sigma, allowed range is from 0 to 1. Default value
1566 is 0. This option controls strength of denoising applied to input
1567 samples. Most useful way to set this option is via decibels, eg.
1568 -45dB.
1569 levels
1571 Set the number of wavelet levels of decomposition. Allowed range
1572 is from 1 to 12. Default value is 10. Setting this too low make
1573 denoising performance very poor.
1574 wavet
1576 Set wavelet type for decomposition of input frame. They are sorted
1577 by number of coefficients, from lowest to highest. More
1578 coefficients means worse filtering speed, but overall better
1579 quality. Available wavelets are:
1580 sym2
1582 sym4
1583 rbior68
1584 deb10
1585 sym10
1586 coif5
1587 bl3
1588 percent
1589 Set percent of full denoising. Allowed range is from 0 to 100
1590 percent. Default value is 85 percent or partial denoising.
1591 profile
1593 If enabled, first input frame will be used as noise profile. If
1594 first frame samples contain non-noise performance will be very
1595 poor.
1596 adaptive
1598 If enabled, input frames are analyzed for presence of noise. If
1599 noise is detected with high possibility then input frame profile
1600 will be used for processing following frames, until new noise frame
1601 is detected.
1602 samples
1604 Set size of single frame in number of samples. Allowed range is
1605 from 512 to 65536. Default frame size is 8192 samples.
1606 softness
1608 Set softness applied inside thresholding function. Allowed range is
1609 from 0 to 10. Default softness is 1.
1610 Commands
1612 This filter supports the all above options as commands.
1614 agate
1616 A gate is mainly used to reduce lower parts of a signal. This kind of
1617 signal processing reduces disturbing noise between useful signals.
1618 Gating is done by detecting the volume below a chosen level threshold
1620 and dividing it by the factor set with ratio. The bottom of the noise
1621 floor is set via range. Because an exact manipulation of the signal
1622 would cause distortion of the waveform the reduction can be levelled
1623 over time. This is done by setting attack and release.
1624 attack determines how long the signal has to fall below the threshold
1626 before any reduction will occur and release sets the time the signal
1627 has to rise above the threshold to reduce the reduction again. Shorter
1628 signals than the chosen attack time will be left untouched.
1629 level_in
1631 Set input level before filtering. Default is 1. Allowed range is
1632 from 0.015625 to 64.
1633 mode
1635 Set the mode of operation. Can be "upward" or "downward". Default
1636 is "downward". If set to "upward" mode, higher parts of signal will
1637 be amplified, expanding dynamic range in upward direction.
1638 Otherwise, in case of "downward" lower parts of signal will be
1639 reduced.
1640 range
1642 Set the level of gain reduction when the signal is below the
1643 threshold. Default is 0.06125. Allowed range is from 0 to 1.
1644 Setting this to 0 disables reduction and then filter behaves like
1645 expander.
1646 threshold
1648 If a signal rises above this level the gain reduction is released.
1649 Default is 0.125. Allowed range is from 0 to 1.
1650 ratio
1652 Set a ratio by which the signal is reduced. Default is 2. Allowed
1653 range is from 1 to 9000.
1654 attack
1656 Amount of milliseconds the signal has to rise above the threshold
1657 before gain reduction stops. Default is 20 milliseconds. Allowed
1658 range is from 0.01 to 9000.
1659 release
1661 Amount of milliseconds the signal has to fall below the threshold
1662 before the reduction is increased again. Default is 250
1663 milliseconds. Allowed range is from 0.01 to 9000.
1664 makeup
1666 Set amount of amplification of signal after processing. Default is
1667 1. Allowed range is from 1 to 64.
1668 knee
1670 Curve the sharp knee around the threshold to enter gain reduction
1671 more softly. Default is 2.828427125. Allowed range is from 1 to 8.
1672 detection
1674 Choose if exact signal should be taken for detection or an RMS like
1675 one. Default is "rms". Can be "peak" or "rms".
1676 link
1678 Choose if the average level between all channels or the louder
1679 channel affects the reduction. Default is "average". Can be
1680 "average" or "maximum".
1681 Commands
1683 This filter supports the all above options as commands.
1685 aiir
1687 Apply an arbitrary Infinite Impulse Response filter.
1688 It accepts the following parameters:
1690 zeros, z
1692 Set B/numerator/zeros/reflection coefficients.
1693 poles, p
1695 Set A/denominator/poles/ladder coefficients.
1696 gains, k
1698 Set channels gains.
1699 dry_gain
1701 Set input gain.
1702 wet_gain
1704 Set output gain.
1705 format, f
1707 Set coefficients format.
1708 ll lattice-ladder function
1710 sf analog transfer function
1712 tf digital transfer function
1714 zp Z-plane zeros/poles, cartesian (default)
1716 pr Z-plane zeros/poles, polar radians
1718 pd Z-plane zeros/poles, polar degrees
1720 sp S-plane zeros/poles
1722 process, r
1724 Set type of processing.
1725 d direct processing
1727 s serial processing
1729 p parallel processing
1731 precision, e
1733 Set filtering precision.
1734 dbl double-precision floating-point (default)
1736 flt single-precision floating-point
1738 i32 32-bit integers
1740 i16 16-bit integers
1742 normalize, n
1744 Normalize filter coefficients, by default is enabled. Enabling it
1745 will normalize magnitude response at DC to 0dB.
1746 mix How much to use filtered signal in output. Default is 1. Range is
1748 between 0 and 1.
1749 response
1751 Show IR frequency response, magnitude(magenta), phase(green) and
1752 group delay(yellow) in additional video stream. By default it is
1753 disabled.
1754 channel
1756 Set for which IR channel to display frequency response. By default
1757 is first channel displayed. This option is used only when response
1758 is enabled.
1759 size
1761 Set video stream size. This option is used only when response is
1762 enabled.
1763 Coefficients in "tf" and "sf" format are separated by spaces and are in
1765 ascending order.
1766 Coefficients in "zp" format are separated by spaces and order of
1768 coefficients doesn't matter. Coefficients in "zp" format are complex
1769 numbers with i imaginary unit.
1770 Different coefficients and gains can be provided for every channel, in
1772 such case use '|' to separate coefficients or gains. Last provided
1773 coefficients will be used for all remaining channels.
1774 Examples
1776 • Apply 2 pole elliptic notch at around 5000Hz for 48000 Hz sample
1778 rate:
1779 aiir=k=1:z=7.957584807809675810E-1 -2.575128568908332300 3.674839853930788710 -2.57512875289799137 7.957586296317130880E-1:p=1 -2.86950072432325953 3.63022088054647218 -2.28075678147272232 6.361362326477423500E-1:f=tf:r=d
1781 • Same as above but in "zp" format:
1783 aiir=k=0.79575848078096756:z=0.80918701+0.58773007i 0.80918701-0.58773007i 0.80884700+0.58784055i 0.80884700-0.58784055i:p=0.63892345+0.59951235i 0.63892345-0.59951235i 0.79582691+0.44198673i 0.79582691-0.44198673i:f=zp:r=s
1785 • Apply 3-rd order analog normalized Butterworth low-pass filter,
1787 using analog transfer function format:
1788 aiir=z=1.3057 0 0 0:p=1.3057 2.3892 2.1860 1:f=sf:r=d
1790 alimiter
1792 The limiter prevents an input signal from rising over a desired
1793 threshold. This limiter uses lookahead technology to prevent your
1794 signal from distorting. It means that there is a small delay after the
1795 signal is processed. Keep in mind that the delay it produces is the
1796 attack time you set.
1797 The filter accepts the following options:
1799 level_in
1801 Set input gain. Default is 1.
1802 level_out
1804 Set output gain. Default is 1.
1805 limit
1807 Don't let signals above this level pass the limiter. Default is 1.
1808 attack
1810 The limiter will reach its attenuation level in this amount of time
1811 in milliseconds. Default is 5 milliseconds.
1812 release
1814 Come back from limiting to attenuation 1.0 in this amount of
1815 milliseconds. Default is 50 milliseconds.
1816 asc When gain reduction is always needed ASC takes care of releasing to
1818 an average reduction level rather than reaching a reduction of 0 in
1819 the release time.
1820 asc_level
1822 Select how much the release time is affected by ASC, 0 means nearly
1823 no changes in release time while 1 produces higher release times.
1824 level
1826 Auto level output signal. Default is enabled. This normalizes
1827 audio back to 0dB if enabled.
1828 latency
1830 Compensate the delay introduced by using the lookahead buffer set
1831 with attack parameter. Also flush the valid audio data in the
1832 lookahead buffer when the stream hits EOF.
1833 Depending on picked setting it is recommended to upsample input 2x or
1835 4x times with aresample before applying this filter.
1836 allpass
1838 Apply a two-pole all-pass filter with central frequency (in Hz)
1839 frequency, and filter-width width. An all-pass filter changes the
1840 audio's frequency to phase relationship without changing its frequency
1841 to amplitude relationship.
1842 The filter accepts the following options:
1844 frequency, f
1846 Set frequency in Hz.
1847 width_type, t
1849 Set method to specify band-width of filter.
1850 h Hz
1852 q Q-Factor
1854 o octave
1856 s slope
1858 k kHz
1860 width, w
1862 Specify the band-width of a filter in width_type units.
1863 mix, m
1865 How much to use filtered signal in output. Default is 1. Range is
1866 between 0 and 1.
1867 channels, c
1869 Specify which channels to filter, by default all available are
1870 filtered.
1871 normalize, n
1873 Normalize biquad coefficients, by default is disabled. Enabling it
1874 will normalize magnitude response at DC to 0dB.
1875 order, o
1877 Set the filter order, can be 1 or 2. Default is 2.
1878 transform, a
1880 Set transform type of IIR filter.
1881 di
1883 dii
1884 tdi
1885 tdii
1886 latt
1887 svf
1888 zdf
1889 precision, r
1890 Set precison of filtering.
1891 auto
1893 Pick automatic sample format depending on surround filters.
1894 s16 Always use signed 16-bit.
1896 s32 Always use signed 32-bit.
1898 f32 Always use float 32-bit.
1900 f64 Always use float 64-bit.
1902 Commands
1904 This filter supports the following commands:
1906 frequency, f
1908 Change allpass frequency. Syntax for the command is : "frequency"
1909 width_type, t
1911 Change allpass width_type. Syntax for the command is :
1912 "width_type"
1913 width, w
1915 Change allpass width. Syntax for the command is : "width"
1916 mix, m
1918 Change allpass mix. Syntax for the command is : "mix"
1919 aloop
1921 Loop audio samples.
1922 The filter accepts the following options:
1924 loop
1926 Set the number of loops. Setting this value to -1 will result in
1927 infinite loops. Default is 0.
1928 size
1930 Set maximal number of samples. Default is 0.
1931 start
1933 Set first sample of loop. Default is 0.
1934 amerge
1936 Merge two or more audio streams into a single multi-channel stream.
1937 The filter accepts the following options:
1939 inputs
1941 Set the number of inputs. Default is 2.
1942 If the channel layouts of the inputs are disjoint, and therefore
1944 compatible, the channel layout of the output will be set accordingly
1945 and the channels will be reordered as necessary. If the channel layouts
1946 of the inputs are not disjoint, the output will have all the channels
1947 of the first input then all the channels of the second input, in that
1948 order, and the channel layout of the output will be the default value
1949 corresponding to the total number of channels.
1950 For example, if the first input is in 2.1 (FL+FR+LF) and the second
1952 input is FC+BL+BR, then the output will be in 5.1, with the channels in
1953 the following order: a1, a2, b1, a3, b2, b3 (a1 is the first channel of
1954 the first input, b1 is the first channel of the second input).
1955 On the other hand, if both input are in stereo, the output channels
1957 will be in the default order: a1, a2, b1, b2, and the channel layout
1958 will be arbitrarily set to 4.0, which may or may not be the expected
1959 value.
1960 All inputs must have the same sample rate, and format.
1962 If inputs do not have the same duration, the output will stop with the
1964 shortest.
1965 Examples
1967 • Merge two mono files into a stereo stream:
1969 amovie=left.wav [l] ; amovie=right.mp3 [r] ; [l] [r] amerge
1971 • Multiple merges assuming 1 video stream and 6 audio streams in
1973 input.mkv:
1974 ffmpeg -i input.mkv -filter_complex "[0:1][0:2][0:3][0:4][0:5][0:6] amerge=inputs=6" -c:a pcm_s16le output.mkv
1976 amix
1978 Mixes multiple audio inputs into a single output.
1979 Note that this filter only supports float samples (the amerge and pan
1981 audio filters support many formats). If the amix input has integer
1982 samples then aresample will be automatically inserted to perform the
1983 conversion to float samples.
1984 For example
1986 ffmpeg -i INPUT1 -i INPUT2 -i INPUT3 -filter_complex amix=inputs=3:duration=first:dropout_transition=3 OUTPUT
1988 will mix 3 input audio streams to a single output with the same
1990 duration as the first input and a dropout transition time of 3 seconds.
1991 It accepts the following parameters:
1993 inputs
1995 The number of inputs. If unspecified, it defaults to 2.
1996 duration
1998 How to determine the end-of-stream.
1999 longest
2001 The duration of the longest input. (default)
2002 shortest
2004 The duration of the shortest input.
2005 first
2007 The duration of the first input.
2008 dropout_transition
2010 The transition time, in seconds, for volume renormalization when an
2011 input stream ends. The default value is 2 seconds.
2012 weights
2014 Specify weight of each input audio stream as sequence. Each weight
2015 is separated by space. By default all inputs have same weight.
2016 normalize
2018 Always scale inputs instead of only doing summation of samples.
2019 Beware of heavy clipping if inputs are not normalized prior or
2020 after filtering by this filter if this option is disabled. By
2021 default is enabled.
2022 Commands
2024 This filter supports the following commands:
2026 weights
2028 normalize
2029 Syntax is same as option with same name.
2030 amultiply
2032 Multiply first audio stream with second audio stream and store result
2033 in output audio stream. Multiplication is done by multiplying each
2034 sample from first stream with sample at same position from second
2035 stream.
2036 With this element-wise multiplication one can create amplitude fades
2038 and amplitude modulations.
2039 anequalizer
2041 High-order parametric multiband equalizer for each channel.
2042 It accepts the following parameters:
2044 params
2046 This option string is in format: "cchn f=cf w=w g=g t=f | ..."
2047 Each equalizer band is separated by '|'.
2048 chn Set channel number to which equalization will be applied. If
2050 input doesn't have that channel the entry is ignored.
2051 f Set central frequency for band. If input doesn't have that
2053 frequency the entry is ignored.
2054 w Set band width in Hertz.
2056 g Set band gain in dB.
2058 t Set filter type for band, optional, can be:
2060 0 Butterworth, this is default.
2062 1 Chebyshev type 1.
2064 2 Chebyshev type 2.
2066 curves
2068 With this option activated frequency response of anequalizer is
2069 displayed in video stream.
2070 size
2072 Set video stream size. Only useful if curves option is activated.
2073 mgain
2075 Set max gain that will be displayed. Only useful if curves option
2076 is activated. Setting this to a reasonable value makes it possible
2077 to display gain which is derived from neighbour bands which are too
2078 close to each other and thus produce higher gain when both are
2079 activated.
2080 fscale
2082 Set frequency scale used to draw frequency response in video
2083 output. Can be linear or logarithmic. Default is logarithmic.
2084 colors
2086 Set color for each channel curve which is going to be displayed in
2087 video stream. This is list of color names separated by space or by
2088 '|'. Unrecognised or missing colors will be replaced by white
2089 color.
2090 Examples
2092 • Lower gain by 10 of central frequency 200Hz and width 100 Hz for
2094 first 2 channels using Chebyshev type 1 filter:
2095 anequalizer=c0 f=200 w=100 g=-10 t=1|c1 f=200 w=100 g=-10 t=1
2097 Commands
2099 This filter supports the following commands:
2101 change
2103 Alter existing filter parameters. Syntax for the commands is :
2104 "fN|f=freq|w=width|g=gain"
2105 fN is existing filter number, starting from 0, if no such filter is
2107 available error is returned. freq set new frequency parameter.
2108 width set new width parameter in Hertz. gain set new gain
2109 parameter in dB.
2110 Full filter invocation with asendcmd may look like this:
2112 asendcmd=c='4.0 anequalizer change
2113 0|f=200|w=50|g=1',anequalizer=...
2114 anlmdn
2116 Reduce broadband noise in audio samples using Non-Local Means
2117 algorithm.
2118 Each sample is adjusted by looking for other samples with similar
2120 contexts. This context similarity is defined by comparing their
2121 surrounding patches of size p. Patches are searched in an area of r
2122 around the sample.
2123 The filter accepts the following options:
2125 strength, s
2127 Set denoising strength. Allowed range is from 0.00001 to 10000.
2128 Default value is 0.00001.
2129 patch, p
2131 Set patch radius duration. Allowed range is from 1 to 100
2132 milliseconds. Default value is 2 milliseconds.
2133 research, r
2135 Set research radius duration. Allowed range is from 2 to 300
2136 milliseconds. Default value is 6 milliseconds.
2137 output, o
2139 Set the output mode.
2140 It accepts the following values:
2142 i Pass input unchanged.
2144 o Pass noise filtered out.
2146 n Pass only noise.
2148 Default value is o.
2150 smooth, m
2152 Set smooth factor. Default value is 11. Allowed range is from 1 to
2153 1000.
2154 Commands
2156 This filter supports the all above options as commands.
2158 anlmf, anlms
2160 Apply Normalized Least-Mean-(Squares|Fourth) algorithm to the first
2161 audio stream using the second audio stream.
2162 This adaptive filter is used to mimic a desired filter by finding the
2164 filter coefficients that relate to producing the least mean square of
2165 the error signal (difference between the desired, 2nd input audio
2166 stream and the actual signal, the 1st input audio stream).
2167 A description of the accepted options follows.
2169 order
2171 Set filter order.
2172 mu Set filter mu.
2174 eps Set the filter eps.
2176 leakage
2178 Set the filter leakage.
2179 out_mode
2181 It accepts the following values:
2182 i Pass the 1st input.
2184 d Pass the 2nd input.
2186 o Pass filtered samples.
2188 n Pass difference between desired and filtered samples.
2190 Default value is o.
2192 Examples
2194 • One of many usages of this filter is noise reduction, input audio
2196 is filtered with same samples that are delayed by fixed amount, one
2197 such example for stereo audio is:
2198 asplit[a][b],[a]adelay=32S|32S[a],[b][a]anlms=order=128:leakage=0.0005:mu=.5:out_mode=o
2200 Commands
2202 This filter supports the same commands as options, excluding option
2204 "order".
2205 anull
2207 Pass the audio source unchanged to the output.
2208 apad
2210 Pad the end of an audio stream with silence.
2211 This can be used together with ffmpeg -shortest to extend audio streams
2213 to the same length as the video stream.
2214 A description of the accepted options follows.
2216 packet_size
2218 Set silence packet size. Default value is 4096.
2219 pad_len
2221 Set the number of samples of silence to add to the end. After the
2222 value is reached, the stream is terminated. This option is mutually
2223 exclusive with whole_len.
2224 whole_len
2226 Set the minimum total number of samples in the output audio stream.
2227 If the value is longer than the input audio length, silence is
2228 added to the end, until the value is reached. This option is
2229 mutually exclusive with pad_len.
2230 pad_dur
2232 Specify the duration of samples of silence to add. See the Time
2233 duration section in the ffmpeg-utils(1) manual for the accepted
2234 syntax. Used only if set to non-negative value.
2235 whole_dur
2237 Specify the minimum total duration in the output audio stream. See
2238 the Time duration section in the ffmpeg-utils(1) manual for the
2239 accepted syntax. Used only if set to non-negative value. If the
2240 value is longer than the input audio length, silence is added to
2241 the end, until the value is reached. This option is mutually
2242 exclusive with pad_dur
2243 If neither the pad_len nor the whole_len nor pad_dur nor whole_dur
2245 option is set, the filter will add silence to the end of the input
2246 stream indefinitely.
2247 Note that for ffmpeg 4.4 and earlier a zero pad_dur or whole_dur also
2249 caused the filter to add silence indefinitely.
2250 Examples
2252 • Add 1024 samples of silence to the end of the input:
2254 apad=pad_len=1024
2256 • Make sure the audio output will contain at least 10000 samples, pad
2258 the input with silence if required:
2259 apad=whole_len=10000
2261 • Use ffmpeg to pad the audio input with silence, so that the video
2263 stream will always result the shortest and will be converted until
2264 the end in the output file when using the shortest option:
2265 ffmpeg -i VIDEO -i AUDIO -filter_complex "[1:0]apad" -shortest OUTPUT
2267 aphaser
2269 Add a phasing effect to the input audio.
2270 A phaser filter creates series of peaks and troughs in the frequency
2272 spectrum. The position of the peaks and troughs are modulated so that
2273 they vary over time, creating a sweeping effect.
2274 A description of the accepted parameters follows.
2276 in_gain
2278 Set input gain. Default is 0.4.
2279 out_gain
2281 Set output gain. Default is 0.74
2282 delay
2284 Set delay in milliseconds. Default is 3.0.
2285 decay
2287 Set decay. Default is 0.4.
2288 speed
2290 Set modulation speed in Hz. Default is 0.5.
2291 type
2293 Set modulation type. Default is triangular.
2294 It accepts the following values:
2296 triangular, t
2298 sinusoidal, s
2299 aphaseshift
2301 Apply phase shift to input audio samples.
2302 The filter accepts the following options:
2304 shift
2306 Specify phase shift. Allowed range is from -1.0 to 1.0. Default
2307 value is 0.0.
2308 level
2310 Set output gain applied to final output. Allowed range is from 0.0
2311 to 1.0. Default value is 1.0.
2312 order
2314 Set filter order used for filtering. Allowed range is from 1 to 16.
2315 Default value is 8.
2316 Commands
2318 This filter supports the all above options as commands.
2320 apsyclip
2322 Apply Psychoacoustic clipper to input audio stream.
2323 The filter accepts the following options:
2325 level_in
2327 Set input gain. By default it is 1. Range is [0.015625 - 64].
2328 level_out
2330 Set output gain. By default it is 1. Range is [0.015625 - 64].
2331 clip
2333 Set the clipping start value. Default value is 0dBFS or 1.
2334 diff
2336 Output only difference samples, useful to hear introduced
2337 distortions. By default is disabled.
2338 adaptive
2340 Set strength of adaptive distortion applied. Default value is 0.5.
2341 Allowed range is from 0 to 1.
2342 iterations
2344 Set number of iterations of psychoacoustic clipper. Allowed range
2345 is from 1 to 20. Default value is 10.
2346 level
2348 Auto level output signal. Default is disabled. This normalizes
2349 audio back to 0dBFS if enabled.
2350 Commands
2352 This filter supports the all above options as commands.
2354 apulsator
2356 Audio pulsator is something between an autopanner and a tremolo. But
2357 it can produce funny stereo effects as well. Pulsator changes the
2358 volume of the left and right channel based on a LFO (low frequency
2359 oscillator) with different waveforms and shifted phases. This filter
2360 have the ability to define an offset between left and right channel. An
2361 offset of 0 means that both LFO shapes match each other. The left and
2362 right channel are altered equally - a conventional tremolo. An offset
2363 of 50% means that the shape of the right channel is exactly shifted in
2364 phase (or moved backwards about half of the frequency) - pulsator acts
2365 as an autopanner. At 1 both curves match again. Every setting in
2366 between moves the phase shift gapless between all stages and produces
2367 some "bypassing" sounds with sine and triangle waveforms. The more you
2368 set the offset near 1 (starting from the 0.5) the faster the signal
2369 passes from the left to the right speaker.
2370 The filter accepts the following options:
2372 level_in
2374 Set input gain. By default it is 1. Range is [0.015625 - 64].
2375 level_out
2377 Set output gain. By default it is 1. Range is [0.015625 - 64].
2378 mode
2380 Set waveform shape the LFO will use. Can be one of: sine, triangle,
2381 square, sawup or sawdown. Default is sine.
2382 amount
2384 Set modulation. Define how much of original signal is affected by
2385 the LFO.
2386 offset_l
2388 Set left channel offset. Default is 0. Allowed range is [0 - 1].
2389 offset_r
2391 Set right channel offset. Default is 0.5. Allowed range is [0 - 1].
2392 width
2394 Set pulse width. Default is 1. Allowed range is [0 - 2].
2395 timing
2397 Set possible timing mode. Can be one of: bpm, ms or hz. Default is
2398 hz.
2399 bpm Set bpm. Default is 120. Allowed range is [30 - 300]. Only used if
2401 timing is set to bpm.
2402 ms Set ms. Default is 500. Allowed range is [10 - 2000]. Only used if
2404 timing is set to ms.
2405 hz Set frequency in Hz. Default is 2. Allowed range is [0.01 - 100].
2407 Only used if timing is set to hz.
2408 aresample
2410 Resample the input audio to the specified parameters, using the
2411 libswresample library. If none are specified then the filter will
2412 automatically convert between its input and output.
2413 This filter is also able to stretch/squeeze the audio data to make it
2415 match the timestamps or to inject silence / cut out audio to make it
2416 match the timestamps, do a combination of both or do neither.
2417 The filter accepts the syntax [sample_rate:]resampler_options, where
2419 sample_rate expresses a sample rate and resampler_options is a list of
2420 key=value pairs, separated by ":". See the "Resampler Options" section
2421 in the ffmpeg-resampler(1) manual for the complete list of supported
2422 options.
2423 Examples
2425 • Resample the input audio to 44100Hz:
2427 aresample=44100
2429 • Stretch/squeeze samples to the given timestamps, with a maximum of
2431 1000 samples per second compensation:
2432 aresample=async=1000
2434 areverse
2436 Reverse an audio clip.
2437 Warning: This filter requires memory to buffer the entire clip, so
2439 trimming is suggested.
2440 Examples
2442 • Take the first 5 seconds of a clip, and reverse it.
2444 atrim=end=5,areverse
2446 arnndn
2448 Reduce noise from speech using Recurrent Neural Networks.
2449 This filter accepts the following options:
2451 model, m
2453 Set train model file to load. This option is always required.
2454 mix Set how much to mix filtered samples into final output. Allowed
2456 range is from -1 to 1. Default value is 1. Negative values are
2457 special, they set how much to keep filtered noise in the final
2458 filter output. Set this option to -1 to hear actual noise removed
2459 from input signal.
2460 Commands
2462 This filter supports the all above options as commands.
2464 asdr
2466 Measure Audio Signal-to-Distortion Ratio.
2467 This filter takes two audio streams for input, and outputs first audio
2469 stream. Results are in dB per channel at end of either input.
2470 asetnsamples
2472 Set the number of samples per each output audio frame.
2473 The last output packet may contain a different number of samples, as
2475 the filter will flush all the remaining samples when the input audio
2476 signals its end.
2477 The filter accepts the following options:
2479 nb_out_samples, n
2481 Set the number of frames per each output audio frame. The number is
2482 intended as the number of samples per each channel. Default value
2483 is 1024.
2484 pad, p
2486 If set to 1, the filter will pad the last audio frame with zeroes,
2487 so that the last frame will contain the same number of samples as
2488 the previous ones. Default value is 1.
2489 For example, to set the number of per-frame samples to 1234 and disable
2491 padding for the last frame, use:
2492 asetnsamples=n=1234:p=0
2494 asetrate
2496 Set the sample rate without altering the PCM data. This will result in
2497 a change of speed and pitch.
2498 The filter accepts the following options:
2500 sample_rate, r
2502 Set the output sample rate. Default is 44100 Hz.
2503 ashowinfo
2505 Show a line containing various information for each input audio frame.
2506 The input audio is not modified.
2507 The shown line contains a sequence of key/value pairs of the form
2509 key:value.
2510 The following values are shown in the output:
2512 n The (sequential) number of the input frame, starting from 0.
2514 pts The presentation timestamp of the input frame, in time base units;
2516 the time base depends on the filter input pad, and is usually
2517 1/sample_rate.
2518 pts_time
2520 The presentation timestamp of the input frame in seconds.
2521 pos position of the frame in the input stream, -1 if this information
2523 in unavailable and/or meaningless (for example in case of synthetic
2524 audio)
2525 fmt The sample format.
2527 chlayout
2529 The channel layout.
2530 rate
2532 The sample rate for the audio frame.
2533 nb_samples
2535 The number of samples (per channel) in the frame.
2536 checksum
2538 The Adler-32 checksum (printed in hexadecimal) of the audio data.
2539 For planar audio, the data is treated as if all the planes were
2540 concatenated.
2541 plane_checksums
2543 A list of Adler-32 checksums for each data plane.
2544 asoftclip
2546 Apply audio soft clipping.
2547 Soft clipping is a type of distortion effect where the amplitude of a
2549 signal is saturated along a smooth curve, rather than the abrupt shape
2550 of hard-clipping.
2551 This filter accepts the following options:
2553 type
2555 Set type of soft-clipping.
2556 It accepts the following values:
2558 hard
2560 tanh
2561 atan
2562 cubic
2563 exp
2564 alg
2565 quintic
2566 sin
2567 erf
2568 threshold
2569 Set threshold from where to start clipping. Default value is 0dB or
2570 1.
2571 output
2573 Set gain applied to output. Default value is 0dB or 1.
2574 param
2576 Set additional parameter which controls sigmoid function.
2577 oversample
2579 Set oversampling factor.
2580 Commands
2582 This filter supports the all above options as commands.
2584 aspectralstats
2586 Display frequency domain statistical information about the audio
2587 channels. Statistics are calculated and stored as metadata for each
2588 audio channel and for each audio frame.
2589 It accepts the following option:
2591 win_size
2593 Set the window length in samples. Default value is 2048. Allowed
2594 range is from 32 to 65536.
2595 win_func
2597 Set window function.
2598 It accepts the following values:
2600 rect
2602 bartlett
2603 hann, hanning
2604 hamming
2605 blackman
2606 welch
2607 flattop
2608 bharris
2609 bnuttall
2610 bhann
2611 sine
2612 nuttall
2613 lanczos
2614 gauss
2615 tukey
2616 dolph
2617 cauchy
2618 parzen
2619 poisson
2620 bohman
2621 Default is "hann".
2623 overlap
2625 Set window overlap. Allowed range is from 0 to 1. Default value is
2626 0.5.
2627 A list of each metadata key follows:
2629 mean
2631 variance
2632 centroid
2633 spread
2634 skewness
2635 kurtosis
2636 entropy
2637 flatness
2638 crest
2639 flux
2640 slope
2641 decrease
2642 rolloff
2643 asr
2645 Automatic Speech Recognition
2646 This filter uses PocketSphinx for speech recognition. To enable
2648 compilation of this filter, you need to configure FFmpeg with
2649 "--enable-pocketsphinx".
2650 It accepts the following options:
2652 rate
2654 Set sampling rate of input audio. Defaults is 16000. This need to
2655 match speech models, otherwise one will get poor results.
2656 hmm Set dictionary containing acoustic model files.
2658 dict
2660 Set pronunciation dictionary.
2661 lm Set language model file.
2663 lmctl
2665 Set language model set.
2666 lmname
2668 Set which language model to use.
2669 logfn
2671 Set output for log messages.
2672 The filter exports recognized speech as the frame metadata
2674 "lavfi.asr.text".
2675 astats
2677 Display time domain statistical information about the audio channels.
2678 Statistics are calculated and displayed for each audio channel and,
2679 where applicable, an overall figure is also given.
2680 It accepts the following option:
2682 length
2684 Short window length in seconds, used for peak and trough RMS
2685 measurement. Default is 0.05 (50 milliseconds). Allowed range is
2686 "[0 - 10]".
2687 metadata
2689 Set metadata injection. All the metadata keys are prefixed with
2690 "lavfi.astats.X", where "X" is channel number starting from 1 or
2691 string "Overall". Default is disabled.
2692 Available keys for each channel are: DC_offset Min_level Max_level
2694 Min_difference Max_difference Mean_difference RMS_difference
2695 Peak_level RMS_peak RMS_trough Crest_factor Flat_factor Peak_count
2696 Noise_floor Noise_floor_count Entropy Bit_depth Dynamic_range
2697 Zero_crossings Zero_crossings_rate Number_of_NaNs Number_of_Infs
2698 Number_of_denormals
2699 and for Overall: DC_offset Min_level Max_level Min_difference
2701 Max_difference Mean_difference RMS_difference Peak_level RMS_level
2702 RMS_peak RMS_trough Flat_factor Peak_count Noise_floor
2703 Noise_floor_count Entropy Bit_depth Number_of_samples
2704 Number_of_NaNs Number_of_Infs Number_of_denormals
2705 For example full key look like this "lavfi.astats.1.DC_offset" or
2707 this "lavfi.astats.Overall.Peak_count".
2708 For description what each key means read below.
2710 reset
2712 Set the number of frames over which cumulative stats are calculated
2713 before being reset Default is disabled.
2714 measure_perchannel
2716 Select the parameters which are measured per channel. The metadata
2717 keys can be used as flags, default is all which measures
2718 everything. none disables all per channel measurement.
2719 measure_overall
2721 Select the parameters which are measured overall. The metadata keys
2722 can be used as flags, default is all which measures everything.
2723 none disables all overall measurement.
2724 A description of each shown parameter follows:
2726 DC offset
2728 Mean amplitude displacement from zero.
2729 Min level
2731 Minimal sample level.
2732 Max level
2734 Maximal sample level.
2735 Min difference
2737 Minimal difference between two consecutive samples.
2738 Max difference
2740 Maximal difference between two consecutive samples.
2741 Mean difference
2743 Mean difference between two consecutive samples. The average of
2744 each difference between two consecutive samples.
2745 RMS difference
2747 Root Mean Square difference between two consecutive samples.
2748 Peak level dB
2750 RMS level dB
2751 Standard peak and RMS level measured in dBFS.
2752 RMS peak dB
2754 RMS trough dB
2755 Peak and trough values for RMS level measured over a short window.
2756 Crest factor
2758 Standard ratio of peak to RMS level (note: not in dB).
2759 Flat factor
2761 Flatness (i.e. consecutive samples with the same value) of the
2762 signal at its peak levels (i.e. either Min level or Max level).
2763 Peak count
2765 Number of occasions (not the number of samples) that the signal
2766 attained either Min level or Max level.
2767 Noise floor dB
2769 Minimum local peak measured in dBFS over a short window.
2770 Noise floor count
2772 Number of occasions (not the number of samples) that the signal
2773 attained Noise floor.
2774 Entropy
2776 Entropy measured across whole audio. Entropy of value near 1.0 is
2777 typically measured for white noise.
2778 Bit depth
2780 Overall bit depth of audio. Number of bits used for each sample.
2781 Dynamic range
2783 Measured dynamic range of audio in dB.
2784 Zero crossings
2786 Number of points where the waveform crosses the zero level axis.
2787 Zero crossings rate
2789 Rate of Zero crossings and number of audio samples.
2790 asubboost
2792 Boost subwoofer frequencies.
2793 The filter accepts the following options:
2795 dry Set dry gain, how much of original signal is kept. Allowed range is
2797 from 0 to 1. Default value is 1.0.
2798 wet Set wet gain, how much of filtered signal is kept. Allowed range is
2800 from 0 to 1. Default value is 1.0.
2801 boost
2803 Set max boost factor. Allowed range is from 1 to 12. Default value
2804 is 2.
2805 decay
2807 Set delay line decay gain value. Allowed range is from 0 to 1.
2808 Default value is 0.0.
2809 feedback
2811 Set delay line feedback gain value. Allowed range is from 0 to 1.
2812 Default value is 0.9.
2813 cutoff
2815 Set cutoff frequency in Hertz. Allowed range is 50 to 900. Default
2816 value is 100.
2817 slope
2819 Set slope amount for cutoff frequency. Allowed range is 0.0001 to
2820 1. Default value is 0.5.
2821 delay
2823 Set delay. Allowed range is from 1 to 100. Default value is 20.
2824 channels
2826 Set the channels to process. Default value is all available.
2827 Commands
2829 This filter supports the all above options as commands.
2831 asubcut
2833 Cut subwoofer frequencies.
2834 This filter allows to set custom, steeper roll off than highpass
2836 filter, and thus is able to more attenuate frequency content in stop-
2837 band.
2838 The filter accepts the following options:
2840 cutoff
2842 Set cutoff frequency in Hertz. Allowed range is 2 to 200. Default
2843 value is 20.
2844 order
2846 Set filter order. Available values are from 3 to 20. Default value
2847 is 10.
2848 level
2850 Set input gain level. Allowed range is from 0 to 1. Default value
2851 is 1.
2852 Commands
2854 This filter supports the all above options as commands.
2856 asupercut
2858 Cut super frequencies.
2859 The filter accepts the following options:
2861 cutoff
2863 Set cutoff frequency in Hertz. Allowed range is 20000 to 192000.
2864 Default value is 20000.
2865 order
2867 Set filter order. Available values are from 3 to 20. Default value
2868 is 10.
2869 level
2871 Set input gain level. Allowed range is from 0 to 1. Default value
2872 is 1.
2873 Commands
2875 This filter supports the all above options as commands.
2877 asuperpass
2879 Apply high order Butterworth band-pass filter.
2880 The filter accepts the following options:
2882 centerf
2884 Set center frequency in Hertz. Allowed range is 2 to 999999.
2885 Default value is 1000.
2886 order
2888 Set filter order. Available values are from 4 to 20. Default value
2889 is 4.
2890 qfactor
2892 Set Q-factor. Allowed range is from 0.01 to 100. Default value is
2893 1.
2894 level
2896 Set input gain level. Allowed range is from 0 to 2. Default value
2897 is 1.
2898 Commands
2900 This filter supports the all above options as commands.
2902 asuperstop
2904 Apply high order Butterworth band-stop filter.
2905 The filter accepts the following options:
2907 centerf
2909 Set center frequency in Hertz. Allowed range is 2 to 999999.
2910 Default value is 1000.
2911 order
2913 Set filter order. Available values are from 4 to 20. Default value
2914 is 4.
2915 qfactor
2917 Set Q-factor. Allowed range is from 0.01 to 100. Default value is
2918 1.
2919 level
2921 Set input gain level. Allowed range is from 0 to 2. Default value
2922 is 1.
2923 Commands
2925 This filter supports the all above options as commands.
2927 atempo
2929 Adjust audio tempo.
2930 The filter accepts exactly one parameter, the audio tempo. If not
2932 specified then the filter will assume nominal 1.0 tempo. Tempo must be
2933 in the [0.5, 100.0] range.
2934 Note that tempo greater than 2 will skip some samples rather than blend
2936 them in. If for any reason this is a concern it is always possible to
2937 daisy-chain several instances of atempo to achieve the desired product
2938 tempo.
2939 Examples
2941 • Slow down audio to 80% tempo:
2943 atempo=0.8
2945 • To speed up audio to 300% tempo:
2947 atempo=3
2949 • To speed up audio to 300% tempo by daisy-chaining two atempo
2951 instances:
2952 atempo=sqrt(3),atempo=sqrt(3)
2954 Commands
2956 This filter supports the following commands:
2958 tempo
2960 Change filter tempo scale factor. Syntax for the command is :
2961 "tempo"
2962 atilt
2964 Apply spectral tilt filter to audio stream.
2965 This filter apply any spectral roll-off slope over any specified
2967 frequency band.
2968 The filter accepts the following options:
2970 freq
2972 Set central frequency of tilt in Hz. Default is 10000 Hz.
2973 slope
2975 Set slope direction of tilt. Default is 0. Allowed range is from -1
2976 to 1.
2977 width
2979 Set width of tilt. Default is 1000. Allowed range is from 100 to
2980 10000.
2981 order
2983 Set order of tilt filter.
2984 level
2986 Set input volume level. Allowed range is from 0 to 4. Defalt is 1.
2987 Commands
2989 This filter supports the all above options as commands.
2991 atrim
2993 Trim the input so that the output contains one continuous subpart of
2994 the input.
2995 It accepts the following parameters:
2997 start
2999 Timestamp (in seconds) of the start of the section to keep. I.e.
3000 the audio sample with the timestamp start will be the first sample
3001 in the output.
3002 end Specify time of the first audio sample that will be dropped, i.e.
3004 the audio sample immediately preceding the one with the timestamp
3005 end will be the last sample in the output.
3006 start_pts
3008 Same as start, except this option sets the start timestamp in
3009 samples instead of seconds.
3010 end_pts
3012 Same as end, except this option sets the end timestamp in samples
3013 instead of seconds.
3014 duration
3016 The maximum duration of the output in seconds.
3017 start_sample
3019 The number of the first sample that should be output.
3020 end_sample
3022 The number of the first sample that should be dropped.
3023 start, end, and duration are expressed as time duration specifications;
3025 see the Time duration section in the ffmpeg-utils(1) manual.
3026 Note that the first two sets of the start/end options and the duration
3028 option look at the frame timestamp, while the _sample options simply
3029 count the samples that pass through the filter. So start/end_pts and
3030 start/end_sample will give different results when the timestamps are
3031 wrong, inexact or do not start at zero. Also note that this filter does
3032 not modify the timestamps. If you wish to have the output timestamps
3033 start at zero, insert the asetpts filter after the atrim filter.
3034 If multiple start or end options are set, this filter tries to be
3036 greedy and keep all samples that match at least one of the specified
3037 constraints. To keep only the part that matches all the constraints at
3038 once, chain multiple atrim filters.
3039 The defaults are such that all the input is kept. So it is possible to
3041 set e.g. just the end values to keep everything before the specified
3042 time.
3043 Examples:
3045 • Drop everything except the second minute of input:
3047 ffmpeg -i INPUT -af atrim=60:120
3049 • Keep only the first 1000 samples:
3051 ffmpeg -i INPUT -af atrim=end_sample=1000
3053 axcorrelate
3055 Calculate normalized windowed cross-correlation between two input audio
3056 streams.
3057 Resulted samples are always between -1 and 1 inclusive. If result is 1
3059 it means two input samples are highly correlated in that selected
3060 segment. Result 0 means they are not correlated at all. If result is
3061 -1 it means two input samples are out of phase, which means they cancel
3062 each other.
3063 The filter accepts the following options:
3065 size
3067 Set size of segment over which cross-correlation is calculated.
3068 Default is 256. Allowed range is from 2 to 131072.
3069 algo
3071 Set algorithm for cross-correlation. Can be "slow" or "fast".
3072 Default is "slow". Fast algorithm assumes mean values over any
3073 given segment are always zero and thus need much less calculations
3074 to make. This is generally not true, but is valid for typical
3075 audio streams.
3076 Examples
3078 • Calculate correlation between channels in stereo audio stream:
3080 ffmpeg -i stereo.wav -af channelsplit,axcorrelate=size=1024:algo=fast correlation.wav
3082 bandpass
3084 Apply a two-pole Butterworth band-pass filter with central frequency
3085 frequency, and (3dB-point) band-width width. The csg option selects a
3086 constant skirt gain (peak gain = Q) instead of the default: constant
3087 0dB peak gain. The filter roll off at 6dB per octave (20dB per
3088 decade).
3089 The filter accepts the following options:
3091 frequency, f
3093 Set the filter's central frequency. Default is 3000.
3094 csg Constant skirt gain if set to 1. Defaults to 0.
3096 width_type, t
3098 Set method to specify band-width of filter.
3099 h Hz
3101 q Q-Factor
3103 o octave
3105 s slope
3107 k kHz
3109 width, w
3111 Specify the band-width of a filter in width_type units.
3112 mix, m
3114 How much to use filtered signal in output. Default is 1. Range is
3115 between 0 and 1.
3116 channels, c
3118 Specify which channels to filter, by default all available are
3119 filtered.
3120 normalize, n
3122 Normalize biquad coefficients, by default is disabled. Enabling it
3123 will normalize magnitude response at DC to 0dB.
3124 transform, a
3126 Set transform type of IIR filter.
3127 di
3129 dii
3130 tdi
3131 tdii
3132 latt
3133 svf
3134 zdf
3135 precision, r
3136 Set precison of filtering.
3137 auto
3139 Pick automatic sample format depending on surround filters.
3140 s16 Always use signed 16-bit.
3142 s32 Always use signed 32-bit.
3144 f32 Always use float 32-bit.
3146 f64 Always use float 64-bit.
3148 block_size, b
3150 Set block size used for reverse IIR processing. If this value is
3151 set to high enough value (higher than impulse response length
3152 truncated when reaches near zero values) filtering will become
3153 linear phase otherwise if not big enough it will just produce nasty
3154 artifacts.
3155 Note that filter delay will be exactly this many samples when set
3157 to non-zero value.
3158 Commands
3160 This filter supports the following commands:
3162 frequency, f
3164 Change bandpass frequency. Syntax for the command is : "frequency"
3165 width_type, t
3167 Change bandpass width_type. Syntax for the command is :
3168 "width_type"
3169 width, w
3171 Change bandpass width. Syntax for the command is : "width"
3172 mix, m
3174 Change bandpass mix. Syntax for the command is : "mix"
3175 bandreject
3177 Apply a two-pole Butterworth band-reject filter with central frequency
3178 frequency, and (3dB-point) band-width width. The filter roll off at
3179 6dB per octave (20dB per decade).
3180 The filter accepts the following options:
3182 frequency, f
3184 Set the filter's central frequency. Default is 3000.
3185 width_type, t
3187 Set method to specify band-width of filter.
3188 h Hz
3190 q Q-Factor
3192 o octave
3194 s slope
3196 k kHz
3198 width, w
3200 Specify the band-width of a filter in width_type units.
3201 mix, m
3203 How much to use filtered signal in output. Default is 1. Range is
3204 between 0 and 1.
3205 channels, c
3207 Specify which channels to filter, by default all available are
3208 filtered.
3209 normalize, n
3211 Normalize biquad coefficients, by default is disabled. Enabling it
3212 will normalize magnitude response at DC to 0dB.
3213 transform, a
3215 Set transform type of IIR filter.
3216 di
3218 dii
3219 tdi
3220 tdii
3221 latt
3222 svf
3223 zdf
3224 precision, r
3225 Set precison of filtering.
3226 auto
3228 Pick automatic sample format depending on surround filters.
3229 s16 Always use signed 16-bit.
3231 s32 Always use signed 32-bit.
3233 f32 Always use float 32-bit.
3235 f64 Always use float 64-bit.
3237 block_size, b
3239 Set block size used for reverse IIR processing. If this value is
3240 set to high enough value (higher than impulse response length
3241 truncated when reaches near zero values) filtering will become
3242 linear phase otherwise if not big enough it will just produce nasty
3243 artifacts.
3244 Note that filter delay will be exactly this many samples when set
3246 to non-zero value.
3247 Commands
3249 This filter supports the following commands:
3251 frequency, f
3253 Change bandreject frequency. Syntax for the command is :
3254 "frequency"
3255 width_type, t
3257 Change bandreject width_type. Syntax for the command is :
3258 "width_type"
3259 width, w
3261 Change bandreject width. Syntax for the command is : "width"
3262 mix, m
3264 Change bandreject mix. Syntax for the command is : "mix"
3265 bass, lowshelf
3267 Boost or cut the bass (lower) frequencies of the audio using a two-pole
3268 shelving filter with a response similar to that of a standard hi-fi's
3269 tone-controls. This is also known as shelving equalisation (EQ).
3270 The filter accepts the following options:
3272 gain, g
3274 Give the gain at 0 Hz. Its useful range is about -20 (for a large
3275 cut) to +20 (for a large boost). Beware of clipping when using a
3276 positive gain.
3277 frequency, f
3279 Set the filter's central frequency and so can be used to extend or
3280 reduce the frequency range to be boosted or cut. The default value
3281 is 100 Hz.
3282 width_type, t
3284 Set method to specify band-width of filter.
3285 h Hz
3287 q Q-Factor
3289 o octave
3291 s slope
3293 k kHz
3295 width, w
3297 Determine how steep is the filter's shelf transition.
3298 poles, p
3300 Set number of poles. Default is 2.
3301 mix, m
3303 How much to use filtered signal in output. Default is 1. Range is
3304 between 0 and 1.
3305 channels, c
3307 Specify which channels to filter, by default all available are
3308 filtered.
3309 normalize, n
3311 Normalize biquad coefficients, by default is disabled. Enabling it
3312 will normalize magnitude response at DC to 0dB.
3313 transform, a
3315 Set transform type of IIR filter.
3316 di
3318 dii
3319 tdi
3320 tdii
3321 latt
3322 svf
3323 zdf
3324 precision, r
3325 Set precison of filtering.
3326 auto
3328 Pick automatic sample format depending on surround filters.
3329 s16 Always use signed 16-bit.
3331 s32 Always use signed 32-bit.
3333 f32 Always use float 32-bit.
3335 f64 Always use float 64-bit.
3337 block_size, b
3339 Set block size used for reverse IIR processing. If this value is
3340 set to high enough value (higher than impulse response length
3341 truncated when reaches near zero values) filtering will become
3342 linear phase otherwise if not big enough it will just produce nasty
3343 artifacts.
3344 Note that filter delay will be exactly this many samples when set
3346 to non-zero value.
3347 Commands
3349 This filter supports the following commands:
3351 frequency, f
3353 Change bass frequency. Syntax for the command is : "frequency"
3354 width_type, t
3356 Change bass width_type. Syntax for the command is : "width_type"
3357 width, w
3359 Change bass width. Syntax for the command is : "width"
3360 gain, g
3362 Change bass gain. Syntax for the command is : "gain"
3363 mix, m
3365 Change bass mix. Syntax for the command is : "mix"
3366 biquad
3368 Apply a biquad IIR filter with the given coefficients. Where b0, b1,
3369 b2 and a0, a1, a2 are the numerator and denominator coefficients
3370 respectively. and channels, c specify which channels to filter, by
3371 default all available are filtered.
3372 Commands
3374 This filter supports the following commands:
3376 a0
3378 a1
3379 a2
3380 b0
3381 b1
3382 b2 Change biquad parameter. Syntax for the command is : "value"
3383 mix, m
3385 How much to use filtered signal in output. Default is 1. Range is
3386 between 0 and 1.
3387 channels, c
3389 Specify which channels to filter, by default all available are
3390 filtered.
3391 normalize, n
3393 Normalize biquad coefficients, by default is disabled. Enabling it
3394 will normalize magnitude response at DC to 0dB.
3395 transform, a
3397 Set transform type of IIR filter.
3398 di
3400 dii
3401 tdi
3402 tdii
3403 latt
3404 svf
3405 zdf
3406 precision, r
3407 Set precison of filtering.
3408 auto
3410 Pick automatic sample format depending on surround filters.
3411 s16 Always use signed 16-bit.
3413 s32 Always use signed 32-bit.
3415 f32 Always use float 32-bit.
3417 f64 Always use float 64-bit.
3419 block_size, b
3421 Set block size used for reverse IIR processing. If this value is
3422 set to high enough value (higher than impulse response length
3423 truncated when reaches near zero values) filtering will become
3424 linear phase otherwise if not big enough it will just produce nasty
3425 artifacts.
3426 Note that filter delay will be exactly this many samples when set
3428 to non-zero value.
3429 bs2b
3431 Bauer stereo to binaural transformation, which improves headphone
3432 listening of stereo audio records.
3433 To enable compilation of this filter you need to configure FFmpeg with
3435 "--enable-libbs2b".
3436 It accepts the following parameters:
3438 profile
3440 Pre-defined crossfeed level.
3441 default
3443 Default level (fcut=700, feed=50).
3444 cmoy
3446 Chu Moy circuit (fcut=700, feed=60).
3447 jmeier
3449 Jan Meier circuit (fcut=650, feed=95).
3450 fcut
3452 Cut frequency (in Hz).
3453 feed
3455 Feed level (in Hz).
3456 channelmap
3458 Remap input channels to new locations.
3459 It accepts the following parameters:
3461 map Map channels from input to output. The argument is a '|'-separated
3463 list of mappings, each in the "in_channel-out_channel" or
3464 in_channel form. in_channel can be either the name of the input
3465 channel (e.g. FL for front left) or its index in the input channel
3466 layout. out_channel is the name of the output channel or its index
3467 in the output channel layout. If out_channel is not given then it
3468 is implicitly an index, starting with zero and increasing by one
3469 for each mapping.
3470 channel_layout
3472 The channel layout of the output stream.
3473 If no mapping is present, the filter will implicitly map input channels
3475 to output channels, preserving indices.
3476 Examples
3478 • For example, assuming a 5.1+downmix input MOV file,
3480 ffmpeg -i in.mov -filter 'channelmap=map=DL-FL|DR-FR' out.wav
3482 will create an output WAV file tagged as stereo from the downmix
3484 channels of the input.
3485 • To fix a 5.1 WAV improperly encoded in AAC's native channel order
3487 ffmpeg -i in.wav -filter 'channelmap=1|2|0|5|3|4:5.1' out.wav
3489 channelsplit
3491 Split each channel from an input audio stream into a separate output
3492 stream.
3493 It accepts the following parameters:
3495 channel_layout
3497 The channel layout of the input stream. The default is "stereo".
3498 channels
3500 A channel layout describing the channels to be extracted as
3501 separate output streams or "all" to extract each input channel as a
3502 separate stream. The default is "all".
3503 Choosing channels not present in channel layout in the input will
3505 result in an error.
3506 Examples
3508 • For example, assuming a stereo input MP3 file,
3510 ffmpeg -i in.mp3 -filter_complex channelsplit out.mkv
3512 will create an output Matroska file with two audio streams, one
3514 containing only the left channel and the other the right channel.
3515 • Split a 5.1 WAV file into per-channel files:
3517 ffmpeg -i in.wav -filter_complex
3519 'channelsplit=channel_layout=5.1[FL][FR][FC][LFE][SL][SR]'
3520 -map '[FL]' front_left.wav -map '[FR]' front_right.wav -map '[FC]'
3521 front_center.wav -map '[LFE]' lfe.wav -map '[SL]' side_left.wav -map '[SR]'
3522 side_right.wav
3523 • Extract only LFE from a 5.1 WAV file:
3525 ffmpeg -i in.wav -filter_complex 'channelsplit=channel_layout=5.1:channels=LFE[LFE]'
3527 -map '[LFE]' lfe.wav
3528 chorus
3530 Add a chorus effect to the audio.
3531 Can make a single vocal sound like a chorus, but can also be applied to
3533 instrumentation.
3534 Chorus resembles an echo effect with a short delay, but whereas with
3536 echo the delay is constant, with chorus, it is varied using using
3537 sinusoidal or triangular modulation. The modulation depth defines the
3538 range the modulated delay is played before or after the delay. Hence
3539 the delayed sound will sound slower or faster, that is the delayed
3540 sound tuned around the original one, like in a chorus where some vocals
3541 are slightly off key.
3542 It accepts the following parameters:
3544 in_gain
3546 Set input gain. Default is 0.4.
3547 out_gain
3549 Set output gain. Default is 0.4.
3550 delays
3552 Set delays. A typical delay is around 40ms to 60ms.
3553 decays
3555 Set decays.
3556 speeds
3558 Set speeds.
3559 depths
3561 Set depths.
3562 Examples
3564 • A single delay:
3566 chorus=0.7:0.9:55:0.4:0.25:2
3568 • Two delays:
3570 chorus=0.6:0.9:50|60:0.4|0.32:0.25|0.4:2|1.3
3572 • Fuller sounding chorus with three delays:
3574 chorus=0.5:0.9:50|60|40:0.4|0.32|0.3:0.25|0.4|0.3:2|2.3|1.3
3576 compand
3578 Compress or expand the audio's dynamic range.
3579 It accepts the following parameters:
3581 attacks
3583 decays
3584 A list of times in seconds for each channel over which the
3585 instantaneous level of the input signal is averaged to determine
3586 its volume. attacks refers to increase of volume and decays refers
3587 to decrease of volume. For most situations, the attack time
3588 (response to the audio getting louder) should be shorter than the
3589 decay time, because the human ear is more sensitive to sudden loud
3590 audio than sudden soft audio. A typical value for attack is 0.3
3591 seconds and a typical value for decay is 0.8 seconds. If specified
3592 number of attacks & decays is lower than number of channels, the
3593 last set attack/decay will be used for all remaining channels.
3594 points
3596 A list of points for the transfer function, specified in dB
3597 relative to the maximum possible signal amplitude. Each key points
3598 list must be defined using the following syntax:
3599 "x0/y0|x1/y1|x2/y2|...." or "x0/y0 x1/y1 x2/y2 ...."
3600 The input values must be in strictly increasing order but the
3602 transfer function does not have to be monotonically rising. The
3603 point "0/0" is assumed but may be overridden (by "0/out-dBn").
3604 Typical values for the transfer function are "-70/-70|-60/-20|1/0".
3605 soft-knee
3607 Set the curve radius in dB for all joints. It defaults to 0.01.
3608 gain
3610 Set the additional gain in dB to be applied at all points on the
3611 transfer function. This allows for easy adjustment of the overall
3612 gain. It defaults to 0.
3613 volume
3615 Set an initial volume, in dB, to be assumed for each channel when
3616 filtering starts. This permits the user to supply a nominal level
3617 initially, so that, for example, a very large gain is not applied
3618 to initial signal levels before the companding has begun to
3619 operate. A typical value for audio which is initially quiet is -90
3620 dB. It defaults to 0.
3621 delay
3623 Set a delay, in seconds. The input audio is analyzed immediately,
3624 but audio is delayed before being fed to the volume adjuster.
3625 Specifying a delay approximately equal to the attack/decay times
3626 allows the filter to effectively operate in predictive rather than
3627 reactive mode. It defaults to 0.
3628 Examples
3630 • Make music with both quiet and loud passages suitable for listening
3632 to in a noisy environment:
3633 compand=.3|.3:1|1:-90/-60|-60/-40|-40/-30|-20/-20:6:0:-90:0.2
3635 Another example for audio with whisper and explosion parts:
3637 compand=0|0:1|1:-90/-900|-70/-70|-30/-9|0/-3:6:0:0:0
3639 • A noise gate for when the noise is at a lower level than the
3641 signal:
3642 compand=.1|.1:.2|.2:-900/-900|-50.1/-900|-50/-50:.01:0:-90:.1
3644 • Here is another noise gate, this time for when the noise is at a
3646 higher level than the signal (making it, in some ways, similar to
3647 squelch):
3648 compand=.1|.1:.1|.1:-45.1/-45.1|-45/-900|0/-900:.01:45:-90:.1
3650 • 2:1 compression starting at -6dB:
3652 compand=points=-80/-80|-6/-6|0/-3.8|20/3.5
3654 • 2:1 compression starting at -9dB:
3656 compand=points=-80/-80|-9/-9|0/-5.3|20/2.9
3658 • 2:1 compression starting at -12dB:
3660 compand=points=-80/-80|-12/-12|0/-6.8|20/1.9
3662 • 2:1 compression starting at -18dB:
3664 compand=points=-80/-80|-18/-18|0/-9.8|20/0.7
3666 • 3:1 compression starting at -15dB:
3668 compand=points=-80/-80|-15/-15|0/-10.8|20/-5.2
3670 • Compressor/Gate:
3672 compand=points=-80/-105|-62/-80|-15.4/-15.4|0/-12|20/-7.6
3674 • Expander:
3676 compand=attacks=0:points=-80/-169|-54/-80|-49.5/-64.6|-41.1/-41.1|-25.8/-15|-10.8/-4.5|0/0|20/8.3
3678 • Hard limiter at -6dB:
3680 compand=attacks=0:points=-80/-80|-6/-6|20/-6
3682 • Hard limiter at -12dB:
3684 compand=attacks=0:points=-80/-80|-12/-12|20/-12
3686 • Hard noise gate at -35 dB:
3688 compand=attacks=0:points=-80/-115|-35.1/-80|-35/-35|20/20
3690 • Soft limiter:
3692 compand=attacks=0:points=-80/-80|-12.4/-12.4|-6/-8|0/-6.8|20/-2.8
3694 compensationdelay
3696 Compensation Delay Line is a metric based delay to compensate differing
3697 positions of microphones or speakers.
3698 For example, you have recorded guitar with two microphones placed in
3700 different locations. Because the front of sound wave has fixed speed in
3701 normal conditions, the phasing of microphones can vary and depends on
3702 their location and interposition. The best sound mix can be achieved
3703 when these microphones are in phase (synchronized). Note that a
3704 distance of ~30 cm between microphones makes one microphone capture the
3705 signal in antiphase to the other microphone. That makes the final mix
3706 sound moody. This filter helps to solve phasing problems by adding
3707 different delays to each microphone track and make them synchronized.
3708 The best result can be reached when you take one track as base and
3710 synchronize other tracks one by one with it. Remember that
3711 synchronization/delay tolerance depends on sample rate, too. Higher
3712 sample rates will give more tolerance.
3713 The filter accepts the following parameters:
3715 mm Set millimeters distance. This is compensation distance for fine
3717 tuning. Default is 0.
3718 cm Set cm distance. This is compensation distance for tightening
3720 distance setup. Default is 0.
3721 m Set meters distance. This is compensation distance for hard
3723 distance setup. Default is 0.
3724 dry Set dry amount. Amount of unprocessed (dry) signal. Default is 0.
3726 wet Set wet amount. Amount of processed (wet) signal. Default is 1.
3728 temp
3730 Set temperature in degrees Celsius. This is the temperature of the
3731 environment. Default is 20.
3732 Commands
3734 This filter supports the all above options as commands.
3736 crossfeed
3738 Apply headphone crossfeed filter.
3739 Crossfeed is the process of blending the left and right channels of
3741 stereo audio recording. It is mainly used to reduce extreme stereo
3742 separation of low frequencies.
3743 The intent is to produce more speaker like sound to the listener.
3745 The filter accepts the following options:
3747 strength
3749 Set strength of crossfeed. Default is 0.2. Allowed range is from 0
3750 to 1. This sets gain of low shelf filter for side part of stereo
3751 image. Default is -6dB. Max allowed is -30db when strength is set
3752 to 1.
3753 range
3755 Set soundstage wideness. Default is 0.5. Allowed range is from 0 to
3756 1. This sets cut off frequency of low shelf filter. Default is cut
3757 off near 1550 Hz. With range set to 1 cut off frequency is set to
3758 2100 Hz.
3759 slope
3761 Set curve slope of low shelf filter. Default is 0.5. Allowed range
3762 is from 0.01 to 1.
3763 level_in
3765 Set input gain. Default is 0.9.
3766 level_out
3768 Set output gain. Default is 1.
3769 block_size
3771 Set block size used for reverse IIR processing. If this value is
3772 set to high enough value (higher than impulse response length
3773 truncated when reaches near zero values) filtering will become
3774 linear phase otherwise if not big enough it will just produce nasty
3775 artifacts.
3776 Note that filter delay will be exactly this many samples when set
3778 to non-zero value.
3779 Commands
3781 This filter supports the all above options as commands.
3783 crystalizer
3785 Simple algorithm for audio noise sharpening.
3786 This filter linearly increases differences betweeen each audio sample.
3788 The filter accepts the following options:
3790 i Sets the intensity of effect (default: 2.0). Must be in range
3792 between -10.0 to 0 (unchanged sound) to 10.0 (maximum effect). To
3793 inverse filtering use negative value.
3794 c Enable clipping. By default is enabled.
3796 Commands
3798 This filter supports the all above options as commands.
3800 dcshift
3802 Apply a DC shift to the audio.
3803 This can be useful to remove a DC offset (caused perhaps by a hardware
3805 problem in the recording chain) from the audio. The effect of a DC
3806 offset is reduced headroom and hence volume. The astats filter can be
3807 used to determine if a signal has a DC offset.
3808 shift
3810 Set the DC shift, allowed range is [-1, 1]. It indicates the amount
3811 to shift the audio.
3812 limitergain
3814 Optional. It should have a value much less than 1 (e.g. 0.05 or
3815 0.02) and is used to prevent clipping.
3816 deesser
3818 Apply de-essing to the audio samples.
3819 i Set intensity for triggering de-essing. Allowed range is from 0 to
3821 1. Default is 0.
3822 m Set amount of ducking on treble part of sound. Allowed range is
3824 from 0 to 1. Default is 0.5.
3825 f How much of original frequency content to keep when de-essing.
3827 Allowed range is from 0 to 1. Default is 0.5.
3828 s Set the output mode.
3830 It accepts the following values:
3832 i Pass input unchanged.
3834 o Pass ess filtered out.
3836 e Pass only ess.
3838 Default value is o.
3840 dialoguenhance
3842 Enhance dialogue in stereo audio.
3843 This filter accepts stereo input and produce surround (3.0) channels
3845 output. The newly produced front center channel have enhanced speech
3846 dialogue originally available in both stereo channels. This filter
3847 outputs front left and front right channels same as available in stereo
3848 input.
3849 The filter accepts the following options:
3851 original
3853 Set the original center factor to keep in front center channel
3854 output. Allowed range is from 0 to 1. Default value is 1.
3855 enhance
3857 Set the dialogue enhance factor to put in front center channel
3858 output. Allowed range is from 0 to 3. Default value is 1.
3859 voice
3861 Set the voice detection factor. Allowed range is from 2 to 32.
3862 Default value is 2.
3863 Commands
3865 This filter supports the all above options as commands.
3867 drmeter
3869 Measure audio dynamic range.
3870 DR values of 14 and higher is found in very dynamic material. DR of 8
3872 to 13 is found in transition material. And anything less that 8 have
3873 very poor dynamics and is very compressed.
3874 The filter accepts the following options:
3876 length
3878 Set window length in seconds used to split audio into segments of
3879 equal length. Default is 3 seconds.
3880 dynaudnorm
3882 Dynamic Audio Normalizer.
3883 This filter applies a certain amount of gain to the input audio in
3885 order to bring its peak magnitude to a target level (e.g. 0 dBFS).
3886 However, in contrast to more "simple" normalization algorithms, the
3887 Dynamic Audio Normalizer *dynamically* re-adjusts the gain factor to
3888 the input audio. This allows for applying extra gain to the "quiet"
3889 sections of the audio while avoiding distortions or clipping the "loud"
3890 sections. In other words: The Dynamic Audio Normalizer will "even out"
3891 the volume of quiet and loud sections, in the sense that the volume of
3892 each section is brought to the same target level. Note, however, that
3893 the Dynamic Audio Normalizer achieves this goal *without* applying
3894 "dynamic range compressing". It will retain 100% of the dynamic range
3895 *within* each section of the audio file.
3896 framelen, f
3898 Set the frame length in milliseconds. In range from 10 to 8000
3899 milliseconds. Default is 500 milliseconds. The Dynamic Audio
3900 Normalizer processes the input audio in small chunks, referred to
3901 as frames. This is required, because a peak magnitude has no
3902 meaning for just a single sample value. Instead, we need to
3903 determine the peak magnitude for a contiguous sequence of sample
3904 values. While a "standard" normalizer would simply use the peak
3905 magnitude of the complete file, the Dynamic Audio Normalizer
3906 determines the peak magnitude individually for each frame. The
3907 length of a frame is specified in milliseconds. By default, the
3908 Dynamic Audio Normalizer uses a frame length of 500 milliseconds,
3909 which has been found to give good results with most files. Note
3910 that the exact frame length, in number of samples, will be
3911 determined automatically, based on the sampling rate of the
3912 individual input audio file.
3913 gausssize, g
3915 Set the Gaussian filter window size. In range from 3 to 301, must
3916 be odd number. Default is 31. Probably the most important
3917 parameter of the Dynamic Audio Normalizer is the "window size" of
3918 the Gaussian smoothing filter. The filter's window size is
3919 specified in frames, centered around the current frame. For the
3920 sake of simplicity, this must be an odd number. Consequently, the
3921 default value of 31 takes into account the current frame, as well
3922 as the 15 preceding frames and the 15 subsequent frames. Using a
3923 larger window results in a stronger smoothing effect and thus in
3924 less gain variation, i.e. slower gain adaptation. Conversely, using
3925 a smaller window results in a weaker smoothing effect and thus in
3926 more gain variation, i.e. faster gain adaptation. In other words,
3927 the more you increase this value, the more the Dynamic Audio
3928 Normalizer will behave like a "traditional" normalization filter.
3929 On the contrary, the more you decrease this value, the more the
3930 Dynamic Audio Normalizer will behave like a dynamic range
3931 compressor.
3932 peak, p
3934 Set the target peak value. This specifies the highest permissible
3935 magnitude level for the normalized audio input. This filter will
3936 try to approach the target peak magnitude as closely as possible,
3937 but at the same time it also makes sure that the normalized signal
3938 will never exceed the peak magnitude. A frame's maximum local gain
3939 factor is imposed directly by the target peak magnitude. The
3940 default value is 0.95 and thus leaves a headroom of 5%*. It is not
3941 recommended to go above this value.
3942 maxgain, m
3944 Set the maximum gain factor. In range from 1.0 to 100.0. Default is
3945 10.0. The Dynamic Audio Normalizer determines the maximum possible
3946 (local) gain factor for each input frame, i.e. the maximum gain
3947 factor that does not result in clipping or distortion. The maximum
3948 gain factor is determined by the frame's highest magnitude sample.
3949 However, the Dynamic Audio Normalizer additionally bounds the
3950 frame's maximum gain factor by a predetermined (global) maximum
3951 gain factor. This is done in order to avoid excessive gain factors
3952 in "silent" or almost silent frames. By default, the maximum gain
3953 factor is 10.0, For most inputs the default value should be
3954 sufficient and it usually is not recommended to increase this
3955 value. Though, for input with an extremely low overall volume
3956 level, it may be necessary to allow even higher gain factors. Note,
3957 however, that the Dynamic Audio Normalizer does not simply apply a
3958 "hard" threshold (i.e. cut off values above the threshold).
3959 Instead, a "sigmoid" threshold function will be applied. This way,
3960 the gain factors will smoothly approach the threshold value, but
3961 never exceed that value.
3962 targetrms, r
3964 Set the target RMS. In range from 0.0 to 1.0. Default is 0.0 -
3965 disabled. By default, the Dynamic Audio Normalizer performs "peak"
3966 normalization. This means that the maximum local gain factor for
3967 each frame is defined (only) by the frame's highest magnitude
3968 sample. This way, the samples can be amplified as much as possible
3969 without exceeding the maximum signal level, i.e. without clipping.
3970 Optionally, however, the Dynamic Audio Normalizer can also take
3971 into account the frame's root mean square, abbreviated RMS. In
3972 electrical engineering, the RMS is commonly used to determine the
3973 power of a time-varying signal. It is therefore considered that the
3974 RMS is a better approximation of the "perceived loudness" than just
3975 looking at the signal's peak magnitude. Consequently, by adjusting
3976 all frames to a constant RMS value, a uniform "perceived loudness"
3977 can be established. If a target RMS value has been specified, a
3978 frame's local gain factor is defined as the factor that would
3979 result in exactly that RMS value. Note, however, that the maximum
3980 local gain factor is still restricted by the frame's highest
3981 magnitude sample, in order to prevent clipping.
3982 coupling, n
3984 Enable channels coupling. By default is enabled. By default, the
3985 Dynamic Audio Normalizer will amplify all channels by the same
3986 amount. This means the same gain factor will be applied to all
3987 channels, i.e. the maximum possible gain factor is determined by
3988 the "loudest" channel. However, in some recordings, it may happen
3989 that the volume of the different channels is uneven, e.g. one
3990 channel may be "quieter" than the other one(s). In this case, this
3991 option can be used to disable the channel coupling. This way, the
3992 gain factor will be determined independently for each channel,
3993 depending only on the individual channel's highest magnitude
3994 sample. This allows for harmonizing the volume of the different
3995 channels.
3996 correctdc, c
3998 Enable DC bias correction. By default is disabled. An audio signal
3999 (in the time domain) is a sequence of sample values. In the
4000 Dynamic Audio Normalizer these sample values are represented in the
4001 -1.0 to 1.0 range, regardless of the original input format.
4002 Normally, the audio signal, or "waveform", should be centered
4003 around the zero point. That means if we calculate the mean value
4004 of all samples in a file, or in a single frame, then the result
4005 should be 0.0 or at least very close to that value. If, however,
4006 there is a significant deviation of the mean value from 0.0, in
4007 either positive or negative direction, this is referred to as a DC
4008 bias or DC offset. Since a DC bias is clearly undesirable, the
4009 Dynamic Audio Normalizer provides optional DC bias correction.
4010 With DC bias correction enabled, the Dynamic Audio Normalizer will
4011 determine the mean value, or "DC correction" offset, of each input
4012 frame and subtract that value from all of the frame's sample values
4013 which ensures those samples are centered around 0.0 again. Also, in
4014 order to avoid "gaps" at the frame boundaries, the DC correction
4015 offset values will be interpolated smoothly between neighbouring
4016 frames.
4017 altboundary, b
4019 Enable alternative boundary mode. By default is disabled. The
4020 Dynamic Audio Normalizer takes into account a certain neighbourhood
4021 around each frame. This includes the preceding frames as well as
4022 the subsequent frames. However, for the "boundary" frames, located
4023 at the very beginning and at the very end of the audio file, not
4024 all neighbouring frames are available. In particular, for the first
4025 few frames in the audio file, the preceding frames are not known.
4026 And, similarly, for the last few frames in the audio file, the
4027 subsequent frames are not known. Thus, the question arises which
4028 gain factors should be assumed for the missing frames in the
4029 "boundary" region. The Dynamic Audio Normalizer implements two
4030 modes to deal with this situation. The default boundary mode
4031 assumes a gain factor of exactly 1.0 for the missing frames,
4032 resulting in a smooth "fade in" and "fade out" at the beginning and
4033 at the end of the input, respectively.
4034 compress, s
4036 Set the compress factor. In range from 0.0 to 30.0. Default is 0.0.
4037 By default, the Dynamic Audio Normalizer does not apply
4038 "traditional" compression. This means that signal peaks will not be
4039 pruned and thus the full dynamic range will be retained within each
4040 local neighbourhood. However, in some cases it may be desirable to
4041 combine the Dynamic Audio Normalizer's normalization algorithm with
4042 a more "traditional" compression. For this purpose, the Dynamic
4043 Audio Normalizer provides an optional compression (thresholding)
4044 function. If (and only if) the compression feature is enabled, all
4045 input frames will be processed by a soft knee thresholding function
4046 prior to the actual normalization process. Put simply, the
4047 thresholding function is going to prune all samples whose magnitude
4048 exceeds a certain threshold value. However, the Dynamic Audio
4049 Normalizer does not simply apply a fixed threshold value. Instead,
4050 the threshold value will be adjusted for each individual frame. In
4051 general, smaller parameters result in stronger compression, and
4052 vice versa. Values below 3.0 are not recommended, because audible
4053 distortion may appear.
4054 threshold, t
4056 Set the target threshold value. This specifies the lowest
4057 permissible magnitude level for the audio input which will be
4058 normalized. If input frame volume is above this value frame will
4059 be normalized. Otherwise frame may not be normalized at all. The
4060 default value is set to 0, which means all input frames will be
4061 normalized. This option is mostly useful if digital noise is not
4062 wanted to be amplified.
4063 channels, h
4065 Specify which channels to filter, by default all available channels
4066 are filtered.
4067 overlap, o
4069 Specify overlap for frames. If set to 0 (default) no frame
4070 overlapping is done. Using >0 and <1 values will make less
4071 conservative gain adjustments, like when framelen option is set to
4072 smaller value, if framelen option value is compensated for non-zero
4073 overlap then gain adjustments will be smoother across time compared
4074 to zero overlap case.
4075 Commands
4077 This filter supports the all above options as commands.
4079 earwax
4081 Make audio easier to listen to on headphones.
4082 This filter adds `cues' to 44.1kHz stereo (i.e. audio CD format) audio
4084 so that when listened to on headphones the stereo image is moved from
4085 inside your head (standard for headphones) to outside and in front of
4086 the listener (standard for speakers).
4087 Ported from SoX.
4089 equalizer
4091 Apply a two-pole peaking equalisation (EQ) filter. With this filter,
4092 the signal-level at and around a selected frequency can be increased or
4093 decreased, whilst (unlike bandpass and bandreject filters) that at all
4094 other frequencies is unchanged.
4095 In order to produce complex equalisation curves, this filter can be
4097 given several times, each with a different central frequency.
4098 The filter accepts the following options:
4100 frequency, f
4102 Set the filter's central frequency in Hz.
4103 width_type, t
4105 Set method to specify band-width of filter.
4106 h Hz
4108 q Q-Factor
4110 o octave
4112 s slope
4114 k kHz
4116 width, w
4118 Specify the band-width of a filter in width_type units.
4119 gain, g
4121 Set the required gain or attenuation in dB. Beware of clipping
4122 when using a positive gain.
4123 mix, m
4125 How much to use filtered signal in output. Default is 1. Range is
4126 between 0 and 1.
4127 channels, c
4129 Specify which channels to filter, by default all available are
4130 filtered.
4131 normalize, n
4133 Normalize biquad coefficients, by default is disabled. Enabling it
4134 will normalize magnitude response at DC to 0dB.
4135 transform, a
4137 Set transform type of IIR filter.
4138 di
4140 dii
4141 tdi
4142 tdii
4143 latt
4144 svf
4145 zdf
4146 precision, r
4147 Set precison of filtering.
4148 auto
4150 Pick automatic sample format depending on surround filters.
4151 s16 Always use signed 16-bit.
4153 s32 Always use signed 32-bit.
4155 f32 Always use float 32-bit.
4157 f64 Always use float 64-bit.
4159 block_size, b
4161 Set block size used for reverse IIR processing. If this value is
4162 set to high enough value (higher than impulse response length
4163 truncated when reaches near zero values) filtering will become
4164 linear phase otherwise if not big enough it will just produce nasty
4165 artifacts.
4166 Note that filter delay will be exactly this many samples when set
4168 to non-zero value.
4169 Examples
4171 • Attenuate 10 dB at 1000 Hz, with a bandwidth of 200 Hz:
4173 equalizer=f=1000:t=h:width=200:g=-10
4175 • Apply 2 dB gain at 1000 Hz with Q 1 and attenuate 5 dB at 100 Hz
4177 with Q 2:
4178 equalizer=f=1000:t=q:w=1:g=2,equalizer=f=100:t=q:w=2:g=-5
4180 Commands
4182 This filter supports the following commands:
4184 frequency, f
4186 Change equalizer frequency. Syntax for the command is :
4187 "frequency"
4188 width_type, t
4190 Change equalizer width_type. Syntax for the command is :
4191 "width_type"
4192 width, w
4194 Change equalizer width. Syntax for the command is : "width"
4195 gain, g
4197 Change equalizer gain. Syntax for the command is : "gain"
4198 mix, m
4200 Change equalizer mix. Syntax for the command is : "mix"
4201 extrastereo
4203 Linearly increases the difference between left and right channels which
4204 adds some sort of "live" effect to playback.
4205 The filter accepts the following options:
4207 m Sets the difference coefficient (default: 2.5). 0.0 means mono
4209 sound (average of both channels), with 1.0 sound will be unchanged,
4210 with -1.0 left and right channels will be swapped.
4211 c Enable clipping. By default is enabled.
4213 Commands
4215 This filter supports the all above options as commands.
4217 firequalizer
4219 Apply FIR Equalization using arbitrary frequency response.
4220 The filter accepts the following option:
4222 gain
4224 Set gain curve equation (in dB). The expression can contain
4225 variables:
4226 f the evaluated frequency
4228 sr sample rate
4230 ch channel number, set to 0 when multichannels evaluation is
4232 disabled
4233 chid
4235 channel id, see libavutil/channel_layout.h, set to the first
4236 channel id when multichannels evaluation is disabled
4237 chs number of channels
4239 chlayout
4241 channel_layout, see libavutil/channel_layout.h
4242 and functions:
4244 gain_interpolate(f)
4246 interpolate gain on frequency f based on gain_entry
4247 cubic_interpolate(f)
4249 same as gain_interpolate, but smoother
4250 This option is also available as command. Default is
4252 gain_interpolate(f).
4253 gain_entry
4255 Set gain entry for gain_interpolate function. The expression can
4256 contain functions:
4257 entry(f, g)
4259 store gain entry at frequency f with value g
4260 This option is also available as command.
4262 delay
4264 Set filter delay in seconds. Higher value means more accurate.
4265 Default is 0.01.
4266 accuracy
4268 Set filter accuracy in Hz. Lower value means more accurate.
4269 Default is 5.
4270 wfunc
4272 Set window function. Acceptable values are:
4273 rectangular
4275 rectangular window, useful when gain curve is already smooth
4276 hann
4278 hann window (default)
4279 hamming
4281 hamming window
4282 blackman
4284 blackman window
4285 nuttall3
4287 3-terms continuous 1st derivative nuttall window
4288 mnuttall3
4290 minimum 3-terms discontinuous nuttall window
4291 nuttall
4293 4-terms continuous 1st derivative nuttall window
4294 bnuttall
4296 minimum 4-terms discontinuous nuttall (blackman-nuttall) window
4297 bharris
4299 blackman-harris window
4300 tukey
4302 tukey window
4303 fixed
4305 If enabled, use fixed number of audio samples. This improves speed
4306 when filtering with large delay. Default is disabled.
4307 multi
4309 Enable multichannels evaluation on gain. Default is disabled.
4310 zero_phase
4312 Enable zero phase mode by subtracting timestamp to compensate
4313 delay. Default is disabled.
4314 scale
4316 Set scale used by gain. Acceptable values are:
4317 linlin
4319 linear frequency, linear gain
4320 linlog
4322 linear frequency, logarithmic (in dB) gain (default)
4323 loglin
4325 logarithmic (in octave scale where 20 Hz is 0) frequency,
4326 linear gain
4327 loglog
4329 logarithmic frequency, logarithmic gain
4330 dumpfile
4332 Set file for dumping, suitable for gnuplot.
4333 dumpscale
4335 Set scale for dumpfile. Acceptable values are same with scale
4336 option. Default is linlog.
4337 fft2
4339 Enable 2-channel convolution using complex FFT. This improves speed
4340 significantly. Default is disabled.
4341 min_phase
4343 Enable minimum phase impulse response. Default is disabled.
4344 Examples
4346 • lowpass at 1000 Hz:
4348 firequalizer=gain='if(lt(f,1000), 0, -INF)'
4350 • lowpass at 1000 Hz with gain_entry:
4352 firequalizer=gain_entry='entry(1000,0); entry(1001, -INF)'
4354 • custom equalization:
4356 firequalizer=gain_entry='entry(100,0); entry(400, -4); entry(1000, -6); entry(2000, 0)'
4358 • higher delay with zero phase to compensate delay:
4360 firequalizer=delay=0.1:fixed=on:zero_phase=on
4362 • lowpass on left channel, highpass on right channel:
4364 firequalizer=gain='if(eq(chid,1), gain_interpolate(f), if(eq(chid,2), gain_interpolate(1e6+f), 0))'
4366 :gain_entry='entry(1000, 0); entry(1001,-INF); entry(1e6+1000,0)':multi=on
4367 flanger
4369 Apply a flanging effect to the audio.
4370 The filter accepts the following options:
4372 delay
4374 Set base delay in milliseconds. Range from 0 to 30. Default value
4375 is 0.
4376 depth
4378 Set added sweep delay in milliseconds. Range from 0 to 10. Default
4379 value is 2.
4380 regen
4382 Set percentage regeneration (delayed signal feedback). Range from
4383 -95 to 95. Default value is 0.
4384 width
4386 Set percentage of delayed signal mixed with original. Range from 0
4387 to 100. Default value is 71.
4388 speed
4390 Set sweeps per second (Hz). Range from 0.1 to 10. Default value is
4391 0.5.
4392 shape
4394 Set swept wave shape, can be triangular or sinusoidal. Default
4395 value is sinusoidal.
4396 phase
4398 Set swept wave percentage-shift for multi channel. Range from 0 to
4399 100. Default value is 25.
4400 interp
4402 Set delay-line interpolation, linear or quadratic. Default is
4403 linear.
4404 haas
4406 Apply Haas effect to audio.
4407 Note that this makes most sense to apply on mono signals. With this
4409 filter applied to mono signals it give some directionality and
4410 stretches its stereo image.
4411 The filter accepts the following options:
4413 level_in
4415 Set input level. By default is 1, or 0dB
4416 level_out
4418 Set output level. By default is 1, or 0dB.
4419 side_gain
4421 Set gain applied to side part of signal. By default is 1.
4422 middle_source
4424 Set kind of middle source. Can be one of the following:
4425 left
4427 Pick left channel.
4428 right
4430 Pick right channel.
4431 mid Pick middle part signal of stereo image.
4433 side
4435 Pick side part signal of stereo image.
4436 middle_phase
4438 Change middle phase. By default is disabled.
4439 left_delay
4441 Set left channel delay. By default is 2.05 milliseconds.
4442 left_balance
4444 Set left channel balance. By default is -1.
4445 left_gain
4447 Set left channel gain. By default is 1.
4448 left_phase
4450 Change left phase. By default is disabled.
4451 right_delay
4453 Set right channel delay. By defaults is 2.12 milliseconds.
4454 right_balance
4456 Set right channel balance. By default is 1.
4457 right_gain
4459 Set right channel gain. By default is 1.
4460 right_phase
4462 Change right phase. By default is enabled.
4463 hdcd
4465 Decodes High Definition Compatible Digital (HDCD) data. A 16-bit PCM
4466 stream with embedded HDCD codes is expanded into a 20-bit PCM stream.
4467 The filter supports the Peak Extend and Low-level Gain Adjustment
4469 features of HDCD, and detects the Transient Filter flag.
4470 ffmpeg -i HDCD16.flac -af hdcd OUT24.flac
4472 When using the filter with wav, note the default encoding for wav is
4474 16-bit, so the resulting 20-bit stream will be truncated back to
4475 16-bit. Use something like -acodec pcm_s24le after the filter to get
4476 24-bit PCM output.
4477 ffmpeg -i HDCD16.wav -af hdcd OUT16.wav
4479 ffmpeg -i HDCD16.wav -af hdcd -c:a pcm_s24le OUT24.wav
4480 The filter accepts the following options:
4482 disable_autoconvert
4484 Disable any automatic format conversion or resampling in the filter
4485 graph.
4486 process_stereo
4488 Process the stereo channels together. If target_gain does not match
4489 between channels, consider it invalid and use the last valid
4490 target_gain.
4491 cdt_ms
4493 Set the code detect timer period in ms.
4494 force_pe
4496 Always extend peaks above -3dBFS even if PE isn't signaled.
4497 analyze_mode
4499 Replace audio with a solid tone and adjust the amplitude to signal
4500 some specific aspect of the decoding process. The output file can
4501 be loaded in an audio editor alongside the original to aid
4502 analysis.
4503 "analyze_mode=pe:force_pe=true" can be used to see all samples
4505 above the PE level.
4506 Modes are:
4508 0, off
4510 Disabled
4511 1, lle
4513 Gain adjustment level at each sample
4514 2, pe
4516 Samples where peak extend occurs
4517 3, cdt
4519 Samples where the code detect timer is active
4520 4, tgm
4522 Samples where the target gain does not match between channels
4523 headphone
4525 Apply head-related transfer functions (HRTFs) to create virtual
4526 loudspeakers around the user for binaural listening via headphones.
4527 The HRIRs are provided via additional streams, for each channel one
4528 stereo input stream is needed.
4529 The filter accepts the following options:
4531 map Set mapping of input streams for convolution. The argument is a
4533 '|'-separated list of channel names in order as they are given as
4534 additional stream inputs for filter. This also specify number of
4535 input streams. Number of input streams must be not less than number
4536 of channels in first stream plus one.
4537 gain
4539 Set gain applied to audio. Value is in dB. Default is 0.
4540 type
4542 Set processing type. Can be time or freq. time is processing audio
4543 in time domain which is slow. freq is processing audio in
4544 frequency domain which is fast. Default is freq.
4545 lfe Set custom gain for LFE channels. Value is in dB. Default is 0.
4547 size
4549 Set size of frame in number of samples which will be processed at
4550 once. Default value is 1024. Allowed range is from 1024 to 96000.
4551 hrir
4553 Set format of hrir stream. Default value is stereo. Alternative
4554 value is multich. If value is set to stereo, number of additional
4555 streams should be greater or equal to number of input channels in
4556 first input stream. Also each additional stream should have stereo
4557 number of channels. If value is set to multich, number of
4558 additional streams should be exactly one. Also number of input
4559 channels of additional stream should be equal or greater than twice
4560 number of channels of first input stream.
4561 Examples
4563 • Full example using wav files as coefficients with amovie filters
4565 for 7.1 downmix, each amovie filter use stereo file with IR
4566 coefficients as input. The files give coefficients for each
4567 position of virtual loudspeaker:
4568 ffmpeg -i input.wav
4570 -filter_complex "amovie=azi_270_ele_0_DFC.wav[sr];amovie=azi_90_ele_0_DFC.wav[sl];amovie=azi_225_ele_0_DFC.wav[br];amovie=azi_135_ele_0_DFC.wav[bl];amovie=azi_0_ele_0_DFC.wav,asplit[fc][lfe];amovie=azi_35_ele_0_DFC.wav[fl];amovie=azi_325_ele_0_DFC.wav[fr];[0:a][fl][fr][fc][lfe][bl][br][sl][sr]headphone=FL|FR|FC|LFE|BL|BR|SL|SR"
4571 output.wav
4572 • Full example using wav files as coefficients with amovie filters
4574 for 7.1 downmix, but now in multich hrir format.
4575 ffmpeg -i input.wav -filter_complex "amovie=minp.wav[hrirs];[0:a][hrirs]headphone=map=FL|FR|FC|LFE|BL|BR|SL|SR:hrir=multich"
4577 output.wav
4578 highpass
4580 Apply a high-pass filter with 3dB point frequency. The filter can be
4581 either single-pole, or double-pole (the default). The filter roll off
4582 at 6dB per pole per octave (20dB per pole per decade).
4583 The filter accepts the following options:
4585 frequency, f
4587 Set frequency in Hz. Default is 3000.
4588 poles, p
4590 Set number of poles. Default is 2.
4591 width_type, t
4593 Set method to specify band-width of filter.
4594 h Hz
4596 q Q-Factor
4598 o octave
4600 s slope
4602 k kHz
4604 width, w
4606 Specify the band-width of a filter in width_type units. Applies
4607 only to double-pole filter. The default is 0.707q and gives a
4608 Butterworth response.
4609 mix, m
4611 How much to use filtered signal in output. Default is 1. Range is
4612 between 0 and 1.
4613 channels, c
4615 Specify which channels to filter, by default all available are
4616 filtered.
4617 normalize, n
4619 Normalize biquad coefficients, by default is disabled. Enabling it
4620 will normalize magnitude response at DC to 0dB.
4621 transform, a
4623 Set transform type of IIR filter.
4624 di
4626 dii
4627 tdi
4628 tdii
4629 latt
4630 svf
4631 zdf
4632 precision, r
4633 Set precison of filtering.
4634 auto
4636 Pick automatic sample format depending on surround filters.
4637 s16 Always use signed 16-bit.
4639 s32 Always use signed 32-bit.
4641 f32 Always use float 32-bit.
4643 f64 Always use float 64-bit.
4645 block_size, b
4647 Set block size used for reverse IIR processing. If this value is
4648 set to high enough value (higher than impulse response length
4649 truncated when reaches near zero values) filtering will become
4650 linear phase otherwise if not big enough it will just produce nasty
4651 artifacts.
4652 Note that filter delay will be exactly this many samples when set
4654 to non-zero value.
4655 Commands
4657 This filter supports the following commands:
4659 frequency, f
4661 Change highpass frequency. Syntax for the command is : "frequency"
4662 width_type, t
4664 Change highpass width_type. Syntax for the command is :
4665 "width_type"
4666 width, w
4668 Change highpass width. Syntax for the command is : "width"
4669 mix, m
4671 Change highpass mix. Syntax for the command is : "mix"
4672 join
4674 Join multiple input streams into one multi-channel stream.
4675 It accepts the following parameters:
4677 inputs
4679 The number of input streams. It defaults to 2.
4680 channel_layout
4682 The desired output channel layout. It defaults to stereo.
4683 map Map channels from inputs to output. The argument is a '|'-separated
4685 list of mappings, each in the "input_idx.in_channel-out_channel"
4686 form. input_idx is the 0-based index of the input stream.
4687 in_channel can be either the name of the input channel (e.g. FL for
4688 front left) or its index in the specified input stream. out_channel
4689 is the name of the output channel.
4690 The filter will attempt to guess the mappings when they are not
4692 specified explicitly. It does so by first trying to find an unused
4693 matching input channel and if that fails it picks the first unused
4694 input channel.
4695 Join 3 inputs (with properly set channel layouts):
4697 ffmpeg -i INPUT1 -i INPUT2 -i INPUT3 -filter_complex join=inputs=3 OUTPUT
4699 Build a 5.1 output from 6 single-channel streams:
4701 ffmpeg -i fl -i fr -i fc -i sl -i sr -i lfe -filter_complex
4703 'join=inputs=6:channel_layout=5.1:map=0.0-FL|1.0-FR|2.0-FC|3.0-SL|4.0-SR|5.0-LFE'
4704 out
4705 ladspa
4707 Load a LADSPA (Linux Audio Developer's Simple Plugin API) plugin.
4708 To enable compilation of this filter you need to configure FFmpeg with
4710 "--enable-ladspa".
4711 file, f
4713 Specifies the name of LADSPA plugin library to load. If the
4714 environment variable LADSPA_PATH is defined, the LADSPA plugin is
4715 searched in each one of the directories specified by the colon
4716 separated list in LADSPA_PATH, otherwise in the standard LADSPA
4717 paths, which are in this order: HOME/.ladspa/lib/,
4718 /usr/local/lib/ladspa/, /usr/lib/ladspa/.
4719 plugin, p
4721 Specifies the plugin within the library. Some libraries contain
4722 only one plugin, but others contain many of them. If this is not
4723 set filter will list all available plugins within the specified
4724 library.
4725 controls, c
4727 Set the '|' separated list of controls which are zero or more
4728 floating point values that determine the behavior of the loaded
4729 plugin (for example delay, threshold or gain). Controls need to be
4730 defined using the following syntax:
4731 c0=value0|c1=value1|c2=value2|..., where valuei is the value set on
4732 the i-th control. Alternatively they can be also defined using the
4733 following syntax: value0|value1|value2|..., where valuei is the
4734 value set on the i-th control. If controls is set to "help", all
4735 available controls and their valid ranges are printed.
4736 sample_rate, s
4738 Specify the sample rate, default to 44100. Only used if plugin have
4739 zero inputs.
4740 nb_samples, n
4742 Set the number of samples per channel per each output frame,
4743 default is 1024. Only used if plugin have zero inputs.
4744 duration, d
4746 Set the minimum duration of the sourced audio. See the Time
4747 duration section in the ffmpeg-utils(1) manual for the accepted
4748 syntax. Note that the resulting duration may be greater than the
4749 specified duration, as the generated audio is always cut at the end
4750 of a complete frame. If not specified, or the expressed duration
4751 is negative, the audio is supposed to be generated forever. Only
4752 used if plugin have zero inputs.
4753 latency, l
4755 Enable latency compensation, by default is disabled. Only used if
4756 plugin have inputs.
4757 Examples
4759 • List all available plugins within amp (LADSPA example plugin)
4761 library:
4762 ladspa=file=amp
4764 • List all available controls and their valid ranges for "vcf_notch"
4766 plugin from "VCF" library:
4767 ladspa=f=vcf:p=vcf_notch:c=help
4769 • Simulate low quality audio equipment using "Computer Music Toolkit"
4771 (CMT) plugin library:
4772 ladspa=file=cmt:plugin=lofi:controls=c0=22|c1=12|c2=12
4774 • Add reverberation to the audio using TAP-plugins (Tom's Audio
4776 Processing plugins):
4777 ladspa=file=tap_reverb:tap_reverb
4779 • Generate white noise, with 0.2 amplitude:
4781 ladspa=file=cmt:noise_source_white:c=c0=.2
4783 • Generate 20 bpm clicks using plugin "C* Click - Metronome" from the
4785 "C* Audio Plugin Suite" (CAPS) library:
4786 ladspa=file=caps:Click:c=c1=20'
4788 • Apply "C* Eq10X2 - Stereo 10-band equaliser" effect:
4790 ladspa=caps:Eq10X2:c=c0=-48|c9=-24|c3=12|c4=2
4792 • Increase volume by 20dB using fast lookahead limiter from Steve
4794 Harris "SWH Plugins" collection:
4795 ladspa=fast_lookahead_limiter_1913:fastLookaheadLimiter:20|0|2
4797 • Attenuate low frequencies using Multiband EQ from Steve Harris "SWH
4799 Plugins" collection:
4800 ladspa=mbeq_1197:mbeq:-24|-24|-24|0|0|0|0|0|0|0|0|0|0|0|0
4802 • Reduce stereo image using "Narrower" from the "C* Audio Plugin
4804 Suite" (CAPS) library:
4805 ladspa=caps:Narrower
4807 • Another white noise, now using "C* Audio Plugin Suite" (CAPS)
4809 library:
4810 ladspa=caps:White:.2
4812 • Some fractal noise, using "C* Audio Plugin Suite" (CAPS) library:
4814 ladspa=caps:Fractal:c=c1=1
4816 • Dynamic volume normalization using "VLevel" plugin:
4818 ladspa=vlevel-ladspa:vlevel_mono
4820 Commands
4822 This filter supports the following commands:
4824 cN Modify the N-th control value.
4826 If the specified value is not valid, it is ignored and prior one is
4828 kept.
4829 loudnorm
4831 EBU R128 loudness normalization. Includes both dynamic and linear
4832 normalization modes. Support for both single pass (livestreams, files)
4833 and double pass (files) modes. This algorithm can target IL, LRA, and
4834 maximum true peak. In dynamic mode, to accurately detect true peaks,
4835 the audio stream will be upsampled to 192 kHz. Use the "-ar" option or
4836 "aresample" filter to explicitly set an output sample rate.
4837 The filter accepts the following options:
4839 I, i
4841 Set integrated loudness target. Range is -70.0 - -5.0. Default
4842 value is -24.0.
4843 LRA, lra
4845 Set loudness range target. Range is 1.0 - 50.0. Default value is
4846 7.0.
4847 TP, tp
4849 Set maximum true peak. Range is -9.0 - +0.0. Default value is
4850 -2.0.
4851 measured_I, measured_i
4853 Measured IL of input file. Range is -99.0 - +0.0.
4854 measured_LRA, measured_lra
4856 Measured LRA of input file. Range is 0.0 - 99.0.
4857 measured_TP, measured_tp
4859 Measured true peak of input file. Range is -99.0 - +99.0.
4860 measured_thresh
4862 Measured threshold of input file. Range is -99.0 - +0.0.
4863 offset
4865 Set offset gain. Gain is applied before the true-peak limiter.
4866 Range is -99.0 - +99.0. Default is +0.0.
4867 linear
4869 Normalize by linearly scaling the source audio. "measured_I",
4870 "measured_LRA", "measured_TP", and "measured_thresh" must all be
4871 specified. Target LRA shouldn't be lower than source LRA and the
4872 change in integrated loudness shouldn't result in a true peak which
4873 exceeds the target TP. If any of these conditions aren't met,
4874 normalization mode will revert to dynamic. Options are "true" or
4875 "false". Default is "true".
4876 dual_mono
4878 Treat mono input files as "dual-mono". If a mono file is intended
4879 for playback on a stereo system, its EBU R128 measurement will be
4880 perceptually incorrect. If set to "true", this option will
4881 compensate for this effect. Multi-channel input files are not
4882 affected by this option. Options are true or false. Default is
4883 false.
4884 print_format
4886 Set print format for stats. Options are summary, json, or none.
4887 Default value is none.
4888 lowpass
4890 Apply a low-pass filter with 3dB point frequency. The filter can be
4891 either single-pole or double-pole (the default). The filter roll off
4892 at 6dB per pole per octave (20dB per pole per decade).
4893 The filter accepts the following options:
4895 frequency, f
4897 Set frequency in Hz. Default is 500.
4898 poles, p
4900 Set number of poles. Default is 2.
4901 width_type, t
4903 Set method to specify band-width of filter.
4904 h Hz
4906 q Q-Factor
4908 o octave
4910 s slope
4912 k kHz
4914 width, w
4916 Specify the band-width of a filter in width_type units. Applies
4917 only to double-pole filter. The default is 0.707q and gives a
4918 Butterworth response.
4919 mix, m
4921 How much to use filtered signal in output. Default is 1. Range is
4922 between 0 and 1.
4923 channels, c
4925 Specify which channels to filter, by default all available are
4926 filtered.
4927 normalize, n
4929 Normalize biquad coefficients, by default is disabled. Enabling it
4930 will normalize magnitude response at DC to 0dB.
4931 transform, a
4933 Set transform type of IIR filter.
4934 di
4936 dii
4937 tdi
4938 tdii
4939 latt
4940 svf
4941 zdf
4942 precision, r
4943 Set precison of filtering.
4944 auto
4946 Pick automatic sample format depending on surround filters.
4947 s16 Always use signed 16-bit.
4949 s32 Always use signed 32-bit.
4951 f32 Always use float 32-bit.
4953 f64 Always use float 64-bit.
4955 block_size, b
4957 Set block size used for reverse IIR processing. If this value is
4958 set to high enough value (higher than impulse response length
4959 truncated when reaches near zero values) filtering will become
4960 linear phase otherwise if not big enough it will just produce nasty
4961 artifacts.
4962 Note that filter delay will be exactly this many samples when set
4964 to non-zero value.
4965 Examples
4967 • Lowpass only LFE channel, it LFE is not present it does nothing:
4969 lowpass=c=LFE
4971 Commands
4973 This filter supports the following commands:
4975 frequency, f
4977 Change lowpass frequency. Syntax for the command is : "frequency"
4978 width_type, t
4980 Change lowpass width_type. Syntax for the command is :
4981 "width_type"
4982 width, w
4984 Change lowpass width. Syntax for the command is : "width"
4985 mix, m
4987 Change lowpass mix. Syntax for the command is : "mix"
4988 lv2
4990 Load a LV2 (LADSPA Version 2) plugin.
4991 To enable compilation of this filter you need to configure FFmpeg with
4993 "--enable-lv2".
4994 plugin, p
4996 Specifies the plugin URI. You may need to escape ':'.
4997 controls, c
4999 Set the '|' separated list of controls which are zero or more
5000 floating point values that determine the behavior of the loaded
5001 plugin (for example delay, threshold or gain). If controls is set
5002 to "help", all available controls and their valid ranges are
5003 printed.
5004 sample_rate, s
5006 Specify the sample rate, default to 44100. Only used if plugin have
5007 zero inputs.
5008 nb_samples, n
5010 Set the number of samples per channel per each output frame,
5011 default is 1024. Only used if plugin have zero inputs.
5012 duration, d
5014 Set the minimum duration of the sourced audio. See the Time
5015 duration section in the ffmpeg-utils(1) manual for the accepted
5016 syntax. Note that the resulting duration may be greater than the
5017 specified duration, as the generated audio is always cut at the end
5018 of a complete frame. If not specified, or the expressed duration
5019 is negative, the audio is supposed to be generated forever. Only
5020 used if plugin have zero inputs.
5021 Examples
5023 • Apply bass enhancer plugin from Calf:
5025 lv2=p=http\\\\://calf.sourceforge.net/plugins/BassEnhancer:c=amount=2
5027 • Apply vinyl plugin from Calf:
5029 lv2=p=http\\\\://calf.sourceforge.net/plugins/Vinyl:c=drone=0.2|aging=0.5
5031 • Apply bit crusher plugin from ArtyFX:
5033 lv2=p=http\\\\://www.openavproductions.com/artyfx#bitta:c=crush=0.3
5035 Commands
5037 This filter supports all options that are exported by plugin as
5039 commands.
5040 mcompand
5042 Multiband Compress or expand the audio's dynamic range.
5043 The input audio is divided into bands using 4th order Linkwitz-Riley
5045 IIRs. This is akin to the crossover of a loudspeaker, and results in
5046 flat frequency response when absent compander action.
5047 It accepts the following parameters:
5049 args
5051 This option syntax is: attack,decay,[attack,decay..] soft-knee
5052 points crossover_frequency [delay [initial_volume [gain]]] |
5053 attack,decay ... For explanation of each item refer to compand
5054 filter documentation.
5055 pan
5057 Mix channels with specific gain levels. The filter accepts the output
5058 channel layout followed by a set of channels definitions.
5059 This filter is also designed to efficiently remap the channels of an
5061 audio stream.
5062 The filter accepts parameters of the form: "l|outdef|outdef|..."
5064 l output channel layout or number of channels
5066 outdef
5068 output channel specification, of the form:
5069 "out_name=[gain*]in_name[(+-)[gain*]in_name...]"
5070 out_name
5072 output channel to define, either a channel name (FL, FR, etc.) or a
5073 channel number (c0, c1, etc.)
5074 gain
5076 multiplicative coefficient for the channel, 1 leaving the volume
5077 unchanged
5078 in_name
5080 input channel to use, see out_name for details; it is not possible
5081 to mix named and numbered input channels
5082 If the `=' in a channel specification is replaced by `<', then the
5084 gains for that specification will be renormalized so that the total is
5085 1, thus avoiding clipping noise.
5086 Mixing examples
5088 For example, if you want to down-mix from stereo to mono, but with a
5090 bigger factor for the left channel:
5091 pan=1c|c0=0.9*c0+0.1*c1
5093 A customized down-mix to stereo that works automatically for 3-, 4-, 5-
5095 and 7-channels surround:
5096 pan=stereo| FL < FL + 0.5*FC + 0.6*BL + 0.6*SL | FR < FR + 0.5*FC + 0.6*BR + 0.6*SR
5098 Note that ffmpeg integrates a default down-mix (and up-mix) system that
5100 should be preferred (see "-ac" option) unless you have very specific
5101 needs.
5102 Remapping examples
5104 The channel remapping will be effective if, and only if:
5106 *<gain coefficients are zeroes or ones,>
5108 *<only one input per channel output,>
5109 If all these conditions are satisfied, the filter will notify the user
5111 ("Pure channel mapping detected"), and use an optimized and lossless
5112 method to do the remapping.
5113 For example, if you have a 5.1 source and want a stereo audio stream by
5115 dropping the extra channels:
5116 pan="stereo| c0=FL | c1=FR"
5118 Given the same source, you can also switch front left and front right
5120 channels and keep the input channel layout:
5121 pan="5.1| c0=c1 | c1=c0 | c2=c2 | c3=c3 | c4=c4 | c5=c5"
5123 If the input is a stereo audio stream, you can mute the front left
5125 channel (and still keep the stereo channel layout) with:
5126 pan="stereo|c1=c1"
5128 Still with a stereo audio stream input, you can copy the right channel
5130 in both front left and right:
5131 pan="stereo| c0=FR | c1=FR"
5133 replaygain
5135 ReplayGain scanner filter. This filter takes an audio stream as an
5136 input and outputs it unchanged. At end of filtering it displays
5137 "track_gain" and "track_peak".
5138 resample
5140 Convert the audio sample format, sample rate and channel layout. It is
5141 not meant to be used directly.
5142 rubberband
5144 Apply time-stretching and pitch-shifting with librubberband.
5145 To enable compilation of this filter, you need to configure FFmpeg with
5147 "--enable-librubberband".
5148 The filter accepts the following options:
5150 tempo
5152 Set tempo scale factor.
5153 pitch
5155 Set pitch scale factor.
5156 transients
5158 Set transients detector. Possible values are:
5159 crisp
5161 mixed
5162 smooth
5163 detector
5164 Set detector. Possible values are:
5165 compound
5167 percussive
5168 soft
5169 phase
5170 Set phase. Possible values are:
5171 laminar
5173 independent
5174 window
5175 Set processing window size. Possible values are:
5176 standard
5178 short
5179 long
5180 smoothing
5181 Set smoothing. Possible values are:
5182 off
5184 on
5185 formant
5186 Enable formant preservation when shift pitching. Possible values
5187 are:
5188 shifted
5190 preserved
5191 pitchq
5192 Set pitch quality. Possible values are:
5193 quality
5195 speed
5196 consistency
5197 channels
5198 Set channels. Possible values are:
5199 apart
5201 together
5202 Commands
5204 This filter supports the following commands:
5206 tempo
5208 Change filter tempo scale factor. Syntax for the command is :
5209 "tempo"
5210 pitch
5212 Change filter pitch scale factor. Syntax for the command is :
5213 "pitch"
5214 sidechaincompress
5216 This filter acts like normal compressor but has the ability to compress
5217 detected signal using second input signal. It needs two input streams
5218 and returns one output stream. First input stream will be processed
5219 depending on second stream signal. The filtered signal then can be
5220 filtered with other filters in later stages of processing. See pan and
5221 amerge filter.
5222 The filter accepts the following options:
5224 level_in
5226 Set input gain. Default is 1. Range is between 0.015625 and 64.
5227 mode
5229 Set mode of compressor operation. Can be "upward" or "downward".
5230 Default is "downward".
5231 threshold
5233 If a signal of second stream raises above this level it will affect
5234 the gain reduction of first stream. By default is 0.125. Range is
5235 between 0.00097563 and 1.
5236 ratio
5238 Set a ratio about which the signal is reduced. 1:2 means that if
5239 the level raised 4dB above the threshold, it will be only 2dB above
5240 after the reduction. Default is 2. Range is between 1 and 20.
5241 attack
5243 Amount of milliseconds the signal has to rise above the threshold
5244 before gain reduction starts. Default is 20. Range is between 0.01
5245 and 2000.
5246 release
5248 Amount of milliseconds the signal has to fall below the threshold
5249 before reduction is decreased again. Default is 250. Range is
5250 between 0.01 and 9000.
5251 makeup
5253 Set the amount by how much signal will be amplified after
5254 processing. Default is 1. Range is from 1 to 64.
5255 knee
5257 Curve the sharp knee around the threshold to enter gain reduction
5258 more softly. Default is 2.82843. Range is between 1 and 8.
5259 link
5261 Choose if the "average" level between all channels of side-chain
5262 stream or the louder("maximum") channel of side-chain stream
5263 affects the reduction. Default is "average".
5264 detection
5266 Should the exact signal be taken in case of "peak" or an RMS one in
5267 case of "rms". Default is "rms" which is mainly smoother.
5268 level_sc
5270 Set sidechain gain. Default is 1. Range is between 0.015625 and 64.
5271 mix How much to use compressed signal in output. Default is 1. Range
5273 is between 0 and 1.
5274 Commands
5276 This filter supports the all above options as commands.
5278 Examples
5280 • Full ffmpeg example taking 2 audio inputs, 1st input to be
5282 compressed depending on the signal of 2nd input and later
5283 compressed signal to be merged with 2nd input:
5284 ffmpeg -i main.flac -i sidechain.flac -filter_complex "[1:a]asplit=2[sc][mix];[0:a][sc]sidechaincompress[compr];[compr][mix]amerge"
5286 sidechaingate
5288 A sidechain gate acts like a normal (wideband) gate but has the ability
5289 to filter the detected signal before sending it to the gain reduction
5290 stage. Normally a gate uses the full range signal to detect a level
5291 above the threshold. For example: If you cut all lower frequencies
5292 from your sidechain signal the gate will decrease the volume of your
5293 track only if not enough highs appear. With this technique you are able
5294 to reduce the resonation of a natural drum or remove "rumbling" of
5295 muted strokes from a heavily distorted guitar. It needs two input
5296 streams and returns one output stream. First input stream will be
5297 processed depending on second stream signal.
5298 The filter accepts the following options:
5300 level_in
5302 Set input level before filtering. Default is 1. Allowed range is
5303 from 0.015625 to 64.
5304 mode
5306 Set the mode of operation. Can be "upward" or "downward". Default
5307 is "downward". If set to "upward" mode, higher parts of signal will
5308 be amplified, expanding dynamic range in upward direction.
5309 Otherwise, in case of "downward" lower parts of signal will be
5310 reduced.
5311 range
5313 Set the level of gain reduction when the signal is below the
5314 threshold. Default is 0.06125. Allowed range is from 0 to 1.
5315 Setting this to 0 disables reduction and then filter behaves like
5316 expander.
5317 threshold
5319 If a signal rises above this level the gain reduction is released.
5320 Default is 0.125. Allowed range is from 0 to 1.
5321 ratio
5323 Set a ratio about which the signal is reduced. Default is 2.
5324 Allowed range is from 1 to 9000.
5325 attack
5327 Amount of milliseconds the signal has to rise above the threshold
5328 before gain reduction stops. Default is 20 milliseconds. Allowed
5329 range is from 0.01 to 9000.
5330 release
5332 Amount of milliseconds the signal has to fall below the threshold
5333 before the reduction is increased again. Default is 250
5334 milliseconds. Allowed range is from 0.01 to 9000.
5335 makeup
5337 Set amount of amplification of signal after processing. Default is
5338 1. Allowed range is from 1 to 64.
5339 knee
5341 Curve the sharp knee around the threshold to enter gain reduction
5342 more softly. Default is 2.828427125. Allowed range is from 1 to 8.
5343 detection
5345 Choose if exact signal should be taken for detection or an RMS like
5346 one. Default is rms. Can be peak or rms.
5347 link
5349 Choose if the average level between all channels or the louder
5350 channel affects the reduction. Default is average. Can be average
5351 or maximum.
5352 level_sc
5354 Set sidechain gain. Default is 1. Range is from 0.015625 to 64.
5355 Commands
5357 This filter supports the all above options as commands.
5359 silencedetect
5361 Detect silence in an audio stream.
5362 This filter logs a message when it detects that the input audio volume
5364 is less or equal to a noise tolerance value for a duration greater or
5365 equal to the minimum detected noise duration.
5366 The printed times and duration are expressed in seconds. The
5368 "lavfi.silence_start" or "lavfi.silence_start.X" metadata key is set on
5369 the first frame whose timestamp equals or exceeds the detection
5370 duration and it contains the timestamp of the first frame of the
5371 silence.
5372 The "lavfi.silence_duration" or "lavfi.silence_duration.X" and
5374 "lavfi.silence_end" or "lavfi.silence_end.X" metadata keys are set on
5375 the first frame after the silence. If mono is enabled, and each channel
5376 is evaluated separately, the ".X" suffixed keys are used, and "X"
5377 corresponds to the channel number.
5378 The filter accepts the following options:
5380 noise, n
5382 Set noise tolerance. Can be specified in dB (in case "dB" is
5383 appended to the specified value) or amplitude ratio. Default is
5384 -60dB, or 0.001.
5385 duration, d
5387 Set silence duration until notification (default is 2 seconds). See
5388 the Time duration section in the ffmpeg-utils(1) manual for the
5389 accepted syntax.
5390 mono, m
5392 Process each channel separately, instead of combined. By default is
5393 disabled.
5394 Examples
5396 • Detect 5 seconds of silence with -50dB noise tolerance:
5398 silencedetect=n=-50dB:d=5
5400 • Complete example with ffmpeg to detect silence with 0.0001 noise
5402 tolerance in silence.mp3:
5403 ffmpeg -i silence.mp3 -af silencedetect=noise=0.0001 -f null -
5405 silenceremove
5407 Remove silence from the beginning, middle or end of the audio.
5408 The filter accepts the following options:
5410 start_periods
5412 This value is used to indicate if audio should be trimmed at
5413 beginning of the audio. A value of zero indicates no silence should
5414 be trimmed from the beginning. When specifying a non-zero value, it
5415 trims audio up until it finds non-silence. Normally, when trimming
5416 silence from beginning of audio the start_periods will be 1 but it
5417 can be increased to higher values to trim all audio up to specific
5418 count of non-silence periods. Default value is 0.
5419 start_duration
5421 Specify the amount of time that non-silence must be detected before
5422 it stops trimming audio. By increasing the duration, bursts of
5423 noises can be treated as silence and trimmed off. Default value is
5424 0.
5425 start_threshold
5427 This indicates what sample value should be treated as silence. For
5428 digital audio, a value of 0 may be fine but for audio recorded from
5429 analog, you may wish to increase the value to account for
5430 background noise. Can be specified in dB (in case "dB" is appended
5431 to the specified value) or amplitude ratio. Default value is 0.
5432 start_silence
5434 Specify max duration of silence at beginning that will be kept
5435 after trimming. Default is 0, which is equal to trimming all
5436 samples detected as silence.
5437 start_mode
5439 Specify mode of detection of silence end in start of multi-channel
5440 audio. Can be any or all. Default is any. With any, any sample
5441 that is detected as non-silence will cause stopped trimming of
5442 silence. With all, only if all channels are detected as non-
5443 silence will cause stopped trimming of silence.
5444 stop_periods
5446 Set the count for trimming silence from the end of audio. To
5447 remove silence from the middle of a file, specify a stop_periods
5448 that is negative. This value is then treated as a positive value
5449 and is used to indicate the effect should restart processing as
5450 specified by start_periods, making it suitable for removing periods
5451 of silence in the middle of the audio. Default value is 0.
5452 stop_duration
5454 Specify a duration of silence that must exist before audio is not
5455 copied any more. By specifying a higher duration, silence that is
5456 wanted can be left in the audio. Default value is 0.
5457 stop_threshold
5459 This is the same as start_threshold but for trimming silence from
5460 the end of audio. Can be specified in dB (in case "dB" is appended
5461 to the specified value) or amplitude ratio. Default value is 0.
5462 stop_silence
5464 Specify max duration of silence at end that will be kept after
5465 trimming. Default is 0, which is equal to trimming all samples
5466 detected as silence.
5467 stop_mode
5469 Specify mode of detection of silence start in end of multi-channel
5470 audio. Can be any or all. Default is any. With any, any sample
5471 that is detected as non-silence will cause stopped trimming of
5472 silence. With all, only if all channels are detected as non-
5473 silence will cause stopped trimming of silence.
5474 detection
5476 Set how is silence detected. Can be "rms" or "peak". Second is
5477 faster and works better with digital silence which is exactly 0.
5478 Default value is "rms".
5479 window
5481 Set duration in number of seconds used to calculate size of window
5482 in number of samples for detecting silence. Default value is 0.02.
5483 Allowed range is from 0 to 10.
5484 Examples
5486 • The following example shows how this filter can be used to start a
5488 recording that does not contain the delay at the start which
5489 usually occurs between pressing the record button and the start of
5490 the performance:
5491 silenceremove=start_periods=1:start_duration=5:start_threshold=0.02
5493 • Trim all silence encountered from beginning to end where there is
5495 more than 1 second of silence in audio:
5496 silenceremove=stop_periods=-1:stop_duration=1:stop_threshold=-90dB
5498 • Trim all digital silence samples, using peak detection, from
5500 beginning to end where there is more than 0 samples of digital
5501 silence in audio and digital silence is detected in all channels at
5502 same positions in stream:
5503 silenceremove=window=0:detection=peak:stop_mode=all:start_mode=all:stop_periods=-1:stop_threshold=0
5505 sofalizer
5507 SOFAlizer uses head-related transfer functions (HRTFs) to create
5508 virtual loudspeakers around the user for binaural listening via
5509 headphones (audio formats up to 9 channels supported). The HRTFs are
5510 stored in SOFA files (see <http://www.sofacoustics.org/> for a
5511 database). SOFAlizer is developed at the Acoustics Research Institute
5512 (ARI) of the Austrian Academy of Sciences.
5513 To enable compilation of this filter you need to configure FFmpeg with
5515 "--enable-libmysofa".
5516 The filter accepts the following options:
5518 sofa
5520 Set the SOFA file used for rendering.
5521 gain
5523 Set gain applied to audio. Value is in dB. Default is 0.
5524 rotation
5526 Set rotation of virtual loudspeakers in deg. Default is 0.
5527 elevation
5529 Set elevation of virtual speakers in deg. Default is 0.
5530 radius
5532 Set distance in meters between loudspeakers and the listener with
5533 near-field HRTFs. Default is 1.
5534 type
5536 Set processing type. Can be time or freq. time is processing audio
5537 in time domain which is slow. freq is processing audio in
5538 frequency domain which is fast. Default is freq.
5539 speakers
5541 Set custom positions of virtual loudspeakers. Syntax for this
5542 option is: <CH> <AZIM> <ELEV>[|<CH> <AZIM> <ELEV>|...]. Each
5543 virtual loudspeaker is described with short channel name following
5544 with azimuth and elevation in degrees. Each virtual loudspeaker
5545 description is separated by '|'. For example to override front
5546 left and front right channel positions use: 'speakers=FL 45 15|FR
5547 345 15'. Descriptions with unrecognised channel names are ignored.
5548 lfegain
5550 Set custom gain for LFE channels. Value is in dB. Default is 0.
5551 framesize
5553 Set custom frame size in number of samples. Default is 1024.
5554 Allowed range is from 1024 to 96000. Only used if option type is
5555 set to freq.
5556 normalize
5558 Should all IRs be normalized upon importing SOFA file. By default
5559 is enabled.
5560 interpolate
5562 Should nearest IRs be interpolated with neighbor IRs if exact
5563 position does not match. By default is disabled.
5564 minphase
5566 Minphase all IRs upon loading of SOFA file. By default is disabled.
5567 anglestep
5569 Set neighbor search angle step. Only used if option interpolate is
5570 enabled.
5571 radstep
5573 Set neighbor search radius step. Only used if option interpolate is
5574 enabled.
5575 Examples
5577 • Using ClubFritz6 sofa file:
5579 sofalizer=sofa=/path/to/ClubFritz6.sofa:type=freq:radius=1
5581 • Using ClubFritz12 sofa file and bigger radius with small rotation:
5583 sofalizer=sofa=/path/to/ClubFritz12.sofa:type=freq:radius=2:rotation=5
5585 • Similar as above but with custom speaker positions for front left,
5587 front right, back left and back right and also with custom gain:
5588 "sofalizer=sofa=/path/to/ClubFritz6.sofa:type=freq:radius=2:speakers=FL 45|FR 315|BL 135|BR 225:gain=28"
5590 speechnorm
5592 Speech Normalizer.
5593 This filter expands or compresses each half-cycle of audio samples
5595 (local set of samples all above or all below zero and between two
5596 nearest zero crossings) depending on threshold value, so audio reaches
5597 target peak value under conditions controlled by below options.
5598 The filter accepts the following options:
5600 peak, p
5602 Set the expansion target peak value. This specifies the highest
5603 allowed absolute amplitude level for the normalized audio input.
5604 Default value is 0.95. Allowed range is from 0.0 to 1.0.
5605 expansion, e
5607 Set the maximum expansion factor. Allowed range is from 1.0 to
5608 50.0. Default value is 2.0. This option controls maximum local
5609 half-cycle of samples expansion. The maximum expansion would be
5610 such that local peak value reaches target peak value but never to
5611 surpass it and that ratio between new and previous peak value does
5612 not surpass this option value.
5613 compression, c
5615 Set the maximum compression factor. Allowed range is from 1.0 to
5616 50.0. Default value is 2.0. This option controls maximum local
5617 half-cycle of samples compression. This option is used only if
5618 threshold option is set to value greater than 0.0, then in such
5619 cases when local peak is lower or same as value set by threshold
5620 all samples belonging to that peak's half-cycle will be compressed
5621 by current compression factor.
5622 threshold, t
5624 Set the threshold value. Default value is 0.0. Allowed range is
5625 from 0.0 to 1.0. This option specifies which half-cycles of
5626 samples will be compressed and which will be expanded. Any half-
5627 cycle samples with their local peak value below or same as this
5628 option value will be compressed by current compression factor,
5629 otherwise, if greater than threshold value they will be expanded
5630 with expansion factor so that it could reach peak target value but
5631 never surpass it.
5632 raise, r
5634 Set the expansion raising amount per each half-cycle of samples.
5635 Default value is 0.001. Allowed range is from 0.0 to 1.0. This
5636 controls how fast expansion factor is raised per each new half-
5637 cycle until it reaches expansion value. Setting this options too
5638 high may lead to distortions.
5639 fall, f
5641 Set the compression raising amount per each half-cycle of samples.
5642 Default value is 0.001. Allowed range is from 0.0 to 1.0. This
5643 controls how fast compression factor is raised per each new half-
5644 cycle until it reaches compression value.
5645 channels, h
5647 Specify which channels to filter, by default all available channels
5648 are filtered.
5649 invert, i
5651 Enable inverted filtering, by default is disabled. This inverts
5652 interpretation of threshold option. When enabled any half-cycle of
5653 samples with their local peak value below or same as threshold
5654 option will be expanded otherwise it will be compressed.
5655 link, l
5657 Link channels when calculating gain applied to each filtered
5658 channel sample, by default is disabled. When disabled each
5659 filtered channel gain calculation is independent, otherwise when
5660 this option is enabled the minimum of all possible gains for each
5661 filtered channel is used.
5662 Commands
5664 This filter supports the all above options as commands.
5666 stereotools
5668 This filter has some handy utilities to manage stereo signals, for
5669 converting M/S stereo recordings to L/R signal while having control
5670 over the parameters or spreading the stereo image of master track.
5671 The filter accepts the following options:
5673 level_in
5675 Set input level before filtering for both channels. Defaults is 1.
5676 Allowed range is from 0.015625 to 64.
5677 level_out
5679 Set output level after filtering for both channels. Defaults is 1.
5680 Allowed range is from 0.015625 to 64.
5681 balance_in
5683 Set input balance between both channels. Default is 0. Allowed
5684 range is from -1 to 1.
5685 balance_out
5687 Set output balance between both channels. Default is 0. Allowed
5688 range is from -1 to 1.
5689 softclip
5691 Enable softclipping. Results in analog distortion instead of harsh
5692 digital 0dB clipping. Disabled by default.
5693 mutel
5695 Mute the left channel. Disabled by default.
5696 muter
5698 Mute the right channel. Disabled by default.
5699 phasel
5701 Change the phase of the left channel. Disabled by default.
5702 phaser
5704 Change the phase of the right channel. Disabled by default.
5705 mode
5707 Set stereo mode. Available values are:
5708 lr>lr
5710 Left/Right to Left/Right, this is default.
5711 lr>ms
5713 Left/Right to Mid/Side.
5714 ms>lr
5716 Mid/Side to Left/Right.
5717 lr>ll
5719 Left/Right to Left/Left.
5720 lr>rr
5722 Left/Right to Right/Right.
5723 lr>l+r
5725 Left/Right to Left + Right.
5726 lr>rl
5728 Left/Right to Right/Left.
5729 ms>ll
5731 Mid/Side to Left/Left.
5732 ms>rr
5734 Mid/Side to Right/Right.
5735 ms>rl
5737 Mid/Side to Right/Left.
5738 lr>l-r
5740 Left/Right to Left - Right.
5741 slev
5743 Set level of side signal. Default is 1. Allowed range is from
5744 0.015625 to 64.
5745 sbal
5747 Set balance of side signal. Default is 0. Allowed range is from -1
5748 to 1.
5749 mlev
5751 Set level of the middle signal. Default is 1. Allowed range is
5752 from 0.015625 to 64.
5753 mpan
5755 Set middle signal pan. Default is 0. Allowed range is from -1 to 1.
5756 base
5758 Set stereo base between mono and inversed channels. Default is 0.
5759 Allowed range is from -1 to 1.
5760 delay
5762 Set delay in milliseconds how much to delay left from right channel
5763 and vice versa. Default is 0. Allowed range is from -20 to 20.
5764 sclevel
5766 Set S/C level. Default is 1. Allowed range is from 1 to 100.
5767 phase
5769 Set the stereo phase in degrees. Default is 0. Allowed range is
5770 from 0 to 360.
5771 bmode_in, bmode_out
5773 Set balance mode for balance_in/balance_out option.
5774 Can be one of the following:
5776 balance
5778 Classic balance mode. Attenuate one channel at time. Gain is
5779 raised up to 1.
5780 amplitude
5782 Similar as classic mode above but gain is raised up to 2.
5783 power
5785 Equal power distribution, from -6dB to +6dB range.
5786 Commands
5788 This filter supports the all above options as commands.
5790 Examples
5792 • Apply karaoke like effect:
5794 stereotools=mlev=0.015625
5796 • Convert M/S signal to L/R:
5798 "stereotools=mode=ms>lr"
5800 stereowiden
5802 This filter enhance the stereo effect by suppressing signal common to
5803 both channels and by delaying the signal of left into right and vice
5804 versa, thereby widening the stereo effect.
5805 The filter accepts the following options:
5807 delay
5809 Time in milliseconds of the delay of left signal into right and
5810 vice versa. Default is 20 milliseconds.
5811 feedback
5813 Amount of gain in delayed signal into right and vice versa. Gives a
5814 delay effect of left signal in right output and vice versa which
5815 gives widening effect. Default is 0.3.
5816 crossfeed
5818 Cross feed of left into right with inverted phase. This helps in
5819 suppressing the mono. If the value is 1 it will cancel all the
5820 signal common to both channels. Default is 0.3.
5821 drymix
5823 Set level of input signal of original channel. Default is 0.8.
5824 Commands
5826 This filter supports the all above options except "delay" as commands.
5828 superequalizer
5830 Apply 18 band equalizer.
5831 The filter accepts the following options:
5833 1b Set 65Hz band gain.
5835 2b Set 92Hz band gain.
5837 3b Set 131Hz band gain.
5839 4b Set 185Hz band gain.
5841 5b Set 262Hz band gain.
5843 6b Set 370Hz band gain.
5845 7b Set 523Hz band gain.
5847 8b Set 740Hz band gain.
5849 9b Set 1047Hz band gain.
5851 10b Set 1480Hz band gain.
5853 11b Set 2093Hz band gain.
5855 12b Set 2960Hz band gain.
5857 13b Set 4186Hz band gain.
5859 14b Set 5920Hz band gain.
5861 15b Set 8372Hz band gain.
5863 16b Set 11840Hz band gain.
5865 17b Set 16744Hz band gain.
5867 18b Set 20000Hz band gain.
5869 surround
5871 Apply audio surround upmix filter.
5872 This filter allows to produce multichannel output from audio stream.
5874 The filter accepts the following options:
5876 chl_out
5878 Set output channel layout. By default, this is 5.1.
5879 See the Channel Layout section in the ffmpeg-utils(1) manual for
5881 the required syntax.
5882 chl_in
5884 Set input channel layout. By default, this is stereo.
5885 See the Channel Layout section in the ffmpeg-utils(1) manual for
5887 the required syntax.
5888 level_in
5890 Set input volume level. By default, this is 1.
5891 level_out
5893 Set output volume level. By default, this is 1.
5894 lfe Enable LFE channel output if output channel layout has it. By
5896 default, this is enabled.
5897 lfe_low
5899 Set LFE low cut off frequency. By default, this is 128 Hz.
5900 lfe_high
5902 Set LFE high cut off frequency. By default, this is 256 Hz.
5903 lfe_mode
5905 Set LFE mode, can be add or sub. Default is add. In add mode, LFE
5906 channel is created from input audio and added to output. In sub
5907 mode, LFE channel is created from input audio and added to output
5908 but also all non-LFE output channels are subtracted with output LFE
5909 channel.
5910 angle
5912 Set angle of stereo surround transform, Allowed range is from 0 to
5913 360. Default is 90.
5914 fc_in
5916 Set front center input volume. By default, this is 1.
5917 fc_out
5919 Set front center output volume. By default, this is 1.
5920 fl_in
5922 Set front left input volume. By default, this is 1.
5923 fl_out
5925 Set front left output volume. By default, this is 1.
5926 fr_in
5928 Set front right input volume. By default, this is 1.
5929 fr_out
5931 Set front right output volume. By default, this is 1.
5932 sl_in
5934 Set side left input volume. By default, this is 1.
5935 sl_out
5937 Set side left output volume. By default, this is 1.
5938 sr_in
5940 Set side right input volume. By default, this is 1.
5941 sr_out
5943 Set side right output volume. By default, this is 1.
5944 bl_in
5946 Set back left input volume. By default, this is 1.
5947 bl_out
5949 Set back left output volume. By default, this is 1.
5950 br_in
5952 Set back right input volume. By default, this is 1.
5953 br_out
5955 Set back right output volume. By default, this is 1.
5956 bc_in
5958 Set back center input volume. By default, this is 1.
5959 bc_out
5961 Set back center output volume. By default, this is 1.
5962 lfe_in
5964 Set LFE input volume. By default, this is 1.
5965 lfe_out
5967 Set LFE output volume. By default, this is 1.
5968 allx
5970 Set spread usage of stereo image across X axis for all channels.
5971 Allowed range is from -1 to 15. By default this value is negative
5972 -1, and thus unused.
5973 ally
5975 Set spread usage of stereo image across Y axis for all channels.
5976 Allowed range is from -1 to 15. By default this value is negative
5977 -1, and thus unused.
5978 fcx, flx, frx, blx, brx, slx, srx, bcx
5980 Set spread usage of stereo image across X axis for each channel.
5981 Allowed range is from 0.06 to 15. By default this value is 0.5.
5982 fcy, fly, fry, bly, bry, sly, sry, bcy
5984 Set spread usage of stereo image across Y axis for each channel.
5985 Allowed range is from 0.06 to 15. By default this value is 0.5.
5986 win_size
5988 Set window size. Allowed range is from 1024 to 65536. Default size
5989 is 4096.
5990 win_func
5992 Set window function.
5993 It accepts the following values:
5995 rect
5997 bartlett
5998 hann, hanning
5999 hamming
6000 blackman
6001 welch
6002 flattop
6003 bharris
6004 bnuttall
6005 bhann
6006 sine
6007 nuttall
6008 lanczos
6009 gauss
6010 tukey
6011 dolph
6012 cauchy
6013 parzen
6014 poisson
6015 bohman
6016 Default is "hann".
6018 overlap
6020 Set window overlap. If set to 1, the recommended overlap for
6021 selected window function will be picked. Default is 0.5.
6022 tiltshelf
6024 Boost or cut the lower frequencies and cut or boost higher frequencies
6025 of the audio using a two-pole shelving filter with a response similar
6026 to that of a standard hi-fi's tone-controls. This is also known as
6027 shelving equalisation (EQ).
6028 The filter accepts the following options:
6030 gain, g
6032 Give the gain at 0 Hz. Its useful range is about -20 (for a large
6033 cut) to +20 (for a large boost). Beware of clipping when using a
6034 positive gain.
6035 frequency, f
6037 Set the filter's central frequency and so can be used to extend or
6038 reduce the frequency range to be boosted or cut. The default value
6039 is 3000 Hz.
6040 width_type, t
6042 Set method to specify band-width of filter.
6043 h Hz
6045 q Q-Factor
6047 o octave
6049 s slope
6051 k kHz
6053 width, w
6055 Determine how steep is the filter's shelf transition.
6056 poles, p
6058 Set number of poles. Default is 2.
6059 mix, m
6061 How much to use filtered signal in output. Default is 1. Range is
6062 between 0 and 1.
6063 channels, c
6065 Specify which channels to filter, by default all available are
6066 filtered.
6067 normalize, n
6069 Normalize biquad coefficients, by default is disabled. Enabling it
6070 will normalize magnitude response at DC to 0dB.
6071 transform, a
6073 Set transform type of IIR filter.
6074 di
6076 dii
6077 tdi
6078 tdii
6079 latt
6080 svf
6081 zdf
6082 precision, r
6083 Set precison of filtering.
6084 auto
6086 Pick automatic sample format depending on surround filters.
6087 s16 Always use signed 16-bit.
6089 s32 Always use signed 32-bit.
6091 f32 Always use float 32-bit.
6093 f64 Always use float 64-bit.
6095 block_size, b
6097 Set block size used for reverse IIR processing. If this value is
6098 set to high enough value (higher than impulse response length
6099 truncated when reaches near zero values) filtering will become
6100 linear phase otherwise if not big enough it will just produce nasty
6101 artifacts.
6102 Note that filter delay will be exactly this many samples when set
6104 to non-zero value.
6105 Commands
6107 This filter supports some options as commands.
6109 treble, highshelf
6111 Boost or cut treble (upper) frequencies of the audio using a two-pole
6112 shelving filter with a response similar to that of a standard hi-fi's
6113 tone-controls. This is also known as shelving equalisation (EQ).
6114 The filter accepts the following options:
6116 gain, g
6118 Give the gain at whichever is the lower of ~22 kHz and the Nyquist
6119 frequency. Its useful range is about -20 (for a large cut) to +20
6120 (for a large boost). Beware of clipping when using a positive gain.
6121 frequency, f
6123 Set the filter's central frequency and so can be used to extend or
6124 reduce the frequency range to be boosted or cut. The default value
6125 is 3000 Hz.
6126 width_type, t
6128 Set method to specify band-width of filter.
6129 h Hz
6131 q Q-Factor
6133 o octave
6135 s slope
6137 k kHz
6139 width, w
6141 Determine how steep is the filter's shelf transition.
6142 poles, p
6144 Set number of poles. Default is 2.
6145 mix, m
6147 How much to use filtered signal in output. Default is 1. Range is
6148 between 0 and 1.
6149 channels, c
6151 Specify which channels to filter, by default all available are
6152 filtered.
6153 normalize, n
6155 Normalize biquad coefficients, by default is disabled. Enabling it
6156 will normalize magnitude response at DC to 0dB.
6157 transform, a
6159 Set transform type of IIR filter.
6160 di
6162 dii
6163 tdi
6164 tdii
6165 latt
6166 svf
6167 zdf
6168 precision, r
6169 Set precison of filtering.
6170 auto
6172 Pick automatic sample format depending on surround filters.
6173 s16 Always use signed 16-bit.
6175 s32 Always use signed 32-bit.
6177 f32 Always use float 32-bit.
6179 f64 Always use float 64-bit.
6181 block_size, b
6183 Set block size used for reverse IIR processing. If this value is
6184 set to high enough value (higher than impulse response length
6185 truncated when reaches near zero values) filtering will become
6186 linear phase otherwise if not big enough it will just produce nasty
6187 artifacts.
6188 Note that filter delay will be exactly this many samples when set
6190 to non-zero value.
6191 Commands
6193 This filter supports the following commands:
6195 frequency, f
6197 Change treble frequency. Syntax for the command is : "frequency"
6198 width_type, t
6200 Change treble width_type. Syntax for the command is : "width_type"
6201 width, w
6203 Change treble width. Syntax for the command is : "width"
6204 gain, g
6206 Change treble gain. Syntax for the command is : "gain"
6207 mix, m
6209 Change treble mix. Syntax for the command is : "mix"
6210 tremolo
6212 Sinusoidal amplitude modulation.
6213 The filter accepts the following options:
6215 f Modulation frequency in Hertz. Modulation frequencies in the
6217 subharmonic range (20 Hz or lower) will result in a tremolo effect.
6218 This filter may also be used as a ring modulator by specifying a
6219 modulation frequency higher than 20 Hz. Range is 0.1 - 20000.0.
6220 Default value is 5.0 Hz.
6221 d Depth of modulation as a percentage. Range is 0.0 - 1.0. Default
6223 value is 0.5.
6224 vibrato
6226 Sinusoidal phase modulation.
6227 The filter accepts the following options:
6229 f Modulation frequency in Hertz. Range is 0.1 - 20000.0. Default
6231 value is 5.0 Hz.
6232 d Depth of modulation as a percentage. Range is 0.0 - 1.0. Default
6234 value is 0.5.
6235 virtualbass
6237 Apply audio Virtual Bass filter.
6238 This filter accepts stereo input and produce stereo with LFE (2.1)
6240 channels output. The newly produced LFE channel have enhanced virtual
6241 bass originally obtained from both stereo channels. This filter
6242 outputs front left and front right channels unchanged as available in
6243 stereo input.
6244 The filter accepts the following options:
6246 cutoff
6248 Set the virtual bass cutoff frequency. Default value is 250 Hz.
6249 Allowed range is from 100 to 500 Hz.
6250 strength
6252 Set the virtual bass strength. Allowed range is from 0.5 to 3.
6253 Default value is 3.
6254 volume
6256 Adjust the input audio volume.
6257 It accepts the following parameters:
6259 volume
6261 Set audio volume expression.
6262 Output values are clipped to the maximum value.
6264 The output audio volume is given by the relation:
6266 <output_volume> = <volume> * <input_volume>
6268 The default value for volume is "1.0".
6270 precision
6272 This parameter represents the mathematical precision.
6273 It determines which input sample formats will be allowed, which
6275 affects the precision of the volume scaling.
6276 fixed
6278 8-bit fixed-point; this limits input sample format to U8, S16,
6279 and S32.
6280 float
6282 32-bit floating-point; this limits input sample format to FLT.
6283 (default)
6284 double
6286 64-bit floating-point; this limits input sample format to DBL.
6287 replaygain
6289 Choose the behaviour on encountering ReplayGain side data in input
6290 frames.
6291 drop
6293 Remove ReplayGain side data, ignoring its contents (the
6294 default).
6295 ignore
6297 Ignore ReplayGain side data, but leave it in the frame.
6298 track
6300 Prefer the track gain, if present.
6301 album
6303 Prefer the album gain, if present.
6304 replaygain_preamp
6306 Pre-amplification gain in dB to apply to the selected replaygain
6307 gain.
6308 Default value for replaygain_preamp is 0.0.
6310 replaygain_noclip
6312 Prevent clipping by limiting the gain applied.
6313 Default value for replaygain_noclip is 1.
6315 eval
6317 Set when the volume expression is evaluated.
6318 It accepts the following values:
6320 once
6322 only evaluate expression once during the filter initialization,
6323 or when the volume command is sent
6324 frame
6326 evaluate expression for each incoming frame
6327 Default value is once.
6329 The volume expression can contain the following parameters.
6331 n frame number (starting at zero)
6333 nb_channels
6335 number of channels
6336 nb_consumed_samples
6338 number of samples consumed by the filter
6339 nb_samples
6341 number of samples in the current frame
6342 pos original frame position in the file
6344 pts frame PTS
6346 sample_rate
6348 sample rate
6349 startpts
6351 PTS at start of stream
6352 startt
6354 time at start of stream
6355 t frame time
6357 tb timestamp timebase
6359 volume
6361 last set volume value
6362 Note that when eval is set to once only the sample_rate and tb
6364 variables are available, all other variables will evaluate to NAN.
6365 Commands
6367 This filter supports the following commands:
6369 volume
6371 Modify the volume expression. The command accepts the same syntax
6372 of the corresponding option.
6373 If the specified expression is not valid, it is kept at its current
6375 value.
6376 Examples
6378 • Halve the input audio volume:
6380 volume=volume=0.5
6382 volume=volume=1/2
6383 volume=volume=-6.0206dB
6384 In all the above example the named key for volume can be omitted,
6386 for example like in:
6387 volume=0.5
6389 • Increase input audio power by 6 decibels using fixed-point
6391 precision:
6392 volume=volume=6dB:precision=fixed
6394 • Fade volume after time 10 with an annihilation period of 5 seconds:
6396 volume='if(lt(t,10),1,max(1-(t-10)/5,0))':eval=frame
6398 volumedetect
6400 Detect the volume of the input video.
6401 The filter has no parameters. It supports only 16-bit signed integer
6403 samples, so the input will be converted when needed. Statistics about
6404 the volume will be printed in the log when the input stream end is
6405 reached.
6406 In particular it will show the mean volume (root mean square), maximum
6408 volume (on a per-sample basis), and the beginning of a histogram of the
6409 registered volume values (from the maximum value to a cumulated 1/1000
6410 of the samples).
6411 All volumes are in decibels relative to the maximum PCM value.
6413 Examples
6415 Here is an excerpt of the output:
6417 [Parsed_volumedetect_0 0xa23120] mean_volume: -27 dB
6419 [Parsed_volumedetect_0 0xa23120] max_volume: -4 dB
6420 [Parsed_volumedetect_0 0xa23120] histogram_4db: 6
6421 [Parsed_volumedetect_0 0xa23120] histogram_5db: 62
6422 [Parsed_volumedetect_0 0xa23120] histogram_6db: 286
6423 [Parsed_volumedetect_0 0xa23120] histogram_7db: 1042
6424 [Parsed_volumedetect_0 0xa23120] histogram_8db: 2551
6425 [Parsed_volumedetect_0 0xa23120] histogram_9db: 4609
6426 [Parsed_volumedetect_0 0xa23120] histogram_10db: 8409
6427 It means that:
6429 • The mean square energy is approximately -27 dB, or 10^-2.7.
6431 • The largest sample is at -4 dB, or more precisely between -4 dB and
6433 -5 dB.
6434 • There are 6 samples at -4 dB, 62 at -5 dB, 286 at -6 dB, etc.
6436 In other words, raising the volume by +4 dB does not cause any
6438 clipping, raising it by +5 dB causes clipping for 6 samples, etc.
6439
6441 Below is a description of the currently available audio sources.
6442 abuffer
6444 Buffer audio frames, and make them available to the filter chain.
6445 This source is mainly intended for a programmatic use, in particular
6447 through the interface defined in libavfilter/buffersrc.h.
6448 It accepts the following parameters:
6450 time_base
6452 The timebase which will be used for timestamps of submitted frames.
6453 It must be either a floating-point number or in
6454 numerator/denominator form.
6455 sample_rate
6457 The sample rate of the incoming audio buffers.
6458 sample_fmt
6460 The sample format of the incoming audio buffers. Either a sample
6461 format name or its corresponding integer representation from the
6462 enum AVSampleFormat in libavutil/samplefmt.h
6463 channel_layout
6465 The channel layout of the incoming audio buffers. Either a channel
6466 layout name from channel_layout_map in libavutil/channel_layout.c
6467 or its corresponding integer representation from the AV_CH_LAYOUT_*
6468 macros in libavutil/channel_layout.h
6469 channels
6471 The number of channels of the incoming audio buffers. If both
6472 channels and channel_layout are specified, then they must be
6473 consistent.
6474 Examples
6476 abuffer=sample_rate=44100:sample_fmt=s16p:channel_layout=stereo
6478 will instruct the source to accept planar 16bit signed stereo at
6480 44100Hz. Since the sample format with name "s16p" corresponds to the
6481 number 6 and the "stereo" channel layout corresponds to the value 0x3,
6482 this is equivalent to:
6483 abuffer=sample_rate=44100:sample_fmt=6:channel_layout=0x3
6485 aevalsrc
6487 Generate an audio signal specified by an expression.
6488 This source accepts in input one or more expressions (one for each
6490 channel), which are evaluated and used to generate a corresponding
6491 audio signal.
6492 This source accepts the following options:
6494 exprs
6496 Set the '|'-separated expressions list for each separate channel.
6497 In case the channel_layout option is not specified, the selected
6498 channel layout depends on the number of provided expressions.
6499 Otherwise the last specified expression is applied to the remaining
6500 output channels.
6501 channel_layout, c
6503 Set the channel layout. The number of channels in the specified
6504 layout must be equal to the number of specified expressions.
6505 duration, d
6507 Set the minimum duration of the sourced audio. See the Time
6508 duration section in the ffmpeg-utils(1) manual for the accepted
6509 syntax. Note that the resulting duration may be greater than the
6510 specified duration, as the generated audio is always cut at the end
6511 of a complete frame.
6512 If not specified, or the expressed duration is negative, the audio
6514 is supposed to be generated forever.
6515 nb_samples, n
6517 Set the number of samples per channel per each output frame,
6518 default to 1024.
6519 sample_rate, s
6521 Specify the sample rate, default to 44100.
6522 Each expression in exprs can contain the following constants:
6524 n number of the evaluated sample, starting from 0
6526 t time of the evaluated sample expressed in seconds, starting from 0
6528 s sample rate
6530 Examples
6532 • Generate silence:
6534 aevalsrc=0
6536 • Generate a sin signal with frequency of 440 Hz, set sample rate to
6538 8000 Hz:
6539 aevalsrc="sin(440*2*PI*t):s=8000"
6541 • Generate a two channels signal, specify the channel layout (Front
6543 Center + Back Center) explicitly:
6544 aevalsrc="sin(420*2*PI*t)|cos(430*2*PI*t):c=FC|BC"
6546 • Generate white noise:
6548 aevalsrc="-2+random(0)"
6550 • Generate an amplitude modulated signal:
6552 aevalsrc="sin(10*2*PI*t)*sin(880*2*PI*t)"
6554 • Generate 2.5 Hz binaural beats on a 360 Hz carrier:
6556 aevalsrc="0.1*sin(2*PI*(360-2.5/2)*t) | 0.1*sin(2*PI*(360+2.5/2)*t)"
6558 afirsrc
6560 Generate a FIR coefficients using frequency sampling method.
6561 The resulting stream can be used with afir filter for filtering the
6563 audio signal.
6564 The filter accepts the following options:
6566 taps, t
6568 Set number of filter coefficents in output audio stream. Default
6569 value is 1025.
6570 frequency, f
6572 Set frequency points from where magnitude and phase are set. This
6573 must be in non decreasing order, and first element must be 0, while
6574 last element must be 1. Elements are separated by white spaces.
6575 magnitude, m
6577 Set magnitude value for every frequency point set by frequency.
6578 Number of values must be same as number of frequency points.
6579 Values are separated by white spaces.
6580 phase, p
6582 Set phase value for every frequency point set by frequency. Number
6583 of values must be same as number of frequency points. Values are
6584 separated by white spaces.
6585 sample_rate, r
6587 Set sample rate, default is 44100.
6588 nb_samples, n
6590 Set number of samples per each frame. Default is 1024.
6591 win_func, w
6593 Set window function. Default is blackman.
6594 anullsrc
6596 The null audio source, return unprocessed audio frames. It is mainly
6597 useful as a template and to be employed in analysis / debugging tools,
6598 or as the source for filters which ignore the input data (for example
6599 the sox synth filter).
6600 This source accepts the following options:
6602 channel_layout, cl
6604 Specifies the channel layout, and can be either an integer or a
6605 string representing a channel layout. The default value of
6606 channel_layout is "stereo".
6607 Check the channel_layout_map definition in
6609 libavutil/channel_layout.c for the mapping between strings and
6610 channel layout values.
6611 sample_rate, r
6613 Specifies the sample rate, and defaults to 44100.
6614 nb_samples, n
6616 Set the number of samples per requested frames.
6617 duration, d
6619 Set the duration of the sourced audio. See the Time duration
6620 section in the ffmpeg-utils(1) manual for the accepted syntax.
6621 If not specified, or the expressed duration is negative, the audio
6623 is supposed to be generated forever.
6624 Examples
6626 • Set the sample rate to 48000 Hz and the channel layout to
6628 AV_CH_LAYOUT_MONO.
6629 anullsrc=r=48000:cl=4
6631 • Do the same operation with a more obvious syntax:
6633 anullsrc=r=48000:cl=mono
6635 All the parameters need to be explicitly defined.
6637 flite
6639 Synthesize a voice utterance using the libflite library.
6640 To enable compilation of this filter you need to configure FFmpeg with
6642 "--enable-libflite".
6643 Note that versions of the flite library prior to 2.0 are not thread-
6645 safe.
6646 The filter accepts the following options:
6648 list_voices
6650 If set to 1, list the names of the available voices and exit
6651 immediately. Default value is 0.
6652 nb_samples, n
6654 Set the maximum number of samples per frame. Default value is 512.
6655 textfile
6657 Set the filename containing the text to speak.
6658 text
6660 Set the text to speak.
6661 voice, v
6663 Set the voice to use for the speech synthesis. Default value is
6664 "kal". See also the list_voices option.
6665 Examples
6667 • Read from file speech.txt, and synthesize the text using the
6669 standard flite voice:
6670 flite=textfile=speech.txt
6672 • Read the specified text selecting the "slt" voice:
6674 flite=text='So fare thee well, poor devil of a Sub-Sub, whose commentator I am':voice=slt
6676 • Input text to ffmpeg:
6678 ffmpeg -f lavfi -i flite=text='So fare thee well, poor devil of a Sub-Sub, whose commentator I am':voice=slt
6680 • Make ffplay speak the specified text, using "flite" and the "lavfi"
6682 device:
6683 ffplay -f lavfi flite=text='No more be grieved for which that thou hast done.'
6685 For more information about libflite, check:
6687 <http://www.festvox.org/flite/>
6688 anoisesrc
6690 Generate a noise audio signal.
6691 The filter accepts the following options:
6693 sample_rate, r
6695 Specify the sample rate. Default value is 48000 Hz.
6696 amplitude, a
6698 Specify the amplitude (0.0 - 1.0) of the generated audio stream.
6699 Default value is 1.0.
6700 duration, d
6702 Specify the duration of the generated audio stream. Not specifying
6703 this option results in noise with an infinite length.
6704 color, colour, c
6706 Specify the color of noise. Available noise colors are white, pink,
6707 brown, blue, violet and velvet. Default color is white.
6708 seed, s
6710 Specify a value used to seed the PRNG.
6711 nb_samples, n
6713 Set the number of samples per each output frame, default is 1024.
6714 Examples
6716 • Generate 60 seconds of pink noise, with a 44.1 kHz sampling rate
6718 and an amplitude of 0.5:
6719 anoisesrc=d=60:c=pink:r=44100:a=0.5
6721 hilbert
6723 Generate odd-tap Hilbert transform FIR coefficients.
6724 The resulting stream can be used with afir filter for phase-shifting
6726 the signal by 90 degrees.
6727 This is used in many matrix coding schemes and for analytic signal
6729 generation. The process is often written as a multiplication by i (or
6730 j), the imaginary unit.
6731 The filter accepts the following options:
6733 sample_rate, s
6735 Set sample rate, default is 44100.
6736 taps, t
6738 Set length of FIR filter, default is 22051.
6739 nb_samples, n
6741 Set number of samples per each frame.
6742 win_func, w
6744 Set window function to be used when generating FIR coefficients.
6745 sinc
6747 Generate a sinc kaiser-windowed low-pass, high-pass, band-pass, or
6748 band-reject FIR coefficients.
6749 The resulting stream can be used with afir filter for filtering the
6751 audio signal.
6752 The filter accepts the following options:
6754 sample_rate, r
6756 Set sample rate, default is 44100.
6757 nb_samples, n
6759 Set number of samples per each frame. Default is 1024.
6760 hp Set high-pass frequency. Default is 0.
6762 lp Set low-pass frequency. Default is 0. If high-pass frequency is
6764 lower than low-pass frequency and low-pass frequency is higher than
6765 0 then filter will create band-pass filter coefficients, otherwise
6766 band-reject filter coefficients.
6767 phase
6769 Set filter phase response. Default is 50. Allowed range is from 0
6770 to 100.
6771 beta
6773 Set Kaiser window beta.
6774 att Set stop-band attenuation. Default is 120dB, allowed range is from
6776 40 to 180 dB.
6777 round
6779 Enable rounding, by default is disabled.
6780 hptaps
6782 Set number of taps for high-pass filter.
6783 lptaps
6785 Set number of taps for low-pass filter.
6786 sine
6788 Generate an audio signal made of a sine wave with amplitude 1/8.
6789 The audio signal is bit-exact.
6791 The filter accepts the following options:
6793 frequency, f
6795 Set the carrier frequency. Default is 440 Hz.
6796 beep_factor, b
6798 Enable a periodic beep every second with frequency beep_factor
6799 times the carrier frequency. Default is 0, meaning the beep is
6800 disabled.
6801 sample_rate, r
6803 Specify the sample rate, default is 44100.
6804 duration, d
6806 Specify the duration of the generated audio stream.
6807 samples_per_frame
6809 Set the number of samples per output frame.
6810 The expression can contain the following constants:
6812 n The (sequential) number of the output audio frame, starting
6814 from 0.
6815 pts The PTS (Presentation TimeStamp) of the output audio frame,
6817 expressed in TB units.
6818 t The PTS of the output audio frame, expressed in seconds.
6820 TB The timebase of the output audio frames.
6822 Default is 1024.
6824 Examples
6826 • Generate a simple 440 Hz sine wave:
6828 sine
6830 • Generate a 220 Hz sine wave with a 880 Hz beep each second, for 5
6832 seconds:
6833 sine=220:4:d=5
6835 sine=f=220:b=4:d=5
6836 sine=frequency=220:beep_factor=4:duration=5
6837 • Generate a 1 kHz sine wave following "1602,1601,1602,1601,1602"
6839 NTSC pattern:
6840 sine=1000:samples_per_frame='st(0,mod(n,5)); 1602-not(not(eq(ld(0),1)+eq(ld(0),3)))'
6842
6844 Below is a description of the currently available audio sinks.
6845 abuffersink
6847 Buffer audio frames, and make them available to the end of filter
6848 chain.
6849 This sink is mainly intended for programmatic use, in particular
6851 through the interface defined in libavfilter/buffersink.h or the
6852 options system.
6853 It accepts a pointer to an AVABufferSinkContext structure, which
6855 defines the incoming buffers' formats, to be passed as the opaque
6856 parameter to "avfilter_init_filter" for initialization.
6857 anullsink
6859 Null audio sink; do absolutely nothing with the input audio. It is
6860 mainly useful as a template and for use in analysis / debugging tools.
6861
6863 When you configure your FFmpeg build, you can disable any of the
6864 existing filters using "--disable-filters". The configure output will
6865 show the video filters included in your build.
6866 Below is a description of the currently available video filters.
6868 addroi
6870 Mark a region of interest in a video frame.
6871 The frame data is passed through unchanged, but metadata is attached to
6873 the frame indicating regions of interest which can affect the behaviour
6874 of later encoding. Multiple regions can be marked by applying the
6875 filter multiple times.
6876 x Region distance in pixels from the left edge of the frame.
6878 y Region distance in pixels from the top edge of the frame.
6880 w Region width in pixels.
6882 h Region height in pixels.
6884 The parameters x, y, w and h are expressions, and may contain the
6886 following variables:
6887 iw Width of the input frame.
6889 ih Height of the input frame.
6891 qoffset
6893 Quantisation offset to apply within the region.
6894 This must be a real value in the range -1 to +1. A value of zero
6896 indicates no quality change. A negative value asks for better
6897 quality (less quantisation), while a positive value asks for worse
6898 quality (greater quantisation).
6899 The range is calibrated so that the extreme values indicate the
6901 largest possible offset - if the rest of the frame is encoded with
6902 the worst possible quality, an offset of -1 indicates that this
6903 region should be encoded with the best possible quality anyway.
6904 Intermediate values are then interpolated in some codec-dependent
6905 way.
6906 For example, in 10-bit H.264 the quantisation parameter varies
6908 between -12 and 51. A typical qoffset value of -1/10 therefore
6909 indicates that this region should be encoded with a QP around one-
6910 tenth of the full range better than the rest of the frame. So, if
6911 most of the frame were to be encoded with a QP of around 30, this
6912 region would get a QP of around 24 (an offset of approximately
6913 -1/10 * (51 - -12) = -6.3). An extreme value of -1 would indicate
6914 that this region should be encoded with the best possible quality
6915 regardless of the treatment of the rest of the frame - that is,
6916 should be encoded at a QP of -12.
6917 clear
6919 If set to true, remove any existing regions of interest marked on
6920 the frame before adding the new one.
6921 Examples
6923 • Mark the centre quarter of the frame as interesting.
6925 addroi=iw/4:ih/4:iw/2:ih/2:-1/10
6927 • Mark the 100-pixel-wide region on the left edge of the frame as
6929 very uninteresting (to be encoded at much lower quality than the
6930 rest of the frame).
6931 addroi=0:0:100:ih:+1/5
6933 alphaextract
6935 Extract the alpha component from the input as a grayscale video. This
6936 is especially useful with the alphamerge filter.
6937 alphamerge
6939 Add or replace the alpha component of the primary input with the
6940 grayscale value of a second input. This is intended for use with
6941 alphaextract to allow the transmission or storage of frame sequences
6942 that have alpha in a format that doesn't support an alpha channel.
6943 For example, to reconstruct full frames from a normal YUV-encoded video
6945 and a separate video created with alphaextract, you might use:
6946 movie=in_alpha.mkv [alpha]; [in][alpha] alphamerge [out]
6948 amplify
6950 Amplify differences between current pixel and pixels of adjacent frames
6951 in same pixel location.
6952 This filter accepts the following options:
6954 radius
6956 Set frame radius. Default is 2. Allowed range is from 1 to 63. For
6957 example radius of 3 will instruct filter to calculate average of 7
6958 frames.
6959 factor
6961 Set factor to amplify difference. Default is 2. Allowed range is
6962 from 0 to 65535.
6963 threshold
6965 Set threshold for difference amplification. Any difference greater
6966 or equal to this value will not alter source pixel. Default is 10.
6967 Allowed range is from 0 to 65535.
6968 tolerance
6970 Set tolerance for difference amplification. Any difference lower to
6971 this value will not alter source pixel. Default is 0. Allowed
6972 range is from 0 to 65535.
6973 low Set lower limit for changing source pixel. Default is 65535.
6975 Allowed range is from 0 to 65535. This option controls maximum
6976 possible value that will decrease source pixel value.
6977 high
6979 Set high limit for changing source pixel. Default is 65535. Allowed
6980 range is from 0 to 65535. This option controls maximum possible
6981 value that will increase source pixel value.
6982 planes
6984 Set which planes to filter. Default is all. Allowed range is from 0
6985 to 15.
6986 Commands
6988 This filter supports the following commands that corresponds to option
6990 of same name:
6991 factor
6993 threshold
6994 tolerance
6995 low
6996 high
6997 planes
6998 ass
7000 Same as the subtitles filter, except that it doesn't require libavcodec
7001 and libavformat to work. On the other hand, it is limited to ASS
7002 (Advanced Substation Alpha) subtitles files.
7003 This filter accepts the following option in addition to the common
7005 options from the subtitles filter:
7006 shaping
7008 Set the shaping engine
7009 Available values are:
7011 auto
7013 The default libass shaping engine, which is the best available.
7014 simple
7016 Fast, font-agnostic shaper that can do only substitutions
7017 complex
7019 Slower shaper using OpenType for substitutions and positioning
7020 The default is "auto".
7022 atadenoise
7024 Apply an Adaptive Temporal Averaging Denoiser to the video input.
7025 The filter accepts the following options:
7027 0a Set threshold A for 1st plane. Default is 0.02. Valid range is 0
7029 to 0.3.
7030 0b Set threshold B for 1st plane. Default is 0.04. Valid range is 0
7032 to 5.
7033 1a Set threshold A for 2nd plane. Default is 0.02. Valid range is 0
7035 to 0.3.
7036 1b Set threshold B for 2nd plane. Default is 0.04. Valid range is 0
7038 to 5.
7039 2a Set threshold A for 3rd plane. Default is 0.02. Valid range is 0
7041 to 0.3.
7042 2b Set threshold B for 3rd plane. Default is 0.04. Valid range is 0
7044 to 5.
7045 Threshold A is designed to react on abrupt changes in the input
7047 signal and threshold B is designed to react on continuous changes
7048 in the input signal.
7049 s Set number of frames filter will use for averaging. Default is 9.
7051 Must be odd number in range [5, 129].
7052 p Set what planes of frame filter will use for averaging. Default is
7054 all.
7055 a Set what variant of algorithm filter will use for averaging.
7057 Default is "p" parallel. Alternatively can be set to "s" serial.
7058 Parallel can be faster then serial, while other way around is never
7060 true. Parallel will abort early on first change being greater then
7061 thresholds, while serial will continue processing other side of
7062 frames if they are equal or below thresholds.
7063 0s
7065 1s
7066 2s Set sigma for 1st plane, 2nd plane or 3rd plane. Default is 32767.
7067 Valid range is from 0 to 32767. This options controls weight for
7068 each pixel in radius defined by size. Default value means every
7069 pixel have same weight. Setting this option to 0 effectively
7070 disables filtering.
7071 Commands
7073 This filter supports same commands as options except option "s". The
7075 command accepts the same syntax of the corresponding option.
7076 avgblur
7078 Apply average blur filter.
7079 The filter accepts the following options:
7081 sizeX
7083 Set horizontal radius size.
7084 planes
7086 Set which planes to filter. By default all planes are filtered.
7087 sizeY
7089 Set vertical radius size, if zero it will be same as "sizeX".
7090 Default is 0.
7091 Commands
7093 This filter supports same commands as options. The command accepts the
7095 same syntax of the corresponding option.
7096 If the specified expression is not valid, it is kept at its current
7098 value.
7099 bbox
7101 Compute the bounding box for the non-black pixels in the input frame
7102 luminance plane.
7103 This filter computes the bounding box containing all the pixels with a
7105 luminance value greater than the minimum allowed value. The parameters
7106 describing the bounding box are printed on the filter log.
7107 The filter accepts the following option:
7109 min_val
7111 Set the minimal luminance value. Default is 16.
7112 Commands
7114 This filter supports the all above options as commands.
7116 bilateral
7118 Apply bilateral filter, spatial smoothing while preserving edges.
7119 The filter accepts the following options:
7121 sigmaS
7123 Set sigma of gaussian function to calculate spatial weight.
7124 Allowed range is 0 to 512. Default is 0.1.
7125 sigmaR
7127 Set sigma of gaussian function to calculate range weight. Allowed
7128 range is 0 to 1. Default is 0.1.
7129 planes
7131 Set planes to filter. Default is first only.
7132 Commands
7134 This filter supports the all above options as commands.
7136 bitplanenoise
7138 Show and measure bit plane noise.
7139 The filter accepts the following options:
7141 bitplane
7143 Set which plane to analyze. Default is 1.
7144 filter
7146 Filter out noisy pixels from "bitplane" set above. Default is
7147 disabled.
7148 blackdetect
7150 Detect video intervals that are (almost) completely black. Can be
7151 useful to detect chapter transitions, commercials, or invalid
7152 recordings.
7153 The filter outputs its detection analysis to both the log as well as
7155 frame metadata. If a black segment of at least the specified minimum
7156 duration is found, a line with the start and end timestamps as well as
7157 duration is printed to the log with level "info". In addition, a log
7158 line with level "debug" is printed per frame showing the black amount
7159 detected for that frame.
7160 The filter also attaches metadata to the first frame of a black segment
7162 with key "lavfi.black_start" and to the first frame after the black
7163 segment ends with key "lavfi.black_end". The value is the frame's
7164 timestamp. This metadata is added regardless of the minimum duration
7165 specified.
7166 The filter accepts the following options:
7168 black_min_duration, d
7170 Set the minimum detected black duration expressed in seconds. It
7171 must be a non-negative floating point number.
7172 Default value is 2.0.
7174 picture_black_ratio_th, pic_th
7176 Set the threshold for considering a picture "black". Express the
7177 minimum value for the ratio:
7178 <nb_black_pixels> / <nb_pixels>
7180 for which a picture is considered black. Default value is 0.98.
7182 pixel_black_th, pix_th
7184 Set the threshold for considering a pixel "black".
7185 The threshold expresses the maximum pixel luminance value for which
7187 a pixel is considered "black". The provided value is scaled
7188 according to the following equation:
7189 <absolute_threshold> = <luminance_minimum_value> + <pixel_black_th> * <luminance_range_size>
7191 luminance_range_size and luminance_minimum_value depend on the
7193 input video format, the range is [0-255] for YUV full-range formats
7194 and [16-235] for YUV non full-range formats.
7195 Default value is 0.10.
7197 The following example sets the maximum pixel threshold to the minimum
7199 value, and detects only black intervals of 2 or more seconds:
7200 blackdetect=d=2:pix_th=0.00
7202 blackframe
7204 Detect frames that are (almost) completely black. Can be useful to
7205 detect chapter transitions or commercials. Output lines consist of the
7206 frame number of the detected frame, the percentage of blackness, the
7207 position in the file if known or -1 and the timestamp in seconds.
7208 In order to display the output lines, you need to set the loglevel at
7210 least to the AV_LOG_INFO value.
7211 This filter exports frame metadata "lavfi.blackframe.pblack". The
7213 value represents the percentage of pixels in the picture that are below
7214 the threshold value.
7215 It accepts the following parameters:
7217 amount
7219 The percentage of the pixels that have to be below the threshold;
7220 it defaults to 98.
7221 threshold, thresh
7223 The threshold below which a pixel value is considered black; it
7224 defaults to 32.
7225 blend
7227 Blend two video frames into each other.
7228 The "blend" filter takes two input streams and outputs one stream, the
7230 first input is the "top" layer and second input is "bottom" layer. By
7231 default, the output terminates when the longest input terminates.
7232 The "tblend" (time blend) filter takes two consecutive frames from one
7234 single stream, and outputs the result obtained by blending the new
7235 frame on top of the old frame.
7236 A description of the accepted options follows.
7238 c0_mode
7240 c1_mode
7241 c2_mode
7242 c3_mode
7243 all_mode
7244 Set blend mode for specific pixel component or all pixel components
7245 in case of all_mode. Default value is "normal".
7246 Available values for component modes are:
7248 addition
7250 and
7251 average
7252 bleach
7253 burn
7254 darken
7255 difference
7256 divide
7257 dodge
7258 exclusion
7259 extremity
7260 freeze
7261 geometric
7262 glow
7263 grainextract
7264 grainmerge
7265 hardlight
7266 hardmix
7267 hardoverlay
7268 harmonic
7269 heat
7270 interpolate
7271 lighten
7272 linearlight
7273 multiply
7274 multiply128
7275 negation
7276 normal
7277 or
7278 overlay
7279 phoenix
7280 pinlight
7281 reflect
7282 screen
7283 softdifference
7284 softlight
7285 stain
7286 subtract
7287 vividlight
7288 xor
7289 c0_opacity
7290 c1_opacity
7291 c2_opacity
7292 c3_opacity
7293 all_opacity
7294 Set blend opacity for specific pixel component or all pixel
7295 components in case of all_opacity. Only used in combination with
7296 pixel component blend modes.
7297 c0_expr
7299 c1_expr
7300 c2_expr
7301 c3_expr
7302 all_expr
7303 Set blend expression for specific pixel component or all pixel
7304 components in case of all_expr. Note that related mode options will
7305 be ignored if those are set.
7306 The expressions can use the following variables:
7308 N The sequential number of the filtered frame, starting from 0.
7310 X
7312 Y the coordinates of the current sample
7313 W
7315 H the width and height of currently filtered plane
7316 SW
7318 SH Width and height scale for the plane being filtered. It is the
7319 ratio between the dimensions of the current plane to the luma
7320 plane, e.g. for a "yuv420p" frame, the values are "1,1" for the
7321 luma plane and "0.5,0.5" for the chroma planes.
7322 T Time of the current frame, expressed in seconds.
7324 TOP, A
7326 Value of pixel component at current location for first video
7327 frame (top layer).
7328 BOTTOM, B
7330 Value of pixel component at current location for second video
7331 frame (bottom layer).
7332 The "blend" filter also supports the framesync options.
7334 Examples
7336 • Apply transition from bottom layer to top layer in first 10
7338 seconds:
7339 blend=all_expr='A*(if(gte(T,10),1,T/10))+B*(1-(if(gte(T,10),1,T/10)))'
7341 • Apply linear horizontal transition from top layer to bottom layer:
7343 blend=all_expr='A*(X/W)+B*(1-X/W)'
7345 • Apply 1x1 checkerboard effect:
7347 blend=all_expr='if(eq(mod(X,2),mod(Y,2)),A,B)'
7349 • Apply uncover left effect:
7351 blend=all_expr='if(gte(N*SW+X,W),A,B)'
7353 • Apply uncover down effect:
7355 blend=all_expr='if(gte(Y-N*SH,0),A,B)'
7357 • Apply uncover up-left effect:
7359 blend=all_expr='if(gte(T*SH*40+Y,H)*gte((T*40*SW+X)*W/H,W),A,B)'
7361 • Split diagonally video and shows top and bottom layer on each side:
7363 blend=all_expr='if(gt(X,Y*(W/H)),A,B)'
7365 • Display differences between the current and the previous frame:
7367 tblend=all_mode=grainextract
7369 Commands
7371 This filter supports same commands as options.
7373 blockdetect
7375 Determines blockiness of frames without altering the input frames.
7376 Based on Remco Muijs and Ihor Kirenko: "A no-reference blocking
7378 artifact measure for adaptive video processing." 2005 13th European
7379 signal processing conference.
7380 The filter accepts the following options:
7382 period_min
7384 period_max
7385 Set minimum and maximum values for determining pixel grids
7386 (periods). Default values are [3,24].
7387 planes
7389 Set planes to filter. Default is first only.
7390 Examples
7392 • Determine blockiness for the first plane and search for periods
7394 within [8,32]:
7395 blockdetect=period_min=8:period_max=32:planes=1
7397 blurdetect
7399 Determines blurriness of frames without altering the input frames.
7400 Based on Marziliano, Pina, et al. "A no-reference perceptual blur
7402 metric." Allows for a block-based abbreviation.
7403 The filter accepts the following options:
7405 low
7407 high
7408 Set low and high threshold values used by the Canny thresholding
7409 algorithm.
7410 The high threshold selects the "strong" edge pixels, which are then
7412 connected through 8-connectivity with the "weak" edge pixels
7413 selected by the low threshold.
7414 low and high threshold values must be chosen in the range [0,1],
7416 and low should be lesser or equal to high.
7417 Default value for low is "20/255", and default value for high is
7419 "50/255".
7420 radius
7422 Define the radius to search around an edge pixel for local maxima.
7423 block_pct
7425 Determine blurriness only for the most significant blocks, given in
7426 percentage.
7427 block_width
7429 Determine blurriness for blocks of width block_width. If set to any
7430 value smaller 1, no blocks are used and the whole image is
7431 processed as one no matter of block_height.
7432 block_height
7434 Determine blurriness for blocks of height block_height. If set to
7435 any value smaller 1, no blocks are used and the whole image is
7436 processed as one no matter of block_width.
7437 planes
7439 Set planes to filter. Default is first only.
7440 Examples
7442 • Determine blur for 80% of most significant 32x32 blocks:
7444 blurdetect=block_width=32:block_height=32:block_pct=80
7446 bm3d
7448 Denoise frames using Block-Matching 3D algorithm.
7449 The filter accepts the following options.
7451 sigma
7453 Set denoising strength. Default value is 1. Allowed range is from
7454 0 to 999.9. The denoising algorithm is very sensitive to sigma, so
7455 adjust it according to the source.
7456 block
7458 Set local patch size. This sets dimensions in 2D.
7459 bstep
7461 Set sliding step for processing blocks. Default value is 4.
7462 Allowed range is from 1 to 64. Smaller values allows processing
7463 more reference blocks and is slower.
7464 group
7466 Set maximal number of similar blocks for 3rd dimension. Default
7467 value is 1. When set to 1, no block matching is done. Larger
7468 values allows more blocks in single group. Allowed range is from 1
7469 to 256.
7470 range
7472 Set radius for search block matching. Default is 9. Allowed range
7473 is from 1 to INT32_MAX.
7474 mstep
7476 Set step between two search locations for block matching. Default
7477 is 1. Allowed range is from 1 to 64. Smaller is slower.
7478 thmse
7480 Set threshold of mean square error for block matching. Valid range
7481 is 0 to INT32_MAX.
7482 hdthr
7484 Set thresholding parameter for hard thresholding in 3D transformed
7485 domain. Larger values results in stronger hard-thresholding
7486 filtering in frequency domain.
7487 estim
7489 Set filtering estimation mode. Can be "basic" or "final". Default
7490 is "basic".
7491 ref If enabled, filter will use 2nd stream for block matching. Default
7493 is disabled for "basic" value of estim option, and always enabled
7494 if value of estim is "final".
7495 planes
7497 Set planes to filter. Default is all available except alpha.
7498 Examples
7500 • Basic filtering with bm3d:
7502 bm3d=sigma=3:block=4:bstep=2:group=1:estim=basic
7504 • Same as above, but filtering only luma:
7506 bm3d=sigma=3:block=4:bstep=2:group=1:estim=basic:planes=1
7508 • Same as above, but with both estimation modes:
7510 split[a][b],[a]bm3d=sigma=3:block=4:bstep=2:group=1:estim=basic[a],[b][a]bm3d=sigma=3:block=4:bstep=2:group=16:estim=final:ref=1
7512 • Same as above, but prefilter with nlmeans filter instead:
7514 split[a][b],[a]nlmeans=s=3:r=7:p=3[a],[b][a]bm3d=sigma=3:block=4:bstep=2:group=16:estim=final:ref=1
7516 boxblur
7518 Apply a boxblur algorithm to the input video.
7519 It accepts the following parameters:
7521 luma_radius, lr
7523 luma_power, lp
7524 chroma_radius, cr
7525 chroma_power, cp
7526 alpha_radius, ar
7527 alpha_power, ap
7528 A description of the accepted options follows.
7530 luma_radius, lr
7532 chroma_radius, cr
7533 alpha_radius, ar
7534 Set an expression for the box radius in pixels used for blurring
7535 the corresponding input plane.
7536 The radius value must be a non-negative number, and must not be
7538 greater than the value of the expression "min(w,h)/2" for the luma
7539 and alpha planes, and of "min(cw,ch)/2" for the chroma planes.
7540 Default value for luma_radius is "2". If not specified,
7542 chroma_radius and alpha_radius default to the corresponding value
7543 set for luma_radius.
7544 The expressions can contain the following constants:
7546 w
7548 h The input width and height in pixels.
7549 cw
7551 ch The input chroma image width and height in pixels.
7552 hsub
7554 vsub
7555 The horizontal and vertical chroma subsample values. For
7556 example, for the pixel format "yuv422p", hsub is 2 and vsub is
7557 1.
7558 luma_power, lp
7560 chroma_power, cp
7561 alpha_power, ap
7562 Specify how many times the boxblur filter is applied to the
7563 corresponding plane.
7564 Default value for luma_power is 2. If not specified, chroma_power
7566 and alpha_power default to the corresponding value set for
7567 luma_power.
7568 A value of 0 will disable the effect.
7570 Examples
7572 • Apply a boxblur filter with the luma, chroma, and alpha radii set
7574 to 2:
7575 boxblur=luma_radius=2:luma_power=1
7577 boxblur=2:1
7578 • Set the luma radius to 2, and alpha and chroma radius to 0:
7580 boxblur=2:1:cr=0:ar=0
7582 • Set the luma and chroma radii to a fraction of the video dimension:
7584 boxblur=luma_radius=min(h\,w)/10:luma_power=1:chroma_radius=min(cw\,ch)/10:chroma_power=1
7586 bwdif
7588 Deinterlace the input video ("bwdif" stands for "Bob Weaver
7589 Deinterlacing Filter").
7590 Motion adaptive deinterlacing based on yadif with the use of w3fdif and
7592 cubic interpolation algorithms. It accepts the following parameters:
7593 mode
7595 The interlacing mode to adopt. It accepts one of the following
7596 values:
7597 0, send_frame
7599 Output one frame for each frame.
7600 1, send_field
7602 Output one frame for each field.
7603 The default value is "send_field".
7605 parity
7607 The picture field parity assumed for the input interlaced video. It
7608 accepts one of the following values:
7609 0, tff
7611 Assume the top field is first.
7612 1, bff
7614 Assume the bottom field is first.
7615 -1, auto
7617 Enable automatic detection of field parity.
7618 The default value is "auto". If the interlacing is unknown or the
7620 decoder does not export this information, top field first will be
7621 assumed.
7622 deint
7624 Specify which frames to deinterlace. Accepts one of the following
7625 values:
7626 0, all
7628 Deinterlace all frames.
7629 1, interlaced
7631 Only deinterlace frames marked as interlaced.
7632 The default value is "all".
7634 cas
7636 Apply Contrast Adaptive Sharpen filter to video stream.
7637 The filter accepts the following options:
7639 strength
7641 Set the sharpening strength. Default value is 0.
7642 planes
7644 Set planes to filter. Default value is to filter all planes except
7645 alpha plane.
7646 Commands
7648 This filter supports same commands as options.
7650 chromahold
7652 Remove all color information for all colors except for certain one.
7653 The filter accepts the following options:
7655 color
7657 The color which will not be replaced with neutral chroma.
7658 similarity
7660 Similarity percentage with the above color. 0.01 matches only the
7661 exact key color, while 1.0 matches everything.
7662 blend
7664 Blend percentage. 0.0 makes pixels either fully gray, or not gray
7665 at all. Higher values result in more preserved color.
7666 yuv Signals that the color passed is already in YUV instead of RGB.
7668 Literal colors like "green" or "red" don't make sense with this
7670 enabled anymore. This can be used to pass exact YUV values as
7671 hexadecimal numbers.
7672 Commands
7674 This filter supports same commands as options. The command accepts the
7676 same syntax of the corresponding option.
7677 If the specified expression is not valid, it is kept at its current
7679 value.
7680 chromakey
7682 YUV colorspace color/chroma keying.
7683 The filter accepts the following options:
7685 color
7687 The color which will be replaced with transparency.
7688 similarity
7690 Similarity percentage with the key color.
7691 0.01 matches only the exact key color, while 1.0 matches
7693 everything.
7694 blend
7696 Blend percentage.
7697 0.0 makes pixels either fully transparent, or not transparent at
7699 all.
7700 Higher values result in semi-transparent pixels, with a higher
7702 transparency the more similar the pixels color is to the key color.
7703 yuv Signals that the color passed is already in YUV instead of RGB.
7705 Literal colors like "green" or "red" don't make sense with this
7707 enabled anymore. This can be used to pass exact YUV values as
7708 hexadecimal numbers.
7709 Commands
7711 This filter supports same commands as options. The command accepts the
7713 same syntax of the corresponding option.
7714 If the specified expression is not valid, it is kept at its current
7716 value.
7717 Examples
7719 • Make every green pixel in the input image transparent:
7721 ffmpeg -i input.png -vf chromakey=green out.png
7723 • Overlay a greenscreen-video on top of a static black background.
7725 ffmpeg -f lavfi -i color=c=black:s=1280x720 -i video.mp4 -shortest -filter_complex "[1:v]chromakey=0x70de77:0.1:0.2[ckout];[0:v][ckout]overlay[out]" -map "[out]" output.mkv
7727 chromakey_cuda
7729 CUDA accelerated YUV colorspace color/chroma keying.
7730 This filter works like normal chromakey filter but operates on CUDA
7732 frames. for more details and parameters see chromakey.
7733 Examples
7735 • Make all the green pixels in the input video transparent and use it
7737 as an overlay for another video:
7738 ./ffmpeg \
7740 -hwaccel cuda -hwaccel_output_format cuda -i input_green.mp4 \
7741 -hwaccel cuda -hwaccel_output_format cuda -i base_video.mp4 \
7742 -init_hw_device cuda \
7743 -filter_complex \
7744 " \
7745 [0:v]chromakey_cuda=0x25302D:0.1:0.12:1[overlay_video]; \
7746 [1:v]scale_cuda=format=yuv420p[base]; \
7747 [base][overlay_video]overlay_cuda" \
7748 -an -sn -c:v h264_nvenc -cq 20 output.mp4
7749 • Process two software sources, explicitly uploading the frames:
7751 ./ffmpeg -init_hw_device cuda=cuda -filter_hw_device cuda \
7753 -f lavfi -i color=size=800x600:color=white,format=yuv420p \
7754 -f lavfi -i yuvtestsrc=size=200x200,format=yuv420p \
7755 -filter_complex \
7756 " \
7757 [0]hwupload[under]; \
7758 [1]hwupload,chromakey_cuda=green:0.1:0.12[over]; \
7759 [under][over]overlay_cuda" \
7760 -c:v hevc_nvenc -cq 18 -preset slow output.mp4
7761 chromanr
7763 Reduce chrominance noise.
7764 The filter accepts the following options:
7766 thres
7768 Set threshold for averaging chrominance values. Sum of absolute
7769 difference of Y, U and V pixel components of current pixel and
7770 neighbour pixels lower than this threshold will be used in
7771 averaging. Luma component is left unchanged and is copied to
7772 output. Default value is 30. Allowed range is from 1 to 200.
7773 sizew
7775 Set horizontal radius of rectangle used for averaging. Allowed
7776 range is from 1 to 100. Default value is 5.
7777 sizeh
7779 Set vertical radius of rectangle used for averaging. Allowed range
7780 is from 1 to 100. Default value is 5.
7781 stepw
7783 Set horizontal step when averaging. Default value is 1. Allowed
7784 range is from 1 to 50. Mostly useful to speed-up filtering.
7785 steph
7787 Set vertical step when averaging. Default value is 1. Allowed
7788 range is from 1 to 50. Mostly useful to speed-up filtering.
7789 threy
7791 Set Y threshold for averaging chrominance values. Set finer
7792 control for max allowed difference between Y components of current
7793 pixel and neigbour pixels. Default value is 200. Allowed range is
7794 from 1 to 200.
7795 threu
7797 Set U threshold for averaging chrominance values. Set finer
7798 control for max allowed difference between U components of current
7799 pixel and neigbour pixels. Default value is 200. Allowed range is
7800 from 1 to 200.
7801 threv
7803 Set V threshold for averaging chrominance values. Set finer
7804 control for max allowed difference between V components of current
7805 pixel and neigbour pixels. Default value is 200. Allowed range is
7806 from 1 to 200.
7807 distance
7809 Set distance type used in calculations.
7810 manhattan
7812 Absolute difference.
7813 euclidean
7815 Difference squared.
7816 Default distance type is manhattan.
7818 Commands
7820 This filter supports same commands as options. The command accepts the
7822 same syntax of the corresponding option.
7823 chromashift
7825 Shift chroma pixels horizontally and/or vertically.
7826 The filter accepts the following options:
7828 cbh Set amount to shift chroma-blue horizontally.
7830 cbv Set amount to shift chroma-blue vertically.
7832 crh Set amount to shift chroma-red horizontally.
7834 crv Set amount to shift chroma-red vertically.
7836 edge
7838 Set edge mode, can be smear, default, or warp.
7839 Commands
7841 This filter supports the all above options as commands.
7843 ciescope
7845 Display CIE color diagram with pixels overlaid onto it.
7846 The filter accepts the following options:
7848 system
7850 Set color system.
7851 ntsc, 470m
7853 ebu, 470bg
7854 smpte
7855 240m
7856 apple
7857 widergb
7858 cie1931
7859 rec709, hdtv
7860 uhdtv, rec2020
7861 dcip3
7862 cie Set CIE system.
7863 xyy
7865 ucs
7866 luv
7867 gamuts
7868 Set what gamuts to draw.
7869 See "system" option for available values.
7871 size, s
7873 Set ciescope size, by default set to 512.
7874 intensity, i
7876 Set intensity used to map input pixel values to CIE diagram.
7877 contrast
7879 Set contrast used to draw tongue colors that are out of active
7880 color system gamut.
7881 corrgamma
7883 Correct gamma displayed on scope, by default enabled.
7884 showwhite
7886 Show white point on CIE diagram, by default disabled.
7887 gamma
7889 Set input gamma. Used only with XYZ input color space.
7890 fill
7892 Fill with CIE colors. By default is enabled.
7893 codecview
7895 Visualize information exported by some codecs.
7896 Some codecs can export information through frames using side-data or
7898 other means. For example, some MPEG based codecs export motion vectors
7899 through the export_mvs flag in the codec flags2 option.
7900 The filter accepts the following option:
7902 block
7904 Display block partition structure using the luma plane.
7905 mv Set motion vectors to visualize.
7907 Available flags for mv are:
7909 pf forward predicted MVs of P-frames
7911 bf forward predicted MVs of B-frames
7913 bb backward predicted MVs of B-frames
7915 qp Display quantization parameters using the chroma planes.
7917 mv_type, mvt
7919 Set motion vectors type to visualize. Includes MVs from all frames
7920 unless specified by frame_type option.
7921 Available flags for mv_type are:
7923 fp forward predicted MVs
7925 bp backward predicted MVs
7927 frame_type, ft
7929 Set frame type to visualize motion vectors of.
7930 Available flags for frame_type are:
7932 if intra-coded frames (I-frames)
7934 pf predicted frames (P-frames)
7936 bf bi-directionally predicted frames (B-frames)
7938 Examples
7940 • Visualize forward predicted MVs of all frames using ffplay:
7942 ffplay -flags2 +export_mvs input.mp4 -vf codecview=mv_type=fp
7944 • Visualize multi-directionals MVs of P and B-Frames using ffplay:
7946 ffplay -flags2 +export_mvs input.mp4 -vf codecview=mv=pf+bf+bb
7948 colorbalance
7950 Modify intensity of primary colors (red, green and blue) of input
7951 frames.
7952 The filter allows an input frame to be adjusted in the shadows,
7954 midtones or highlights regions for the red-cyan, green-magenta or blue-
7955 yellow balance.
7956 A positive adjustment value shifts the balance towards the primary
7958 color, a negative value towards the complementary color.
7959 The filter accepts the following options:
7961 rs
7963 gs
7964 bs Adjust red, green and blue shadows (darkest pixels).
7965 rm
7967 gm
7968 bm Adjust red, green and blue midtones (medium pixels).
7969 rh
7971 gh
7972 bh Adjust red, green and blue highlights (brightest pixels).
7973 Allowed ranges for options are "[-1.0, 1.0]". Defaults are 0.
7975 pl Preserve lightness when changing color balance. Default is
7977 disabled.
7978 Examples
7980 • Add red color cast to shadows:
7982 colorbalance=rs=.3
7984 Commands
7986 This filter supports the all above options as commands.
7988 colorcontrast
7990 Adjust color contrast between RGB components.
7991 The filter accepts the following options:
7993 rc Set the red-cyan contrast. Defaults is 0.0. Allowed range is from
7995 -1.0 to 1.0.
7996 gm Set the green-magenta contrast. Defaults is 0.0. Allowed range is
7998 from -1.0 to 1.0.
7999 by Set the blue-yellow contrast. Defaults is 0.0. Allowed range is
8001 from -1.0 to 1.0.
8002 rcw
8004 gmw
8005 byw Set the weight of each "rc", "gm", "by" option value. Default value
8006 is 0.0. Allowed range is from 0.0 to 1.0. If all weights are 0.0
8007 filtering is disabled.
8008 pl Set the amount of preserving lightness. Default value is 0.0.
8010 Allowed range is from 0.0 to 1.0.
8011 Commands
8013 This filter supports the all above options as commands.
8015 colorcorrect
8017 Adjust color white balance selectively for blacks and whites. This
8018 filter operates in YUV colorspace.
8019 The filter accepts the following options:
8021 rl Set the red shadow spot. Allowed range is from -1.0 to 1.0.
8023 Default value is 0.
8024 bl Set the blue shadow spot. Allowed range is from -1.0 to 1.0.
8026 Default value is 0.
8027 rh Set the red highlight spot. Allowed range is from -1.0 to 1.0.
8029 Default value is 0.
8030 bh Set the red highlight spot. Allowed range is from -1.0 to 1.0.
8032 Default value is 0.
8033 saturation
8035 Set the amount of saturation. Allowed range is from -3.0 to 3.0.
8036 Default value is 1.
8037 analyze
8039 If set to anything other than "manual" it will analyze every frame
8040 and use derived parameters for filtering output frame.
8041 Possible values are:
8043 manual
8045 average
8046 minmax
8047 median
8048 Default value is "manual".
8050 Commands
8052 This filter supports the all above options as commands.
8054 colorchannelmixer
8056 Adjust video input frames by re-mixing color channels.
8057 This filter modifies a color channel by adding the values associated to
8059 the other channels of the same pixels. For example if the value to
8060 modify is red, the output value will be:
8061 <red>=<red>*<rr> + <blue>*<rb> + <green>*<rg> + <alpha>*<ra>
8063 The filter accepts the following options:
8065 rr
8067 rg
8068 rb
8069 ra Adjust contribution of input red, green, blue and alpha channels
8070 for output red channel. Default is 1 for rr, and 0 for rg, rb and
8071 ra.
8072 gr
8074 gg
8075 gb
8076 ga Adjust contribution of input red, green, blue and alpha channels
8077 for output green channel. Default is 1 for gg, and 0 for gr, gb
8078 and ga.
8079 br
8081 bg
8082 bb
8083 ba Adjust contribution of input red, green, blue and alpha channels
8084 for output blue channel. Default is 1 for bb, and 0 for br, bg and
8085 ba.
8086 ar
8088 ag
8089 ab
8090 aa Adjust contribution of input red, green, blue and alpha channels
8091 for output alpha channel. Default is 1 for aa, and 0 for ar, ag
8092 and ab.
8093 Allowed ranges for options are "[-2.0, 2.0]".
8095 pc Set preserve color mode. The accepted values are:
8097 none
8099 Disable color preserving, this is default.
8100 lum Preserve luminance.
8102 max Preserve max value of RGB triplet.
8104 avg Preserve average value of RGB triplet.
8106 sum Preserve sum value of RGB triplet.
8108 nrm Preserve normalized value of RGB triplet.
8110 pwr Preserve power value of RGB triplet.
8112 pa Set the preserve color amount when changing colors. Allowed range
8114 is from "[0.0, 1.0]". Default is 0.0, thus disabled.
8115 Examples
8117 • Convert source to grayscale:
8119 colorchannelmixer=.3:.4:.3:0:.3:.4:.3:0:.3:.4:.3
8121 • Simulate sepia tones:
8123 colorchannelmixer=.393:.769:.189:0:.349:.686:.168:0:.272:.534:.131
8125 Commands
8127 This filter supports the all above options as commands.
8129 colorize
8131 Overlay a solid color on the video stream.
8132 The filter accepts the following options:
8134 hue Set the color hue. Allowed range is from 0 to 360. Default value
8136 is 0.
8137 saturation
8139 Set the color saturation. Allowed range is from 0 to 1. Default
8140 value is 0.5.
8141 lightness
8143 Set the color lightness. Allowed range is from 0 to 1. Default
8144 value is 0.5.
8145 mix Set the mix of source lightness. By default is set to 1.0. Allowed
8147 range is from 0.0 to 1.0.
8148 Commands
8150 This filter supports the all above options as commands.
8152 colorkey
8154 RGB colorspace color keying. This filter operates on 8-bit RGB format
8155 frames by setting the alpha component of each pixel which falls within
8156 the similarity radius of the key color to 0. The alpha value for pixels
8157 outside the similarity radius depends on the value of the blend option.
8158 The filter accepts the following options:
8160 color
8162 Set the color for which alpha will be set to 0 (full transparency).
8163 See "Color" section in the ffmpeg-utils manual. Default is
8164 "black".
8165 similarity
8167 Set the radius from the key color within which other colors also
8168 have full transparency. The computed distance is related to the
8169 unit fractional distance in 3D space between the RGB values of the
8170 key color and the pixel's color. Range is 0.01 to 1.0. 0.01 matches
8171 within a very small radius around the exact key color, while 1.0
8172 matches everything. Default is 0.01.
8173 blend
8175 Set how the alpha value for pixels that fall outside the similarity
8176 radius is computed. 0.0 makes pixels either fully transparent or
8177 fully opaque. Higher values result in semi-transparent pixels,
8178 with greater transparency the more similar the pixel color is to
8179 the key color. Range is 0.0 to 1.0. Default is 0.0.
8180 Examples
8182 • Make every green pixel in the input image transparent:
8184 ffmpeg -i input.png -vf colorkey=green out.png
8186 • Overlay a greenscreen-video on top of a static background image.
8188 ffmpeg -i background.png -i video.mp4 -filter_complex "[1:v]colorkey=0x3BBD1E:0.3:0.2[ckout];[0:v][ckout]overlay[out]" -map "[out]" output.flv
8190 Commands
8192 This filter supports same commands as options. The command accepts the
8194 same syntax of the corresponding option.
8195 If the specified expression is not valid, it is kept at its current
8197 value.
8198 colorhold
8200 Remove all color information for all RGB colors except for certain one.
8201 The filter accepts the following options:
8203 color
8205 The color which will not be replaced with neutral gray.
8206 similarity
8208 Similarity percentage with the above color. 0.01 matches only the
8209 exact key color, while 1.0 matches everything.
8210 blend
8212 Blend percentage. 0.0 makes pixels fully gray. Higher values
8213 result in more preserved color.
8214 Commands
8216 This filter supports same commands as options. The command accepts the
8218 same syntax of the corresponding option.
8219 If the specified expression is not valid, it is kept at its current
8221 value.
8222 colorlevels
8224 Adjust video input frames using levels.
8225 The filter accepts the following options:
8227 rimin
8229 gimin
8230 bimin
8231 aimin
8232 Adjust red, green, blue and alpha input black point. Allowed
8233 ranges for options are "[-1.0, 1.0]". Defaults are 0.
8234 rimax
8236 gimax
8237 bimax
8238 aimax
8239 Adjust red, green, blue and alpha input white point. Allowed
8240 ranges for options are "[-1.0, 1.0]". Defaults are 1.
8241 Input levels are used to lighten highlights (bright tones), darken
8243 shadows (dark tones), change the balance of bright and dark tones.
8244 romin
8246 gomin
8247 bomin
8248 aomin
8249 Adjust red, green, blue and alpha output black point. Allowed
8250 ranges for options are "[0, 1.0]". Defaults are 0.
8251 romax
8253 gomax
8254 bomax
8255 aomax
8256 Adjust red, green, blue and alpha output white point. Allowed
8257 ranges for options are "[0, 1.0]". Defaults are 1.
8258 Output levels allows manual selection of a constrained output level
8260 range.
8261 preserve
8263 Set preserve color mode. The accepted values are:
8264 none
8266 Disable color preserving, this is default.
8267 lum Preserve luminance.
8269 max Preserve max value of RGB triplet.
8271 avg Preserve average value of RGB triplet.
8273 sum Preserve sum value of RGB triplet.
8275 nrm Preserve normalized value of RGB triplet.
8277 pwr Preserve power value of RGB triplet.
8279 Examples
8281 • Make video output darker:
8283 colorlevels=rimin=0.058:gimin=0.058:bimin=0.058
8285 • Increase contrast:
8287 colorlevels=rimin=0.039:gimin=0.039:bimin=0.039:rimax=0.96:gimax=0.96:bimax=0.96
8289 • Make video output lighter:
8291 colorlevels=rimax=0.902:gimax=0.902:bimax=0.902
8293 • Increase brightness:
8295 colorlevels=romin=0.5:gomin=0.5:bomin=0.5
8297 Commands
8299 This filter supports the all above options as commands.
8301 colormap
8303 Apply custom color maps to video stream.
8304 This filter needs three input video streams. First stream is video
8306 stream that is going to be filtered out. Second and third video stream
8307 specify color patches for source color to target color mapping.
8308 The filter accepts the following options:
8310 patch_size
8312 Set the source and target video stream patch size in pixels.
8313 nb_patches
8315 Set the max number of used patches from source and target video
8316 stream. Default value is number of patches available in additional
8317 video streams. Max allowed number of patches is 64.
8318 type
8320 Set the adjustments used for target colors. Can be "relative" or
8321 "absolute". Defaults is "absolute".
8322 kernel
8324 Set the kernel used to measure color differences between mapped
8325 colors.
8326 The accepted values are:
8328 euclidean
8330 weuclidean
8331 Default is "euclidean".
8333 colormatrix
8335 Convert color matrix.
8336 The filter accepts the following options:
8338 src
8340 dst Specify the source and destination color matrix. Both values must
8341 be specified.
8342 The accepted values are:
8344 bt709
8346 BT.709
8347 fcc FCC
8349 bt601
8351 BT.601
8352 bt470
8354 BT.470
8355 bt470bg
8357 BT.470BG
8358 smpte170m
8360 SMPTE-170M
8361 smpte240m
8363 SMPTE-240M
8364 bt2020
8366 BT.2020
8367 For example to convert from BT.601 to SMPTE-240M, use the command:
8369 colormatrix=bt601:smpte240m
8371 colorspace
8373 Convert colorspace, transfer characteristics or color primaries. Input
8374 video needs to have an even size.
8375 The filter accepts the following options:
8377 all Specify all color properties at once.
8379 The accepted values are:
8381 bt470m
8383 BT.470M
8384 bt470bg
8386 BT.470BG
8387 bt601-6-525
8389 BT.601-6 525
8390 bt601-6-625
8392 BT.601-6 625
8393 bt709
8395 BT.709
8396 smpte170m
8398 SMPTE-170M
8399 smpte240m
8401 SMPTE-240M
8402 bt2020
8404 BT.2020
8405 space
8407 Specify output colorspace.
8408 The accepted values are:
8410 bt709
8412 BT.709
8413 fcc FCC
8415 bt470bg
8417 BT.470BG or BT.601-6 625
8418 smpte170m
8420 SMPTE-170M or BT.601-6 525
8421 smpte240m
8423 SMPTE-240M
8424 ycgco
8426 YCgCo
8427 bt2020ncl
8429 BT.2020 with non-constant luminance
8430 trc Specify output transfer characteristics.
8432 The accepted values are:
8434 bt709
8436 BT.709
8437 bt470m
8439 BT.470M
8440 bt470bg
8442 BT.470BG
8443 gamma22
8445 Constant gamma of 2.2
8446 gamma28
8448 Constant gamma of 2.8
8449 smpte170m
8451 SMPTE-170M, BT.601-6 625 or BT.601-6 525
8452 smpte240m
8454 SMPTE-240M
8455 srgb
8457 SRGB
8458 iec61966-2-1
8460 iec61966-2-1
8461 iec61966-2-4
8463 iec61966-2-4
8464 xvycc
8466 xvycc
8467 bt2020-10
8469 BT.2020 for 10-bits content
8470 bt2020-12
8472 BT.2020 for 12-bits content
8473 primaries
8475 Specify output color primaries.
8476 The accepted values are:
8478 bt709
8480 BT.709
8481 bt470m
8483 BT.470M
8484 bt470bg
8486 BT.470BG or BT.601-6 625
8487 smpte170m
8489 SMPTE-170M or BT.601-6 525
8490 smpte240m
8492 SMPTE-240M
8493 film
8495 film
8496 smpte431
8498 SMPTE-431
8499 smpte432
8501 SMPTE-432
8502 bt2020
8504 BT.2020
8505 jedec-p22
8507 JEDEC P22 phosphors
8508 range
8510 Specify output color range.
8511 The accepted values are:
8513 tv TV (restricted) range
8515 mpeg
8517 MPEG (restricted) range
8518 pc PC (full) range
8520 jpeg
8522 JPEG (full) range
8523 format
8525 Specify output color format.
8526 The accepted values are:
8528 yuv420p
8530 YUV 4:2:0 planar 8-bits
8531 yuv420p10
8533 YUV 4:2:0 planar 10-bits
8534 yuv420p12
8536 YUV 4:2:0 planar 12-bits
8537 yuv422p
8539 YUV 4:2:2 planar 8-bits
8540 yuv422p10
8542 YUV 4:2:2 planar 10-bits
8543 yuv422p12
8545 YUV 4:2:2 planar 12-bits
8546 yuv444p
8548 YUV 4:4:4 planar 8-bits
8549 yuv444p10
8551 YUV 4:4:4 planar 10-bits
8552 yuv444p12
8554 YUV 4:4:4 planar 12-bits
8555 fast
8557 Do a fast conversion, which skips gamma/primary correction. This
8558 will take significantly less CPU, but will be mathematically
8559 incorrect. To get output compatible with that produced by the
8560 colormatrix filter, use fast=1.
8561 dither
8563 Specify dithering mode.
8564 The accepted values are:
8566 none
8568 No dithering
8569 fsb Floyd-Steinberg dithering
8571 wpadapt
8573 Whitepoint adaptation mode.
8574 The accepted values are:
8576 bradford
8578 Bradford whitepoint adaptation
8579 vonkries
8581 von Kries whitepoint adaptation
8582 identity
8584 identity whitepoint adaptation (i.e. no whitepoint adaptation)
8585 iall
8587 Override all input properties at once. Same accepted values as all.
8588 ispace
8590 Override input colorspace. Same accepted values as space.
8591 iprimaries
8593 Override input color primaries. Same accepted values as primaries.
8594 itrc
8596 Override input transfer characteristics. Same accepted values as
8597 trc.
8598 irange
8600 Override input color range. Same accepted values as range.
8601 The filter converts the transfer characteristics, color space and color
8603 primaries to the specified user values. The output value, if not
8604 specified, is set to a default value based on the "all" property. If
8605 that property is also not specified, the filter will log an error. The
8606 output color range and format default to the same value as the input
8607 color range and format. The input transfer characteristics, color
8608 space, color primaries and color range should be set on the input data.
8609 If any of these are missing, the filter will log an error and no
8610 conversion will take place.
8611 For example to convert the input to SMPTE-240M, use the command:
8613 colorspace=smpte240m
8615 colortemperature
8617 Adjust color temperature in video to simulate variations in ambient
8618 color temperature.
8619 The filter accepts the following options:
8621 temperature
8623 Set the temperature in Kelvin. Allowed range is from 1000 to 40000.
8624 Default value is 6500 K.
8625 mix Set mixing with filtered output. Allowed range is from 0 to 1.
8627 Default value is 1.
8628 pl Set the amount of preserving lightness. Allowed range is from 0 to
8630 1. Default value is 0.
8631 Commands
8633 This filter supports same commands as options.
8635 convolution
8637 Apply convolution of 3x3, 5x5, 7x7 or horizontal/vertical up to 49
8638 elements.
8639 The filter accepts the following options:
8641 0m
8643 1m
8644 2m
8645 3m Set matrix for each plane. Matrix is sequence of 9, 25 or 49
8646 signed integers in square mode, and from 1 to 49 odd number of
8647 signed integers in row mode.
8648 0rdiv
8650 1rdiv
8651 2rdiv
8652 3rdiv
8653 Set multiplier for calculated value for each plane. If unset or 0,
8654 it will be sum of all matrix elements.
8655 0bias
8657 1bias
8658 2bias
8659 3bias
8660 Set bias for each plane. This value is added to the result of the
8661 multiplication. Useful for making the overall image brighter or
8662 darker. Default is 0.0.
8663 0mode
8665 1mode
8666 2mode
8667 3mode
8668 Set matrix mode for each plane. Can be square, row or column.
8669 Default is square.
8670 Commands
8672 This filter supports the all above options as commands.
8674 Examples
8676 • Apply sharpen:
8678 convolution="0 -1 0 -1 5 -1 0 -1 0:0 -1 0 -1 5 -1 0 -1 0:0 -1 0 -1 5 -1 0 -1 0:0 -1 0 -1 5 -1 0 -1 0"
8680 • Apply blur:
8682 convolution="1 1 1 1 1 1 1 1 1:1 1 1 1 1 1 1 1 1:1 1 1 1 1 1 1 1 1:1 1 1 1 1 1 1 1 1:1/9:1/9:1/9:1/9"
8684 • Apply edge enhance:
8686 convolution="0 0 0 -1 1 0 0 0 0:0 0 0 -1 1 0 0 0 0:0 0 0 -1 1 0 0 0 0:0 0 0 -1 1 0 0 0 0:5:1:1:1:0:128:128:128"
8688 • Apply edge detect:
8690 convolution="0 1 0 1 -4 1 0 1 0:0 1 0 1 -4 1 0 1 0:0 1 0 1 -4 1 0 1 0:0 1 0 1 -4 1 0 1 0:5:5:5:1:0:128:128:128"
8692 • Apply laplacian edge detector which includes diagonals:
8694 convolution="1 1 1 1 -8 1 1 1 1:1 1 1 1 -8 1 1 1 1:1 1 1 1 -8 1 1 1 1:1 1 1 1 -8 1 1 1 1:5:5:5:1:0:128:128:0"
8696 • Apply emboss:
8698 convolution="-2 -1 0 -1 1 1 0 1 2:-2 -1 0 -1 1 1 0 1 2:-2 -1 0 -1 1 1 0 1 2:-2 -1 0 -1 1 1 0 1 2"
8700 convolve
8702 Apply 2D convolution of video stream in frequency domain using second
8703 stream as impulse.
8704 The filter accepts the following options:
8706 planes
8708 Set which planes to process.
8709 impulse
8711 Set which impulse video frames will be processed, can be first or
8712 all. Default is all.
8713 The "convolve" filter also supports the framesync options.
8715 copy
8717 Copy the input video source unchanged to the output. This is mainly
8718 useful for testing purposes.
8719 coreimage
8721 Video filtering on GPU using Apple's CoreImage API on OSX.
8722 Hardware acceleration is based on an OpenGL context. Usually, this
8724 means it is processed by video hardware. However, software-based OpenGL
8725 implementations exist which means there is no guarantee for hardware
8726 processing. It depends on the respective OSX.
8727 There are many filters and image generators provided by Apple that come
8729 with a large variety of options. The filter has to be referenced by its
8730 name along with its options.
8731 The coreimage filter accepts the following options:
8733 list_filters
8735 List all available filters and generators along with all their
8736 respective options as well as possible minimum and maximum values
8737 along with the default values.
8738 list_filters=true
8740 filter
8742 Specify all filters by their respective name and options. Use
8743 list_filters to determine all valid filter names and options.
8744 Numerical options are specified by a float value and are
8745 automatically clamped to their respective value range. Vector and
8746 color options have to be specified by a list of space separated
8747 float values. Character escaping has to be done. A special option
8748 name "default" is available to use default options for a filter.
8749 It is required to specify either "default" or at least one of the
8751 filter options. All omitted options are used with their default
8752 values. The syntax of the filter string is as follows:
8753 filter=<NAME>@<OPTION>=<VALUE>[@<OPTION>=<VALUE>][@...][#<NAME>@<OPTION>=<VALUE>[@<OPTION>=<VALUE>][@...]][#...]
8755 output_rect
8757 Specify a rectangle where the output of the filter chain is copied
8758 into the input image. It is given by a list of space separated
8759 float values:
8760 output_rect=x\ y\ width\ height
8762 If not given, the output rectangle equals the dimensions of the
8764 input image. The output rectangle is automatically cropped at the
8765 borders of the input image. Negative values are valid for each
8766 component.
8767 output_rect=25\ 25\ 100\ 100
8769 Several filters can be chained for successive processing without GPU-
8771 HOST transfers allowing for fast processing of complex filter chains.
8772 Currently, only filters with zero (generators) or exactly one (filters)
8773 input image and one output image are supported. Also, transition
8774 filters are not yet usable as intended.
8775 Some filters generate output images with additional padding depending
8777 on the respective filter kernel. The padding is automatically removed
8778 to ensure the filter output has the same size as the input image.
8779 For image generators, the size of the output image is determined by the
8781 previous output image of the filter chain or the input image of the
8782 whole filterchain, respectively. The generators do not use the pixel
8783 information of this image to generate their output. However, the
8784 generated output is blended onto this image, resulting in partial or
8785 complete coverage of the output image.
8786 The coreimagesrc video source can be used for generating input images
8788 which are directly fed into the filter chain. By using it, providing
8789 input images by another video source or an input video is not required.
8790 Examples
8792 • List all filters available:
8794 coreimage=list_filters=true
8796 • Use the CIBoxBlur filter with default options to blur an image:
8798 coreimage=filter=CIBoxBlur@default
8800 • Use a filter chain with CISepiaTone at default values and
8802 CIVignetteEffect with its center at 100x100 and a radius of 50
8803 pixels:
8804 coreimage=filter=CIBoxBlur@default#CIVignetteEffect@inputCenter=100\ 100@inputRadius=50
8806 • Use nullsrc and CIQRCodeGenerator to create a QR code for the
8808 FFmpeg homepage, given as complete and escaped command-line for
8809 Apple's standard bash shell:
8810 ffmpeg -f lavfi -i nullsrc=s=100x100,coreimage=filter=CIQRCodeGenerator@inputMessage=https\\\\\://FFmpeg.org/@inputCorrectionLevel=H -frames:v 1 QRCode.png
8812 cover_rect
8814 Cover a rectangular object
8815 It accepts the following options:
8817 cover
8819 Filepath of the optional cover image, needs to be in yuv420.
8820 mode
8822 Set covering mode.
8823 It accepts the following values:
8825 cover
8827 cover it by the supplied image
8828 blur
8830 cover it by interpolating the surrounding pixels
8831 Default value is blur.
8833 Examples
8835 • Cover a rectangular object by the supplied image of a given video
8837 using ffmpeg:
8838 ffmpeg -i file.ts -vf find_rect=newref.pgm,cover_rect=cover.jpg:mode=cover new.mkv
8840 crop
8842 Crop the input video to given dimensions.
8843 It accepts the following parameters:
8845 w, out_w
8847 The width of the output video. It defaults to "iw". This
8848 expression is evaluated only once during the filter configuration,
8849 or when the w or out_w command is sent.
8850 h, out_h
8852 The height of the output video. It defaults to "ih". This
8853 expression is evaluated only once during the filter configuration,
8854 or when the h or out_h command is sent.
8855 x The horizontal position, in the input video, of the left edge of
8857 the output video. It defaults to "(in_w-out_w)/2". This expression
8858 is evaluated per-frame.
8859 y The vertical position, in the input video, of the top edge of the
8861 output video. It defaults to "(in_h-out_h)/2". This expression is
8862 evaluated per-frame.
8863 keep_aspect
8865 If set to 1 will force the output display aspect ratio to be the
8866 same of the input, by changing the output sample aspect ratio. It
8867 defaults to 0.
8868 exact
8870 Enable exact cropping. If enabled, subsampled videos will be
8871 cropped at exact width/height/x/y as specified and will not be
8872 rounded to nearest smaller value. It defaults to 0.
8873 The out_w, out_h, x, y parameters are expressions containing the
8875 following constants:
8876 x
8878 y The computed values for x and y. They are evaluated for each new
8879 frame.
8880 in_w
8882 in_h
8883 The input width and height.
8884 iw
8886 ih These are the same as in_w and in_h.
8887 out_w
8889 out_h
8890 The output (cropped) width and height.
8891 ow
8893 oh These are the same as out_w and out_h.
8894 a same as iw / ih
8896 sar input sample aspect ratio
8898 dar input display aspect ratio, it is the same as (iw / ih) * sar
8900 hsub
8902 vsub
8903 horizontal and vertical chroma subsample values. For example for
8904 the pixel format "yuv422p" hsub is 2 and vsub is 1.
8905 n The number of the input frame, starting from 0.
8907 pos the position in the file of the input frame, NAN if unknown
8909 t The timestamp expressed in seconds. It's NAN if the input timestamp
8911 is unknown.
8912 The expression for out_w may depend on the value of out_h, and the
8914 expression for out_h may depend on out_w, but they cannot depend on x
8915 and y, as x and y are evaluated after out_w and out_h.
8916 The x and y parameters specify the expressions for the position of the
8918 top-left corner of the output (non-cropped) area. They are evaluated
8919 for each frame. If the evaluated value is not valid, it is approximated
8920 to the nearest valid value.
8921 The expression for x may depend on y, and the expression for y may
8923 depend on x.
8924 Examples
8926 • Crop area with size 100x100 at position (12,34).
8928 crop=100:100:12:34
8930 Using named options, the example above becomes:
8932 crop=w=100:h=100:x=12:y=34
8934 • Crop the central input area with size 100x100:
8936 crop=100:100
8938 • Crop the central input area with size 2/3 of the input video:
8940 crop=2/3*in_w:2/3*in_h
8942 • Crop the input video central square:
8944 crop=out_w=in_h
8946 crop=in_h
8947 • Delimit the rectangle with the top-left corner placed at position
8949 100:100 and the right-bottom corner corresponding to the right-
8950 bottom corner of the input image.
8951 crop=in_w-100:in_h-100:100:100
8953 • Crop 10 pixels from the left and right borders, and 20 pixels from
8955 the top and bottom borders
8956 crop=in_w-2*10:in_h-2*20
8958 • Keep only the bottom right quarter of the input image:
8960 crop=in_w/2:in_h/2:in_w/2:in_h/2
8962 • Crop height for getting Greek harmony:
8964 crop=in_w:1/PHI*in_w
8966 • Apply trembling effect:
8968 crop=in_w/2:in_h/2:(in_w-out_w)/2+((in_w-out_w)/2)*sin(n/10):(in_h-out_h)/2 +((in_h-out_h)/2)*sin(n/7)
8970 • Apply erratic camera effect depending on timestamp:
8972 crop=in_w/2:in_h/2:(in_w-out_w)/2+((in_w-out_w)/2)*sin(t*10):(in_h-out_h)/2 +((in_h-out_h)/2)*sin(t*13)"
8974 • Set x depending on the value of y:
8976 crop=in_w/2:in_h/2:y:10+10*sin(n/10)
8978 Commands
8980 This filter supports the following commands:
8982 w, out_w
8984 h, out_h
8985 x
8986 y Set width/height of the output video and the horizontal/vertical
8987 position in the input video. The command accepts the same syntax
8988 of the corresponding option.
8989 If the specified expression is not valid, it is kept at its current
8991 value.
8992 cropdetect
8994 Auto-detect the crop size.
8995 It calculates the necessary cropping parameters and prints the
8997 recommended parameters via the logging system. The detected dimensions
8998 correspond to the non-black area of the input video.
8999 It accepts the following parameters:
9001 limit
9003 Set higher black value threshold, which can be optionally specified
9004 from nothing (0) to everything (255 for 8-bit based formats). An
9005 intensity value greater to the set value is considered non-black.
9006 It defaults to 24. You can also specify a value between 0.0 and
9007 1.0 which will be scaled depending on the bitdepth of the pixel
9008 format.
9009 round
9011 The value which the width/height should be divisible by. It
9012 defaults to 16. The offset is automatically adjusted to center the
9013 video. Use 2 to get only even dimensions (needed for 4:2:2 video).
9014 16 is best when encoding to most video codecs.
9015 skip
9017 Set the number of initial frames for which evaluation is skipped.
9018 Default is 2. Range is 0 to INT_MAX.
9019 reset_count, reset
9021 Set the counter that determines after how many frames cropdetect
9022 will reset the previously detected largest video area and start
9023 over to detect the current optimal crop area. Default value is 0.
9024 This can be useful when channel logos distort the video area. 0
9026 indicates 'never reset', and returns the largest area encountered
9027 during playback.
9028 cue
9030 Delay video filtering until a given wallclock timestamp. The filter
9031 first passes on preroll amount of frames, then it buffers at most
9032 buffer amount of frames and waits for the cue. After reaching the cue
9033 it forwards the buffered frames and also any subsequent frames coming
9034 in its input.
9035 The filter can be used synchronize the output of multiple ffmpeg
9037 processes for realtime output devices like decklink. By putting the
9038 delay in the filtering chain and pre-buffering frames the process can
9039 pass on data to output almost immediately after the target wallclock
9040 timestamp is reached.
9041 Perfect frame accuracy cannot be guaranteed, but the result is good
9043 enough for some use cases.
9044 cue The cue timestamp expressed in a UNIX timestamp in microseconds.
9046 Default is 0.
9047 preroll
9049 The duration of content to pass on as preroll expressed in seconds.
9050 Default is 0.
9051 buffer
9053 The maximum duration of content to buffer before waiting for the
9054 cue expressed in seconds. Default is 0.
9055 curves
9057 Apply color adjustments using curves.
9058 This filter is similar to the Adobe Photoshop and GIMP curves tools.
9060 Each component (red, green and blue) has its values defined by N key
9061 points tied from each other using a smooth curve. The x-axis represents
9062 the pixel values from the input frame, and the y-axis the new pixel
9063 values to be set for the output frame.
9064 By default, a component curve is defined by the two points (0;0) and
9066 (1;1). This creates a straight line where each original pixel value is
9067 "adjusted" to its own value, which means no change to the image.
9068 The filter allows you to redefine these two points and add some more. A
9070 new curve (using a natural cubic spline interpolation) will be define
9071 to pass smoothly through all these new coordinates. The new defined
9072 points needs to be strictly increasing over the x-axis, and their x and
9073 y values must be in the [0;1] interval. If the computed curves
9074 happened to go outside the vector spaces, the values will be clipped
9075 accordingly.
9076 The filter accepts the following options:
9078 preset
9080 Select one of the available color presets. This option can be used
9081 in addition to the r, g, b parameters; in this case, the later
9082 options takes priority on the preset values. Available presets
9083 are:
9084 none
9086 color_negative
9087 cross_process
9088 darker
9089 increase_contrast
9090 lighter
9091 linear_contrast
9092 medium_contrast
9093 negative
9094 strong_contrast
9095 vintage
9096 Default is "none".
9098 master, m
9100 Set the master key points. These points will define a second pass
9101 mapping. It is sometimes called a "luminance" or "value" mapping.
9102 It can be used with r, g, b or all since it acts like a post-
9103 processing LUT.
9104 red, r
9106 Set the key points for the red component.
9107 green, g
9109 Set the key points for the green component.
9110 blue, b
9112 Set the key points for the blue component.
9113 all Set the key points for all components (not including master). Can
9115 be used in addition to the other key points component options. In
9116 this case, the unset component(s) will fallback on this all
9117 setting.
9118 psfile
9120 Specify a Photoshop curves file (".acv") to import the settings
9121 from.
9122 plot
9124 Save Gnuplot script of the curves in specified file.
9125 To avoid some filtergraph syntax conflicts, each key points list need
9127 to be defined using the following syntax: "x0/y0 x1/y1 x2/y2 ...".
9128 Commands
9130 This filter supports same commands as options.
9132 Examples
9134 • Increase slightly the middle level of blue:
9136 curves=blue='0/0 0.5/0.58 1/1'
9138 • Vintage effect:
9140 curves=r='0/0.11 .42/.51 1/0.95':g='0/0 0.50/0.48 1/1':b='0/0.22 .49/.44 1/0.8'
9142 Here we obtain the following coordinates for each components:
9144 red "(0;0.11) (0.42;0.51) (1;0.95)"
9146 green
9148 "(0;0) (0.50;0.48) (1;1)"
9149 blue
9151 "(0;0.22) (0.49;0.44) (1;0.80)"
9152 • The previous example can also be achieved with the associated
9154 built-in preset:
9155 curves=preset=vintage
9157 • Or simply:
9159 curves=vintage
9161 • Use a Photoshop preset and redefine the points of the green
9163 component:
9164 curves=psfile='MyCurvesPresets/purple.acv':green='0/0 0.45/0.53 1/1'
9166 • Check out the curves of the "cross_process" profile using ffmpeg
9168 and gnuplot:
9169 ffmpeg -f lavfi -i color -vf curves=cross_process:plot=/tmp/curves.plt -frames:v 1 -f null -
9171 gnuplot -p /tmp/curves.plt
9172 datascope
9174 Video data analysis filter.
9175 This filter shows hexadecimal pixel values of part of video.
9177 The filter accepts the following options:
9179 size, s
9181 Set output video size.
9182 x Set x offset from where to pick pixels.
9184 y Set y offset from where to pick pixels.
9186 mode
9188 Set scope mode, can be one of the following:
9189 mono
9191 Draw hexadecimal pixel values with white color on black
9192 background.
9193 color
9195 Draw hexadecimal pixel values with input video pixel color on
9196 black background.
9197 color2
9199 Draw hexadecimal pixel values on color background picked from
9200 input video, the text color is picked in such way so its always
9201 visible.
9202 axis
9204 Draw rows and columns numbers on left and top of video.
9205 opacity
9207 Set background opacity.
9208 format
9210 Set display number format. Can be "hex", or "dec". Default is
9211 "hex".
9212 components
9214 Set pixel components to display. By default all pixel components
9215 are displayed.
9216 Commands
9218 This filter supports same commands as options excluding "size" option.
9220 dblur
9222 Apply Directional blur filter.
9223 The filter accepts the following options:
9225 angle
9227 Set angle of directional blur. Default is 45.
9228 radius
9230 Set radius of directional blur. Default is 5.
9231 planes
9233 Set which planes to filter. By default all planes are filtered.
9234 Commands
9236 This filter supports same commands as options. The command accepts the
9238 same syntax of the corresponding option.
9239 If the specified expression is not valid, it is kept at its current
9241 value.
9242 dctdnoiz
9244 Denoise frames using 2D DCT (frequency domain filtering).
9245 This filter is not designed for real time.
9247 The filter accepts the following options:
9249 sigma, s
9251 Set the noise sigma constant.
9252 This sigma defines a hard threshold of "3 * sigma"; every DCT
9254 coefficient (absolute value) below this threshold with be dropped.
9255 If you need a more advanced filtering, see expr.
9257 Default is 0.
9259 overlap
9261 Set number overlapping pixels for each block. Since the filter can
9262 be slow, you may want to reduce this value, at the cost of a less
9263 effective filter and the risk of various artefacts.
9264 If the overlapping value doesn't permit processing the whole input
9266 width or height, a warning will be displayed and according borders
9267 won't be denoised.
9268 Default value is blocksize-1, which is the best possible setting.
9270 expr, e
9272 Set the coefficient factor expression.
9273 For each coefficient of a DCT block, this expression will be
9275 evaluated as a multiplier value for the coefficient.
9276 If this is option is set, the sigma option will be ignored.
9278 The absolute value of the coefficient can be accessed through the c
9280 variable.
9281 n Set the blocksize using the number of bits. "1<<n" defines the
9283 blocksize, which is the width and height of the processed blocks.
9284 The default value is 3 (8x8) and can be raised to 4 for a blocksize
9286 of 16x16. Note that changing this setting has huge consequences on
9287 the speed processing. Also, a larger block size does not
9288 necessarily means a better de-noising.
9289 Examples
9291 Apply a denoise with a sigma of 4.5:
9293 dctdnoiz=4.5
9295 The same operation can be achieved using the expression system:
9297 dctdnoiz=e='gte(c, 4.5*3)'
9299 Violent denoise using a block size of "16x16":
9301 dctdnoiz=15:n=4
9303 deband
9305 Remove banding artifacts from input video. It works by replacing
9306 banded pixels with average value of referenced pixels.
9307 The filter accepts the following options:
9309 1thr
9311 2thr
9312 3thr
9313 4thr
9314 Set banding detection threshold for each plane. Default is 0.02.
9315 Valid range is 0.00003 to 0.5. If difference between current pixel
9316 and reference pixel is less than threshold, it will be considered
9317 as banded.
9318 range, r
9320 Banding detection range in pixels. Default is 16. If positive,
9321 random number in range 0 to set value will be used. If negative,
9322 exact absolute value will be used. The range defines square of
9323 four pixels around current pixel.
9324 direction, d
9326 Set direction in radians from which four pixel will be compared. If
9327 positive, random direction from 0 to set direction will be picked.
9328 If negative, exact of absolute value will be picked. For example
9329 direction 0, -PI or -2*PI radians will pick only pixels on same row
9330 and -PI/2 will pick only pixels on same column.
9331 blur, b
9333 If enabled, current pixel is compared with average value of all
9334 four surrounding pixels. The default is enabled. If disabled
9335 current pixel is compared with all four surrounding pixels. The
9336 pixel is considered banded if only all four differences with
9337 surrounding pixels are less than threshold.
9338 coupling, c
9340 If enabled, current pixel is changed if and only if all pixel
9341 components are banded, e.g. banding detection threshold is
9342 triggered for all color components. The default is disabled.
9343 Commands
9345 This filter supports the all above options as commands.
9347 deblock
9349 Remove blocking artifacts from input video.
9350 The filter accepts the following options:
9352 filter
9354 Set filter type, can be weak or strong. Default is strong. This
9355 controls what kind of deblocking is applied.
9356 block
9358 Set size of block, allowed range is from 4 to 512. Default is 8.
9359 alpha
9361 beta
9362 gamma
9363 delta
9364 Set blocking detection thresholds. Allowed range is 0 to 1.
9365 Defaults are: 0.098 for alpha and 0.05 for the rest. Using higher
9366 threshold gives more deblocking strength. Setting alpha controls
9367 threshold detection at exact edge of block. Remaining options
9368 controls threshold detection near the edge. Each one for
9369 below/above or left/right. Setting any of those to 0 disables
9370 deblocking.
9371 planes
9373 Set planes to filter. Default is to filter all available planes.
9374 Examples
9376 • Deblock using weak filter and block size of 4 pixels.
9378 deblock=filter=weak:block=4
9380 • Deblock using strong filter, block size of 4 pixels and custom
9382 thresholds for deblocking more edges.
9383 deblock=filter=strong:block=4:alpha=0.12:beta=0.07:gamma=0.06:delta=0.05
9385 • Similar as above, but filter only first plane.
9387 deblock=filter=strong:block=4:alpha=0.12:beta=0.07:gamma=0.06:delta=0.05:planes=1
9389 • Similar as above, but filter only second and third plane.
9391 deblock=filter=strong:block=4:alpha=0.12:beta=0.07:gamma=0.06:delta=0.05:planes=6
9393 Commands
9395 This filter supports the all above options as commands.
9397 decimate
9399 Drop duplicated frames at regular intervals.
9400 The filter accepts the following options:
9402 cycle
9404 Set the number of frames from which one will be dropped. Setting
9405 this to N means one frame in every batch of N frames will be
9406 dropped. Default is 5.
9407 dupthresh
9409 Set the threshold for duplicate detection. If the difference metric
9410 for a frame is less than or equal to this value, then it is
9411 declared as duplicate. Default is 1.1
9412 scthresh
9414 Set scene change threshold. Default is 15.
9415 blockx
9417 blocky
9418 Set the size of the x and y-axis blocks used during metric
9419 calculations. Larger blocks give better noise suppression, but
9420 also give worse detection of small movements. Must be a power of
9421 two. Default is 32.
9422 ppsrc
9424 Mark main input as a pre-processed input and activate clean source
9425 input stream. This allows the input to be pre-processed with
9426 various filters to help the metrics calculation while keeping the
9427 frame selection lossless. When set to 1, the first stream is for
9428 the pre-processed input, and the second stream is the clean source
9429 from where the kept frames are chosen. Default is 0.
9430 chroma
9432 Set whether or not chroma is considered in the metric calculations.
9433 Default is 1.
9434 deconvolve
9436 Apply 2D deconvolution of video stream in frequency domain using second
9437 stream as impulse.
9438 The filter accepts the following options:
9440 planes
9442 Set which planes to process.
9443 impulse
9445 Set which impulse video frames will be processed, can be first or
9446 all. Default is all.
9447 noise
9449 Set noise when doing divisions. Default is 0.0000001. Useful when
9450 width and height are not same and not power of 2 or if stream prior
9451 to convolving had noise.
9452 The "deconvolve" filter also supports the framesync options.
9454 dedot
9456 Reduce cross-luminance (dot-crawl) and cross-color (rainbows) from
9457 video.
9458 It accepts the following options:
9460 m Set mode of operation. Can be combination of dotcrawl for cross-
9462 luminance reduction and/or rainbows for cross-color reduction.
9463 lt Set spatial luma threshold. Lower values increases reduction of
9465 cross-luminance.
9466 tl Set tolerance for temporal luma. Higher values increases reduction
9468 of cross-luminance.
9469 tc Set tolerance for chroma temporal variation. Higher values
9471 increases reduction of cross-color.
9472 ct Set temporal chroma threshold. Lower values increases reduction of
9474 cross-color.
9475 deflate
9477 Apply deflate effect to the video.
9478 This filter replaces the pixel by the local(3x3) average by taking into
9480 account only values lower than the pixel.
9481 It accepts the following options:
9483 threshold0
9485 threshold1
9486 threshold2
9487 threshold3
9488 Limit the maximum change for each plane, default is 65535. If 0,
9489 plane will remain unchanged.
9490 Commands
9492 This filter supports the all above options as commands.
9494 deflicker
9496 Remove temporal frame luminance variations.
9497 It accepts the following options:
9499 size, s
9501 Set moving-average filter size in frames. Default is 5. Allowed
9502 range is 2 - 129.
9503 mode, m
9505 Set averaging mode to smooth temporal luminance variations.
9506 Available values are:
9508 am Arithmetic mean
9510 gm Geometric mean
9512 hm Harmonic mean
9514 qm Quadratic mean
9516 cm Cubic mean
9518 pm Power mean
9520 median
9522 Median
9523 bypass
9525 Do not actually modify frame. Useful when one only wants metadata.
9526 dejudder
9528 Remove judder produced by partially interlaced telecined content.
9529 Judder can be introduced, for instance, by pullup filter. If the
9531 original source was partially telecined content then the output of
9532 "pullup,dejudder" will have a variable frame rate. May change the
9533 recorded frame rate of the container. Aside from that change, this
9534 filter will not affect constant frame rate video.
9535 The option available in this filter is:
9537 cycle
9539 Specify the length of the window over which the judder repeats.
9540 Accepts any integer greater than 1. Useful values are:
9542 4 If the original was telecined from 24 to 30 fps (Film to NTSC).
9544 5 If the original was telecined from 25 to 30 fps (PAL to NTSC).
9546 20 If a mixture of the two.
9548 The default is 4.
9550 delogo
9552 Suppress a TV station logo by a simple interpolation of the surrounding
9553 pixels. Just set a rectangle covering the logo and watch it disappear
9554 (and sometimes something even uglier appear - your mileage may vary).
9555 It accepts the following parameters:
9557 x
9559 y Specify the top left corner coordinates of the logo. They must be
9560 specified.
9561 w
9563 h Specify the width and height of the logo to clear. They must be
9564 specified.
9565 show
9567 When set to 1, a green rectangle is drawn on the screen to simplify
9568 finding the right x, y, w, and h parameters. The default value is
9569 0.
9570 The rectangle is drawn on the outermost pixels which will be
9572 (partly) replaced with interpolated values. The values of the next
9573 pixels immediately outside this rectangle in each direction will be
9574 used to compute the interpolated pixel values inside the rectangle.
9575 Examples
9577 • Set a rectangle covering the area with top left corner coordinates
9579 0,0 and size 100x77:
9580 delogo=x=0:y=0:w=100:h=77
9582 derain
9584 Remove the rain in the input image/video by applying the derain methods
9585 based on convolutional neural networks. Supported models:
9586 • Recurrent Squeeze-and-Excitation Context Aggregation Net (RESCAN).
9588 See
9589 <http://openaccess.thecvf.com/content_ECCV_2018/papers/Xia_Li_Recurrent_Squeeze-and-Excitation_Context_ECCV_2018_paper.pdf>.
9590 Training as well as model generation scripts are provided in the
9592 repository at <https://github.com/XueweiMeng/derain_filter.git>.
9593 Native model files (.model) can be generated from TensorFlow model
9595 files (.pb) by using tools/python/convert.py
9596 The filter accepts the following options:
9598 filter_type
9600 Specify which filter to use. This option accepts the following
9601 values:
9602 derain
9604 Derain filter. To conduct derain filter, you need to use a
9605 derain model.
9606 dehaze
9608 Dehaze filter. To conduct dehaze filter, you need to use a
9609 dehaze model.
9610 Default value is derain.
9612 dnn_backend
9614 Specify which DNN backend to use for model loading and execution.
9615 This option accepts the following values:
9616 native
9618 Native implementation of DNN loading and execution.
9619 tensorflow
9621 TensorFlow backend. To enable this backend you need to install
9622 the TensorFlow for C library (see
9623 <https://www.tensorflow.org/install/lang_c>) and configure
9624 FFmpeg with "--enable-libtensorflow"
9625 Default value is native.
9627 model
9629 Set path to model file specifying network architecture and its
9630 parameters. Note that different backends use different file
9631 formats. TensorFlow and native backend can load files for only its
9632 format.
9633 To get full functionality (such as async execution), please use the
9635 dnn_processing filter.
9636 deshake
9638 Attempt to fix small changes in horizontal and/or vertical shift. This
9639 filter helps remove camera shake from hand-holding a camera, bumping a
9640 tripod, moving on a vehicle, etc.
9641 The filter accepts the following options:
9643 x
9645 y
9646 w
9647 h Specify a rectangular area where to limit the search for motion
9648 vectors. If desired the search for motion vectors can be limited
9649 to a rectangular area of the frame defined by its top left corner,
9650 width and height. These parameters have the same meaning as the
9651 drawbox filter which can be used to visualise the position of the
9652 bounding box.
9653 This is useful when simultaneous movement of subjects within the
9655 frame might be confused for camera motion by the motion vector
9656 search.
9657 If any or all of x, y, w and h are set to -1 then the full frame is
9659 used. This allows later options to be set without specifying the
9660 bounding box for the motion vector search.
9661 Default - search the whole frame.
9663 rx
9665 ry Specify the maximum extent of movement in x and y directions in the
9666 range 0-64 pixels. Default 16.
9667 edge
9669 Specify how to generate pixels to fill blanks at the edge of the
9670 frame. Available values are:
9671 blank, 0
9673 Fill zeroes at blank locations
9674 original, 1
9676 Original image at blank locations
9677 clamp, 2
9679 Extruded edge value at blank locations
9680 mirror, 3
9682 Mirrored edge at blank locations
9683 Default value is mirror.
9685 blocksize
9687 Specify the blocksize to use for motion search. Range 4-128 pixels,
9688 default 8.
9689 contrast
9691 Specify the contrast threshold for blocks. Only blocks with more
9692 than the specified contrast (difference between darkest and
9693 lightest pixels) will be considered. Range 1-255, default 125.
9694 search
9696 Specify the search strategy. Available values are:
9697 exhaustive, 0
9699 Set exhaustive search
9700 less, 1
9702 Set less exhaustive search.
9703 Default value is exhaustive.
9705 filename
9707 If set then a detailed log of the motion search is written to the
9708 specified file.
9709 despill
9711 Remove unwanted contamination of foreground colors, caused by reflected
9712 color of greenscreen or bluescreen.
9713 This filter accepts the following options:
9715 type
9717 Set what type of despill to use.
9718 mix Set how spillmap will be generated.
9720 expand
9722 Set how much to get rid of still remaining spill.
9723 red Controls amount of red in spill area.
9725 green
9727 Controls amount of green in spill area. Should be -1 for
9728 greenscreen.
9729 blue
9731 Controls amount of blue in spill area. Should be -1 for
9732 bluescreen.
9733 brightness
9735 Controls brightness of spill area, preserving colors.
9736 alpha
9738 Modify alpha from generated spillmap.
9739 Commands
9741 This filter supports the all above options as commands.
9743 detelecine
9745 Apply an exact inverse of the telecine operation. It requires a
9746 predefined pattern specified using the pattern option which must be the
9747 same as that passed to the telecine filter.
9748 This filter accepts the following options:
9750 first_field
9752 top, t
9753 top field first
9754 bottom, b
9756 bottom field first The default value is "top".
9757 pattern
9759 A string of numbers representing the pulldown pattern you wish to
9760 apply. The default value is 23.
9761 start_frame
9763 A number representing position of the first frame with respect to
9764 the telecine pattern. This is to be used if the stream is cut. The
9765 default value is 0.
9766 dilation
9768 Apply dilation effect to the video.
9769 This filter replaces the pixel by the local(3x3) maximum.
9771 It accepts the following options:
9773 threshold0
9775 threshold1
9776 threshold2
9777 threshold3
9778 Limit the maximum change for each plane, default is 65535. If 0,
9779 plane will remain unchanged.
9780 coordinates
9782 Flag which specifies the pixel to refer to. Default is 255 i.e. all
9783 eight pixels are used.
9784 Flags to local 3x3 coordinates maps like this:
9786 1 2 3
9788 4 5
9789 6 7 8
9790 Commands
9792 This filter supports the all above options as commands.
9794 displace
9796 Displace pixels as indicated by second and third input stream.
9797 It takes three input streams and outputs one stream, the first input is
9799 the source, and second and third input are displacement maps.
9800 The second input specifies how much to displace pixels along the
9802 x-axis, while the third input specifies how much to displace pixels
9803 along the y-axis. If one of displacement map streams terminates, last
9804 frame from that displacement map will be used.
9805 Note that once generated, displacements maps can be reused over and
9807 over again.
9808 A description of the accepted options follows.
9810 edge
9812 Set displace behavior for pixels that are out of range.
9813 Available values are:
9815 blank
9817 Missing pixels are replaced by black pixels.
9818 smear
9820 Adjacent pixels will spread out to replace missing pixels.
9821 wrap
9823 Out of range pixels are wrapped so they point to pixels of
9824 other side.
9825 mirror
9827 Out of range pixels will be replaced with mirrored pixels.
9828 Default is smear.
9830 Examples
9832 • Add ripple effect to rgb input of video size hd720:
9834 ffmpeg -i INPUT -f lavfi -i nullsrc=s=hd720,lutrgb=128:128:128 -f lavfi -i nullsrc=s=hd720,geq='r=128+30*sin(2*PI*X/400+T):g=128+30*sin(2*PI*X/400+T):b=128+30*sin(2*PI*X/400+T)' -lavfi '[0][1][2]displace' OUTPUT
9836 • Add wave effect to rgb input of video size hd720:
9838 ffmpeg -i INPUT -f lavfi -i nullsrc=hd720,geq='r=128+80*(sin(sqrt((X-W/2)*(X-W/2)+(Y-H/2)*(Y-H/2))/220*2*PI+T)):g=128+80*(sin(sqrt((X-W/2)*(X-W/2)+(Y-H/2)*(Y-H/2))/220*2*PI+T)):b=128+80*(sin(sqrt((X-W/2)*(X-W/2)+(Y-H/2)*(Y-H/2))/220*2*PI+T))' -lavfi '[1]split[x][y],[0][x][y]displace' OUTPUT
9840 dnn_classify
9842 Do classification with deep neural networks based on bounding boxes.
9843 The filter accepts the following options:
9845 dnn_backend
9847 Specify which DNN backend to use for model loading and execution.
9848 This option accepts only openvino now, tensorflow backends will be
9849 added.
9850 model
9852 Set path to model file specifying network architecture and its
9853 parameters. Note that different backends use different file
9854 formats.
9855 input
9857 Set the input name of the dnn network.
9858 output
9860 Set the output name of the dnn network.
9861 confidence
9863 Set the confidence threshold (default: 0.5).
9864 labels
9866 Set path to label file specifying the mapping between label id and
9867 name. Each label name is written in one line, tailing spaces and
9868 empty lines are skipped. The first line is the name of label id 0,
9869 and the second line is the name of label id 1, etc. The label id
9870 is considered as name if the label file is not provided.
9871 backend_configs
9873 Set the configs to be passed into backend
9874 For tensorflow backend, you can set its configs with sess_config
9876 options, please use tools/python/tf_sess_config.py to get the
9877 configs for your system.
9878 dnn_detect
9880 Do object detection with deep neural networks.
9881 The filter accepts the following options:
9883 dnn_backend
9885 Specify which DNN backend to use for model loading and execution.
9886 This option accepts only openvino now, tensorflow backends will be
9887 added.
9888 model
9890 Set path to model file specifying network architecture and its
9891 parameters. Note that different backends use different file
9892 formats.
9893 input
9895 Set the input name of the dnn network.
9896 output
9898 Set the output name of the dnn network.
9899 confidence
9901 Set the confidence threshold (default: 0.5).
9902 labels
9904 Set path to label file specifying the mapping between label id and
9905 name. Each label name is written in one line, tailing spaces and
9906 empty lines are skipped. The first line is the name of label id 0
9907 (usually it is 'background'), and the second line is the name of
9908 label id 1, etc. The label id is considered as name if the label
9909 file is not provided.
9910 backend_configs
9912 Set the configs to be passed into backend. To use async execution,
9913 set async (default: set). Roll back to sync execution if the
9914 backend does not support async.
9915 dnn_processing
9917 Do image processing with deep neural networks. It works together with
9918 another filter which converts the pixel format of the Frame to what the
9919 dnn network requires.
9920 The filter accepts the following options:
9922 dnn_backend
9924 Specify which DNN backend to use for model loading and execution.
9925 This option accepts the following values:
9926 native
9928 Native implementation of DNN loading and execution.
9929 tensorflow
9931 TensorFlow backend. To enable this backend you need to install
9932 the TensorFlow for C library (see
9933 <https://www.tensorflow.org/install/lang_c>) and configure
9934 FFmpeg with "--enable-libtensorflow"
9935 openvino
9937 OpenVINO backend. To enable this backend you need to build and
9938 install the OpenVINO for C library (see
9939 <https://github.com/openvinotoolkit/openvino/blob/master/build-instruction.md>)
9940 and configure FFmpeg with "--enable-libopenvino"
9941 (--extra-cflags=-I... --extra-ldflags=-L... might be needed if
9942 the header files and libraries are not installed into system
9943 path)
9944 Default value is native.
9946 model
9948 Set path to model file specifying network architecture and its
9949 parameters. Note that different backends use different file
9950 formats. TensorFlow, OpenVINO and native backend can load files for
9951 only its format.
9952 Native model file (.model) can be generated from TensorFlow model
9954 file (.pb) by using tools/python/convert.py
9955 input
9957 Set the input name of the dnn network.
9958 output
9960 Set the output name of the dnn network.
9961 backend_configs
9963 Set the configs to be passed into backend. To use async execution,
9964 set async (default: set). Roll back to sync execution if the
9965 backend does not support async.
9966 For tensorflow backend, you can set its configs with sess_config
9968 options, please use tools/python/tf_sess_config.py to get the
9969 configs of TensorFlow backend for your system.
9970 Examples
9972 • Remove rain in rgb24 frame with can.pb (see derain filter):
9974 ./ffmpeg -i rain.jpg -vf format=rgb24,dnn_processing=dnn_backend=tensorflow:model=can.pb:input=x:output=y derain.jpg
9976 • Halve the pixel value of the frame with format gray32f:
9978 ffmpeg -i input.jpg -vf format=grayf32,dnn_processing=model=halve_gray_float.model:input=dnn_in:output=dnn_out:dnn_backend=native -y out.native.png
9980 • Handle the Y channel with srcnn.pb (see sr filter) for frame with
9982 yuv420p (planar YUV formats supported):
9983 ./ffmpeg -i 480p.jpg -vf format=yuv420p,scale=w=iw*2:h=ih*2,dnn_processing=dnn_backend=tensorflow:model=srcnn.pb:input=x:output=y -y srcnn.jpg
9985 • Handle the Y channel with espcn.pb (see sr filter), which changes
9987 frame size, for format yuv420p (planar YUV formats supported),
9988 please use tools/python/tf_sess_config.py to get the configs of
9989 TensorFlow backend for your system.
9990 ./ffmpeg -i 480p.jpg -vf format=yuv420p,dnn_processing=dnn_backend=tensorflow:model=espcn.pb:input=x:output=y:backend_configs=sess_config=0x10022805320e09cdccccccccccec3f20012a01303801 -y tmp.espcn.jpg
9992 drawbox
9994 Draw a colored box on the input image.
9995 It accepts the following parameters:
9997 x
9999 y The expressions which specify the top left corner coordinates of
10000 the box. It defaults to 0.
10001 width, w
10003 height, h
10004 The expressions which specify the width and height of the box; if 0
10005 they are interpreted as the input width and height. It defaults to
10006 0.
10007 color, c
10009 Specify the color of the box to write. For the general syntax of
10010 this option, check the "Color" section in the ffmpeg-utils manual.
10011 If the special value "invert" is used, the box edge color is the
10012 same as the video with inverted luma.
10013 thickness, t
10015 The expression which sets the thickness of the box edge. A value
10016 of "fill" will create a filled box. Default value is 3.
10017 See below for the list of accepted constants.
10019 replace
10021 Applicable if the input has alpha. With value 1, the pixels of the
10022 painted box will overwrite the video's color and alpha pixels.
10023 Default is 0, which composites the box onto the input, leaving the
10024 video's alpha intact.
10025 The parameters for x, y, w and h and t are expressions containing the
10027 following constants:
10028 dar The input display aspect ratio, it is the same as (w / h) * sar.
10030 hsub
10032 vsub
10033 horizontal and vertical chroma subsample values. For example for
10034 the pixel format "yuv422p" hsub is 2 and vsub is 1.
10035 in_h, ih
10037 in_w, iw
10038 The input width and height.
10039 sar The input sample aspect ratio.
10041 x
10043 y The x and y offset coordinates where the box is drawn.
10044 w
10046 h The width and height of the drawn box.
10047 box_source
10049 Box source can be set as side_data_detection_bboxes if you want to
10050 use box data in detection bboxes of side data.
10051 If box_source is set, the x, y, width and height will be ignored
10053 and still use box data in detection bboxes of side data. So please
10054 do not use this parameter if you were not sure about the box
10055 source.
10056 t The thickness of the drawn box.
10058 These constants allow the x, y, w, h and t expressions to refer to
10060 each other, so you may for example specify "y=x/dar" or "h=w/dar".
10061 Examples
10063 • Draw a black box around the edge of the input image:
10065 drawbox
10067 • Draw a box with color red and an opacity of 50%:
10069 drawbox=10:20:200:60:red@0.5
10071 The previous example can be specified as:
10073 drawbox=x=10:y=20:w=200:h=60:color=red@0.5
10075 • Fill the box with pink color:
10077 drawbox=x=10:y=10:w=100:h=100:color=pink@0.5:t=fill
10079 • Draw a 2-pixel red 2.40:1 mask:
10081 drawbox=x=-t:y=0.5*(ih-iw/2.4)-t:w=iw+t*2:h=iw/2.4+t*2:t=2:c=red
10083 Commands
10085 This filter supports same commands as options. The command accepts the
10087 same syntax of the corresponding option.
10088 If the specified expression is not valid, it is kept at its current
10090 value.
10091 drawgraph
10093 Draw a graph using input video metadata.
10094 It accepts the following parameters:
10096 m1 Set 1st frame metadata key from which metadata values will be used
10098 to draw a graph.
10099 fg1 Set 1st foreground color expression.
10101 m2 Set 2nd frame metadata key from which metadata values will be used
10103 to draw a graph.
10104 fg2 Set 2nd foreground color expression.
10106 m3 Set 3rd frame metadata key from which metadata values will be used
10108 to draw a graph.
10109 fg3 Set 3rd foreground color expression.
10111 m4 Set 4th frame metadata key from which metadata values will be used
10113 to draw a graph.
10114 fg4 Set 4th foreground color expression.
10116 min Set minimal value of metadata value.
10118 max Set maximal value of metadata value.
10120 bg Set graph background color. Default is white.
10122 mode
10124 Set graph mode.
10125 Available values for mode is:
10127 bar
10129 dot
10130 line
10131 Default is "line".
10133 slide
10135 Set slide mode.
10136 Available values for slide is:
10138 frame
10140 Draw new frame when right border is reached.
10141 replace
10143 Replace old columns with new ones.
10144 scroll
10146 Scroll from right to left.
10147 rscroll
10149 Scroll from left to right.
10150 picture
10152 Draw single picture.
10153 Default is "frame".
10155 size
10157 Set size of graph video. For the syntax of this option, check the
10158 "Video size" section in the ffmpeg-utils manual. The default value
10159 is "900x256".
10160 rate, r
10162 Set the output frame rate. Default value is 25.
10163 The foreground color expressions can use the following variables:
10165 MIN Minimal value of metadata value.
10167 MAX Maximal value of metadata value.
10169 VAL Current metadata key value.
10171 The color is defined as 0xAABBGGRR.
10173 Example using metadata from signalstats filter:
10175 signalstats,drawgraph=lavfi.signalstats.YAVG:min=0:max=255
10177 Example using metadata from ebur128 filter:
10179 ebur128=metadata=1,adrawgraph=lavfi.r128.M:min=-120:max=5
10181 drawgrid
10183 Draw a grid on the input image.
10184 It accepts the following parameters:
10186 x
10188 y The expressions which specify the coordinates of some point of grid
10189 intersection (meant to configure offset). Both default to 0.
10190 width, w
10192 height, h
10193 The expressions which specify the width and height of the grid
10194 cell, if 0 they are interpreted as the input width and height,
10195 respectively, minus "thickness", so image gets framed. Default to
10196 0.
10197 color, c
10199 Specify the color of the grid. For the general syntax of this
10200 option, check the "Color" section in the ffmpeg-utils manual. If
10201 the special value "invert" is used, the grid color is the same as
10202 the video with inverted luma.
10203 thickness, t
10205 The expression which sets the thickness of the grid line. Default
10206 value is 1.
10207 See below for the list of accepted constants.
10209 replace
10211 Applicable if the input has alpha. With 1 the pixels of the painted
10212 grid will overwrite the video's color and alpha pixels. Default is
10213 0, which composites the grid onto the input, leaving the video's
10214 alpha intact.
10215 The parameters for x, y, w and h and t are expressions containing the
10217 following constants:
10218 dar The input display aspect ratio, it is the same as (w / h) * sar.
10220 hsub
10222 vsub
10223 horizontal and vertical chroma subsample values. For example for
10224 the pixel format "yuv422p" hsub is 2 and vsub is 1.
10225 in_h, ih
10227 in_w, iw
10228 The input grid cell width and height.
10229 sar The input sample aspect ratio.
10231 x
10233 y The x and y coordinates of some point of grid intersection (meant
10234 to configure offset).
10235 w
10237 h The width and height of the drawn cell.
10238 t The thickness of the drawn cell.
10240 These constants allow the x, y, w, h and t expressions to refer to
10242 each other, so you may for example specify "y=x/dar" or "h=w/dar".
10243 Examples
10245 • Draw a grid with cell 100x100 pixels, thickness 2 pixels, with
10247 color red and an opacity of 50%:
10248 drawgrid=width=100:height=100:thickness=2:color=red@0.5
10250 • Draw a white 3x3 grid with an opacity of 50%:
10252 drawgrid=w=iw/3:h=ih/3:t=2:c=white@0.5
10254 Commands
10256 This filter supports same commands as options. The command accepts the
10258 same syntax of the corresponding option.
10259 If the specified expression is not valid, it is kept at its current
10261 value.
10262 drawtext
10264 Draw a text string or text from a specified file on top of a video,
10265 using the libfreetype library.
10266 To enable compilation of this filter, you need to configure FFmpeg with
10268 "--enable-libfreetype". To enable default font fallback and the font
10269 option you need to configure FFmpeg with "--enable-libfontconfig". To
10270 enable the text_shaping option, you need to configure FFmpeg with
10271 "--enable-libfribidi".
10272 Syntax
10274 It accepts the following parameters:
10276 box Used to draw a box around text using the background color. The
10278 value must be either 1 (enable) or 0 (disable). The default value
10279 of box is 0.
10280 boxborderw
10282 Set the width of the border to be drawn around the box using
10283 boxcolor. The default value of boxborderw is 0.
10284 boxcolor
10286 The color to be used for drawing box around text. For the syntax of
10287 this option, check the "Color" section in the ffmpeg-utils manual.
10288 The default value of boxcolor is "white".
10290 line_spacing
10292 Set the line spacing in pixels of the border to be drawn around the
10293 box using box. The default value of line_spacing is 0.
10294 borderw
10296 Set the width of the border to be drawn around the text using
10297 bordercolor. The default value of borderw is 0.
10298 bordercolor
10300 Set the color to be used for drawing border around text. For the
10301 syntax of this option, check the "Color" section in the ffmpeg-
10302 utils manual.
10303 The default value of bordercolor is "black".
10305 expansion
10307 Select how the text is expanded. Can be either "none", "strftime"
10308 (deprecated) or "normal" (default). See the drawtext_expansion,
10309 Text expansion section below for details.
10310 basetime
10312 Set a start time for the count. Value is in microseconds. Only
10313 applied in the deprecated strftime expansion mode. To emulate in
10314 normal expansion mode use the "pts" function, supplying the start
10315 time (in seconds) as the second argument.
10316 fix_bounds
10318 If true, check and fix text coords to avoid clipping.
10319 fontcolor
10321 The color to be used for drawing fonts. For the syntax of this
10322 option, check the "Color" section in the ffmpeg-utils manual.
10323 The default value of fontcolor is "black".
10325 fontcolor_expr
10327 String which is expanded the same way as text to obtain dynamic
10328 fontcolor value. By default this option has empty value and is not
10329 processed. When this option is set, it overrides fontcolor option.
10330 font
10332 The font family to be used for drawing text. By default Sans.
10333 fontfile
10335 The font file to be used for drawing text. The path must be
10336 included. This parameter is mandatory if the fontconfig support is
10337 disabled.
10338 alpha
10340 Draw the text applying alpha blending. The value can be a number
10341 between 0.0 and 1.0. The expression accepts the same variables x,
10342 y as well. The default value is 1. Please see fontcolor_expr.
10343 fontsize
10345 The font size to be used for drawing text. The default value of
10346 fontsize is 16.
10347 text_shaping
10349 If set to 1, attempt to shape the text (for example, reverse the
10350 order of right-to-left text and join Arabic characters) before
10351 drawing it. Otherwise, just draw the text exactly as given. By
10352 default 1 (if supported).
10353 ft_load_flags
10355 The flags to be used for loading the fonts.
10356 The flags map the corresponding flags supported by libfreetype, and
10358 are a combination of the following values:
10359 default
10361 no_scale
10362 no_hinting
10363 render
10364 no_bitmap
10365 vertical_layout
10366 force_autohint
10367 crop_bitmap
10368 pedantic
10369 ignore_global_advance_width
10370 no_recurse
10371 ignore_transform
10372 monochrome
10373 linear_design
10374 no_autohint
10375 Default value is "default".
10377 For more information consult the documentation for the FT_LOAD_*
10379 libfreetype flags.
10380 shadowcolor
10382 The color to be used for drawing a shadow behind the drawn text.
10383 For the syntax of this option, check the "Color" section in the
10384 ffmpeg-utils manual.
10385 The default value of shadowcolor is "black".
10387 shadowx
10389 shadowy
10390 The x and y offsets for the text shadow position with respect to
10391 the position of the text. They can be either positive or negative
10392 values. The default value for both is "0".
10393 start_number
10395 The starting frame number for the n/frame_num variable. The default
10396 value is "0".
10397 tabsize
10399 The size in number of spaces to use for rendering the tab. Default
10400 value is 4.
10401 timecode
10403 Set the initial timecode representation in "hh:mm:ss[:;.]ff"
10404 format. It can be used with or without text parameter.
10405 timecode_rate option must be specified.
10406 timecode_rate, rate, r
10408 Set the timecode frame rate (timecode only). Value will be rounded
10409 to nearest integer. Minimum value is "1". Drop-frame timecode is
10410 supported for frame rates 30 & 60.
10411 tc24hmax
10413 If set to 1, the output of the timecode option will wrap around at
10414 24 hours. Default is 0 (disabled).
10415 text
10417 The text string to be drawn. The text must be a sequence of UTF-8
10418 encoded characters. This parameter is mandatory if no file is
10419 specified with the parameter textfile.
10420 textfile
10422 A text file containing text to be drawn. The text must be a
10423 sequence of UTF-8 encoded characters.
10424 This parameter is mandatory if no text string is specified with the
10426 parameter text.
10427 If both text and textfile are specified, an error is thrown.
10429 text_source
10431 Text source should be set as side_data_detection_bboxes if you want
10432 to use text data in detection bboxes of side data.
10433 If text source is set, text and textfile will be ignored and still
10435 use text data in detection bboxes of side data. So please do not
10436 use this parameter if you are not sure about the text source.
10437 reload
10439 The textfile will be reloaded at specified frame interval. Be sure
10440 to update textfile atomically, or it may be read partially, or even
10441 fail. Range is 0 to INT_MAX. Default is 0.
10442 x
10444 y The expressions which specify the offsets where text will be drawn
10445 within the video frame. They are relative to the top/left border of
10446 the output image.
10447 The default value of x and y is "0".
10449 See below for the list of accepted constants and functions.
10451 The parameters for x and y are expressions containing the following
10453 constants and functions:
10454 dar input display aspect ratio, it is the same as (w / h) * sar
10456 hsub
10458 vsub
10459 horizontal and vertical chroma subsample values. For example for
10460 the pixel format "yuv422p" hsub is 2 and vsub is 1.
10461 line_h, lh
10463 the height of each text line
10464 main_h, h, H
10466 the input height
10467 main_w, w, W
10469 the input width
10470 max_glyph_a, ascent
10472 the maximum distance from the baseline to the highest/upper grid
10473 coordinate used to place a glyph outline point, for all the
10474 rendered glyphs. It is a positive value, due to the grid's
10475 orientation with the Y axis upwards.
10476 max_glyph_d, descent
10478 the maximum distance from the baseline to the lowest grid
10479 coordinate used to place a glyph outline point, for all the
10480 rendered glyphs. This is a negative value, due to the grid's
10481 orientation, with the Y axis upwards.
10482 max_glyph_h
10484 maximum glyph height, that is the maximum height for all the glyphs
10485 contained in the rendered text, it is equivalent to ascent -
10486 descent.
10487 max_glyph_w
10489 maximum glyph width, that is the maximum width for all the glyphs
10490 contained in the rendered text
10491 n the number of input frame, starting from 0
10493 rand(min, max)
10495 return a random number included between min and max
10496 sar The input sample aspect ratio.
10498 t timestamp expressed in seconds, NAN if the input timestamp is
10500 unknown
10501 text_h, th
10503 the height of the rendered text
10504 text_w, tw
10506 the width of the rendered text
10507 x
10509 y the x and y offset coordinates where the text is drawn.
10510 These parameters allow the x and y expressions to refer to each
10512 other, so you can for example specify "y=x/dar".
10513 pict_type
10515 A one character description of the current frame's picture type.
10516 pkt_pos
10518 The current packet's position in the input file or stream (in
10519 bytes, from the start of the input). A value of -1 indicates this
10520 info is not available.
10521 pkt_duration
10523 The current packet's duration, in seconds.
10524 pkt_size
10526 The current packet's size (in bytes).
10527 Text expansion
10529 If expansion is set to "strftime", the filter recognizes strftime()
10531 sequences in the provided text and expands them accordingly. Check the
10532 documentation of strftime(). This feature is deprecated.
10533 If expansion is set to "none", the text is printed verbatim.
10535 If expansion is set to "normal" (which is the default), the following
10537 expansion mechanism is used.
10538 The backslash character \, followed by any character, always expands to
10540 the second character.
10541 Sequences of the form "%{...}" are expanded. The text between the
10543 braces is a function name, possibly followed by arguments separated by
10544 ':'. If the arguments contain special characters or delimiters (':' or
10545 '}'), they should be escaped.
10546 Note that they probably must also be escaped as the value for the text
10548 option in the filter argument string and as the filter argument in the
10549 filtergraph description, and possibly also for the shell, that makes up
10550 to four levels of escaping; using a text file avoids these problems.
10551 The following functions are available:
10553 expr, e
10555 The expression evaluation result.
10556 It must take one argument specifying the expression to be
10558 evaluated, which accepts the same constants and functions as the x
10559 and y values. Note that not all constants should be used, for
10560 example the text size is not known when evaluating the expression,
10561 so the constants text_w and text_h will have an undefined value.
10562 expr_int_format, eif
10564 Evaluate the expression's value and output as formatted integer.
10565 The first argument is the expression to be evaluated, just as for
10567 the expr function. The second argument specifies the output
10568 format. Allowed values are x, X, d and u. They are treated exactly
10569 as in the "printf" function. The third parameter is optional and
10570 sets the number of positions taken by the output. It can be used
10571 to add padding with zeros from the left.
10572 gmtime
10574 The time at which the filter is running, expressed in UTC. It can
10575 accept an argument: a strftime() format string. The format string
10576 is extended to support the variable %[1-6]N which prints fractions
10577 of the second with optionally specified number of digits.
10578 localtime
10580 The time at which the filter is running, expressed in the local
10581 time zone. It can accept an argument: a strftime() format string.
10582 The format string is extended to support the variable %[1-6]N which
10583 prints fractions of the second with optionally specified number of
10584 digits.
10585 metadata
10587 Frame metadata. Takes one or two arguments.
10588 The first argument is mandatory and specifies the metadata key.
10590 The second argument is optional and specifies a default value, used
10592 when the metadata key is not found or empty.
10593 Available metadata can be identified by inspecting entries starting
10595 with TAG included within each frame section printed by running
10596 "ffprobe -show_frames".
10597 String metadata generated in filters leading to the drawtext filter
10599 are also available.
10600 n, frame_num
10602 The frame number, starting from 0.
10603 pict_type
10605 A one character description of the current picture type.
10606 pts The timestamp of the current frame. It can take up to three
10608 arguments.
10609 The first argument is the format of the timestamp; it defaults to
10611 "flt" for seconds as a decimal number with microsecond accuracy;
10612 "hms" stands for a formatted [-]HH:MM:SS.mmm timestamp with
10613 millisecond accuracy. "gmtime" stands for the timestamp of the
10614 frame formatted as UTC time; "localtime" stands for the timestamp
10615 of the frame formatted as local time zone time.
10616 The second argument is an offset added to the timestamp.
10618 If the format is set to "hms", a third argument "24HH" may be
10620 supplied to present the hour part of the formatted timestamp in 24h
10621 format (00-23).
10622 If the format is set to "localtime" or "gmtime", a third argument
10624 may be supplied: a strftime() format string. By default, YYYY-MM-
10625 DD HH:MM:SS format will be used.
10626 Commands
10628 This filter supports altering parameters via commands:
10630 reinit
10632 Alter existing filter parameters.
10633 Syntax for the argument is the same as for filter invocation, e.g.
10635 fontsize=56:fontcolor=green:text='Hello World'
10637 Full filter invocation with sendcmd would look like this:
10639 sendcmd=c='56.0 drawtext reinit fontsize=56\:fontcolor=green\:text=Hello\\ World'
10641 If the entire argument can't be parsed or applied as valid values then
10643 the filter will continue with its existing parameters.
10644 Examples
10646 • Draw "Test Text" with font FreeSerif, using the default values for
10648 the optional parameters.
10649 drawtext="fontfile=/usr/share/fonts/truetype/freefont/FreeSerif.ttf: text='Test Text'"
10651 • Draw 'Test Text' with font FreeSerif of size 24 at position x=100
10653 and y=50 (counting from the top-left corner of the screen), text is
10654 yellow with a red box around it. Both the text and the box have an
10655 opacity of 20%.
10656 drawtext="fontfile=/usr/share/fonts/truetype/freefont/FreeSerif.ttf: text='Test Text':\
10658 x=100: y=50: fontsize=24: fontcolor=yellow@0.2: box=1: boxcolor=red@0.2"
10659 Note that the double quotes are not necessary if spaces are not
10661 used within the parameter list.
10662 • Show the text at the center of the video frame:
10664 drawtext="fontsize=30:fontfile=FreeSerif.ttf:text='hello world':x=(w-text_w)/2:y=(h-text_h)/2"
10666 • Show the text at a random position, switching to a new position
10668 every 30 seconds:
10669 drawtext="fontsize=30:fontfile=FreeSerif.ttf:text='hello world':x=if(eq(mod(t\,30)\,0)\,rand(0\,(w-text_w))\,x):y=if(eq(mod(t\,30)\,0)\,rand(0\,(h-text_h))\,y)"
10671 • Show a text line sliding from right to left in the last row of the
10673 video frame. The file LONG_LINE is assumed to contain a single line
10674 with no newlines.
10675 drawtext="fontsize=15:fontfile=FreeSerif.ttf:text=LONG_LINE:y=h-line_h:x=-50*t"
10677 • Show the content of file CREDITS off the bottom of the frame and
10679 scroll up.
10680 drawtext="fontsize=20:fontfile=FreeSerif.ttf:textfile=CREDITS:y=h-20*t"
10682 • Draw a single green letter "g", at the center of the input video.
10684 The glyph baseline is placed at half screen height.
10685 drawtext="fontsize=60:fontfile=FreeSerif.ttf:fontcolor=green:text=g:x=(w-max_glyph_w)/2:y=h/2-ascent"
10687 • Show text for 1 second every 3 seconds:
10689 drawtext="fontfile=FreeSerif.ttf:fontcolor=white:x=100:y=x/dar:enable=lt(mod(t\,3)\,1):text='blink'"
10691 • Use fontconfig to set the font. Note that the colons need to be
10693 escaped.
10694 drawtext='fontfile=Linux Libertine O-40\:style=Semibold:text=FFmpeg'
10696 • Draw "Test Text" with font size dependent on height of the video.
10698 drawtext="text='Test Text': fontsize=h/30: x=(w-text_w)/2: y=(h-text_h*2)"
10700 • Print the date of a real-time encoding (see strftime(3)):
10702 drawtext='fontfile=FreeSans.ttf:text=%{localtime\:%a %b %d %Y}'
10704 • Show text fading in and out (appearing/disappearing):
10706 #!/bin/sh
10708 DS=1.0 # display start
10709 DE=10.0 # display end
10710 FID=1.5 # fade in duration
10711 FOD=5 # fade out duration
10712 ffplay -f lavfi "color,drawtext=text=TEST:fontsize=50:fontfile=FreeSerif.ttf:fontcolor_expr=ff0000%{eif\\\\: clip(255*(1*between(t\\, $DS + $FID\\, $DE - $FOD) + ((t - $DS)/$FID)*between(t\\, $DS\\, $DS + $FID) + (-(t - $DE)/$FOD)*between(t\\, $DE - $FOD\\, $DE) )\\, 0\\, 255) \\\\: x\\\\: 2 }"
10713 • Horizontally align multiple separate texts. Note that max_glyph_a
10715 and the fontsize value are included in the y offset.
10716 drawtext=fontfile=FreeSans.ttf:text=DOG:fontsize=24:x=10:y=20+24-max_glyph_a,
10718 drawtext=fontfile=FreeSans.ttf:text=cow:fontsize=24:x=80:y=20+24-max_glyph_a
10719 • Plot special lavf.image2dec.source_basename metadata onto each
10721 frame if such metadata exists. Otherwise, plot the string "NA".
10722 Note that image2 demuxer must have option -export_path_metadata 1
10723 for the special metadata fields to be available for filters.
10724 drawtext="fontsize=20:fontcolor=white:fontfile=FreeSans.ttf:text='%{metadata\:lavf.image2dec.source_basename\:NA}':x=10:y=10"
10726 For more information about libfreetype, check:
10728 <http://www.freetype.org/>.
10729 For more information about fontconfig, check:
10731 <http://freedesktop.org/software/fontconfig/fontconfig-user.html>.
10732 For more information about libfribidi, check: <http://fribidi.org/>.
10734 edgedetect
10736 Detect and draw edges. The filter uses the Canny Edge Detection
10737 algorithm.
10738 The filter accepts the following options:
10740 low
10742 high
10743 Set low and high threshold values used by the Canny thresholding
10744 algorithm.
10745 The high threshold selects the "strong" edge pixels, which are then
10747 connected through 8-connectivity with the "weak" edge pixels
10748 selected by the low threshold.
10749 low and high threshold values must be chosen in the range [0,1],
10751 and low should be lesser or equal to high.
10752 Default value for low is "20/255", and default value for high is
10754 "50/255".
10755 mode
10757 Define the drawing mode.
10758 wires
10760 Draw white/gray wires on black background.
10761 colormix
10763 Mix the colors to create a paint/cartoon effect.
10764 canny
10766 Apply Canny edge detector on all selected planes.
10767 Default value is wires.
10769 planes
10771 Select planes for filtering. By default all available planes are
10772 filtered.
10773 Examples
10775 • Standard edge detection with custom values for the hysteresis
10777 thresholding:
10778 edgedetect=low=0.1:high=0.4
10780 • Painting effect without thresholding:
10782 edgedetect=mode=colormix:high=0
10784 elbg
10786 Apply a posterize effect using the ELBG (Enhanced LBG) algorithm.
10787 For each input image, the filter will compute the optimal mapping from
10789 the input to the output given the codebook length, that is the number
10790 of distinct output colors.
10791 This filter accepts the following options.
10793 codebook_length, l
10795 Set codebook length. The value must be a positive integer, and
10796 represents the number of distinct output colors. Default value is
10797 256.
10798 nb_steps, n
10800 Set the maximum number of iterations to apply for computing the
10801 optimal mapping. The higher the value the better the result and the
10802 higher the computation time. Default value is 1.
10803 seed, s
10805 Set a random seed, must be an integer included between 0 and
10806 UINT32_MAX. If not specified, or if explicitly set to -1, the
10807 filter will try to use a good random seed on a best effort basis.
10808 pal8
10810 Set pal8 output pixel format. This option does not work with
10811 codebook length greater than 256. Default is disabled.
10812 use_alpha
10814 Include alpha values in the quantization calculation. Allows
10815 creating palettized output images (e.g. PNG8) with multiple alpha
10816 smooth blending.
10817 entropy
10819 Measure graylevel entropy in histogram of color channels of video
10820 frames.
10821 It accepts the following parameters:
10823 mode
10825 Can be either normal or diff. Default is normal.
10826 diff mode measures entropy of histogram delta values, absolute
10828 differences between neighbour histogram values.
10829 epx
10831 Apply the EPX magnification filter which is designed for pixel art.
10832 It accepts the following option:
10834 n Set the scaling dimension: 2 for "2xEPX", 3 for "3xEPX". Default
10836 is 3.
10837 eq
10839 Set brightness, contrast, saturation and approximate gamma adjustment.
10840 The filter accepts the following options:
10842 contrast
10844 Set the contrast expression. The value must be a float value in
10845 range "-1000.0" to 1000.0. The default value is "1".
10846 brightness
10848 Set the brightness expression. The value must be a float value in
10849 range "-1.0" to 1.0. The default value is "0".
10850 saturation
10852 Set the saturation expression. The value must be a float in range
10853 0.0 to 3.0. The default value is "1".
10854 gamma
10856 Set the gamma expression. The value must be a float in range 0.1 to
10857 10.0. The default value is "1".
10858 gamma_r
10860 Set the gamma expression for red. The value must be a float in
10861 range 0.1 to 10.0. The default value is "1".
10862 gamma_g
10864 Set the gamma expression for green. The value must be a float in
10865 range 0.1 to 10.0. The default value is "1".
10866 gamma_b
10868 Set the gamma expression for blue. The value must be a float in
10869 range 0.1 to 10.0. The default value is "1".
10870 gamma_weight
10872 Set the gamma weight expression. It can be used to reduce the
10873 effect of a high gamma value on bright image areas, e.g. keep them
10874 from getting overamplified and just plain white. The value must be
10875 a float in range 0.0 to 1.0. A value of 0.0 turns the gamma
10876 correction all the way down while 1.0 leaves it at its full
10877 strength. Default is "1".
10878 eval
10880 Set when the expressions for brightness, contrast, saturation and
10881 gamma expressions are evaluated.
10882 It accepts the following values:
10884 init
10886 only evaluate expressions once during the filter initialization
10887 or when a command is processed
10888 frame
10890 evaluate expressions for each incoming frame
10891 Default value is init.
10893 The expressions accept the following parameters:
10895 n frame count of the input frame starting from 0
10897 pos byte position of the corresponding packet in the input file, NAN if
10899 unspecified
10900 r frame rate of the input video, NAN if the input frame rate is
10902 unknown
10903 t timestamp expressed in seconds, NAN if the input timestamp is
10905 unknown
10906 Commands
10908 The filter supports the following commands:
10910 contrast
10912 Set the contrast expression.
10913 brightness
10915 Set the brightness expression.
10916 saturation
10918 Set the saturation expression.
10919 gamma
10921 Set the gamma expression.
10922 gamma_r
10924 Set the gamma_r expression.
10925 gamma_g
10927 Set gamma_g expression.
10928 gamma_b
10930 Set gamma_b expression.
10931 gamma_weight
10933 Set gamma_weight expression.
10934 The command accepts the same syntax of the corresponding option.
10936 If the specified expression is not valid, it is kept at its current
10938 value.
10939 erosion
10941 Apply erosion effect to the video.
10942 This filter replaces the pixel by the local(3x3) minimum.
10944 It accepts the following options:
10946 threshold0
10948 threshold1
10949 threshold2
10950 threshold3
10951 Limit the maximum change for each plane, default is 65535. If 0,
10952 plane will remain unchanged.
10953 coordinates
10955 Flag which specifies the pixel to refer to. Default is 255 i.e. all
10956 eight pixels are used.
10957 Flags to local 3x3 coordinates maps like this:
10959 1 2 3
10961 4 5
10962 6 7 8
10963 Commands
10965 This filter supports the all above options as commands.
10967 estdif
10969 Deinterlace the input video ("estdif" stands for "Edge Slope Tracing
10970 Deinterlacing Filter").
10971 Spatial only filter that uses edge slope tracing algorithm to
10973 interpolate missing lines. It accepts the following parameters:
10974 mode
10976 The interlacing mode to adopt. It accepts one of the following
10977 values:
10978 frame
10980 Output one frame for each frame.
10981 field
10983 Output one frame for each field.
10984 The default value is "field".
10986 parity
10988 The picture field parity assumed for the input interlaced video. It
10989 accepts one of the following values:
10990 tff Assume the top field is first.
10992 bff Assume the bottom field is first.
10994 auto
10996 Enable automatic detection of field parity.
10997 The default value is "auto". If the interlacing is unknown or the
10999 decoder does not export this information, top field first will be
11000 assumed.
11001 deint
11003 Specify which frames to deinterlace. Accepts one of the following
11004 values:
11005 all Deinterlace all frames.
11007 interlaced
11009 Only deinterlace frames marked as interlaced.
11010 The default value is "all".
11012 rslope
11014 Specify the search radius for edge slope tracing. Default value is
11015 1. Allowed range is from 1 to 15.
11016 redge
11018 Specify the search radius for best edge matching. Default value is
11019 2. Allowed range is from 0 to 15.
11020 ecost
11022 Specify the edge cost for edge matching. Default value is 1.0.
11023 Allowed range is from 0 to 9.
11024 mcost
11026 Specify the middle cost for edge matching. Default value is 0.5.
11027 Allowed range is from 0 to 1.
11028 dcost
11030 Specify the distance cost for edge matching. Default value is 0.5.
11031 Allowed range is from 0 to 1.
11032 interp
11034 Specify the interpolation used. Default is 4-point interpolation.
11035 It accepts one of the following values:
11036 2p Two-point interpolation.
11038 4p Four-point interpolation.
11040 6p Six-point interpolation.
11042 Commands
11044 This filter supports same commands as options.
11046 exposure
11048 Adjust exposure of the video stream.
11049 The filter accepts the following options:
11051 exposure
11053 Set the exposure correction in EV. Allowed range is from -3.0 to
11054 3.0 EV Default value is 0 EV.
11055 black
11057 Set the black level correction. Allowed range is from -1.0 to 1.0.
11058 Default value is 0.
11059 Commands
11061 This filter supports same commands as options.
11063 extractplanes
11065 Extract color channel components from input video stream into separate
11066 grayscale video streams.
11067 The filter accepts the following option:
11069 planes
11071 Set plane(s) to extract.
11072 Available values for planes are:
11074 y
11076 u
11077 v
11078 a
11079 r
11080 g
11081 b
11082 Choosing planes not available in the input will result in an error.
11084 That means you cannot select "r", "g", "b" planes with "y", "u",
11085 "v" planes at same time.
11086 Examples
11088 • Extract luma, u and v color channel component from input video
11090 frame into 3 grayscale outputs:
11091 ffmpeg -i video.avi -filter_complex 'extractplanes=y+u+v[y][u][v]' -map '[y]' y.avi -map '[u]' u.avi -map '[v]' v.avi
11093 fade
11095 Apply a fade-in/out effect to the input video.
11096 It accepts the following parameters:
11098 type, t
11100 The effect type can be either "in" for a fade-in, or "out" for a
11101 fade-out effect. Default is "in".
11102 start_frame, s
11104 Specify the number of the frame to start applying the fade effect
11105 at. Default is 0.
11106 nb_frames, n
11108 The number of frames that the fade effect lasts. At the end of the
11109 fade-in effect, the output video will have the same intensity as
11110 the input video. At the end of the fade-out transition, the output
11111 video will be filled with the selected color. Default is 25.
11112 alpha
11114 If set to 1, fade only alpha channel, if one exists on the input.
11115 Default value is 0.
11116 start_time, st
11118 Specify the timestamp (in seconds) of the frame to start to apply
11119 the fade effect. If both start_frame and start_time are specified,
11120 the fade will start at whichever comes last. Default is 0.
11121 duration, d
11123 The number of seconds for which the fade effect has to last. At the
11124 end of the fade-in effect the output video will have the same
11125 intensity as the input video, at the end of the fade-out transition
11126 the output video will be filled with the selected color. If both
11127 duration and nb_frames are specified, duration is used. Default is
11128 0 (nb_frames is used by default).
11129 color, c
11131 Specify the color of the fade. Default is "black".
11132 Examples
11134 • Fade in the first 30 frames of video:
11136 fade=in:0:30
11138 The command above is equivalent to:
11140 fade=t=in:s=0:n=30
11142 • Fade out the last 45 frames of a 200-frame video:
11144 fade=out:155:45
11146 fade=type=out:start_frame=155:nb_frames=45
11147 • Fade in the first 25 frames and fade out the last 25 frames of a
11149 1000-frame video:
11150 fade=in:0:25, fade=out:975:25
11152 • Make the first 5 frames yellow, then fade in from frame 5-24:
11154 fade=in:5:20:color=yellow
11156 • Fade in alpha over first 25 frames of video:
11158 fade=in:0:25:alpha=1
11160 • Make the first 5.5 seconds black, then fade in for 0.5 seconds:
11162 fade=t=in:st=5.5:d=0.5
11164 feedback
11166 Apply feedback video filter.
11167 This filter pass cropped input frames to 2nd output. From there it can
11169 be filtered with other video filters. After filter receives frame from
11170 2nd input, that frame is combined on top of original frame from 1st
11171 input and passed to 1st output.
11172 The typical usage is filter only part of frame.
11174 The filter accepts the following options:
11176 x
11178 y Set the top left crop position.
11179 w
11181 h Set the crop size.
11182 Examples
11184 • Blur only top left rectangular part of video frame size 100x100
11186 with gblur filter.
11187 [in][blurin]feedback=x=0:y=0:w=100:h=100[out][blurout];[blurout]gblur=8[blurin]
11189 • Draw black box on top left part of video frame of size 100x100 with
11191 drawbox filter.
11192 [in][blurin]feedback=x=0:y=0:w=100:h=100[out][blurout];[blurout]drawbox=x=0:y=0:w=100:h=100:t=100[blurin]
11194 fftdnoiz
11196 Denoise frames using 3D FFT (frequency domain filtering).
11197 The filter accepts the following options:
11199 sigma
11201 Set the noise sigma constant. This sets denoising strength.
11202 Default value is 1. Allowed range is from 0 to 30. Using very high
11203 sigma with low overlap may give blocking artifacts.
11204 amount
11206 Set amount of denoising. By default all detected noise is reduced.
11207 Default value is 1. Allowed range is from 0 to 1.
11208 block
11210 Set size of block in pixels, Default is 32, can be 8 to 256.
11211 overlap
11213 Set block overlap. Default is 0.5. Allowed range is from 0.2 to
11214 0.8.
11215 method
11217 Set denoising method. Default is "wiener", can also be "hard".
11218 prev
11220 Set number of previous frames to use for denoising. By default is
11221 set to 0.
11222 next
11224 Set number of next frames to to use for denoising. By default is
11225 set to 0.
11226 planes
11228 Set planes which will be filtered, by default are all available
11229 filtered except alpha.
11230 fftfilt
11232 Apply arbitrary expressions to samples in frequency domain
11233 dc_Y
11235 Adjust the dc value (gain) of the luma plane of the image. The
11236 filter accepts an integer value in range 0 to 1000. The default
11237 value is set to 0.
11238 dc_U
11240 Adjust the dc value (gain) of the 1st chroma plane of the image.
11241 The filter accepts an integer value in range 0 to 1000. The default
11242 value is set to 0.
11243 dc_V
11245 Adjust the dc value (gain) of the 2nd chroma plane of the image.
11246 The filter accepts an integer value in range 0 to 1000. The default
11247 value is set to 0.
11248 weight_Y
11250 Set the frequency domain weight expression for the luma plane.
11251 weight_U
11253 Set the frequency domain weight expression for the 1st chroma
11254 plane.
11255 weight_V
11257 Set the frequency domain weight expression for the 2nd chroma
11258 plane.
11259 eval
11261 Set when the expressions are evaluated.
11262 It accepts the following values:
11264 init
11266 Only evaluate expressions once during the filter
11267 initialization.
11268 frame
11270 Evaluate expressions for each incoming frame.
11271 Default value is init.
11273 The filter accepts the following variables:
11275 X
11277 Y The coordinates of the current sample.
11278 W
11280 H The width and height of the image.
11281 N The number of input frame, starting from 0.
11283 WS
11285 HS The size of FFT array for horizontal and vertical processing.
11286 Examples
11288 • High-pass:
11290 fftfilt=dc_Y=128:weight_Y='squish(1-(Y+X)/100)'
11292 • Low-pass:
11294 fftfilt=dc_Y=0:weight_Y='squish((Y+X)/100-1)'
11296 • Sharpen:
11298 fftfilt=dc_Y=0:weight_Y='1+squish(1-(Y+X)/100)'
11300 • Blur:
11302 fftfilt=dc_Y=0:weight_Y='exp(-4 * ((Y+X)/(W+H)))'
11304 field
11306 Extract a single field from an interlaced image using stride arithmetic
11307 to avoid wasting CPU time. The output frames are marked as non-
11308 interlaced.
11309 The filter accepts the following options:
11311 type
11313 Specify whether to extract the top (if the value is 0 or "top") or
11314 the bottom field (if the value is 1 or "bottom").
11315 fieldhint
11317 Create new frames by copying the top and bottom fields from surrounding
11318 frames supplied as numbers by the hint file.
11319 hint
11321 Set file containing hints: absolute/relative frame numbers.
11322 There must be one line for each frame in a clip. Each line must
11324 contain two numbers separated by the comma, optionally followed by
11325 "-" or "+". Numbers supplied on each line of file can not be out
11326 of [N-1,N+1] where N is current frame number for "absolute" mode or
11327 out of [-1, 1] range for "relative" mode. First number tells from
11328 which frame to pick up top field and second number tells from which
11329 frame to pick up bottom field.
11330 If optionally followed by "+" output frame will be marked as
11332 interlaced, else if followed by "-" output frame will be marked as
11333 progressive, else it will be marked same as input frame. If
11334 optionally followed by "t" output frame will use only top field, or
11335 in case of "b" it will use only bottom field. If line starts with
11336 "#" or ";" that line is skipped.
11337 mode
11339 Can be item "absolute" or "relative" or "pattern". Default is
11340 "absolute". The "pattern" mode is same as "relative" mode, except
11341 at last entry of file if there are more frames to process than
11342 "hint" file is seek back to start.
11343 Example of first several lines of "hint" file for "relative" mode:
11345 0,0 - # first frame
11347 1,0 - # second frame, use third's frame top field and second's frame bottom field
11348 1,0 - # third frame, use fourth's frame top field and third's frame bottom field
11349 1,0 -
11350 0,0 -
11351 0,0 -
11352 1,0 -
11353 1,0 -
11354 1,0 -
11355 0,0 -
11356 0,0 -
11357 1,0 -
11358 1,0 -
11359 1,0 -
11360 0,0 -
11361 fieldmatch
11363 Field matching filter for inverse telecine. It is meant to reconstruct
11364 the progressive frames from a telecined stream. The filter does not
11365 drop duplicated frames, so to achieve a complete inverse telecine
11366 "fieldmatch" needs to be followed by a decimation filter such as
11367 decimate in the filtergraph.
11368 The separation of the field matching and the decimation is notably
11370 motivated by the possibility of inserting a de-interlacing filter
11371 fallback between the two. If the source has mixed telecined and real
11372 interlaced content, "fieldmatch" will not be able to match fields for
11373 the interlaced parts. But these remaining combed frames will be marked
11374 as interlaced, and thus can be de-interlaced by a later filter such as
11375 yadif before decimation.
11376 In addition to the various configuration options, "fieldmatch" can take
11378 an optional second stream, activated through the ppsrc option. If
11379 enabled, the frames reconstruction will be based on the fields and
11380 frames from this second stream. This allows the first input to be pre-
11381 processed in order to help the various algorithms of the filter, while
11382 keeping the output lossless (assuming the fields are matched properly).
11383 Typically, a field-aware denoiser, or brightness/contrast adjustments
11384 can help.
11385 Note that this filter uses the same algorithms as TIVTC/TFM (AviSynth
11387 project) and VIVTC/VFM (VapourSynth project). The later is a light
11388 clone of TFM from which "fieldmatch" is based on. While the semantic
11389 and usage are very close, some behaviour and options names can differ.
11390 The decimate filter currently only works for constant frame rate input.
11392 If your input has mixed telecined (30fps) and progressive content with
11393 a lower framerate like 24fps use the following filterchain to produce
11394 the necessary cfr stream:
11395 "dejudder,fps=30000/1001,fieldmatch,decimate".
11396 The filter accepts the following options:
11398 order
11400 Specify the assumed field order of the input stream. Available
11401 values are:
11402 auto
11404 Auto detect parity (use FFmpeg's internal parity value).
11405 bff Assume bottom field first.
11407 tff Assume top field first.
11409 Note that it is sometimes recommended not to trust the parity
11411 announced by the stream.
11412 Default value is auto.
11414 mode
11416 Set the matching mode or strategy to use. pc mode is the safest in
11417 the sense that it won't risk creating jerkiness due to duplicate
11418 frames when possible, but if there are bad edits or blended fields
11419 it will end up outputting combed frames when a good match might
11420 actually exist. On the other hand, pcn_ub mode is the most risky in
11421 terms of creating jerkiness, but will almost always find a good
11422 frame if there is one. The other values are all somewhere in
11423 between pc and pcn_ub in terms of risking jerkiness and creating
11424 duplicate frames versus finding good matches in sections with bad
11425 edits, orphaned fields, blended fields, etc.
11426 More details about p/c/n/u/b are available in p/c/n/u/b meaning
11428 section.
11429 Available values are:
11431 pc 2-way matching (p/c)
11433 pc_n
11435 2-way matching, and trying 3rd match if still combed (p/c + n)
11436 pc_u
11438 2-way matching, and trying 3rd match (same order) if still
11439 combed (p/c + u)
11440 pc_n_ub
11442 2-way matching, trying 3rd match if still combed, and trying
11443 4th/5th matches if still combed (p/c + n + u/b)
11444 pcn 3-way matching (p/c/n)
11446 pcn_ub
11448 3-way matching, and trying 4th/5th matches if all 3 of the
11449 original matches are detected as combed (p/c/n + u/b)
11450 The parenthesis at the end indicate the matches that would be used
11452 for that mode assuming order=tff (and field on auto or top).
11453 In terms of speed pc mode is by far the fastest and pcn_ub is the
11455 slowest.
11456 Default value is pc_n.
11458 ppsrc
11460 Mark the main input stream as a pre-processed input, and enable the
11461 secondary input stream as the clean source to pick the fields from.
11462 See the filter introduction for more details. It is similar to the
11463 clip2 feature from VFM/TFM.
11464 Default value is 0 (disabled).
11466 field
11468 Set the field to match from. It is recommended to set this to the
11469 same value as order unless you experience matching failures with
11470 that setting. In certain circumstances changing the field that is
11471 used to match from can have a large impact on matching performance.
11472 Available values are:
11473 auto
11475 Automatic (same value as order).
11476 bottom
11478 Match from the bottom field.
11479 top Match from the top field.
11481 Default value is auto.
11483 mchroma
11485 Set whether or not chroma is included during the match comparisons.
11486 In most cases it is recommended to leave this enabled. You should
11487 set this to 0 only if your clip has bad chroma problems such as
11488 heavy rainbowing or other artifacts. Setting this to 0 could also
11489 be used to speed things up at the cost of some accuracy.
11490 Default value is 1.
11492 y0
11494 y1 These define an exclusion band which excludes the lines between y0
11495 and y1 from being included in the field matching decision. An
11496 exclusion band can be used to ignore subtitles, a logo, or other
11497 things that may interfere with the matching. y0 sets the starting
11498 scan line and y1 sets the ending line; all lines in between y0 and
11499 y1 (including y0 and y1) will be ignored. Setting y0 and y1 to the
11500 same value will disable the feature. y0 and y1 defaults to 0.
11501 scthresh
11503 Set the scene change detection threshold as a percentage of maximum
11504 change on the luma plane. Good values are in the "[8.0, 14.0]"
11505 range. Scene change detection is only relevant in case
11506 combmatch=sc. The range for scthresh is "[0.0, 100.0]".
11507 Default value is 12.0.
11509 combmatch
11511 When combatch is not none, "fieldmatch" will take into account the
11512 combed scores of matches when deciding what match to use as the
11513 final match. Available values are:
11514 none
11516 No final matching based on combed scores.
11517 sc Combed scores are only used when a scene change is detected.
11519 full
11521 Use combed scores all the time.
11522 Default is sc.
11524 combdbg
11526 Force "fieldmatch" to calculate the combed metrics for certain
11527 matches and print them. This setting is known as micout in TFM/VFM
11528 vocabulary. Available values are:
11529 none
11531 No forced calculation.
11532 pcn Force p/c/n calculations.
11534 pcnub
11536 Force p/c/n/u/b calculations.
11537 Default value is none.
11539 cthresh
11541 This is the area combing threshold used for combed frame detection.
11542 This essentially controls how "strong" or "visible" combing must be
11543 to be detected. Larger values mean combing must be more visible
11544 and smaller values mean combing can be less visible or strong and
11545 still be detected. Valid settings are from "-1" (every pixel will
11546 be detected as combed) to 255 (no pixel will be detected as
11547 combed). This is basically a pixel difference value. A good range
11548 is "[8, 12]".
11549 Default value is 9.
11551 chroma
11553 Sets whether or not chroma is considered in the combed frame
11554 decision. Only disable this if your source has chroma problems
11555 (rainbowing, etc.) that are causing problems for the combed frame
11556 detection with chroma enabled. Actually, using chroma=0 is usually
11557 more reliable, except for the case where there is chroma only
11558 combing in the source.
11559 Default value is 0.
11561 blockx
11563 blocky
11564 Respectively set the x-axis and y-axis size of the window used
11565 during combed frame detection. This has to do with the size of the
11566 area in which combpel pixels are required to be detected as combed
11567 for a frame to be declared combed. See the combpel parameter
11568 description for more info. Possible values are any number that is
11569 a power of 2 starting at 4 and going up to 512.
11570 Default value is 16.
11572 combpel
11574 The number of combed pixels inside any of the blocky by blockx size
11575 blocks on the frame for the frame to be detected as combed. While
11576 cthresh controls how "visible" the combing must be, this setting
11577 controls "how much" combing there must be in any localized area (a
11578 window defined by the blockx and blocky settings) on the frame.
11579 Minimum value is 0 and maximum is "blocky x blockx" (at which point
11580 no frames will ever be detected as combed). This setting is known
11581 as MI in TFM/VFM vocabulary.
11582 Default value is 80.
11584 p/c/n/u/b meaning
11586 p/c/n
11588 We assume the following telecined stream:
11590 Top fields: 1 2 2 3 4
11592 Bottom fields: 1 2 3 4 4
11593 The numbers correspond to the progressive frame the fields relate to.
11595 Here, the first two frames are progressive, the 3rd and 4th are combed,
11596 and so on.
11597 When "fieldmatch" is configured to run a matching from bottom
11599 (field=bottom) this is how this input stream get transformed:
11600 Input stream:
11602 T 1 2 2 3 4
11603 B 1 2 3 4 4 <-- matching reference
11604 Matches: c c n n c
11606 Output stream:
11608 T 1 2 3 4 4
11609 B 1 2 3 4 4
11610 As a result of the field matching, we can see that some frames get
11612 duplicated. To perform a complete inverse telecine, you need to rely
11613 on a decimation filter after this operation. See for instance the
11614 decimate filter.
11615 The same operation now matching from top fields (field=top) looks like
11617 this:
11618 Input stream:
11620 T 1 2 2 3 4 <-- matching reference
11621 B 1 2 3 4 4
11622 Matches: c c p p c
11624 Output stream:
11626 T 1 2 2 3 4
11627 B 1 2 2 3 4
11628 In these examples, we can see what p, c and n mean; basically, they
11630 refer to the frame and field of the opposite parity:
11631 *<p matches the field of the opposite parity in the previous frame>
11633 *<c matches the field of the opposite parity in the current frame>
11634 *<n matches the field of the opposite parity in the next frame>
11635 u/b
11637 The u and b matching are a bit special in the sense that they match
11639 from the opposite parity flag. In the following examples, we assume
11640 that we are currently matching the 2nd frame (Top:2, bottom:2).
11641 According to the match, a 'x' is placed above and below each matched
11642 fields.
11643 With bottom matching (field=bottom):
11645 Match: c p n b u
11647 x x x x x
11649 Top 1 2 2 1 2 2 1 2 2 1 2 2 1 2 2
11650 Bottom 1 2 3 1 2 3 1 2 3 1 2 3 1 2 3
11651 x x x x x
11652 Output frames:
11654 2 1 2 2 2
11655 2 2 2 1 3
11656 With top matching (field=top):
11658 Match: c p n b u
11660 x x x x x
11662 Top 1 2 2 1 2 2 1 2 2 1 2 2 1 2 2
11663 Bottom 1 2 3 1 2 3 1 2 3 1 2 3 1 2 3
11664 x x x x x
11665 Output frames:
11667 2 2 2 1 2
11668 2 1 3 2 2
11669 Examples
11671 Simple IVTC of a top field first telecined stream:
11673 fieldmatch=order=tff:combmatch=none, decimate
11675 Advanced IVTC, with fallback on yadif for still combed frames:
11677 fieldmatch=order=tff:combmatch=full, yadif=deint=interlaced, decimate
11679 fieldorder
11681 Transform the field order of the input video.
11682 It accepts the following parameters:
11684 order
11686 The output field order. Valid values are tff for top field first or
11687 bff for bottom field first.
11688 The default value is tff.
11690 The transformation is done by shifting the picture content up or down
11692 by one line, and filling the remaining line with appropriate picture
11693 content. This method is consistent with most broadcast field order
11694 converters.
11695 If the input video is not flagged as being interlaced, or it is already
11697 flagged as being of the required output field order, then this filter
11698 does not alter the incoming video.
11699 It is very useful when converting to or from PAL DV material, which is
11701 bottom field first.
11702 For example:
11704 ffmpeg -i in.vob -vf "fieldorder=bff" out.dv
11706 fifo, afifo
11708 Buffer input images and send them when they are requested.
11709 It is mainly useful when auto-inserted by the libavfilter framework.
11711 It does not take parameters.
11713 fillborders
11715 Fill borders of the input video, without changing video stream
11716 dimensions. Sometimes video can have garbage at the four edges and you
11717 may not want to crop video input to keep size multiple of some number.
11718 This filter accepts the following options:
11720 left
11722 Number of pixels to fill from left border.
11723 right
11725 Number of pixels to fill from right border.
11726 top Number of pixels to fill from top border.
11728 bottom
11730 Number of pixels to fill from bottom border.
11731 mode
11733 Set fill mode.
11734 It accepts the following values:
11736 smear
11738 fill pixels using outermost pixels
11739 mirror
11741 fill pixels using mirroring (half sample symmetric)
11742 fixed
11744 fill pixels with constant value
11745 reflect
11747 fill pixels using reflecting (whole sample symmetric)
11748 wrap
11750 fill pixels using wrapping
11751 fade
11753 fade pixels to constant value
11754 margins
11756 fill pixels at top and bottom with weighted averages pixels
11757 near borders
11758 Default is smear.
11760 color
11762 Set color for pixels in fixed or fade mode. Default is black.
11763 Commands
11765 This filter supports same commands as options. The command accepts the
11767 same syntax of the corresponding option.
11768 If the specified expression is not valid, it is kept at its current
11770 value.
11771 find_rect
11773 Find a rectangular object
11774 It accepts the following options:
11776 object
11778 Filepath of the object image, needs to be in gray8.
11779 threshold
11781 Detection threshold, default is 0.5.
11782 mipmaps
11784 Number of mipmaps, default is 3.
11785 xmin, ymin, xmax, ymax
11787 Specifies the rectangle in which to search.
11788 discard
11790 Discard frames where object is not detected. Default is disabled.
11791 Examples
11793 • Cover a rectangular object by the supplied image of a given video
11795 using ffmpeg:
11796 ffmpeg -i file.ts -vf find_rect=newref.pgm,cover_rect=cover.jpg:mode=cover new.mkv
11798 floodfill
11800 Flood area with values of same pixel components with another values.
11801 It accepts the following options:
11803 x Set pixel x coordinate.
11805 y Set pixel y coordinate.
11807 s0 Set source #0 component value.
11809 s1 Set source #1 component value.
11811 s2 Set source #2 component value.
11813 s3 Set source #3 component value.
11815 d0 Set destination #0 component value.
11817 d1 Set destination #1 component value.
11819 d2 Set destination #2 component value.
11821 d3 Set destination #3 component value.
11823 format
11825 Convert the input video to one of the specified pixel formats.
11826 Libavfilter will try to pick one that is suitable as input to the next
11827 filter.
11828 It accepts the following parameters:
11830 pix_fmts
11832 A '|'-separated list of pixel format names, such as
11833 "pix_fmts=yuv420p|monow|rgb24".
11834 Examples
11836 • Convert the input video to the yuv420p format
11838 format=pix_fmts=yuv420p
11840 Convert the input video to any of the formats in the list
11842 format=pix_fmts=yuv420p|yuv444p|yuv410p
11844 fps
11846 Convert the video to specified constant frame rate by duplicating or
11847 dropping frames as necessary.
11848 It accepts the following parameters:
11850 fps The desired output frame rate. It accepts expressions containing
11852 the following constants:
11853 source_fps
11855 The input's frame rate
11856 ntsc
11858 NTSC frame rate of "30000/1001"
11859 pal PAL frame rate of 25.0
11861 film
11863 Film frame rate of 24.0
11864 ntsc_film
11866 NTSC-film frame rate of "24000/1001"
11867 The default is 25.
11869 start_time
11871 Assume the first PTS should be the given value, in seconds. This
11872 allows for padding/trimming at the start of stream. By default, no
11873 assumption is made about the first frame's expected PTS, so no
11874 padding or trimming is done. For example, this could be set to 0
11875 to pad the beginning with duplicates of the first frame if a video
11876 stream starts after the audio stream or to trim any frames with a
11877 negative PTS.
11878 round
11880 Timestamp (PTS) rounding method.
11881 Possible values are:
11883 zero
11885 round towards 0
11886 inf round away from 0
11888 down
11890 round towards -infinity
11891 up round towards +infinity
11893 near
11895 round to nearest
11896 The default is "near".
11898 eof_action
11900 Action performed when reading the last frame.
11901 Possible values are:
11903 round
11905 Use same timestamp rounding method as used for other frames.
11906 pass
11908 Pass through last frame if input duration has not been reached
11909 yet.
11910 The default is "round".
11912 Alternatively, the options can be specified as a flat string:
11914 fps[:start_time[:round]].
11915 See also the setpts filter.
11917 Examples
11919 • A typical usage in order to set the fps to 25:
11921 fps=fps=25
11923 • Sets the fps to 24, using abbreviation and rounding method to round
11925 to nearest:
11926 fps=fps=film:round=near
11928 framepack
11930 Pack two different video streams into a stereoscopic video, setting
11931 proper metadata on supported codecs. The two views should have the same
11932 size and framerate and processing will stop when the shorter video
11933 ends. Please note that you may conveniently adjust view properties with
11934 the scale and fps filters.
11935 It accepts the following parameters:
11937 format
11939 The desired packing format. Supported values are:
11940 sbs The views are next to each other (default).
11942 tab The views are on top of each other.
11944 lines
11946 The views are packed by line.
11947 columns
11949 The views are packed by column.
11950 frameseq
11952 The views are temporally interleaved.
11953 Some examples:
11955 # Convert left and right views into a frame-sequential video
11957 ffmpeg -i LEFT -i RIGHT -filter_complex framepack=frameseq OUTPUT
11958 # Convert views into a side-by-side video with the same output resolution as the input
11960 ffmpeg -i LEFT -i RIGHT -filter_complex [0:v]scale=w=iw/2[left],[1:v]scale=w=iw/2[right],[left][right]framepack=sbs OUTPUT
11961 framerate
11963 Change the frame rate by interpolating new video output frames from the
11964 source frames.
11965 This filter is not designed to function correctly with interlaced
11967 media. If you wish to change the frame rate of interlaced media then
11968 you are required to deinterlace before this filter and re-interlace
11969 after this filter.
11970 A description of the accepted options follows.
11972 fps Specify the output frames per second. This option can also be
11974 specified as a value alone. The default is 50.
11975 interp_start
11977 Specify the start of a range where the output frame will be created
11978 as a linear interpolation of two frames. The range is [0-255], the
11979 default is 15.
11980 interp_end
11982 Specify the end of a range where the output frame will be created
11983 as a linear interpolation of two frames. The range is [0-255], the
11984 default is 240.
11985 scene
11987 Specify the level at which a scene change is detected as a value
11988 between 0 and 100 to indicate a new scene; a low value reflects a
11989 low probability for the current frame to introduce a new scene,
11990 while a higher value means the current frame is more likely to be
11991 one. The default is 8.2.
11992 flags
11994 Specify flags influencing the filter process.
11995 Available value for flags is:
11997 scene_change_detect, scd
11999 Enable scene change detection using the value of the option
12000 scene. This flag is enabled by default.
12001 framestep
12003 Select one frame every N-th frame.
12004 This filter accepts the following option:
12006 step
12008 Select frame after every "step" frames. Allowed values are
12009 positive integers higher than 0. Default value is 1.
12010 freezedetect
12012 Detect frozen video.
12013 This filter logs a message and sets frame metadata when it detects that
12015 the input video has no significant change in content during a specified
12016 duration. Video freeze detection calculates the mean average absolute
12017 difference of all the components of video frames and compares it to a
12018 noise floor.
12019 The printed times and duration are expressed in seconds. The
12021 "lavfi.freezedetect.freeze_start" metadata key is set on the first
12022 frame whose timestamp equals or exceeds the detection duration and it
12023 contains the timestamp of the first frame of the freeze. The
12024 "lavfi.freezedetect.freeze_duration" and
12025 "lavfi.freezedetect.freeze_end" metadata keys are set on the first
12026 frame after the freeze.
12027 The filter accepts the following options:
12029 noise, n
12031 Set noise tolerance. Can be specified in dB (in case "dB" is
12032 appended to the specified value) or as a difference ratio between 0
12033 and 1. Default is -60dB, or 0.001.
12034 duration, d
12036 Set freeze duration until notification (default is 2 seconds).
12037 freezeframes
12039 Freeze video frames.
12040 This filter freezes video frames using frame from 2nd input.
12042 The filter accepts the following options:
12044 first
12046 Set number of first frame from which to start freeze.
12047 last
12049 Set number of last frame from which to end freeze.
12050 replace
12052 Set number of frame from 2nd input which will be used instead of
12053 replaced frames.
12054 frei0r
12056 Apply a frei0r effect to the input video.
12057 To enable the compilation of this filter, you need to install the
12059 frei0r header and configure FFmpeg with "--enable-frei0r".
12060 It accepts the following parameters:
12062 filter_name
12064 The name of the frei0r effect to load. If the environment variable
12065 FREI0R_PATH is defined, the frei0r effect is searched for in each
12066 of the directories specified by the colon-separated list in
12067 FREI0R_PATH. Otherwise, the standard frei0r paths are searched, in
12068 this order: HOME/.frei0r-1/lib/, /usr/local/lib/frei0r-1/,
12069 /usr/lib/frei0r-1/.
12070 filter_params
12072 A '|'-separated list of parameters to pass to the frei0r effect.
12073 A frei0r effect parameter can be a boolean (its value is either "y" or
12075 "n"), a double, a color (specified as R/G/B, where R, G, and B are
12076 floating point numbers between 0.0 and 1.0, inclusive) or a color
12077 description as specified in the "Color" section in the ffmpeg-utils
12078 manual, a position (specified as X/Y, where X and Y are floating point
12079 numbers) and/or a string.
12080 The number and types of parameters depend on the loaded effect. If an
12082 effect parameter is not specified, the default value is set.
12083 Examples
12085 • Apply the distort0r effect, setting the first two double
12087 parameters:
12088 frei0r=filter_name=distort0r:filter_params=0.5|0.01
12090 • Apply the colordistance effect, taking a color as the first
12092 parameter:
12093 frei0r=colordistance:0.2/0.3/0.4
12095 frei0r=colordistance:violet
12096 frei0r=colordistance:0x112233
12097 • Apply the perspective effect, specifying the top left and top right
12099 image positions:
12100 frei0r=perspective:0.2/0.2|0.8/0.2
12102 For more information, see <http://frei0r.dyne.org>
12104 Commands
12106 This filter supports the filter_params option as commands.
12108 fspp
12110 Apply fast and simple postprocessing. It is a faster version of spp.
12111 It splits (I)DCT into horizontal/vertical passes. Unlike the simple
12113 post- processing filter, one of them is performed once per block, not
12114 per pixel. This allows for much higher speed.
12115 The filter accepts the following options:
12117 quality
12119 Set quality. This option defines the number of levels for
12120 averaging. It accepts an integer in the range 4-5. Default value is
12121 4.
12122 qp Force a constant quantization parameter. It accepts an integer in
12124 range 0-63. If not set, the filter will use the QP from the video
12125 stream (if available).
12126 strength
12128 Set filter strength. It accepts an integer in range -15 to 32.
12129 Lower values mean more details but also more artifacts, while
12130 higher values make the image smoother but also blurrier. Default
12131 value is 0 X PSNR optimal.
12132 use_bframe_qp
12134 Enable the use of the QP from the B-Frames if set to 1. Using this
12135 option may cause flicker since the B-Frames have often larger QP.
12136 Default is 0 (not enabled).
12137 gblur
12139 Apply Gaussian blur filter.
12140 The filter accepts the following options:
12142 sigma
12144 Set horizontal sigma, standard deviation of Gaussian blur. Default
12145 is 0.5.
12146 steps
12148 Set number of steps for Gaussian approximation. Default is 1.
12149 planes
12151 Set which planes to filter. By default all planes are filtered.
12152 sigmaV
12154 Set vertical sigma, if negative it will be same as "sigma".
12155 Default is "-1".
12156 Commands
12158 This filter supports same commands as options. The command accepts the
12160 same syntax of the corresponding option.
12161 If the specified expression is not valid, it is kept at its current
12163 value.
12164 geq
12166 Apply generic equation to each pixel.
12167 The filter accepts the following options:
12169 lum_expr, lum
12171 Set the luminance expression.
12172 cb_expr, cb
12174 Set the chrominance blue expression.
12175 cr_expr, cr
12177 Set the chrominance red expression.
12178 alpha_expr, a
12180 Set the alpha expression.
12181 red_expr, r
12183 Set the red expression.
12184 green_expr, g
12186 Set the green expression.
12187 blue_expr, b
12189 Set the blue expression.
12190 The colorspace is selected according to the specified options. If one
12192 of the lum_expr, cb_expr, or cr_expr options is specified, the filter
12193 will automatically select a YCbCr colorspace. If one of the red_expr,
12194 green_expr, or blue_expr options is specified, it will select an RGB
12195 colorspace.
12196 If one of the chrominance expression is not defined, it falls back on
12198 the other one. If no alpha expression is specified it will evaluate to
12199 opaque value. If none of chrominance expressions are specified, they
12200 will evaluate to the luminance expression.
12201 The expressions can use the following variables and functions:
12203 N The sequential number of the filtered frame, starting from 0.
12205 X
12207 Y The coordinates of the current sample.
12208 W
12210 H The width and height of the image.
12211 SW
12213 SH Width and height scale depending on the currently filtered plane.
12214 It is the ratio between the corresponding luma plane number of
12215 pixels and the current plane ones. E.g. for YUV4:2:0 the values are
12216 "1,1" for the luma plane, and "0.5,0.5" for chroma planes.
12217 T Time of the current frame, expressed in seconds.
12219 p(x, y)
12221 Return the value of the pixel at location (x,y) of the current
12222 plane.
12223 lum(x, y)
12225 Return the value of the pixel at location (x,y) of the luminance
12226 plane.
12227 cb(x, y)
12229 Return the value of the pixel at location (x,y) of the blue-
12230 difference chroma plane. Return 0 if there is no such plane.
12231 cr(x, y)
12233 Return the value of the pixel at location (x,y) of the red-
12234 difference chroma plane. Return 0 if there is no such plane.
12235 r(x, y)
12237 g(x, y)
12238 b(x, y)
12239 Return the value of the pixel at location (x,y) of the
12240 red/green/blue component. Return 0 if there is no such component.
12241 alpha(x, y)
12243 Return the value of the pixel at location (x,y) of the alpha plane.
12244 Return 0 if there is no such plane.
12245 psum(x,y), lumsum(x, y), cbsum(x,y), crsum(x,y), rsum(x,y), gsum(x,y),
12247 bsum(x,y), alphasum(x,y)
12248 Sum of sample values in the rectangle from (0,0) to (x,y), this
12249 allows obtaining sums of samples within a rectangle. See the
12250 functions without the sum postfix.
12251 interpolation
12253 Set one of interpolation methods:
12254 nearest, n
12256 bilinear, b
12257 Default is bilinear.
12259 For functions, if x and y are outside the area, the value will be
12261 automatically clipped to the closer edge.
12262 Please note that this filter can use multiple threads in which case
12264 each slice will have its own expression state. If you want to use only
12265 a single expression state because your expressions depend on previous
12266 state then you should limit the number of filter threads to 1.
12267 Examples
12269 • Flip the image horizontally:
12271 geq=p(W-X\,Y)
12273 • Generate a bidimensional sine wave, with angle "PI/3" and a
12275 wavelength of 100 pixels:
12276 geq=128 + 100*sin(2*(PI/100)*(cos(PI/3)*(X-50*T) + sin(PI/3)*Y)):128:128
12278 • Generate a fancy enigmatic moving light:
12280 nullsrc=s=256x256,geq=random(1)/hypot(X-cos(N*0.07)*W/2-W/2\,Y-sin(N*0.09)*H/2-H/2)^2*1000000*sin(N*0.02):128:128
12282 • Generate a quick emboss effect:
12284 format=gray,geq=lum_expr='(p(X,Y)+(256-p(X-4,Y-4)))/2'
12286 • Modify RGB components depending on pixel position:
12288 geq=r='X/W*r(X,Y)':g='(1-X/W)*g(X,Y)':b='(H-Y)/H*b(X,Y)'
12290 • Create a radial gradient that is the same size as the input (also
12292 see the vignette filter):
12293 geq=lum=255*gauss((X/W-0.5)*3)*gauss((Y/H-0.5)*3)/gauss(0)/gauss(0),format=gray
12295 gradfun
12297 Fix the banding artifacts that are sometimes introduced into nearly
12298 flat regions by truncation to 8-bit color depth. Interpolate the
12299 gradients that should go where the bands are, and dither them.
12300 It is designed for playback only. Do not use it prior to lossy
12302 compression, because compression tends to lose the dither and bring
12303 back the bands.
12304 It accepts the following parameters:
12306 strength
12308 The maximum amount by which the filter will change any one pixel.
12309 This is also the threshold for detecting nearly flat regions.
12310 Acceptable values range from .51 to 64; the default value is 1.2.
12311 Out-of-range values will be clipped to the valid range.
12312 radius
12314 The neighborhood to fit the gradient to. A larger radius makes for
12315 smoother gradients, but also prevents the filter from modifying the
12316 pixels near detailed regions. Acceptable values are 8-32; the
12317 default value is 16. Out-of-range values will be clipped to the
12318 valid range.
12319 Alternatively, the options can be specified as a flat string:
12321 strength[:radius]
12322 Examples
12324 • Apply the filter with a 3.5 strength and radius of 8:
12326 gradfun=3.5:8
12328 • Specify radius, omitting the strength (which will fall-back to the
12330 default value):
12331 gradfun=radius=8
12333 graphmonitor
12335 Show various filtergraph stats.
12336 With this filter one can debug complete filtergraph. Especially issues
12338 with links filling with queued frames.
12339 The filter accepts the following options:
12341 size, s
12343 Set video output size. Default is hd720.
12344 opacity, o
12346 Set video opacity. Default is 0.9. Allowed range is from 0 to 1.
12347 mode, m
12349 Set output mode, can be fulll or compact. In compact mode only
12350 filters with some queued frames have displayed stats.
12351 flags, f
12353 Set flags which enable which stats are shown in video.
12354 Available values for flags are:
12356 queue
12358 Display number of queued frames in each link.
12359 frame_count_in
12361 Display number of frames taken from filter.
12362 frame_count_out
12364 Display number of frames given out from filter.
12365 frame_count_delta
12367 Display delta number of frames between above two values.
12368 pts Display current filtered frame pts.
12370 pts_delta
12372 Display pts delta between current and previous frame.
12373 time
12375 Display current filtered frame time.
12376 time_delta
12378 Display time delta between current and previous frame.
12379 timebase
12381 Display time base for filter link.
12382 format
12384 Display used format for filter link.
12385 size
12387 Display video size or number of audio channels in case of audio
12388 used by filter link.
12389 rate
12391 Display video frame rate or sample rate in case of audio used
12392 by filter link.
12393 eof Display link output status.
12395 sample_count_in
12397 Display number of samples taken from filter.
12398 sample_count_out
12400 Display number of samples given out from filter.
12401 sample_count_delta
12403 Display delta number of samples between above two values.
12404 rate, r
12406 Set upper limit for video rate of output stream, Default value is
12407 25. This guarantee that output video frame rate will not be higher
12408 than this value.
12409 grayworld
12411 A color constancy filter that applies color correction based on the
12412 grayworld assumption
12413 See:
12415 <https://www.researchgate.net/publication/275213614_A_New_Color_Correction_Method_for_Underwater_Imaging>
12416 The algorithm uses linear light, so input data should be linearized
12418 beforehand (and possibly correctly tagged).
12419 ffmpeg -i INPUT -vf zscale=transfer=linear,grayworld,zscale=transfer=bt709,format=yuv420p OUTPUT
12421 greyedge
12423 A color constancy variation filter which estimates scene illumination
12424 via grey edge algorithm and corrects the scene colors accordingly.
12425 See: <https://staff.science.uva.nl/th.gevers/pub/GeversTIP07.pdf>
12427 The filter accepts the following options:
12429 difford
12431 The order of differentiation to be applied on the scene. Must be
12432 chosen in the range [0,2] and default value is 1.
12433 minknorm
12435 The Minkowski parameter to be used for calculating the Minkowski
12436 distance. Must be chosen in the range [0,20] and default value is
12437 1. Set to 0 for getting max value instead of calculating Minkowski
12438 distance.
12439 sigma
12441 The standard deviation of Gaussian blur to be applied on the scene.
12442 Must be chosen in the range [0,1024.0] and default value = 1.
12443 floor( sigma * break_off_sigma(3) ) can't be equal to 0 if difford
12444 is greater than 0.
12445 Examples
12447 • Grey Edge:
12449 greyedge=difford=1:minknorm=5:sigma=2
12451 • Max Edge:
12453 greyedge=difford=1:minknorm=0:sigma=2
12455 guided
12457 Apply guided filter for edge-preserving smoothing, dehazing and so on.
12458 The filter accepts the following options:
12460 radius
12462 Set the box radius in pixels. Allowed range is 1 to 20. Default is
12463 3.
12464 eps Set regularization parameter (with square). Allowed range is 0 to
12466 1. Default is 0.01.
12467 mode
12469 Set filter mode. Can be "basic" or "fast". Default is "basic".
12470 sub Set subsampling ratio for "fast" mode. Range is 2 to 64. Default
12472 is 4. No subsampling occurs in "basic" mode.
12473 guidance
12475 Set guidance mode. Can be "off" or "on". Default is "off". If
12476 "off", single input is required. If "on", two inputs of the same
12477 resolution and pixel format are required. The second input serves
12478 as the guidance.
12479 planes
12481 Set planes to filter. Default is first only.
12482 Commands
12484 This filter supports the all above options as commands.
12486 Examples
12488 • Edge-preserving smoothing with guided filter:
12490 ffmpeg -i in.png -vf guided out.png
12492 • Dehazing, structure-transferring filtering, detail enhancement with
12494 guided filter. For the generation of guidance image, refer to
12495 paper "Guided Image Filtering". See:
12496 <http://kaiminghe.com/publications/pami12guidedfilter.pdf>.
12497 ffmpeg -i in.png -i guidance.png -filter_complex guided=guidance=on out.png
12499 haldclut
12501 Apply a Hald CLUT to a video stream.
12502 First input is the video stream to process, and second one is the Hald
12504 CLUT. The Hald CLUT input can be a simple picture or a complete video
12505 stream.
12506 The filter accepts the following options:
12508 clut
12510 Set which CLUT video frames will be processed from second input
12511 stream, can be first or all. Default is all.
12512 shortest
12514 Force termination when the shortest input terminates. Default is 0.
12515 repeatlast
12517 Continue applying the last CLUT after the end of the stream. A
12518 value of 0 disable the filter after the last frame of the CLUT is
12519 reached. Default is 1.
12520 "haldclut" also has the same interpolation options as lut3d (both
12522 filters share the same internals).
12523 This filter also supports the framesync options.
12525 More information about the Hald CLUT can be found on Eskil Steenberg's
12527 website (Hald CLUT author) at
12528 <http://www.quelsolaar.com/technology/clut.html>.
12529 Commands
12531 This filter supports the "interp" option as commands.
12533 Workflow examples
12535 Hald CLUT video stream
12537 Generate an identity Hald CLUT stream altered with various effects:
12539 ffmpeg -f lavfi -i B<haldclutsrc>=8 -vf "hue=H=2*PI*t:s=sin(2*PI*t)+1, curves=cross_process" -t 10 -c:v ffv1 clut.nut
12541 Note: make sure you use a lossless codec.
12543 Then use it with "haldclut" to apply it on some random stream:
12545 ffmpeg -f lavfi -i mandelbrot -i clut.nut -filter_complex '[0][1] haldclut' -t 20 mandelclut.mkv
12547 The Hald CLUT will be applied to the 10 first seconds (duration of
12549 clut.nut), then the latest picture of that CLUT stream will be applied
12550 to the remaining frames of the "mandelbrot" stream.
12551 Hald CLUT with preview
12553 A Hald CLUT is supposed to be a squared image of "Level*Level*Level" by
12555 "Level*Level*Level" pixels. For a given Hald CLUT, FFmpeg will select
12556 the biggest possible square starting at the top left of the picture.
12557 The remaining padding pixels (bottom or right) will be ignored. This
12558 area can be used to add a preview of the Hald CLUT.
12559 Typically, the following generated Hald CLUT will be supported by the
12561 "haldclut" filter:
12562 ffmpeg -f lavfi -i B<haldclutsrc>=8 -vf "
12564 pad=iw+320 [padded_clut];
12565 smptebars=s=320x256, split [a][b];
12566 [padded_clut][a] overlay=W-320:h, curves=color_negative [main];
12567 [main][b] overlay=W-320" -frames:v 1 clut.png
12568 It contains the original and a preview of the effect of the CLUT: SMPTE
12570 color bars are displayed on the right-top, and below the same color
12571 bars processed by the color changes.
12572 Then, the effect of this Hald CLUT can be visualized with:
12574 ffplay input.mkv -vf "movie=clut.png, [in] haldclut"
12576 hflip
12578 Flip the input video horizontally.
12579 For example, to horizontally flip the input video with ffmpeg:
12581 ffmpeg -i in.avi -vf "hflip" out.avi
12583 histeq
12585 This filter applies a global color histogram equalization on a per-
12586 frame basis.
12587 It can be used to correct video that has a compressed range of pixel
12589 intensities. The filter redistributes the pixel intensities to
12590 equalize their distribution across the intensity range. It may be
12591 viewed as an "automatically adjusting contrast filter". This filter is
12592 useful only for correcting degraded or poorly captured source video.
12593 The filter accepts the following options:
12595 strength
12597 Determine the amount of equalization to be applied. As the
12598 strength is reduced, the distribution of pixel intensities more-
12599 and-more approaches that of the input frame. The value must be a
12600 float number in the range [0,1] and defaults to 0.200.
12601 intensity
12603 Set the maximum intensity that can generated and scale the output
12604 values appropriately. The strength should be set as desired and
12605 then the intensity can be limited if needed to avoid washing-out.
12606 The value must be a float number in the range [0,1] and defaults to
12607 0.210.
12608 antibanding
12610 Set the antibanding level. If enabled the filter will randomly vary
12611 the luminance of output pixels by a small amount to avoid banding
12612 of the histogram. Possible values are "none", "weak" or "strong".
12613 It defaults to "none".
12614 histogram
12616 Compute and draw a color distribution histogram for the input video.
12617 The computed histogram is a representation of the color component
12619 distribution in an image.
12620 Standard histogram displays the color components distribution in an
12622 image. Displays color graph for each color component. Shows
12623 distribution of the Y, U, V, A or R, G, B components, depending on
12624 input format, in the current frame. Below each graph a color component
12625 scale meter is shown.
12626 The filter accepts the following options:
12628 level_height
12630 Set height of level. Default value is 200. Allowed range is [50,
12631 2048].
12632 scale_height
12634 Set height of color scale. Default value is 12. Allowed range is
12635 [0, 40].
12636 display_mode
12638 Set display mode. It accepts the following values:
12639 stack
12641 Per color component graphs are placed below each other.
12642 parade
12644 Per color component graphs are placed side by side.
12645 overlay
12647 Presents information identical to that in the "parade", except
12648 that the graphs representing color components are superimposed
12649 directly over one another.
12650 Default is "stack".
12652 levels_mode
12654 Set mode. Can be either "linear", or "logarithmic". Default is
12655 "linear".
12656 components
12658 Set what color components to display. Default is 7.
12659 fgopacity
12661 Set foreground opacity. Default is 0.7.
12662 bgopacity
12664 Set background opacity. Default is 0.5.
12665 colors_mode
12667 Set colors mode. It accepts the following values:
12668 whiteonblack
12670 blackonwhite
12671 whiteongray
12672 blackongray
12673 coloronblack
12674 coloronwhite
12675 colorongray
12676 blackoncolor
12677 whiteoncolor
12678 grayoncolor
12679 Default is "whiteonblack".
12681 Examples
12683 • Calculate and draw histogram:
12685 ffplay -i input -vf histogram
12687 hqdn3d
12689 This is a high precision/quality 3d denoise filter. It aims to reduce
12690 image noise, producing smooth images and making still images really
12691 still. It should enhance compressibility.
12692 It accepts the following optional parameters:
12694 luma_spatial
12696 A non-negative floating point number which specifies spatial luma
12697 strength. It defaults to 4.0.
12698 chroma_spatial
12700 A non-negative floating point number which specifies spatial chroma
12701 strength. It defaults to 3.0*luma_spatial/4.0.
12702 luma_tmp
12704 A floating point number which specifies luma temporal strength. It
12705 defaults to 6.0*luma_spatial/4.0.
12706 chroma_tmp
12708 A floating point number which specifies chroma temporal strength.
12709 It defaults to luma_tmp*chroma_spatial/luma_spatial.
12710 Commands
12712 This filter supports same commands as options. The command accepts the
12714 same syntax of the corresponding option.
12715 If the specified expression is not valid, it is kept at its current
12717 value.
12718 hwdownload
12720 Download hardware frames to system memory.
12721 The input must be in hardware frames, and the output a non-hardware
12723 format. Not all formats will be supported on the output - it may be
12724 necessary to insert an additional format filter immediately following
12725 in the graph to get the output in a supported format.
12726 hwmap
12728 Map hardware frames to system memory or to another device.
12729 This filter has several different modes of operation; which one is used
12731 depends on the input and output formats:
12732 • Hardware frame input, normal frame output
12734 Map the input frames to system memory and pass them to the output.
12736 If the original hardware frame is later required (for example,
12737 after overlaying something else on part of it), the hwmap filter
12738 can be used again in the next mode to retrieve it.
12739 • Normal frame input, hardware frame output
12741 If the input is actually a software-mapped hardware frame, then
12743 unmap it - that is, return the original hardware frame.
12744 Otherwise, a device must be provided. Create new hardware surfaces
12746 on that device for the output, then map them back to the software
12747 format at the input and give those frames to the preceding filter.
12748 This will then act like the hwupload filter, but may be able to
12749 avoid an additional copy when the input is already in a compatible
12750 format.
12751 • Hardware frame input and output
12753 A device must be supplied for the output, either directly or with
12755 the derive_device option. The input and output devices must be of
12756 different types and compatible - the exact meaning of this is
12757 system-dependent, but typically it means that they must refer to
12758 the same underlying hardware context (for example, refer to the
12759 same graphics card).
12760 If the input frames were originally created on the output device,
12762 then unmap to retrieve the original frames.
12763 Otherwise, map the frames to the output device - create new
12765 hardware frames on the output corresponding to the frames on the
12766 input.
12767 The following additional parameters are accepted:
12769 mode
12771 Set the frame mapping mode. Some combination of:
12772 read
12774 The mapped frame should be readable.
12775 write
12777 The mapped frame should be writeable.
12778 overwrite
12780 The mapping will always overwrite the entire frame.
12781 This may improve performance in some cases, as the original
12783 contents of the frame need not be loaded.
12784 direct
12786 The mapping must not involve any copying.
12787 Indirect mappings to copies of frames are created in some cases
12789 where either direct mapping is not possible or it would have
12790 unexpected properties. Setting this flag ensures that the
12791 mapping is direct and will fail if that is not possible.
12792 Defaults to read+write if not specified.
12794 derive_device type
12796 Rather than using the device supplied at initialisation, instead
12797 derive a new device of type type from the device the input frames
12798 exist on.
12799 reverse
12801 In a hardware to hardware mapping, map in reverse - create frames
12802 in the sink and map them back to the source. This may be necessary
12803 in some cases where a mapping in one direction is required but only
12804 the opposite direction is supported by the devices being used.
12805 This option is dangerous - it may break the preceding filter in
12807 undefined ways if there are any additional constraints on that
12808 filter's output. Do not use it without fully understanding the
12809 implications of its use.
12810 hwupload
12812 Upload system memory frames to hardware surfaces.
12813 The device to upload to must be supplied when the filter is
12815 initialised. If using ffmpeg, select the appropriate device with the
12816 -filter_hw_device option or with the derive_device option. The input
12817 and output devices must be of different types and compatible - the
12818 exact meaning of this is system-dependent, but typically it means that
12819 they must refer to the same underlying hardware context (for example,
12820 refer to the same graphics card).
12821 The following additional parameters are accepted:
12823 derive_device type
12825 Rather than using the device supplied at initialisation, instead
12826 derive a new device of type type from the device the input frames
12827 exist on.
12828 hwupload_cuda
12830 Upload system memory frames to a CUDA device.
12831 It accepts the following optional parameters:
12833 device
12835 The number of the CUDA device to use
12836 hqx
12838 Apply a high-quality magnification filter designed for pixel art. This
12839 filter was originally created by Maxim Stepin.
12840 It accepts the following option:
12842 n Set the scaling dimension: 2 for "hq2x", 3 for "hq3x" and 4 for
12844 "hq4x". Default is 3.
12845 hstack
12847 Stack input videos horizontally.
12848 All streams must be of same pixel format and of same height.
12850 Note that this filter is faster than using overlay and pad filter to
12852 create same output.
12853 The filter accepts the following option:
12855 inputs
12857 Set number of input streams. Default is 2.
12858 shortest
12860 If set to 1, force the output to terminate when the shortest input
12861 terminates. Default value is 0.
12862 hsvhold
12864 Turns a certain HSV range into gray values.
12865 This filter measures color difference between set HSV color in options
12867 and ones measured in video stream. Depending on options, output colors
12868 can be changed to be gray or not.
12869 The filter accepts the following options:
12871 hue Set the hue value which will be used in color difference
12873 calculation. Allowed range is from -360 to 360. Default value is
12874 0.
12875 sat Set the saturation value which will be used in color difference
12877 calculation. Allowed range is from -1 to 1. Default value is 0.
12878 val Set the value which will be used in color difference calculation.
12880 Allowed range is from -1 to 1. Default value is 0.
12881 similarity
12883 Set similarity percentage with the key color. Allowed range is
12884 from 0 to 1. Default value is 0.01.
12885 0.00001 matches only the exact key color, while 1.0 matches
12887 everything.
12888 blend
12890 Blend percentage. Allowed range is from 0 to 1. Default value is
12891 0.
12892 0.0 makes pixels either fully gray, or not gray at all.
12894 Higher values result in more gray pixels, with a higher gray pixel
12896 the more similar the pixels color is to the key color.
12897 hsvkey
12899 Turns a certain HSV range into transparency.
12900 This filter measures color difference between set HSV color in options
12902 and ones measured in video stream. Depending on options, output colors
12903 can be changed to transparent by adding alpha channel.
12904 The filter accepts the following options:
12906 hue Set the hue value which will be used in color difference
12908 calculation. Allowed range is from -360 to 360. Default value is
12909 0.
12910 sat Set the saturation value which will be used in color difference
12912 calculation. Allowed range is from -1 to 1. Default value is 0.
12913 val Set the value which will be used in color difference calculation.
12915 Allowed range is from -1 to 1. Default value is 0.
12916 similarity
12918 Set similarity percentage with the key color. Allowed range is
12919 from 0 to 1. Default value is 0.01.
12920 0.00001 matches only the exact key color, while 1.0 matches
12922 everything.
12923 blend
12925 Blend percentage. Allowed range is from 0 to 1. Default value is
12926 0.
12927 0.0 makes pixels either fully transparent, or not transparent at
12929 all.
12930 Higher values result in semi-transparent pixels, with a higher
12932 transparency the more similar the pixels color is to the key color.
12933 hue
12935 Modify the hue and/or the saturation of the input.
12936 It accepts the following parameters:
12938 h Specify the hue angle as a number of degrees. It accepts an
12940 expression, and defaults to "0".
12941 s Specify the saturation in the [-10,10] range. It accepts an
12943 expression and defaults to "1".
12944 H Specify the hue angle as a number of radians. It accepts an
12946 expression, and defaults to "0".
12947 b Specify the brightness in the [-10,10] range. It accepts an
12949 expression and defaults to "0".
12950 h and H are mutually exclusive, and can't be specified at the same
12952 time.
12953 The b, h, H and s option values are expressions containing the
12955 following constants:
12956 n frame count of the input frame starting from 0
12958 pts presentation timestamp of the input frame expressed in time base
12960 units
12961 r frame rate of the input video, NAN if the input frame rate is
12963 unknown
12964 t timestamp expressed in seconds, NAN if the input timestamp is
12966 unknown
12967 tb time base of the input video
12969 Examples
12971 • Set the hue to 90 degrees and the saturation to 1.0:
12973 hue=h=90:s=1
12975 • Same command but expressing the hue in radians:
12977 hue=H=PI/2:s=1
12979 • Rotate hue and make the saturation swing between 0 and 2 over a
12981 period of 1 second:
12982 hue="H=2*PI*t: s=sin(2*PI*t)+1"
12984 • Apply a 3 seconds saturation fade-in effect starting at 0:
12986 hue="s=min(t/3\,1)"
12988 The general fade-in expression can be written as:
12990 hue="s=min(0\, max((t-START)/DURATION\, 1))"
12992 • Apply a 3 seconds saturation fade-out effect starting at 5 seconds:
12994 hue="s=max(0\, min(1\, (8-t)/3))"
12996 The general fade-out expression can be written as:
12998 hue="s=max(0\, min(1\, (START+DURATION-t)/DURATION))"
13000 Commands
13002 This filter supports the following commands:
13004 b
13006 s
13007 h
13008 H Modify the hue and/or the saturation and/or brightness of the input
13009 video. The command accepts the same syntax of the corresponding
13010 option.
13011 If the specified expression is not valid, it is kept at its current
13013 value.
13014 huesaturation
13016 Apply hue-saturation-intensity adjustments to input video stream.
13017 This filter operates in RGB colorspace.
13019 This filter accepts the following options:
13021 hue Set the hue shift in degrees to apply. Default is 0. Allowed range
13023 is from -180 to 180.
13024 saturation
13026 Set the saturation shift. Default is 0. Allowed range is from -1
13027 to 1.
13028 intensity
13030 Set the intensity shift. Default is 0. Allowed range is from -1 to
13031 1.
13032 colors
13034 Set which primary and complementary colors are going to be
13035 adjusted. This options is set by providing one or multiple values.
13036 This can select multiple colors at once. By default all colors are
13037 selected.
13038 r Adjust reds.
13040 y Adjust yellows.
13042 g Adjust greens.
13044 c Adjust cyans.
13046 b Adjust blues.
13048 m Adjust magentas.
13050 a Adjust all colors.
13052 strength
13054 Set strength of filtering. Allowed range is from 0 to 100. Default
13055 value is 1.
13056 rw, gw, bw
13058 Set weight for each RGB component. Allowed range is from 0 to 1.
13059 By default is set to 0.333, 0.334, 0.333. Those options are used
13060 in saturation and lightess processing.
13061 lightness
13063 Set preserving lightness, by default is disabled. Adjusting hues
13064 can change lightness from original RGB triplet, with this option
13065 enabled lightness is kept at same value.
13066 hysteresis
13068 Grow first stream into second stream by connecting components. This
13069 makes it possible to build more robust edge masks.
13070 This filter accepts the following options:
13072 planes
13074 Set which planes will be processed as bitmap, unprocessed planes
13075 will be copied from first stream. By default value 0xf, all planes
13076 will be processed.
13077 threshold
13079 Set threshold which is used in filtering. If pixel component value
13080 is higher than this value filter algorithm for connecting
13081 components is activated. By default value is 0.
13082 The "hysteresis" filter also supports the framesync options.
13084 iccdetect
13086 Detect the colorspace from an embedded ICC profile (if present), and
13087 update the frame's tags accordingly.
13088 This filter accepts the following options:
13090 force
13092 If true, the frame's existing colorspace tags will always be
13093 overridden by values detected from an ICC profile. Otherwise, they
13094 will only be assigned if they contain "unknown". Enabled by
13095 default.
13096 iccgen
13098 Generate ICC profiles and attach them to frames.
13099 This filter accepts the following options:
13101 color_primaries
13103 color_trc
13104 Configure the colorspace that the ICC profile will be generated
13105 for. The default value of "auto" infers the value from the input
13106 frame's metadata, defaulting to BT.709/sRGB as appropriate.
13107 See the setparams filter for a list of possible values, but note
13109 that "unknown" are not valid values for this filter.
13110 force
13112 If true, an ICC profile will be generated even if it would
13113 overwrite an already existing ICC profile. Disabled by default.
13114 identity
13116 Obtain the identity score between two input videos.
13117 This filter takes two input videos.
13119 Both input videos must have the same resolution and pixel format for
13121 this filter to work correctly. Also it assumes that both inputs have
13122 the same number of frames, which are compared one by one.
13123 The obtained per component, average, min and max identity score is
13125 printed through the logging system.
13126 The filter stores the calculated identity scores of each frame in frame
13128 metadata.
13129 In the below example the input file main.mpg being processed is
13131 compared with the reference file ref.mpg.
13132 ffmpeg -i main.mpg -i ref.mpg -lavfi identity -f null -
13134 idet
13136 Detect video interlacing type.
13137 This filter tries to detect if the input frames are interlaced,
13139 progressive, top or bottom field first. It will also try to detect
13140 fields that are repeated between adjacent frames (a sign of telecine).
13141 Single frame detection considers only immediately adjacent frames when
13143 classifying each frame. Multiple frame detection incorporates the
13144 classification history of previous frames.
13145 The filter will log these metadata values:
13147 single.current_frame
13149 Detected type of current frame using single-frame detection. One
13150 of: ``tff'' (top field first), ``bff'' (bottom field first),
13151 ``progressive'', or ``undetermined''
13152 single.tff
13154 Cumulative number of frames detected as top field first using
13155 single-frame detection.
13156 multiple.tff
13158 Cumulative number of frames detected as top field first using
13159 multiple-frame detection.
13160 single.bff
13162 Cumulative number of frames detected as bottom field first using
13163 single-frame detection.
13164 multiple.current_frame
13166 Detected type of current frame using multiple-frame detection. One
13167 of: ``tff'' (top field first), ``bff'' (bottom field first),
13168 ``progressive'', or ``undetermined''
13169 multiple.bff
13171 Cumulative number of frames detected as bottom field first using
13172 multiple-frame detection.
13173 single.progressive
13175 Cumulative number of frames detected as progressive using single-
13176 frame detection.
13177 multiple.progressive
13179 Cumulative number of frames detected as progressive using multiple-
13180 frame detection.
13181 single.undetermined
13183 Cumulative number of frames that could not be classified using
13184 single-frame detection.
13185 multiple.undetermined
13187 Cumulative number of frames that could not be classified using
13188 multiple-frame detection.
13189 repeated.current_frame
13191 Which field in the current frame is repeated from the last. One of
13192 ``neither'', ``top'', or ``bottom''.
13193 repeated.neither
13195 Cumulative number of frames with no repeated field.
13196 repeated.top
13198 Cumulative number of frames with the top field repeated from the
13199 previous frame's top field.
13200 repeated.bottom
13202 Cumulative number of frames with the bottom field repeated from the
13203 previous frame's bottom field.
13204 The filter accepts the following options:
13206 intl_thres
13208 Set interlacing threshold.
13209 prog_thres
13211 Set progressive threshold.
13212 rep_thres
13214 Threshold for repeated field detection.
13215 half_life
13217 Number of frames after which a given frame's contribution to the
13218 statistics is halved (i.e., it contributes only 0.5 to its
13219 classification). The default of 0 means that all frames seen are
13220 given full weight of 1.0 forever.
13221 analyze_interlaced_flag
13223 When this is not 0 then idet will use the specified number of
13224 frames to determine if the interlaced flag is accurate, it will not
13225 count undetermined frames. If the flag is found to be accurate it
13226 will be used without any further computations, if it is found to be
13227 inaccurate it will be cleared without any further computations.
13228 This allows inserting the idet filter as a low computational method
13229 to clean up the interlaced flag
13230 il
13232 Deinterleave or interleave fields.
13233 This filter allows one to process interlaced images fields without
13235 deinterlacing them. Deinterleaving splits the input frame into 2 fields
13236 (so called half pictures). Odd lines are moved to the top half of the
13237 output image, even lines to the bottom half. You can process (filter)
13238 them independently and then re-interleave them.
13239 The filter accepts the following options:
13241 luma_mode, l
13243 chroma_mode, c
13244 alpha_mode, a
13245 Available values for luma_mode, chroma_mode and alpha_mode are:
13246 none
13248 Do nothing.
13249 deinterleave, d
13251 Deinterleave fields, placing one above the other.
13252 interleave, i
13254 Interleave fields. Reverse the effect of deinterleaving.
13255 Default value is "none".
13257 luma_swap, ls
13259 chroma_swap, cs
13260 alpha_swap, as
13261 Swap luma/chroma/alpha fields. Exchange even & odd lines. Default
13262 value is 0.
13263 Commands
13265 This filter supports the all above options as commands.
13267 inflate
13269 Apply inflate effect to the video.
13270 This filter replaces the pixel by the local(3x3) average by taking into
13272 account only values higher than the pixel.
13273 It accepts the following options:
13275 threshold0
13277 threshold1
13278 threshold2
13279 threshold3
13280 Limit the maximum change for each plane, default is 65535. If 0,
13281 plane will remain unchanged.
13282 Commands
13284 This filter supports the all above options as commands.
13286 interlace
13288 Simple interlacing filter from progressive contents. This interleaves
13289 upper (or lower) lines from odd frames with lower (or upper) lines from
13290 even frames, halving the frame rate and preserving image height.
13291 Original Original New Frame
13293 Frame 'j' Frame 'j+1' (tff)
13294 ========== =========== ==================
13295 Line 0 --------------------> Frame 'j' Line 0
13296 Line 1 Line 1 ----> Frame 'j+1' Line 1
13297 Line 2 ---------------------> Frame 'j' Line 2
13298 Line 3 Line 3 ----> Frame 'j+1' Line 3
13299 ... ... ...
13300 New Frame + 1 will be generated by Frame 'j+2' and Frame 'j+3' and so on
13301 It accepts the following optional parameters:
13303 scan
13305 This determines whether the interlaced frame is taken from the even
13306 (tff - default) or odd (bff) lines of the progressive frame.
13307 lowpass
13309 Vertical lowpass filter to avoid twitter interlacing and reduce
13310 moire patterns.
13311 0, off
13313 Disable vertical lowpass filter
13314 1, linear
13316 Enable linear filter (default)
13317 2, complex
13319 Enable complex filter. This will slightly less reduce twitter
13320 and moire but better retain detail and subjective sharpness
13321 impression.
13322 kerndeint
13324 Deinterlace input video by applying Donald Graft's adaptive kernel
13325 deinterling. Work on interlaced parts of a video to produce progressive
13326 frames.
13327 The description of the accepted parameters follows.
13329 thresh
13331 Set the threshold which affects the filter's tolerance when
13332 determining if a pixel line must be processed. It must be an
13333 integer in the range [0,255] and defaults to 10. A value of 0 will
13334 result in applying the process on every pixels.
13335 map Paint pixels exceeding the threshold value to white if set to 1.
13337 Default is 0.
13338 order
13340 Set the fields order. Swap fields if set to 1, leave fields alone
13341 if 0. Default is 0.
13342 sharp
13344 Enable additional sharpening if set to 1. Default is 0.
13345 twoway
13347 Enable twoway sharpening if set to 1. Default is 0.
13348 Examples
13350 • Apply default values:
13352 kerndeint=thresh=10:map=0:order=0:sharp=0:twoway=0
13354 • Enable additional sharpening:
13356 kerndeint=sharp=1
13358 • Paint processed pixels in white:
13360 kerndeint=map=1
13362 kirsch
13364 Apply kirsch operator to input video stream.
13365 The filter accepts the following option:
13367 planes
13369 Set which planes will be processed, unprocessed planes will be
13370 copied. By default value 0xf, all planes will be processed.
13371 scale
13373 Set value which will be multiplied with filtered result.
13374 delta
13376 Set value which will be added to filtered result.
13377 Commands
13379 This filter supports the all above options as commands.
13381 lagfun
13383 Slowly update darker pixels.
13384 This filter makes short flashes of light appear longer. This filter
13386 accepts the following options:
13387 decay
13389 Set factor for decaying. Default is .95. Allowed range is from 0 to
13390 1.
13391 planes
13393 Set which planes to filter. Default is all. Allowed range is from 0
13394 to 15.
13395 Commands
13397 This filter supports the all above options as commands.
13399 lenscorrection
13401 Correct radial lens distortion
13402 This filter can be used to correct for radial distortion as can result
13404 from the use of wide angle lenses, and thereby re-rectify the image. To
13405 find the right parameters one can use tools available for example as
13406 part of opencv or simply trial-and-error. To use opencv use the
13407 calibration sample (under samples/cpp) from the opencv sources and
13408 extract the k1 and k2 coefficients from the resulting matrix.
13409 Note that effectively the same filter is available in the open-source
13411 tools Krita and Digikam from the KDE project.
13412 In contrast to the vignette filter, which can also be used to
13414 compensate lens errors, this filter corrects the distortion of the
13415 image, whereas vignette corrects the brightness distribution, so you
13416 may want to use both filters together in certain cases, though you will
13417 have to take care of ordering, i.e. whether vignetting should be
13418 applied before or after lens correction.
13419 Options
13421 The filter accepts the following options:
13423 cx Relative x-coordinate of the focal point of the image, and thereby
13425 the center of the distortion. This value has a range [0,1] and is
13426 expressed as fractions of the image width. Default is 0.5.
13427 cy Relative y-coordinate of the focal point of the image, and thereby
13429 the center of the distortion. This value has a range [0,1] and is
13430 expressed as fractions of the image height. Default is 0.5.
13431 k1 Coefficient of the quadratic correction term. This value has a
13433 range [-1,1]. 0 means no correction. Default is 0.
13434 k2 Coefficient of the double quadratic correction term. This value has
13436 a range [-1,1]. 0 means no correction. Default is 0.
13437 i Set interpolation type. Can be "nearest" or "bilinear". Default is
13439 "nearest".
13440 fc Specify the color of the unmapped pixels. For the syntax of this
13442 option, check the "Color" section in the ffmpeg-utils manual.
13443 Default color is "black@0".
13444 The formula that generates the correction is:
13446 r_src = r_tgt * (1 + k1 * (r_tgt / r_0)^2 + k2 * (r_tgt / r_0)^4)
13448 where r_0 is halve of the image diagonal and r_src and r_tgt are the
13450 distances from the focal point in the source and target images,
13451 respectively.
13452 Commands
13454 This filter supports the all above options as commands.
13456 lensfun
13458 Apply lens correction via the lensfun library
13459 (<http://lensfun.sourceforge.net/>).
13460 The "lensfun" filter requires the camera make, camera model, and lens
13462 model to apply the lens correction. The filter will load the lensfun
13463 database and query it to find the corresponding camera and lens entries
13464 in the database. As long as these entries can be found with the given
13465 options, the filter can perform corrections on frames. Note that
13466 incomplete strings will result in the filter choosing the best match
13467 with the given options, and the filter will output the chosen camera
13468 and lens models (logged with level "info"). You must provide the make,
13469 camera model, and lens model as they are required.
13470 To obtain a list of available makes and models, leave out one or both
13472 of "make" and "model" options. The filter will send the full list to
13473 the log with level "INFO". The first column is the make and the second
13474 column is the model. To obtain a list of available lenses, set any
13475 values for make and model and leave out the "lens_model" option. The
13476 filter will send the full list of lenses in the log with level "INFO".
13477 The ffmpeg tool will exit after the list is printed.
13478 The filter accepts the following options:
13480 make
13482 The make of the camera (for example, "Canon"). This option is
13483 required.
13484 model
13486 The model of the camera (for example, "Canon EOS 100D"). This
13487 option is required.
13488 lens_model
13490 The model of the lens (for example, "Canon EF-S 18-55mm f/3.5-5.6
13491 IS STM"). This option is required.
13492 db_path
13494 The full path to the lens database folder. If not set, the filter
13495 will attempt to load the database from the install path when the
13496 library was built. Default is unset.
13497 mode
13499 The type of correction to apply. The following values are valid
13500 options:
13501 vignetting
13503 Enables fixing lens vignetting.
13504 geometry
13506 Enables fixing lens geometry. This is the default.
13507 subpixel
13509 Enables fixing chromatic aberrations.
13510 vig_geo
13512 Enables fixing lens vignetting and lens geometry.
13513 vig_subpixel
13515 Enables fixing lens vignetting and chromatic aberrations.
13516 distortion
13518 Enables fixing both lens geometry and chromatic aberrations.
13519 all Enables all possible corrections.
13521 focal_length
13523 The focal length of the image/video (zoom; expected constant for
13524 video). For example, a 18--55mm lens has focal length range of
13525 [18--55], so a value in that range should be chosen when using that
13526 lens. Default 18.
13527 aperture
13529 The aperture of the image/video (expected constant for video). Note
13530 that aperture is only used for vignetting correction. Default 3.5.
13531 focus_distance
13533 The focus distance of the image/video (expected constant for
13534 video). Note that focus distance is only used for vignetting and
13535 only slightly affects the vignetting correction process. If
13536 unknown, leave it at the default value (which is 1000).
13537 scale
13539 The scale factor which is applied after transformation. After
13540 correction the video is no longer necessarily rectangular. This
13541 parameter controls how much of the resulting image is visible. The
13542 value 0 means that a value will be chosen automatically such that
13543 there is little or no unmapped area in the output image. 1.0 means
13544 that no additional scaling is done. Lower values may result in more
13545 of the corrected image being visible, while higher values may avoid
13546 unmapped areas in the output.
13547 target_geometry
13549 The target geometry of the output image/video. The following values
13550 are valid options:
13551 rectilinear (default)
13553 fisheye
13554 panoramic
13555 equirectangular
13556 fisheye_orthographic
13557 fisheye_stereographic
13558 fisheye_equisolid
13559 fisheye_thoby
13560 reverse
13561 Apply the reverse of image correction (instead of correcting
13562 distortion, apply it).
13563 interpolation
13565 The type of interpolation used when correcting distortion. The
13566 following values are valid options:
13567 nearest
13569 linear (default)
13570 lanczos
13571 Examples
13573 • Apply lens correction with make "Canon", camera model "Canon EOS
13575 100D", and lens model "Canon EF-S 18-55mm f/3.5-5.6 IS STM" with
13576 focal length of "18" and aperture of "8.0".
13577 ffmpeg -i input.mov -vf lensfun=make=Canon:model="Canon EOS 100D":lens_model="Canon EF-S 18-55mm f/3.5-5.6 IS STM":focal_length=18:aperture=8 -c:v h264 -b:v 8000k output.mov
13579 • Apply the same as before, but only for the first 5 seconds of
13581 video.
13582 ffmpeg -i input.mov -vf lensfun=make=Canon:model="Canon EOS 100D":lens_model="Canon EF-S 18-55mm f/3.5-5.6 IS STM":focal_length=18:aperture=8:enable='lte(t\,5)' -c:v h264 -b:v 8000k output.mov
13584 libplacebo
13586 Flexible GPU-accelerated processing filter based on libplacebo
13587 (<https://code.videolan.org/videolan/libplacebo>). Note that this
13588 filter currently only accepts Vulkan input frames.
13589 Options
13591 The options for this filter are divided into the following sections:
13593 Output mode
13595 These options control the overall output mode. By default, libplacebo
13597 will try to preserve the source colorimetry and size as best as it can,
13598 but it will apply any embedded film grain, dolby vision metadata or
13599 anamorphic SAR present in source frames.
13600 w
13602 h Set the output video dimension expression. Default value is the
13603 input dimension.
13604 Allows for the same expressions as the scale filter.
13606 format
13608 Set the output format override. If unset (the default), frames will
13609 be output in the same format as the respective input frames.
13610 Otherwise, format conversion will be performed.
13611 force_original_aspect_ratio
13613 force_divisible_by
13614 Work the same as the identical scale filter options.
13615 normalize_sar
13617 If enabled (the default), output frames will always have a pixel
13618 aspect ratio of 1:1. If disabled, any aspect ratio mismatches,
13619 including those from e.g. anamorphic video sources, are forwarded
13620 to the output pixel aspect ratio.
13621 pad_crop_ratio
13623 Specifies a ratio (between 0.0 and 1.0) between padding and
13624 cropping when the input aspect ratio does not match the output
13625 aspect ratio and normalize_sar is in effect. The default of 0.0
13626 always pads the content with black borders, while a value of 1.0
13627 always crops off parts of the content. Intermediate values are
13628 possible, leading to a mix of the two approaches.
13629 colorspace
13631 color_primaries
13632 color_trc
13633 range
13634 Configure the colorspace that output frames will be delivered in.
13635 The default value of "auto" outputs frames in the same format as
13636 the input frames, leading to no change. For any other value,
13637 conversion will be performed.
13638 See the setparams filter for a list of possible values.
13640 apply_filmgrain
13642 Apply film grain (e.g. AV1 or H.274) if present in source frames,
13643 and strip it from the output. Enabled by default.
13644 apply_dolbyvision
13646 Apply Dolby Vision RPU metadata if present in source frames, and
13647 strip it from the output. Enabled by default. Note that Dolby
13648 Vision will always output BT.2020+PQ, overriding the usual input
13649 frame metadata. These will also be picked as the values of "auto"
13650 for the respective frame output options.
13651 Scaling
13653 The options in this section control how libplacebo performs upscaling
13655 and (if necessary) downscaling. Note that libplacebo will always
13656 internally operate on 4:4:4 content, so any sub-sampled chroma formats
13657 such as "yuv420p" will necessarily be upsampled and downsampled as part
13658 of the rendering process. That means scaling might be in effect even if
13659 the source and destination resolution are the same.
13660 upscaler
13662 downscaler
13663 Configure the filter kernel used for upscaling and downscaling. The
13664 respective defaults are "spline36" and "mitchell". For a full list
13665 of possible values, pass "help" to these options. The most
13666 important values are:
13667 none
13669 Forces the use of built-in GPU texture sampling (typically
13670 bilinear). Extremely fast but poor quality, especially when
13671 downscaling.
13672 bilinear
13674 Bilinear interpolation. Can generally be done for free on GPUs,
13675 except when doing so would lead to aliasing. Fast and low
13676 quality.
13677 nearest
13679 Nearest-neighbour interpolation. Sharp but highly aliasing.
13680 oversample
13682 Algorithm that looks visually similar to nearest-neighbour
13683 interpolation but tries to preserve pixel aspect ratio. Good
13684 for pixel art, since it results in minimal distortion of the
13685 artistic appearance.
13686 lanczos
13688 Standard sinc-sinc interpolation kernel.
13689 spline36
13691 Cubic spline approximation of lanczos. No difference in
13692 performance, but has very slightly less ringing.
13693 ewa_lanczos
13695 Elliptically weighted average version of lanczos, based on a
13696 jinc-sinc kernel. This is also popularly referred to as just
13697 "Jinc scaling". Slow but very high quality.
13698 gaussian
13700 Gaussian kernel. Has certain ideal mathematical properties, but
13701 subjectively very blurry.
13702 mitchell
13704 Cubic BC spline with parameters recommended by Mitchell and
13705 Netravali. Very little ringing.
13706 lut_entries
13708 Configures the size of scaler LUTs, ranging from 1 to 256. The
13709 default of 0 will pick libplacebo's internal default, typically 64.
13710 antiringing
13712 Enables anti-ringing (for non-EWA filters). The value (between 0.0
13713 and 1.0) configures the strength of the anti-ringing algorithm. May
13714 increase aliasing if set too high. Disabled by default.
13715 sigmoid
13717 Enable sigmoidal compression during upscaling. Reduces ringing
13718 slightly. Enabled by default.
13719 Debanding
13721 Libplacebo comes with a built-in debanding filter that is good at
13723 counteracting many common sources of banding and blocking. Turning this
13724 on is highly recommended whenever quality is desired.
13725 deband
13727 Enable (fast) debanding algorithm. Disabled by default.
13728 deband_iterations
13730 Number of deband iterations of the debanding algorithm. Each
13731 iteration is performed with progressively increased radius (and
13732 diminished threshold). Recommended values are in the range 1 to 4.
13733 Defaults to 1.
13734 deband_threshold
13736 Debanding filter strength. Higher numbers lead to more aggressive
13737 debanding. Defaults to 4.0.
13738 deband_radius
13740 Debanding filter radius. A higher radius is better for slow
13741 gradients, while a lower radius is better for steep gradients.
13742 Defaults to 16.0.
13743 deband_grain
13745 Amount of extra output grain to add. Helps hide imperfections.
13746 Defaults to 6.0.
13747 Color adjustment
13749 A collection of subjective color controls. Not very rigorous, so the
13751 exact effect will vary somewhat depending on the input primaries and
13752 colorspace.
13753 brightness
13755 Brightness boost, between "-1.0" and 1.0. Defaults to 0.0.
13756 contrast
13758 Contrast gain, between 0.0 and 16.0. Defaults to 1.0.
13759 saturation
13761 Saturation gain, between 0.0 and 16.0. Defaults to 1.0.
13762 hue Hue shift in radians, between "-3.14" and 3.14. Defaults to 0.0.
13764 This will rotate the UV subvector, defaulting to BT.709
13765 coefficients for RGB inputs.
13766 gamma
13768 Gamma adjustment, between 0.0 and 16.0. Defaults to 1.0.
13769 cones
13771 Cone model to use for color blindness simulation. Accepts any
13772 combination of "l", "m" and "s". Here are some examples:
13773 m Deuteranomaly / deuteranopia (affecting 3%-4% of the
13775 population)
13776 l Protanomaly / protanopia (affecting 1%-2% of the population)
13778 l+m Monochromacy (very rare)
13780 l+m+s
13782 Achromatopsy (complete loss of daytime vision, extremely rare)
13783 cone-strength
13785 Gain factor for the cones specified by "cones", between 0.0 and
13786 10.0. A value of 1.0 results in no change to color vision. A value
13787 of 0.0 (the default) simulates complete loss of those cones. Values
13788 above 1.0 result in exaggerating the differences between cones,
13789 which may help compensate for reduced color vision.
13790 Peak detection
13792 To help deal with sources that only have static HDR10 metadata (or no
13794 tagging whatsoever), libplacebo uses its own internal frame analysis
13795 compute shader to analyze source frames and adapt the tone mapping
13796 function in realtime. If this is too slow, or if exactly reproducible
13797 frame-perfect results are needed, it's recommended to turn this feature
13798 off.
13799 peak_detect
13801 Enable HDR peak detection. Ignores static MaxCLL/MaxFALL values in
13802 favor of dynamic detection from the input. Note that the detected
13803 values do not get written back to the output frames, they merely
13804 guide the internal tone mapping process. Enabled by default.
13805 smoothing_period
13807 Peak detection smoothing period, between 0.0 and 1000.0. Higher
13808 values result in peak detection becoming less responsive to changes
13809 in the input. Defaults to 100.0.
13810 minimum_peak
13812 Lower bound on the detected peak (relative to SDR white), between
13813 0.0 and 100.0. Defaults to 1.0.
13814 scene_threshold_low
13816 scene_threshold_high
13817 Lower and upper thresholds for scene change detection. Expressed in
13818 a logarithmic scale between 0.0 and 100.0. Default to 5.5 and 10.0,
13819 respectively. Setting either to a negative value disables this
13820 functionality.
13821 overshoot
13823 Peak smoothing overshoot margin, between 0.0 and 1.0. Provides a
13824 safety margin to prevent clipping as a result of peak smoothing.
13825 Defaults to 0.05, corresponding to a margin of 5%.
13826 Tone mapping
13828 The options in this section control how libplacebo performs tone-
13830 mapping and gamut-mapping when dealing with mismatches between wide-
13831 gamut or HDR content. In general, libplacebo relies on accurate source
13832 tagging and mastering display gamut information to produce the best
13833 results.
13834 intent
13836 Rendering intent to use when adapting between different primary
13837 color gamuts (after tone-mapping).
13838 perceptual
13840 Perceptual gamut mapping. Currently equivalent to relative
13841 colorimetric.
13842 relative
13844 Relative colorimetric. This is the default.
13845 absolute
13847 Absolute colorimetric.
13848 saturation
13850 Saturation mapping. Forcibly stretches the source gamut to the
13851 target gamut.
13852 gamut_mode
13854 How to handle out-of-gamut colors that can occur as a result of
13855 colorimetric gamut mapping.
13856 clip
13858 Do nothing, simply clip out-of-range colors to the RGB volume.
13859 This is the default.
13860 warn
13862 Highlight out-of-gamut pixels (by coloring them pink).
13863 darken
13865 Linearly reduces content brightness to preserves saturated
13866 details, followed by clipping the remaining out-of-gamut
13867 colors. As the name implies, this makes everything darker, but
13868 provides a good balance between preserving details and colors.
13869 desaturate
13871 Hard-desaturates out-of-gamut colors towards white, while
13872 preserving the luminance. Has a tendency to shift colors.
13873 tonemapping
13875 Tone-mapping algorithm to use. Available values are:
13876 auto
13878 Automatic selection based on internal heuristics. This is the
13879 default.
13880 clip
13882 Performs no tone-mapping, just clips out-of-range colors.
13883 Retains perfect color accuracy for in-range colors but
13884 completely destroys out-of-range information. Does not perform
13885 any black point adaptation. Not configurable.
13886 bt.2390
13888 EETF from the ITU-R Report BT.2390, a hermite spline roll-off
13889 with linear segment. The knee point offset is configurable.
13890 Note that this parameter defaults to 1.0, rather than the value
13891 of 0.5 from the ITU-R spec.
13892 bt.2446a
13894 EETF from ITU-R Report BT.2446, method A. Designed for well-
13895 mastered HDR sources. Can be used for both forward and inverse
13896 tone mapping. Not configurable.
13897 spline
13899 Simple spline consisting of two polynomials, joined by a single
13900 pivot point. The parameter gives the pivot point (in PQ
13901 space), defaulting to 0.30. Can be used for both forward and
13902 inverse tone mapping.
13903 reinhard
13905 Simple non-linear, global tone mapping algorithm. The parameter
13906 specifies the local contrast coefficient at the display peak.
13907 Essentially, a parameter of 0.5 implies that the reference
13908 white will be about half as bright as when clipping. Defaults
13909 to 0.5, which results in the simplest formulation of this
13910 function.
13911 mobius
13913 Generalization of the reinhard tone mapping algorithm to
13914 support an additional linear slope near black. The tone mapping
13915 parameter indicates the trade-off between the linear section
13916 and the non-linear section. Essentially, for a given parameter
13917 x, every color value below x will be mapped linearly, while
13918 higher values get non-linearly tone-mapped. Values near 1.0
13919 make this curve behave like "clip", while values near 0.0 make
13920 this curve behave like "reinhard". The default value is 0.3,
13921 which provides a good balance between colorimetric accuracy and
13922 preserving out-of-gamut details.
13923 hable
13925 Piece-wise, filmic tone-mapping algorithm developed by John
13926 Hable for use in Uncharted 2, inspired by a similar tone-
13927 mapping algorithm used by Kodak. Popularized by its use in
13928 video games with HDR rendering. Preserves both dark and bright
13929 details very well, but comes with the drawback of changing the
13930 average brightness quite significantly. This is sort of similar
13931 to "reinhard" with parameter 0.24.
13932 gamma
13934 Fits a gamma (power) function to transfer between the source
13935 and target color spaces, effectively resulting in a perceptual
13936 hard-knee joining two roughly linear sections. This preserves
13937 details at all scales fairly accurately, but can result in an
13938 image with a muted or dull appearance. The parameter is used as
13939 the cutoff point, defaulting to 0.5.
13940 linear
13942 Linearly stretches the input range to the output range, in PQ
13943 space. This will preserve all details accurately, but results
13944 in a significantly different average brightness. Can be used
13945 for inverse tone-mapping in addition to regular tone-mapping.
13946 The parameter can be used as an additional linear gain
13947 coefficient (defaulting to 1.0).
13948 tonemapping_param
13950 For tunable tone mapping functions, this parameter can be used to
13951 fine-tune the curve behavior. Refer to the documentation of
13952 "tonemapping". The default value of 0.0 is replaced by the curve's
13953 preferred default setting.
13954 tonemapping_mode
13956 This option determines how the tone mapping function specified by
13957 "tonemapping" is applied to the colors in a scene. Possible values
13958 are:
13959 auto
13961 Automatic selection based on internal heuristics. This is the
13962 default.
13963 rgb Apply the function per-channel in the RGB colorspace. Per-
13965 channel tone-mapping in RGB. Guarantees no clipping and heavily
13966 desaturates the output, but distorts the colors quite
13967 significantly. Very similar to the "Hollywood" look and feel.
13968 max Tone-mapping is performed on the brightest component found in
13970 the signal. Good at preserving details in highlights, but has a
13971 tendency to crush blacks.
13972 hybrid
13974 Tone-map per-channel for highlights and linearly (luma-based)
13975 for midtones/shadows, based on a fixed gamma 2.4 coefficient
13976 curve.
13977 luma
13979 Tone-map linearly on the luma component (CIE Y), and adjust
13980 (desaturate) the chromaticities to compensate using a simple
13981 constant factor. This is essentially the mode used in ITU-R
13982 BT.2446 method A.
13983 inverse_tonemapping
13985 If enabled, this filter will also attempt stretching SDR signals to
13986 fill HDR output color volumes. Disabled by default.
13987 tonemapping_crosstalk
13989 Extra tone-mapping crosstalk factor, between 0.0 and 0.3. This can
13990 help reduce issues tone-mapping certain bright spectral colors.
13991 Defaults to 0.04.
13992 tonemapping_lut_size
13994 Size of the tone-mapping LUT, between 2 and 1024. Defaults to 256.
13995 Note that this figure is squared when combined with "peak_detect".
13996 Dithering
13998 By default, libplacebo will dither whenever necessary, which includes
14000 rendering to any integer format below 16-bit precision. It's
14001 recommended to always leave this on, since not doing so may result in
14002 visible banding in the output, even if the "debanding" filter is
14003 enabled. If maximum performance is needed, use "ordered_fixed" instead
14004 of disabling dithering.
14005 dithering
14007 Dithering method to use. Accepts the following values:
14008 none
14010 Disables dithering completely. May result in visible banding.
14011 blue
14013 Dither with pseudo-blue noise. This is the default.
14014 ordered
14016 Tunable ordered dither pattern.
14017 ordered_fixed
14019 Faster ordered dither with a fixed size of 6. Texture-less.
14020 white
14022 Dither with white noise. Texture-less.
14023 dither_lut_size
14025 Dither LUT size, as log base2 between 1 and 8. Defaults to 6,
14026 corresponding to a LUT size of "64x64".
14027 dither_temporal
14029 Enables temporal dithering. Disabled by default.
14030 Custom shaders
14032 libplacebo supports a number of custom shaders based on the mpv .hook
14034 GLSL syntax. A collection of such shaders can be found here:
14035 <https://github.com/mpv-player/mpv/wiki/User-Scripts#user-shaders>
14036 A full description of the mpv shader format is beyond the scope of this
14038 section, but a summary can be found here:
14039 <https://mpv.io/manual/master/#options-glsl-shader>
14040 custom_shader_path
14042 Specifies a path to a custom shader file to load at runtime.
14043 custom_shader_bin
14045 Specifies a complete custom shader as a raw string.
14046 Debugging / performance
14048 All of the options in this section default off. They may be of
14050 assistance when attempting to squeeze the maximum performance at the
14051 cost of quality.
14052 skip_aa
14054 Disable anti-aliasing when downscaling.
14055 polar_cutoff
14057 Truncate polar (EWA) scaler kernels below this absolute magnitude,
14058 between 0.0 and 1.0.
14059 disable_linear
14061 Disable linear light scaling.
14062 disable_builtin
14064 Disable built-in GPU sampling (forces LUT).
14065 force_icc_lut
14067 Force the use of a full ICC 3DLUT for gamut mapping.
14068 disable_fbos
14070 Forcibly disable FBOs, resulting in loss of almost all
14071 functionality, but offering the maximum possible speed.
14072 Commands
14074 This filter supports almost all of the above options as commands.
14076 Examples
14078 • Complete example for how to initialize the Vulkan device, upload
14080 frames to the GPU, perform filter conversion to yuv420p, and
14081 download frames back to the CPU for output. Note that in specific
14082 cases you can get around the need to perform format conversion by
14083 specifying the correct "format" filter option corresponding to the
14084 input frames.
14085 ffmpeg -i $INPUT -init_hw_device vulkan -vf hwupload,libplacebo=format=yuv420p,hwdownload,format=yuv420p $OUTPUT
14087 • Tone-map input to standard gamut BT.709 output:
14089 libplacebo=colorspace=bt709:color_primaries=bt709:color_trc=bt709:range=tv
14091 • Rescale input to fit into standard 1080p, with high quality
14093 scaling:
14094 libplacebo=w=1920:h=1080:force_original_aspect_ratio=decrease:normalize_sar=true:upscaler=ewa_lanczos:downscaler=ewa_lanczos
14096 • Convert input to standard sRGB JPEG:
14098 libplacebo=format=yuv420p:colorspace=bt470bg:color_primaries=bt709:color_trc=iec61966-2-1:range=pc
14100 • Use higher quality debanding settings:
14102 libplacebo=deband=true:deband_iterations=3:deband_radius=8:deband_threshold=6
14104 • Run this filter on the CPU, on systems with Mesa installed (and
14106 with the most expensive options disabled):
14107 ffmpeg ... -init_hw_device vulkan:llvmpipe ... -vf libplacebo=upscaler=none:downscaler=none:peak_detect=false
14109 • Suppress CPU-based AV1/H.274 film grain application in the decoder,
14111 in favor of doing it with this filter. Note that this is only a
14112 gain if the frames are either already on the GPU, or if you're
14113 using libplacebo for other purposes, since otherwise the VRAM
14114 roundtrip will more than offset any expected speedup.
14115 ffmpeg -export_side_data +film_grain ... -vf libplacebo=apply_filmgrain=true
14117 libvmaf
14119 Calulate the VMAF (Video Multi-Method Assessment Fusion) score for a
14120 reference/distorted pair of input videos.
14121 The first input is the distorted video, and the second input is the
14123 reference video.
14124 The obtained VMAF score is printed through the logging system.
14126 It requires Netflix's vmaf library (libvmaf) as a pre-requisite. After
14128 installing the library it can be enabled using: "./configure
14129 --enable-libvmaf".
14130 The filter has following options:
14132 model
14134 A `|` delimited list of vmaf models. Each model can be configured
14135 with a number of parameters. Default value: "version=vmaf_v0.6.1"
14136 model_path
14138 Deprecated, use model='path=...'.
14139 enable_transform
14141 Deprecated, use model='enable_transform=true'.
14142 phone_model
14144 Deprecated, use model='enable_transform=true'.
14145 enable_conf_interval
14147 Deprecated, use model='enable_conf_interval=true'.
14148 feature
14150 A `|` delimited list of features. Each feature can be configured
14151 with a number of parameters.
14152 psnr
14154 Deprecated, use feature='name=psnr'.
14155 ssim
14157 Deprecated, use feature='name=ssim'.
14158 ms_ssim
14160 Deprecated, use feature='name=ms_ssim'.
14161 log_path
14163 Set the file path to be used to store log files.
14164 log_fmt
14166 Set the format of the log file (xml, json, csv, or sub).
14167 n_threads
14169 Set number of threads to be used when initializing libvmaf.
14170 Default value: 0, no threads.
14171 n_subsample
14173 Set frame subsampling interval to be used.
14174 This filter also supports the framesync options.
14176 Examples
14178 • In the examples below, a distorted video distorted.mpg is compared
14180 with a reference file reference.mpg.
14181 • Basic usage:
14183 ffmpeg -i distorted.mpg -i reference.mpg -lavfi libvmaf=log_path=output.xml -f null -
14185 • Example with multiple models:
14187 ffmpeg -i distorted.mpg -i reference.mpg -lavfi libvmaf='model=version=vmaf_v0.6.1\\:name=vmaf|version=vmaf_v0.6.1neg\\:name=vmaf_neg' -f null -
14189 • Example with multiple addtional features:
14191 ffmpeg -i distorted.mpg -i reference.mpg -lavfi libvmaf='feature=name=psnr|name=ciede' -f null -
14193 • Example with options and different containers:
14195 ffmpeg -i distorted.mpg -i reference.mkv -lavfi "[0:v]settb=AVTB,setpts=PTS-STARTPTS[main];[1:v]settb=AVTB,setpts=PTS-STARTPTS[ref];[main][ref]libvmaf=log_fmt=json:log_path=output.json" -f null -
14197 limitdiff
14199 Apply limited difference filter using second and optionally third video
14200 stream.
14201 The filter accepts the following options:
14203 threshold
14205 Set the threshold to use when allowing certain differences between
14206 video streams. Any absolute difference value lower or exact than
14207 this threshold will pick pixel components from first video stream.
14208 elasticity
14210 Set the elasticity of soft thresholding when processing video
14211 streams. This value multiplied with first one sets second
14212 threshold. Any absolute difference value greater or exact than
14213 second threshold will pick pixel components from second video
14214 stream. For values between those two threshold linear interpolation
14215 between first and second video stream will be used.
14216 reference
14218 Enable the reference (third) video stream processing. By default is
14219 disabled. If set, this video stream will be used for calculating
14220 absolute difference with first video stream.
14221 planes
14223 Specify which planes will be processed. Defaults to all available.
14224 Commands
14226 This filter supports the all above options as commands except option
14228 reference.
14229 limiter
14231 Limits the pixel components values to the specified range [min, max].
14232 The filter accepts the following options:
14234 min Lower bound. Defaults to the lowest allowed value for the input.
14236 max Upper bound. Defaults to the highest allowed value for the input.
14238 planes
14240 Specify which planes will be processed. Defaults to all available.
14241 Commands
14243 This filter supports the all above options as commands.
14245 loop
14247 Loop video frames.
14248 The filter accepts the following options:
14250 loop
14252 Set the number of loops. Setting this value to -1 will result in
14253 infinite loops. Default is 0.
14254 size
14256 Set maximal size in number of frames. Default is 0.
14257 start
14259 Set first frame of loop. Default is 0.
14260 Examples
14262 • Loop single first frame infinitely:
14264 loop=loop=-1:size=1:start=0
14266 • Loop single first frame 10 times:
14268 loop=loop=10:size=1:start=0
14270 • Loop 10 first frames 5 times:
14272 loop=loop=5:size=10:start=0
14274 lut1d
14276 Apply a 1D LUT to an input video.
14277 The filter accepts the following options:
14279 file
14281 Set the 1D LUT file name.
14282 Currently supported formats:
14284 cube
14286 Iridas
14287 csp cineSpace
14289 interp
14291 Select interpolation mode.
14292 Available values are:
14294 nearest
14296 Use values from the nearest defined point.
14297 linear
14299 Interpolate values using the linear interpolation.
14300 cosine
14302 Interpolate values using the cosine interpolation.
14303 cubic
14305 Interpolate values using the cubic interpolation.
14306 spline
14308 Interpolate values using the spline interpolation.
14309 Commands
14311 This filter supports the all above options as commands.
14313 lut3d
14315 Apply a 3D LUT to an input video.
14316 The filter accepts the following options:
14318 file
14320 Set the 3D LUT file name.
14321 Currently supported formats:
14323 3dl AfterEffects
14325 cube
14327 Iridas
14328 dat DaVinci
14330 m3d Pandora
14332 csp cineSpace
14334 interp
14336 Select interpolation mode.
14337 Available values are:
14339 nearest
14341 Use values from the nearest defined point.
14342 trilinear
14344 Interpolate values using the 8 points defining a cube.
14345 tetrahedral
14347 Interpolate values using a tetrahedron.
14348 pyramid
14350 Interpolate values using a pyramid.
14351 prism
14353 Interpolate values using a prism.
14354 Commands
14356 This filter supports the "interp" option as commands.
14358 lumakey
14360 Turn certain luma values into transparency.
14361 The filter accepts the following options:
14363 threshold
14365 Set the luma which will be used as base for transparency. Default
14366 value is 0.
14367 tolerance
14369 Set the range of luma values to be keyed out. Default value is
14370 0.01.
14371 softness
14373 Set the range of softness. Default value is 0. Use this to control
14374 gradual transition from zero to full transparency.
14375 Commands
14377 This filter supports same commands as options. The command accepts the
14379 same syntax of the corresponding option.
14380 If the specified expression is not valid, it is kept at its current
14382 value.
14383 lut, lutrgb, lutyuv
14385 Compute a look-up table for binding each pixel component input value to
14386 an output value, and apply it to the input video.
14387 lutyuv applies a lookup table to a YUV input video, lutrgb to an RGB
14389 input video.
14390 These filters accept the following parameters:
14392 c0 set first pixel component expression
14394 c1 set second pixel component expression
14396 c2 set third pixel component expression
14398 c3 set fourth pixel component expression, corresponds to the alpha
14400 component
14401 r set red component expression
14403 g set green component expression
14405 b set blue component expression
14407 a alpha component expression
14409 y set Y/luminance component expression
14411 u set U/Cb component expression
14413 v set V/Cr component expression
14415 Each of them specifies the expression to use for computing the lookup
14417 table for the corresponding pixel component values.
14418 The exact component associated to each of the c* options depends on the
14420 format in input.
14421 The lut filter requires either YUV or RGB pixel formats in input,
14423 lutrgb requires RGB pixel formats in input, and lutyuv requires YUV.
14424 The expressions can contain the following constants and functions:
14426 w
14428 h The input width and height.
14429 val The input value for the pixel component.
14431 clipval
14433 The input value, clipped to the minval-maxval range.
14434 maxval
14436 The maximum value for the pixel component.
14437 minval
14439 The minimum value for the pixel component.
14440 negval
14442 The negated value for the pixel component value, clipped to the
14443 minval-maxval range; it corresponds to the expression
14444 "maxval-clipval+minval".
14445 clip(val)
14447 The computed value in val, clipped to the minval-maxval range.
14448 gammaval(gamma)
14450 The computed gamma correction value of the pixel component value,
14451 clipped to the minval-maxval range. It corresponds to the
14452 expression
14453 "pow((clipval-minval)/(maxval-minval)\,gamma)*(maxval-minval)+minval"
14454 All expressions default to "clipval".
14456 Commands
14458 This filter supports same commands as options.
14460 Examples
14462 • Negate input video:
14464 lutrgb="r=maxval+minval-val:g=maxval+minval-val:b=maxval+minval-val"
14466 lutyuv="y=maxval+minval-val:u=maxval+minval-val:v=maxval+minval-val"
14467 The above is the same as:
14469 lutrgb="r=negval:g=negval:b=negval"
14471 lutyuv="y=negval:u=negval:v=negval"
14472 • Negate luminance:
14474 lutyuv=y=negval
14476 • Remove chroma components, turning the video into a graytone image:
14478 lutyuv="u=128:v=128"
14480 • Apply a luma burning effect:
14482 lutyuv="y=2*val"
14484 • Remove green and blue components:
14486 lutrgb="g=0:b=0"
14488 • Set a constant alpha channel value on input:
14490 format=rgba,lutrgb=a="maxval-minval/2"
14492 • Correct luminance gamma by a factor of 0.5:
14494 lutyuv=y=gammaval(0.5)
14496 • Discard least significant bits of luma:
14498 lutyuv=y='bitand(val, 128+64+32)'
14500 • Technicolor like effect:
14502 lutyuv=u='(val-maxval/2)*2+maxval/2':v='(val-maxval/2)*2+maxval/2'
14504 lut2, tlut2
14506 The "lut2" filter takes two input streams and outputs one stream.
14507 The "tlut2" (time lut2) filter takes two consecutive frames from one
14509 single stream.
14510 This filter accepts the following parameters:
14512 c0 set first pixel component expression
14514 c1 set second pixel component expression
14516 c2 set third pixel component expression
14518 c3 set fourth pixel component expression, corresponds to the alpha
14520 component
14521 d set output bit depth, only available for "lut2" filter. By default
14523 is 0, which means bit depth is automatically picked from first
14524 input format.
14525 The "lut2" filter also supports the framesync options.
14527 Each of them specifies the expression to use for computing the lookup
14529 table for the corresponding pixel component values.
14530 The exact component associated to each of the c* options depends on the
14532 format in inputs.
14533 The expressions can contain the following constants:
14535 w
14537 h The input width and height.
14538 x The first input value for the pixel component.
14540 y The second input value for the pixel component.
14542 bdx The first input video bit depth.
14544 bdy The second input video bit depth.
14546 All expressions default to "x".
14548 Commands
14550 This filter supports the all above options as commands except option
14552 "d".
14553 Examples
14555 • Highlight differences between two RGB video streams:
14557 lut2='ifnot(x-y,0,pow(2,bdx)-1):ifnot(x-y,0,pow(2,bdx)-1):ifnot(x-y,0,pow(2,bdx)-1)'
14559 • Highlight differences between two YUV video streams:
14561 lut2='ifnot(x-y,0,pow(2,bdx)-1):ifnot(x-y,pow(2,bdx-1),pow(2,bdx)-1):ifnot(x-y,pow(2,bdx-1),pow(2,bdx)-1)'
14563 • Show max difference between two video streams:
14565 lut2='if(lt(x,y),0,if(gt(x,y),pow(2,bdx)-1,pow(2,bdx-1))):if(lt(x,y),0,if(gt(x,y),pow(2,bdx)-1,pow(2,bdx-1))):if(lt(x,y),0,if(gt(x,y),pow(2,bdx)-1,pow(2,bdx-1)))'
14567 maskedclamp
14569 Clamp the first input stream with the second input and third input
14570 stream.
14571 Returns the value of first stream to be between second input stream -
14573 "undershoot" and third input stream + "overshoot".
14574 This filter accepts the following options:
14576 undershoot
14578 Default value is 0.
14579 overshoot
14581 Default value is 0.
14582 planes
14584 Set which planes will be processed as bitmap, unprocessed planes
14585 will be copied from first stream. By default value 0xf, all planes
14586 will be processed.
14587 Commands
14589 This filter supports the all above options as commands.
14591 maskedmax
14593 Merge the second and third input stream into output stream using
14594 absolute differences between second input stream and first input stream
14595 and absolute difference between third input stream and first input
14596 stream. The picked value will be from second input stream if second
14597 absolute difference is greater than first one or from third input
14598 stream otherwise.
14599 This filter accepts the following options:
14601 planes
14603 Set which planes will be processed as bitmap, unprocessed planes
14604 will be copied from first stream. By default value 0xf, all planes
14605 will be processed.
14606 Commands
14608 This filter supports the all above options as commands.
14610 maskedmerge
14612 Merge the first input stream with the second input stream using per
14613 pixel weights in the third input stream.
14614 A value of 0 in the third stream pixel component means that pixel
14616 component from first stream is returned unchanged, while maximum value
14617 (eg. 255 for 8-bit videos) means that pixel component from second
14618 stream is returned unchanged. Intermediate values define the amount of
14619 merging between both input stream's pixel components.
14620 This filter accepts the following options:
14622 planes
14624 Set which planes will be processed as bitmap, unprocessed planes
14625 will be copied from first stream. By default value 0xf, all planes
14626 will be processed.
14627 Commands
14629 This filter supports the all above options as commands.
14631 maskedmin
14633 Merge the second and third input stream into output stream using
14634 absolute differences between second input stream and first input stream
14635 and absolute difference between third input stream and first input
14636 stream. The picked value will be from second input stream if second
14637 absolute difference is less than first one or from third input stream
14638 otherwise.
14639 This filter accepts the following options:
14641 planes
14643 Set which planes will be processed as bitmap, unprocessed planes
14644 will be copied from first stream. By default value 0xf, all planes
14645 will be processed.
14646 Commands
14648 This filter supports the all above options as commands.
14650 maskedthreshold
14652 Pick pixels comparing absolute difference of two video streams with
14653 fixed threshold.
14654 If absolute difference between pixel component of first and second
14656 video stream is equal or lower than user supplied threshold than pixel
14657 component from first video stream is picked, otherwise pixel component
14658 from second video stream is picked.
14659 This filter accepts the following options:
14661 threshold
14663 Set threshold used when picking pixels from absolute difference
14664 from two input video streams.
14665 planes
14667 Set which planes will be processed as bitmap, unprocessed planes
14668 will be copied from second stream. By default value 0xf, all
14669 planes will be processed.
14670 Commands
14672 This filter supports the all above options as commands.
14674 maskfun
14676 Create mask from input video.
14677 For example it is useful to create motion masks after "tblend" filter.
14679 This filter accepts the following options:
14681 low Set low threshold. Any pixel component lower or exact than this
14683 value will be set to 0.
14684 high
14686 Set high threshold. Any pixel component higher than this value will
14687 be set to max value allowed for current pixel format.
14688 planes
14690 Set planes to filter, by default all available planes are filtered.
14691 fill
14693 Fill all frame pixels with this value.
14694 sum Set max average pixel value for frame. If sum of all pixel
14696 components is higher that this average, output frame will be
14697 completely filled with value set by fill option. Typically useful
14698 for scene changes when used in combination with "tblend" filter.
14699 Commands
14701 This filter supports the all above options as commands.
14703 mcdeint
14705 Apply motion-compensation deinterlacing.
14706 It needs one field per frame as input and must thus be used together
14708 with yadif=1/3 or equivalent.
14709 This filter is only available in ffmpeg version 4.4 or earlier.
14711 This filter accepts the following options:
14713 mode
14715 Set the deinterlacing mode.
14716 It accepts one of the following values:
14718 fast
14720 medium
14721 slow
14722 use iterative motion estimation
14723 extra_slow
14725 like slow, but use multiple reference frames.
14726 Default value is fast.
14728 parity
14730 Set the picture field parity assumed for the input video. It must
14731 be one of the following values:
14732 0, tff
14734 assume top field first
14735 1, bff
14737 assume bottom field first
14738 Default value is bff.
14740 qp Set per-block quantization parameter (QP) used by the internal
14742 encoder.
14743 Higher values should result in a smoother motion vector field but
14745 less optimal individual vectors. Default value is 1.
14746 median
14748 Pick median pixel from certain rectangle defined by radius.
14749 This filter accepts the following options:
14751 radius
14753 Set horizontal radius size. Default value is 1. Allowed range is
14754 integer from 1 to 127.
14755 planes
14757 Set which planes to process. Default is 15, which is all available
14758 planes.
14759 radiusV
14761 Set vertical radius size. Default value is 0. Allowed range is
14762 integer from 0 to 127. If it is 0, value will be picked from
14763 horizontal "radius" option.
14764 percentile
14766 Set median percentile. Default value is 0.5. Default value of 0.5
14767 will pick always median values, while 0 will pick minimum values,
14768 and 1 maximum values.
14769 Commands
14771 This filter supports same commands as options. The command accepts the
14773 same syntax of the corresponding option.
14774 If the specified expression is not valid, it is kept at its current
14776 value.
14777 mergeplanes
14779 Merge color channel components from several video streams.
14780 The filter accepts up to 4 input streams, and merge selected input
14782 planes to the output video.
14783 This filter accepts the following options:
14785 mapping
14787 Set input to output plane mapping. Default is 0.
14788 The mappings is specified as a bitmap. It should be specified as a
14790 hexadecimal number in the form 0xAa[Bb[Cc[Dd]]]. 'Aa' describes the
14791 mapping for the first plane of the output stream. 'A' sets the
14792 number of the input stream to use (from 0 to 3), and 'a' the plane
14793 number of the corresponding input to use (from 0 to 3). The rest of
14794 the mappings is similar, 'Bb' describes the mapping for the output
14795 stream second plane, 'Cc' describes the mapping for the output
14796 stream third plane and 'Dd' describes the mapping for the output
14797 stream fourth plane.
14798 format
14800 Set output pixel format. Default is "yuva444p".
14801 map0s
14803 map1s
14804 map2s
14805 map3s
14806 Set input to output stream mapping for output Nth plane. Default is
14807 0.
14808 map0p
14810 map1p
14811 map2p
14812 map3p
14813 Set input to output plane mapping for output Nth plane. Default is
14814 0.
14815 Examples
14817 • Merge three gray video streams of same width and height into single
14819 video stream:
14820 [a0][a1][a2]mergeplanes=0x001020:yuv444p
14822 • Merge 1st yuv444p stream and 2nd gray video stream into yuva444p
14824 video stream:
14825 [a0][a1]mergeplanes=0x00010210:yuva444p
14827 • Swap Y and A plane in yuva444p stream:
14829 format=yuva444p,mergeplanes=0x03010200:yuva444p
14831 • Swap U and V plane in yuv420p stream:
14833 format=yuv420p,mergeplanes=0x000201:yuv420p
14835 • Cast a rgb24 clip to yuv444p:
14837 format=rgb24,mergeplanes=0x000102:yuv444p
14839 mestimate
14841 Estimate and export motion vectors using block matching algorithms.
14842 Motion vectors are stored in frame side data to be used by other
14843 filters.
14844 This filter accepts the following options:
14846 method
14848 Specify the motion estimation method. Accepts one of the following
14849 values:
14850 esa Exhaustive search algorithm.
14852 tss Three step search algorithm.
14854 tdls
14856 Two dimensional logarithmic search algorithm.
14857 ntss
14859 New three step search algorithm.
14860 fss Four step search algorithm.
14862 ds Diamond search algorithm.
14864 hexbs
14866 Hexagon-based search algorithm.
14867 epzs
14869 Enhanced predictive zonal search algorithm.
14870 umh Uneven multi-hexagon search algorithm.
14872 Default value is esa.
14874 mb_size
14876 Macroblock size. Default 16.
14877 search_param
14879 Search parameter. Default 7.
14880 midequalizer
14882 Apply Midway Image Equalization effect using two video streams.
14883 Midway Image Equalization adjusts a pair of images to have the same
14885 histogram, while maintaining their dynamics as much as possible. It's
14886 useful for e.g. matching exposures from a pair of stereo cameras.
14887 This filter has two inputs and one output, which must be of same pixel
14889 format, but may be of different sizes. The output of filter is first
14890 input adjusted with midway histogram of both inputs.
14891 This filter accepts the following option:
14893 planes
14895 Set which planes to process. Default is 15, which is all available
14896 planes.
14897 minterpolate
14899 Convert the video to specified frame rate using motion interpolation.
14900 This filter accepts the following options:
14902 fps Specify the output frame rate. This can be rational e.g.
14904 "60000/1001". Frames are dropped if fps is lower than source fps.
14905 Default 60.
14906 mi_mode
14908 Motion interpolation mode. Following values are accepted:
14909 dup Duplicate previous or next frame for interpolating new ones.
14911 blend
14913 Blend source frames. Interpolated frame is mean of previous and
14914 next frames.
14915 mci Motion compensated interpolation. Following options are
14917 effective when this mode is selected:
14918 mc_mode
14920 Motion compensation mode. Following values are accepted:
14921 obmc
14923 Overlapped block motion compensation.
14924 aobmc
14926 Adaptive overlapped block motion compensation. Window
14927 weighting coefficients are controlled adaptively
14928 according to the reliabilities of the neighboring
14929 motion vectors to reduce oversmoothing.
14930 Default mode is obmc.
14932 me_mode
14934 Motion estimation mode. Following values are accepted:
14935 bidir
14937 Bidirectional motion estimation. Motion vectors are
14938 estimated for each source frame in both forward and
14939 backward directions.
14940 bilat
14942 Bilateral motion estimation. Motion vectors are
14943 estimated directly for interpolated frame.
14944 Default mode is bilat.
14946 me The algorithm to be used for motion estimation. Following
14948 values are accepted:
14949 esa Exhaustive search algorithm.
14951 tss Three step search algorithm.
14953 tdls
14955 Two dimensional logarithmic search algorithm.
14956 ntss
14958 New three step search algorithm.
14959 fss Four step search algorithm.
14961 ds Diamond search algorithm.
14963 hexbs
14965 Hexagon-based search algorithm.
14966 epzs
14968 Enhanced predictive zonal search algorithm.
14969 umh Uneven multi-hexagon search algorithm.
14971 Default algorithm is epzs.
14973 mb_size
14975 Macroblock size. Default 16.
14976 search_param
14978 Motion estimation search parameter. Default 32.
14979 vsbmc
14981 Enable variable-size block motion compensation. Motion
14982 estimation is applied with smaller block sizes at object
14983 boundaries in order to make the them less blur. Default is
14984 0 (disabled).
14985 scd Scene change detection method. Scene change leads motion vectors to
14987 be in random direction. Scene change detection replace interpolated
14988 frames by duplicate ones. May not be needed for other modes.
14989 Following values are accepted:
14990 none
14992 Disable scene change detection.
14993 fdiff
14995 Frame difference. Corresponding pixel values are compared and
14996 if it satisfies scd_threshold scene change is detected.
14997 Default method is fdiff.
14999 scd_threshold
15001 Scene change detection threshold. Default is 10..
15002 mix
15004 Mix several video input streams into one video stream.
15005 A description of the accepted options follows.
15007 inputs
15009 The number of inputs. If unspecified, it defaults to 2.
15010 weights
15012 Specify weight of each input video stream as sequence. Each weight
15013 is separated by space. If number of weights is smaller than number
15014 of frames last specified weight will be used for all remaining
15015 unset weights.
15016 scale
15018 Specify scale, if it is set it will be multiplied with sum of each
15019 weight multiplied with pixel values to give final destination pixel
15020 value. By default scale is auto scaled to sum of weights.
15021 planes
15023 Set which planes to filter. Default is all. Allowed range is from 0
15024 to 15.
15025 duration
15027 Specify how end of stream is determined.
15028 longest
15030 The duration of the longest input. (default)
15031 shortest
15033 The duration of the shortest input.
15034 first
15036 The duration of the first input.
15037 Commands
15039 This filter supports the following commands:
15041 weights
15043 scale
15044 planes
15045 Syntax is same as option with same name.
15046 monochrome
15048 Convert video to gray using custom color filter.
15049 A description of the accepted options follows.
15051 cb Set the chroma blue spot. Allowed range is from -1 to 1. Default
15053 value is 0.
15054 cr Set the chroma red spot. Allowed range is from -1 to 1. Default
15056 value is 0.
15057 size
15059 Set the color filter size. Allowed range is from .1 to 10. Default
15060 value is 1.
15061 high
15063 Set the highlights strength. Allowed range is from 0 to 1. Default
15064 value is 0.
15065 Commands
15067 This filter supports the all above options as commands.
15069 morpho
15071 This filter allows to apply main morphological grayscale transforms,
15072 erode and dilate with arbitrary structures set in second input stream.
15073 Unlike naive implementation and much slower performance in erosion and
15075 dilation filters, when speed is critical "morpho" filter should be used
15076 instead.
15077 A description of accepted options follows,
15079 mode
15081 Set morphological transform to apply, can be:
15082 erode
15084 dilate
15085 open
15086 close
15087 gradient
15088 tophat
15089 blackhat
15090 Default is "erode".
15092 planes
15094 Set planes to filter, by default all planes except alpha are
15095 filtered.
15096 structure
15098 Set which structure video frames will be processed from second
15099 input stream, can be first or all. Default is all.
15100 The "morpho" filter also supports the framesync options.
15102 Commands
15104 This filter supports same commands as options.
15106 mpdecimate
15108 Drop frames that do not differ greatly from the previous frame in order
15109 to reduce frame rate.
15110 The main use of this filter is for very-low-bitrate encoding (e.g.
15112 streaming over dialup modem), but it could in theory be used for fixing
15113 movies that were inverse-telecined incorrectly.
15114 A description of the accepted options follows.
15116 max Set the maximum number of consecutive frames which can be dropped
15118 (if positive), or the minimum interval between dropped frames (if
15119 negative). If the value is 0, the frame is dropped disregarding the
15120 number of previous sequentially dropped frames.
15121 Default value is 0.
15123 hi
15125 lo
15126 frac
15127 Set the dropping threshold values.
15128 Values for hi and lo are for 8x8 pixel blocks and represent actual
15130 pixel value differences, so a threshold of 64 corresponds to 1 unit
15131 of difference for each pixel, or the same spread out differently
15132 over the block.
15133 A frame is a candidate for dropping if no 8x8 blocks differ by more
15135 than a threshold of hi, and if no more than frac blocks (1 meaning
15136 the whole image) differ by more than a threshold of lo.
15137 Default value for hi is 64*12, default value for lo is 64*5, and
15139 default value for frac is 0.33.
15140 msad
15142 Obtain the MSAD (Mean Sum of Absolute Differences) between two input
15143 videos.
15144 This filter takes two input videos.
15146 Both input videos must have the same resolution and pixel format for
15148 this filter to work correctly. Also it assumes that both inputs have
15149 the same number of frames, which are compared one by one.
15150 The obtained per component, average, min and max MSAD is printed
15152 through the logging system.
15153 The filter stores the calculated MSAD of each frame in frame metadata.
15155 In the below example the input file main.mpg being processed is
15157 compared with the reference file ref.mpg.
15158 ffmpeg -i main.mpg -i ref.mpg -lavfi msad -f null -
15160 multiply
15162 Multiply first video stream pixels values with second video stream
15163 pixels values.
15164 The filter accepts the following options:
15166 scale
15168 Set the scale applied to second video stream. By default is 1.
15169 Allowed range is from 0 to 9.
15170 offset
15172 Set the offset applied to second video stream. By default is 0.5.
15173 Allowed range is from "-1" to 1.
15174 planes
15176 Specify planes from input video stream that will be processed. By
15177 default all planes are processed.
15178 Commands
15180 This filter supports same commands as options.
15182 negate
15184 Negate (invert) the input video.
15185 It accepts the following option:
15187 components
15189 Set components to negate.
15190 Available values for components are:
15192 y
15194 u
15195 v
15196 a
15197 r
15198 g
15199 b
15200 negate_alpha
15201 With value 1, it negates the alpha component, if present. Default
15202 value is 0.
15203 Commands
15205 This filter supports same commands as options.
15207 nlmeans
15209 Denoise frames using Non-Local Means algorithm.
15210 Each pixel is adjusted by looking for other pixels with similar
15212 contexts. This context similarity is defined by comparing their
15213 surrounding patches of size pxp. Patches are searched in an area of rxr
15214 around the pixel.
15215 Note that the research area defines centers for patches, which means
15217 some patches will be made of pixels outside that research area.
15218 The filter accepts the following options.
15220 s Set denoising strength. Default is 1.0. Must be in range [1.0,
15222 30.0].
15223 p Set patch size. Default is 7. Must be odd number in range [0, 99].
15225 pc Same as p but for chroma planes.
15227 The default value is 0 and means automatic.
15229 r Set research size. Default is 15. Must be odd number in range [0,
15231 99].
15232 rc Same as r but for chroma planes.
15234 The default value is 0 and means automatic.
15236 nnedi
15238 Deinterlace video using neural network edge directed interpolation.
15239 This filter accepts the following options:
15241 weights
15243 Mandatory option, without binary file filter can not work.
15244 Currently file can be found here:
15245 https://github.com/dubhater/vapoursynth-nnedi3/blob/master/src/nnedi3_weights.bin
15246 deint
15248 Set which frames to deinterlace, by default it is "all". Can be
15249 "all" or "interlaced".
15250 field
15252 Set mode of operation.
15253 Can be one of the following:
15255 af Use frame flags, both fields.
15257 a Use frame flags, single field.
15259 t Use top field only.
15261 b Use bottom field only.
15263 tf Use both fields, top first.
15265 bf Use both fields, bottom first.
15267 planes
15269 Set which planes to process, by default filter process all frames.
15270 nsize
15272 Set size of local neighborhood around each pixel, used by the
15273 predictor neural network.
15274 Can be one of the following:
15276 s8x6
15278 s16x6
15279 s32x6
15280 s48x6
15281 s8x4
15282 s16x4
15283 s32x4
15284 nns Set the number of neurons in predictor neural network. Can be one
15285 of the following:
15286 n16
15288 n32
15289 n64
15290 n128
15291 n256
15292 qual
15293 Controls the number of different neural network predictions that
15294 are blended together to compute the final output value. Can be
15295 "fast", default or "slow".
15296 etype
15298 Set which set of weights to use in the predictor. Can be one of
15299 the following:
15300 a, abs
15302 weights trained to minimize absolute error
15303 s, mse
15305 weights trained to minimize squared error
15306 pscrn
15308 Controls whether or not the prescreener neural network is used to
15309 decide which pixels should be processed by the predictor neural
15310 network and which can be handled by simple cubic interpolation.
15311 The prescreener is trained to know whether cubic interpolation will
15312 be sufficient for a pixel or whether it should be predicted by the
15313 predictor nn. The computational complexity of the prescreener nn
15314 is much less than that of the predictor nn. Since most pixels can
15315 be handled by cubic interpolation, using the prescreener generally
15316 results in much faster processing. The prescreener is pretty
15317 accurate, so the difference between using it and not using it is
15318 almost always unnoticeable.
15319 Can be one of the following:
15321 none
15323 original
15324 new
15325 new2
15326 new3
15327 Default is "new".
15329 Commands
15331 This filter supports same commands as options, excluding weights
15333 option.
15334 noformat
15336 Force libavfilter not to use any of the specified pixel formats for the
15337 input to the next filter.
15338 It accepts the following parameters:
15340 pix_fmts
15342 A '|'-separated list of pixel format names, such as
15343 pix_fmts=yuv420p|monow|rgb24".
15344 Examples
15346 • Force libavfilter to use a format different from yuv420p for the
15348 input to the vflip filter:
15349 noformat=pix_fmts=yuv420p,vflip
15351 • Convert the input video to any of the formats not contained in the
15353 list:
15354 noformat=yuv420p|yuv444p|yuv410p
15356 noise
15358 Add noise on video input frame.
15359 The filter accepts the following options:
15361 all_seed
15363 c0_seed
15364 c1_seed
15365 c2_seed
15366 c3_seed
15367 Set noise seed for specific pixel component or all pixel components
15368 in case of all_seed. Default value is 123457.
15369 all_strength, alls
15371 c0_strength, c0s
15372 c1_strength, c1s
15373 c2_strength, c2s
15374 c3_strength, c3s
15375 Set noise strength for specific pixel component or all pixel
15376 components in case all_strength. Default value is 0. Allowed range
15377 is [0, 100].
15378 all_flags, allf
15380 c0_flags, c0f
15381 c1_flags, c1f
15382 c2_flags, c2f
15383 c3_flags, c3f
15384 Set pixel component flags or set flags for all components if
15385 all_flags. Available values for component flags are:
15386 a averaged temporal noise (smoother)
15388 p mix random noise with a (semi)regular pattern
15390 t temporal noise (noise pattern changes between frames)
15392 u uniform noise (gaussian otherwise)
15394 Examples
15396 Add temporal and uniform noise to input video:
15398 noise=alls=20:allf=t+u
15400 normalize
15402 Normalize RGB video (aka histogram stretching, contrast stretching).
15403 See: https://en.wikipedia.org/wiki/Normalization_(image_processing)
15404 For each channel of each frame, the filter computes the input range and
15406 maps it linearly to the user-specified output range. The output range
15407 defaults to the full dynamic range from pure black to pure white.
15408 Temporal smoothing can be used on the input range to reduce flickering
15410 (rapid changes in brightness) caused when small dark or bright objects
15411 enter or leave the scene. This is similar to the auto-exposure
15412 (automatic gain control) on a video camera, and, like a video camera,
15413 it may cause a period of over- or under-exposure of the video.
15414 The R,G,B channels can be normalized independently, which may cause
15416 some color shifting, or linked together as a single channel, which
15417 prevents color shifting. Linked normalization preserves hue.
15418 Independent normalization does not, so it can be used to remove some
15419 color casts. Independent and linked normalization can be combined in
15420 any ratio.
15421 The normalize filter accepts the following options:
15423 blackpt
15425 whitept
15426 Colors which define the output range. The minimum input value is
15427 mapped to the blackpt. The maximum input value is mapped to the
15428 whitept. The defaults are black and white respectively. Specifying
15429 white for blackpt and black for whitept will give color-inverted,
15430 normalized video. Shades of grey can be used to reduce the dynamic
15431 range (contrast). Specifying saturated colors here can create some
15432 interesting effects.
15433 smoothing
15435 The number of previous frames to use for temporal smoothing. The
15436 input range of each channel is smoothed using a rolling average
15437 over the current frame and the smoothing previous frames. The
15438 default is 0 (no temporal smoothing).
15439 independence
15441 Controls the ratio of independent (color shifting) channel
15442 normalization to linked (color preserving) normalization. 0.0 is
15443 fully linked, 1.0 is fully independent. Defaults to 1.0 (fully
15444 independent).
15445 strength
15447 Overall strength of the filter. 1.0 is full strength. 0.0 is a
15448 rather expensive no-op. Defaults to 1.0 (full strength).
15449 Commands
15451 This filter supports same commands as options, excluding smoothing
15453 option. The command accepts the same syntax of the corresponding
15454 option.
15455 If the specified expression is not valid, it is kept at its current
15457 value.
15458 Examples
15460 Stretch video contrast to use the full dynamic range, with no temporal
15462 smoothing; may flicker depending on the source content:
15463 normalize=blackpt=black:whitept=white:smoothing=0
15465 As above, but with 50 frames of temporal smoothing; flicker should be
15467 reduced, depending on the source content:
15468 normalize=blackpt=black:whitept=white:smoothing=50
15470 As above, but with hue-preserving linked channel normalization:
15472 normalize=blackpt=black:whitept=white:smoothing=50:independence=0
15474 As above, but with half strength:
15476 normalize=blackpt=black:whitept=white:smoothing=50:independence=0:strength=0.5
15478 Map the darkest input color to red, the brightest input color to cyan:
15480 normalize=blackpt=red:whitept=cyan
15482 null
15484 Pass the video source unchanged to the output.
15485 ocr
15487 Optical Character Recognition
15488 This filter uses Tesseract for optical character recognition. To enable
15490 compilation of this filter, you need to configure FFmpeg with
15491 "--enable-libtesseract".
15492 It accepts the following options:
15494 datapath
15496 Set datapath to tesseract data. Default is to use whatever was set
15497 at installation.
15498 language
15500 Set language, default is "eng".
15501 whitelist
15503 Set character whitelist.
15504 blacklist
15506 Set character blacklist.
15507 The filter exports recognized text as the frame metadata
15509 "lavfi.ocr.text". The filter exports confidence of recognized words as
15510 the frame metadata "lavfi.ocr.confidence".
15511 ocv
15513 Apply a video transform using libopencv.
15514 To enable this filter, install the libopencv library and headers and
15516 configure FFmpeg with "--enable-libopencv".
15517 It accepts the following parameters:
15519 filter_name
15521 The name of the libopencv filter to apply.
15522 filter_params
15524 The parameters to pass to the libopencv filter. If not specified,
15525 the default values are assumed.
15526 Refer to the official libopencv documentation for more precise
15528 information:
15529 <http://docs.opencv.org/master/modules/imgproc/doc/filtering.html>
15530 Several libopencv filters are supported; see the following subsections.
15532 dilate
15534 Dilate an image by using a specific structuring element. It
15536 corresponds to the libopencv function "cvDilate".
15537 It accepts the parameters: struct_el|nb_iterations.
15539 struct_el represents a structuring element, and has the syntax:
15541 colsxrows+anchor_xxanchor_y/shape
15542 cols and rows represent the number of columns and rows of the
15544 structuring element, anchor_x and anchor_y the anchor point, and shape
15545 the shape for the structuring element. shape must be "rect", "cross",
15546 "ellipse", or "custom".
15547 If the value for shape is "custom", it must be followed by a string of
15549 the form "=filename". The file with name filename is assumed to
15550 represent a binary image, with each printable character corresponding
15551 to a bright pixel. When a custom shape is used, cols and rows are
15552 ignored, the number or columns and rows of the read file are assumed
15553 instead.
15554 The default value for struct_el is "3x3+0x0/rect".
15556 nb_iterations specifies the number of times the transform is applied to
15558 the image, and defaults to 1.
15559 Some examples:
15561 # Use the default values
15563 ocv=dilate
15564 # Dilate using a structuring element with a 5x5 cross, iterating two times
15566 ocv=filter_name=dilate:filter_params=5x5+2x2/cross|2
15567 # Read the shape from the file diamond.shape, iterating two times.
15569 # The file diamond.shape may contain a pattern of characters like this
15570 # *
15571 # ***
15572 # *****
15573 # ***
15574 # *
15575 # The specified columns and rows are ignored
15576 # but the anchor point coordinates are not
15577 ocv=dilate:0x0+2x2/custom=diamond.shape|2
15578 erode
15580 Erode an image by using a specific structuring element. It corresponds
15582 to the libopencv function "cvErode".
15583 It accepts the parameters: struct_el:nb_iterations, with the same
15585 syntax and semantics as the dilate filter.
15586 smooth
15588 Smooth the input video.
15590 The filter takes the following parameters:
15592 type|param1|param2|param3|param4.
15593 type is the type of smooth filter to apply, and must be one of the
15595 following values: "blur", "blur_no_scale", "median", "gaussian", or
15596 "bilateral". The default value is "gaussian".
15597 The meaning of param1, param2, param3, and param4 depends on the smooth
15599 type. param1 and param2 accept integer positive values or 0. param3 and
15600 param4 accept floating point values.
15601 The default value for param1 is 3. The default value for the other
15603 parameters is 0.
15604 These parameters correspond to the parameters assigned to the libopencv
15606 function "cvSmooth".
15607 oscilloscope
15609 2D Video Oscilloscope.
15610 Useful to measure spatial impulse, step responses, chroma delays, etc.
15612 It accepts the following parameters:
15614 x Set scope center x position.
15616 y Set scope center y position.
15618 s Set scope size, relative to frame diagonal.
15620 t Set scope tilt/rotation.
15622 o Set trace opacity.
15624 tx Set trace center x position.
15626 ty Set trace center y position.
15628 tw Set trace width, relative to width of frame.
15630 th Set trace height, relative to height of frame.
15632 c Set which components to trace. By default it traces first three
15634 components.
15635 g Draw trace grid. By default is enabled.
15637 st Draw some statistics. By default is enabled.
15639 sc Draw scope. By default is enabled.
15641 Commands
15643 This filter supports same commands as options. The command accepts the
15645 same syntax of the corresponding option.
15646 If the specified expression is not valid, it is kept at its current
15648 value.
15649 Examples
15651 • Inspect full first row of video frame.
15653 oscilloscope=x=0.5:y=0:s=1
15655 • Inspect full last row of video frame.
15657 oscilloscope=x=0.5:y=1:s=1
15659 • Inspect full 5th line of video frame of height 1080.
15661 oscilloscope=x=0.5:y=5/1080:s=1
15663 • Inspect full last column of video frame.
15665 oscilloscope=x=1:y=0.5:s=1:t=1
15667 overlay
15669 Overlay one video on top of another.
15670 It takes two inputs and has one output. The first input is the "main"
15672 video on which the second input is overlaid.
15673 It accepts the following parameters:
15675 A description of the accepted options follows.
15677 x
15679 y Set the expression for the x and y coordinates of the overlaid
15680 video on the main video. Default value is "0" for both expressions.
15681 In case the expression is invalid, it is set to a huge value
15682 (meaning that the overlay will not be displayed within the output
15683 visible area).
15684 eof_action
15686 See framesync.
15687 eval
15689 Set when the expressions for x, and y are evaluated.
15690 It accepts the following values:
15692 init
15694 only evaluate expressions once during the filter initialization
15695 or when a command is processed
15696 frame
15698 evaluate expressions for each incoming frame
15699 Default value is frame.
15701 shortest
15703 See framesync.
15704 format
15706 Set the format for the output video.
15707 It accepts the following values:
15709 yuv420
15711 force YUV420 output
15712 yuv420p10
15714 force YUV420p10 output
15715 yuv422
15717 force YUV422 output
15718 yuv422p10
15720 force YUV422p10 output
15721 yuv444
15723 force YUV444 output
15724 rgb force packed RGB output
15726 gbrp
15728 force planar RGB output
15729 auto
15731 automatically pick format
15732 Default value is yuv420.
15734 repeatlast
15736 See framesync.
15737 alpha
15739 Set format of alpha of the overlaid video, it can be straight or
15740 premultiplied. Default is straight.
15741 The x, and y expressions can contain the following parameters.
15743 main_w, W
15745 main_h, H
15746 The main input width and height.
15747 overlay_w, w
15749 overlay_h, h
15750 The overlay input width and height.
15751 x
15753 y The computed values for x and y. They are evaluated for each new
15754 frame.
15755 hsub
15757 vsub
15758 horizontal and vertical chroma subsample values of the output
15759 format. For example for the pixel format "yuv422p" hsub is 2 and
15760 vsub is 1.
15761 n the number of input frame, starting from 0
15763 pos the position in the file of the input frame, NAN if unknown
15765 t The timestamp, expressed in seconds. It's NAN if the input
15767 timestamp is unknown.
15768 This filter also supports the framesync options.
15770 Note that the n, pos, t variables are available only when evaluation is
15772 done per frame, and will evaluate to NAN when eval is set to init.
15773 Be aware that frames are taken from each input video in timestamp
15775 order, hence, if their initial timestamps differ, it is a good idea to
15776 pass the two inputs through a setpts=PTS-STARTPTS filter to have them
15777 begin in the same zero timestamp, as the example for the movie filter
15778 does.
15779 You can chain together more overlays but you should test the efficiency
15781 of such approach.
15782 Commands
15784 This filter supports the following commands:
15786 x
15788 y Modify the x and y of the overlay input. The command accepts the
15789 same syntax of the corresponding option.
15790 If the specified expression is not valid, it is kept at its current
15792 value.
15793 Examples
15795 • Draw the overlay at 10 pixels from the bottom right corner of the
15797 main video:
15798 overlay=main_w-overlay_w-10:main_h-overlay_h-10
15800 Using named options the example above becomes:
15802 overlay=x=main_w-overlay_w-10:y=main_h-overlay_h-10
15804 • Insert a transparent PNG logo in the bottom left corner of the
15806 input, using the ffmpeg tool with the "-filter_complex" option:
15807 ffmpeg -i input -i logo -filter_complex 'overlay=10:main_h-overlay_h-10' output
15809 • Insert 2 different transparent PNG logos (second logo on bottom
15811 right corner) using the ffmpeg tool:
15812 ffmpeg -i input -i logo1 -i logo2 -filter_complex 'overlay=x=10:y=H-h-10,overlay=x=W-w-10:y=H-h-10' output
15814 • Add a transparent color layer on top of the main video; "WxH" must
15816 specify the size of the main input to the overlay filter:
15817 color=color=red@.3:size=WxH [over]; [in][over] overlay [out]
15819 • Play an original video and a filtered version (here with the
15821 deshake filter) side by side using the ffplay tool:
15822 ffplay input.avi -vf 'split[a][b]; [a]pad=iw*2:ih[src]; [b]deshake[filt]; [src][filt]overlay=w'
15824 The above command is the same as:
15826 ffplay input.avi -vf 'split[b], pad=iw*2[src], [b]deshake, [src]overlay=w'
15828 • Make a sliding overlay appearing from the left to the right top
15830 part of the screen starting since time 2:
15831 overlay=x='if(gte(t,2), -w+(t-2)*20, NAN)':y=0
15833 • Compose output by putting two input videos side to side:
15835 ffmpeg -i left.avi -i right.avi -filter_complex "
15837 nullsrc=size=200x100 [background];
15838 [0:v] setpts=PTS-STARTPTS, scale=100x100 [left];
15839 [1:v] setpts=PTS-STARTPTS, scale=100x100 [right];
15840 [background][left] overlay=shortest=1 [background+left];
15841 [background+left][right] overlay=shortest=1:x=100 [left+right]
15842 "
15843 • Mask 10-20 seconds of a video by applying the delogo filter to a
15845 section
15846 ffmpeg -i test.avi -codec:v:0 wmv2 -ar 11025 -b:v 9000k
15848 -vf '[in]split[split_main][split_delogo];[split_delogo]trim=start=360:end=371,delogo=0:0:640:480[delogoed];[split_main][delogoed]overlay=eof_action=pass[out]'
15849 masked.avi
15850 • Chain several overlays in cascade:
15852 nullsrc=s=200x200 [bg];
15854 testsrc=s=100x100, split=4 [in0][in1][in2][in3];
15855 [in0] lutrgb=r=0, [bg] overlay=0:0 [mid0];
15856 [in1] lutrgb=g=0, [mid0] overlay=100:0 [mid1];
15857 [in2] lutrgb=b=0, [mid1] overlay=0:100 [mid2];
15858 [in3] null, [mid2] overlay=100:100 [out0]
15859 overlay_cuda
15861 Overlay one video on top of another.
15862 This is the CUDA variant of the overlay filter. It only accepts CUDA
15864 frames. The underlying input pixel formats have to match.
15865 It takes two inputs and has one output. The first input is the "main"
15867 video on which the second input is overlaid.
15868 It accepts the following parameters:
15870 x
15872 y Set expressions for the x and y coordinates of the overlaid video
15873 on the main video.
15874 They can contain the following parameters:
15876 main_w, W
15878 main_h, H
15879 The main input width and height.
15880 overlay_w, w
15882 overlay_h, h
15883 The overlay input width and height.
15884 x
15886 y The computed values for x and y. They are evaluated for each
15887 new frame.
15888 n The ordinal index of the main input frame, starting from 0.
15890 pos The byte offset position in the file of the main input frame,
15892 NAN if unknown.
15893 t The timestamp of the main input frame, expressed in seconds,
15895 NAN if unknown.
15896 Default value is "0" for both expressions.
15898 eval
15900 Set when the expressions for x and y are evaluated.
15901 It accepts the following values:
15903 init
15905 Evaluate expressions once during filter initialization or when
15906 a command is processed.
15907 frame
15909 Evaluate expressions for each incoming frame
15910 Default value is frame.
15912 eof_action
15914 See framesync.
15915 shortest
15917 See framesync.
15918 repeatlast
15920 See framesync.
15921 This filter also supports the framesync options.
15923 owdenoise
15925 Apply Overcomplete Wavelet denoiser.
15926 The filter accepts the following options:
15928 depth
15930 Set depth.
15931 Larger depth values will denoise lower frequency components more,
15933 but slow down filtering.
15934 Must be an int in the range 8-16, default is 8.
15936 luma_strength, ls
15938 Set luma strength.
15939 Must be a double value in the range 0-1000, default is 1.0.
15941 chroma_strength, cs
15943 Set chroma strength.
15944 Must be a double value in the range 0-1000, default is 1.0.
15946 pad
15948 Add paddings to the input image, and place the original input at the
15949 provided x, y coordinates.
15950 It accepts the following parameters:
15952 width, w
15954 height, h
15955 Specify an expression for the size of the output image with the
15956 paddings added. If the value for width or height is 0, the
15957 corresponding input size is used for the output.
15958 The width expression can reference the value set by the height
15960 expression, and vice versa.
15961 The default value of width and height is 0.
15963 x
15965 y Specify the offsets to place the input image at within the padded
15966 area, with respect to the top/left border of the output image.
15967 The x expression can reference the value set by the y expression,
15969 and vice versa.
15970 The default value of x and y is 0.
15972 If x or y evaluate to a negative number, they'll be changed so the
15974 input image is centered on the padded area.
15975 color
15977 Specify the color of the padded area. For the syntax of this
15978 option, check the "Color" section in the ffmpeg-utils manual.
15979 The default value of color is "black".
15981 eval
15983 Specify when to evaluate width, height, x and y expression.
15984 It accepts the following values:
15986 init
15988 Only evaluate expressions once during the filter initialization
15989 or when a command is processed.
15990 frame
15992 Evaluate expressions for each incoming frame.
15993 Default value is init.
15995 aspect
15997 Pad to aspect instead to a resolution.
15998 The value for the width, height, x, and y options are expressions
16000 containing the following constants:
16001 in_w
16003 in_h
16004 The input video width and height.
16005 iw
16007 ih These are the same as in_w and in_h.
16008 out_w
16010 out_h
16011 The output width and height (the size of the padded area), as
16012 specified by the width and height expressions.
16013 ow
16015 oh These are the same as out_w and out_h.
16016 x
16018 y The x and y offsets as specified by the x and y expressions, or NAN
16019 if not yet specified.
16020 a same as iw / ih
16022 sar input sample aspect ratio
16024 dar input display aspect ratio, it is the same as (iw / ih) * sar
16026 hsub
16028 vsub
16029 The horizontal and vertical chroma subsample values. For example
16030 for the pixel format "yuv422p" hsub is 2 and vsub is 1.
16031 Examples
16033 • Add paddings with the color "violet" to the input video. The output
16035 video size is 640x480, and the top-left corner of the input video
16036 is placed at column 0, row 40
16037 pad=640:480:0:40:violet
16039 The example above is equivalent to the following command:
16041 pad=width=640:height=480:x=0:y=40:color=violet
16043 • Pad the input to get an output with dimensions increased by 3/2,
16045 and put the input video at the center of the padded area:
16046 pad="3/2*iw:3/2*ih:(ow-iw)/2:(oh-ih)/2"
16048 • Pad the input to get a squared output with size equal to the
16050 maximum value between the input width and height, and put the input
16051 video at the center of the padded area:
16052 pad="max(iw\,ih):ow:(ow-iw)/2:(oh-ih)/2"
16054 • Pad the input to get a final w/h ratio of 16:9:
16056 pad="ih*16/9:ih:(ow-iw)/2:(oh-ih)/2"
16058 • In case of anamorphic video, in order to set the output display
16060 aspect correctly, it is necessary to use sar in the expression,
16061 according to the relation:
16062 (ih * X / ih) * sar = output_dar
16064 X = output_dar / sar
16065 Thus the previous example needs to be modified to:
16067 pad="ih*16/9/sar:ih:(ow-iw)/2:(oh-ih)/2"
16069 • Double the output size and put the input video in the bottom-right
16071 corner of the output padded area:
16072 pad="2*iw:2*ih:ow-iw:oh-ih"
16074 palettegen
16076 Generate one palette for a whole video stream.
16077 It accepts the following options:
16079 max_colors
16081 Set the maximum number of colors to quantize in the palette. Note:
16082 the palette will still contain 256 colors; the unused palette
16083 entries will be black.
16084 reserve_transparent
16086 Create a palette of 255 colors maximum and reserve the last one for
16087 transparency. Reserving the transparency color is useful for GIF
16088 optimization. If not set, the maximum of colors in the palette
16089 will be 256. You probably want to disable this option for a
16090 standalone image. Set by default.
16091 transparency_color
16093 Set the color that will be used as background for transparency.
16094 stats_mode
16096 Set statistics mode.
16097 It accepts the following values:
16099 full
16101 Compute full frame histograms.
16102 diff
16104 Compute histograms only for the part that differs from previous
16105 frame. This might be relevant to give more importance to the
16106 moving part of your input if the background is static.
16107 single
16109 Compute new histogram for each frame.
16110 Default value is full.
16112 use_alpha
16114 Create a palette of colors with alpha components. Setting this,
16115 will automatically disable 'reserve_transparent'.
16116 The filter also exports the frame metadata "lavfi.color_quant_ratio"
16118 ("nb_color_in / nb_color_out") which you can use to evaluate the degree
16119 of color quantization of the palette. This information is also visible
16120 at info logging level.
16121 Examples
16123 • Generate a representative palette of a given video using ffmpeg:
16125 ffmpeg -i input.mkv -vf palettegen palette.png
16127 paletteuse
16129 Use a palette to downsample an input video stream.
16130 The filter takes two inputs: one video stream and a palette. The
16132 palette must be a 256 pixels image.
16133 It accepts the following options:
16135 dither
16137 Select dithering mode. Available algorithms are:
16138 bayer
16140 Ordered 8x8 bayer dithering (deterministic)
16141 heckbert
16143 Dithering as defined by Paul Heckbert in 1982 (simple error
16144 diffusion). Note: this dithering is sometimes considered
16145 "wrong" and is included as a reference.
16146 floyd_steinberg
16148 Floyd and Steingberg dithering (error diffusion)
16149 sierra2
16151 Frankie Sierra dithering v2 (error diffusion)
16152 sierra2_4a
16154 Frankie Sierra dithering v2 "Lite" (error diffusion)
16155 Default is sierra2_4a.
16157 bayer_scale
16159 When bayer dithering is selected, this option defines the scale of
16160 the pattern (how much the crosshatch pattern is visible). A low
16161 value means more visible pattern for less banding, and higher value
16162 means less visible pattern at the cost of more banding.
16163 The option must be an integer value in the range [0,5]. Default is
16165 2.
16166 diff_mode
16168 If set, define the zone to process
16169 rectangle
16171 Only the changing rectangle will be reprocessed. This is
16172 similar to GIF cropping/offsetting compression mechanism. This
16173 option can be useful for speed if only a part of the image is
16174 changing, and has use cases such as limiting the scope of the
16175 error diffusal dither to the rectangle that bounds the moving
16176 scene (it leads to more deterministic output if the scene
16177 doesn't change much, and as a result less moving noise and
16178 better GIF compression).
16179 Default is none.
16181 new Take new palette for each output frame.
16183 alpha_threshold
16185 Sets the alpha threshold for transparency. Alpha values above this
16186 threshold will be treated as completely opaque, and values below
16187 this threshold will be treated as completely transparent.
16188 The option must be an integer value in the range [0,255]. Default
16190 is 128.
16191 use_alpha
16193 Apply the palette by taking alpha values into account. Only useful
16194 with palettes that are containing multiple colors with alpha
16195 components. Setting this will automatically disable
16196 'alpha_treshold'.
16197 Examples
16199 • Use a palette (generated for example with palettegen) to encode a
16201 GIF using ffmpeg:
16202 ffmpeg -i input.mkv -i palette.png -lavfi paletteuse output.gif
16204 perspective
16206 Correct perspective of video not recorded perpendicular to the screen.
16207 A description of the accepted parameters follows.
16209 x0
16211 y0
16212 x1
16213 y1
16214 x2
16215 y2
16216 x3
16217 y3 Set coordinates expression for top left, top right, bottom left and
16218 bottom right corners. Default values are "0:0:W:0:0:H:W:H" with
16219 which perspective will remain unchanged. If the "sense" option is
16220 set to "source", then the specified points will be sent to the
16221 corners of the destination. If the "sense" option is set to
16222 "destination", then the corners of the source will be sent to the
16223 specified coordinates.
16224 The expressions can use the following variables:
16226 W
16228 H the width and height of video frame.
16229 in Input frame count.
16231 on Output frame count.
16233 interpolation
16235 Set interpolation for perspective correction.
16236 It accepts the following values:
16238 linear
16240 cubic
16241 Default value is linear.
16243 sense
16245 Set interpretation of coordinate options.
16246 It accepts the following values:
16248 0, source
16250 Send point in the source specified by the given coordinates to
16251 the corners of the destination.
16252 1, destination
16254 Send the corners of the source to the point in the destination
16255 specified by the given coordinates.
16256 Default value is source.
16258 eval
16260 Set when the expressions for coordinates x0,y0,...x3,y3 are
16261 evaluated.
16262 It accepts the following values:
16264 init
16266 only evaluate expressions once during the filter initialization
16267 or when a command is processed
16268 frame
16270 evaluate expressions for each incoming frame
16271 Default value is init.
16273 phase
16275 Delay interlaced video by one field time so that the field order
16276 changes.
16277 The intended use is to fix PAL movies that have been captured with the
16279 opposite field order to the film-to-video transfer.
16280 A description of the accepted parameters follows.
16282 mode
16284 Set phase mode.
16285 It accepts the following values:
16287 t Capture field order top-first, transfer bottom-first. Filter
16289 will delay the bottom field.
16290 b Capture field order bottom-first, transfer top-first. Filter
16292 will delay the top field.
16293 p Capture and transfer with the same field order. This mode only
16295 exists for the documentation of the other options to refer to,
16296 but if you actually select it, the filter will faithfully do
16297 nothing.
16298 a Capture field order determined automatically by field flags,
16300 transfer opposite. Filter selects among t and b modes on a
16301 frame by frame basis using field flags. If no field information
16302 is available, then this works just like u.
16303 u Capture unknown or varying, transfer opposite. Filter selects
16305 among t and b on a frame by frame basis by analyzing the images
16306 and selecting the alternative that produces best match between
16307 the fields.
16308 T Capture top-first, transfer unknown or varying. Filter selects
16310 among t and p using image analysis.
16311 B Capture bottom-first, transfer unknown or varying. Filter
16313 selects among b and p using image analysis.
16314 A Capture determined by field flags, transfer unknown or varying.
16316 Filter selects among t, b and p using field flags and image
16317 analysis. If no field information is available, then this works
16318 just like U. This is the default mode.
16319 U Both capture and transfer unknown or varying. Filter selects
16321 among t, b and p using image analysis only.
16322 Commands
16324 This filter supports the all above options as commands.
16326 photosensitivity
16328 Reduce various flashes in video, so to help users with epilepsy.
16329 It accepts the following options:
16331 frames, f
16333 Set how many frames to use when filtering. Default is 30.
16334 threshold, t
16336 Set detection threshold factor. Default is 1. Lower is stricter.
16337 skip
16339 Set how many pixels to skip when sampling frames. Default is 1.
16340 Allowed range is from 1 to 1024.
16341 bypass
16343 Leave frames unchanged. Default is disabled.
16344 pixdesctest
16346 Pixel format descriptor test filter, mainly useful for internal
16347 testing. The output video should be equal to the input video.
16348 For example:
16350 format=monow, pixdesctest
16352 can be used to test the monowhite pixel format descriptor definition.
16354 pixelize
16356 Apply pixelization to video stream.
16357 The filter accepts the following options:
16359 width, w
16361 height, h
16362 Set block dimensions that will be used for pixelization. Default
16363 value is 16.
16364 mode, m
16366 Set the mode of pixelization used.
16367 Possible values are:
16369 avg
16371 min
16372 max
16373 Default value is "avg".
16375 planes, p
16377 Set what planes to filter. Default is to filter all planes.
16378 Commands
16380 This filter supports all options as commands.
16382 pixscope
16384 Display sample values of color channels. Mainly useful for checking
16385 color and levels. Minimum supported resolution is 640x480.
16386 The filters accept the following options:
16388 x Set scope X position, relative offset on X axis.
16390 y Set scope Y position, relative offset on Y axis.
16392 w Set scope width.
16394 h Set scope height.
16396 o Set window opacity. This window also holds statistics about pixel
16398 area.
16399 wx Set window X position, relative offset on X axis.
16401 wy Set window Y position, relative offset on Y axis.
16403 Commands
16405 This filter supports same commands as options.
16407 pp
16409 Enable the specified chain of postprocessing subfilters using
16410 libpostproc. This library should be automatically selected with a GPL
16411 build ("--enable-gpl"). Subfilters must be separated by '/' and can be
16412 disabled by prepending a '-'. Each subfilter and some options have a
16413 short and a long name that can be used interchangeably, i.e. dr/dering
16414 are the same.
16415 The filters accept the following options:
16417 subfilters
16419 Set postprocessing subfilters string.
16420 All subfilters share common options to determine their scope:
16422 a/autoq
16424 Honor the quality commands for this subfilter.
16425 c/chrom
16427 Do chrominance filtering, too (default).
16428 y/nochrom
16430 Do luminance filtering only (no chrominance).
16431 n/noluma
16433 Do chrominance filtering only (no luminance).
16434 These options can be appended after the subfilter name, separated by a
16436 '|'.
16437 Available subfilters are:
16439 hb/hdeblock[|difference[|flatness]]
16441 Horizontal deblocking filter
16442 difference
16444 Difference factor where higher values mean more deblocking
16445 (default: 32).
16446 flatness
16448 Flatness threshold where lower values mean more deblocking
16449 (default: 39).
16450 vb/vdeblock[|difference[|flatness]]
16452 Vertical deblocking filter
16453 difference
16455 Difference factor where higher values mean more deblocking
16456 (default: 32).
16457 flatness
16459 Flatness threshold where lower values mean more deblocking
16460 (default: 39).
16461 ha/hadeblock[|difference[|flatness]]
16463 Accurate horizontal deblocking filter
16464 difference
16466 Difference factor where higher values mean more deblocking
16467 (default: 32).
16468 flatness
16470 Flatness threshold where lower values mean more deblocking
16471 (default: 39).
16472 va/vadeblock[|difference[|flatness]]
16474 Accurate vertical deblocking filter
16475 difference
16477 Difference factor where higher values mean more deblocking
16478 (default: 32).
16479 flatness
16481 Flatness threshold where lower values mean more deblocking
16482 (default: 39).
16483 The horizontal and vertical deblocking filters share the difference and
16485 flatness values so you cannot set different horizontal and vertical
16486 thresholds.
16487 h1/x1hdeblock
16489 Experimental horizontal deblocking filter
16490 v1/x1vdeblock
16492 Experimental vertical deblocking filter
16493 dr/dering
16495 Deringing filter
16496 tn/tmpnoise[|threshold1[|threshold2[|threshold3]]], temporal noise
16498 reducer
16499 threshold1
16500 larger -> stronger filtering
16501 threshold2
16503 larger -> stronger filtering
16504 threshold3
16506 larger -> stronger filtering
16507 al/autolevels[:f/fullyrange], automatic brightness / contrast
16509 correction
16510 f/fullyrange
16511 Stretch luminance to "0-255".
16512 lb/linblenddeint
16514 Linear blend deinterlacing filter that deinterlaces the given block
16515 by filtering all lines with a "(1 2 1)" filter.
16516 li/linipoldeint
16518 Linear interpolating deinterlacing filter that deinterlaces the
16519 given block by linearly interpolating every second line.
16520 ci/cubicipoldeint
16522 Cubic interpolating deinterlacing filter deinterlaces the given
16523 block by cubically interpolating every second line.
16524 md/mediandeint
16526 Median deinterlacing filter that deinterlaces the given block by
16527 applying a median filter to every second line.
16528 fd/ffmpegdeint
16530 FFmpeg deinterlacing filter that deinterlaces the given block by
16531 filtering every second line with a "(-1 4 2 4 -1)" filter.
16532 l5/lowpass5
16534 Vertically applied FIR lowpass deinterlacing filter that
16535 deinterlaces the given block by filtering all lines with a "(-1 2 6
16536 2 -1)" filter.
16537 fq/forceQuant[|quantizer]
16539 Overrides the quantizer table from the input with the constant
16540 quantizer you specify.
16541 quantizer
16543 Quantizer to use
16544 de/default
16546 Default pp filter combination ("hb|a,vb|a,dr|a")
16547 fa/fast
16549 Fast pp filter combination ("h1|a,v1|a,dr|a")
16550 ac High quality pp filter combination ("ha|a|128|7,va|a,dr|a")
16552 Examples
16554 • Apply horizontal and vertical deblocking, deringing and automatic
16556 brightness/contrast:
16557 pp=hb/vb/dr/al
16559 • Apply default filters without brightness/contrast correction:
16561 pp=de/-al
16563 • Apply default filters and temporal denoiser:
16565 pp=default/tmpnoise|1|2|3
16567 • Apply deblocking on luminance only, and switch vertical deblocking
16569 on or off automatically depending on available CPU time:
16570 pp=hb|y/vb|a
16572 pp7
16574 Apply Postprocessing filter 7. It is variant of the spp filter, similar
16575 to spp = 6 with 7 point DCT, where only the center sample is used after
16576 IDCT.
16577 The filter accepts the following options:
16579 qp Force a constant quantization parameter. It accepts an integer in
16581 range 0 to 63. If not set, the filter will use the QP from the
16582 video stream (if available).
16583 mode
16585 Set thresholding mode. Available modes are:
16586 hard
16588 Set hard thresholding.
16589 soft
16591 Set soft thresholding (better de-ringing effect, but likely
16592 blurrier).
16593 medium
16595 Set medium thresholding (good results, default).
16596 premultiply
16598 Apply alpha premultiply effect to input video stream using first plane
16599 of second stream as alpha.
16600 Both streams must have same dimensions and same pixel format.
16602 The filter accepts the following option:
16604 planes
16606 Set which planes will be processed, unprocessed planes will be
16607 copied. By default value 0xf, all planes will be processed.
16608 inplace
16610 Do not require 2nd input for processing, instead use alpha plane
16611 from input stream.
16612 prewitt
16614 Apply prewitt operator to input video stream.
16615 The filter accepts the following option:
16617 planes
16619 Set which planes will be processed, unprocessed planes will be
16620 copied. By default value 0xf, all planes will be processed.
16621 scale
16623 Set value which will be multiplied with filtered result.
16624 delta
16626 Set value which will be added to filtered result.
16627 Commands
16629 This filter supports the all above options as commands.
16631 pseudocolor
16633 Alter frame colors in video with pseudocolors.
16634 This filter accepts the following options:
16636 c0 set pixel first component expression
16638 c1 set pixel second component expression
16640 c2 set pixel third component expression
16642 c3 set pixel fourth component expression, corresponds to the alpha
16644 component
16645 index, i
16647 set component to use as base for altering colors
16648 preset, p
16650 Pick one of built-in LUTs. By default is set to none.
16651 Available LUTs:
16653 magma
16655 inferno
16656 plasma
16657 viridis
16658 turbo
16659 cividis
16660 range1
16661 range2
16662 shadows
16663 highlights
16664 solar
16665 nominal
16666 preferred
16667 total
16668 opacity
16669 Set opacity of output colors. Allowed range is from 0 to 1.
16670 Default value is set to 1.
16671 Each of the expression options specifies the expression to use for
16673 computing the lookup table for the corresponding pixel component
16674 values.
16675 The expressions can contain the following constants and functions:
16677 w
16679 h The input width and height.
16680 val The input value for the pixel component.
16682 ymin, umin, vmin, amin
16684 The minimum allowed component value.
16685 ymax, umax, vmax, amax
16687 The maximum allowed component value.
16688 All expressions default to "val".
16690 Commands
16692 This filter supports the all above options as commands.
16694 Examples
16696 • Change too high luma values to gradient:
16698 pseudocolor="'if(between(val,ymax,amax),lerp(ymin,ymax,(val-ymax)/(amax-ymax)),-1):if(between(val,ymax,amax),lerp(umax,umin,(val-ymax)/(amax-ymax)),-1):if(between(val,ymax,amax),lerp(vmin,vmax,(val-ymax)/(amax-ymax)),-1):-1'"
16700 psnr
16702 Obtain the average, maximum and minimum PSNR (Peak Signal to Noise
16703 Ratio) between two input videos.
16704 This filter takes in input two input videos, the first input is
16706 considered the "main" source and is passed unchanged to the output. The
16707 second input is used as a "reference" video for computing the PSNR.
16708 Both video inputs must have the same resolution and pixel format for
16710 this filter to work correctly. Also it assumes that both inputs have
16711 the same number of frames, which are compared one by one.
16712 The obtained average PSNR is printed through the logging system.
16714 The filter stores the accumulated MSE (mean squared error) of each
16716 frame, and at the end of the processing it is averaged across all
16717 frames equally, and the following formula is applied to obtain the
16718 PSNR:
16719 PSNR = 10*log10(MAX^2/MSE)
16721 Where MAX is the average of the maximum values of each component of the
16723 image.
16724 The description of the accepted parameters follows.
16726 stats_file, f
16728 If specified the filter will use the named file to save the PSNR of
16729 each individual frame. When filename equals "-" the data is sent to
16730 standard output.
16731 stats_version
16733 Specifies which version of the stats file format to use. Details of
16734 each format are written below. Default value is 1.
16735 stats_add_max
16737 Determines whether the max value is output to the stats log.
16738 Default value is 0. Requires stats_version >= 2. If this is set
16739 and stats_version < 2, the filter will return an error.
16740 This filter also supports the framesync options.
16742 The file printed if stats_file is selected, contains a sequence of
16744 key/value pairs of the form key:value for each compared couple of
16745 frames.
16746 If a stats_version greater than 1 is specified, a header line precedes
16748 the list of per-frame-pair stats, with key value pairs following the
16749 frame format with the following parameters:
16750 psnr_log_version
16752 The version of the log file format. Will match stats_version.
16753 fields
16755 A comma separated list of the per-frame-pair parameters included in
16756 the log.
16757 A description of each shown per-frame-pair parameter follows:
16759 n sequential number of the input frame, starting from 1
16761 mse_avg
16763 Mean Square Error pixel-by-pixel average difference of the compared
16764 frames, averaged over all the image components.
16765 mse_y, mse_u, mse_v, mse_r, mse_g, mse_b, mse_a
16767 Mean Square Error pixel-by-pixel average difference of the compared
16768 frames for the component specified by the suffix.
16769 psnr_y, psnr_u, psnr_v, psnr_r, psnr_g, psnr_b, psnr_a
16771 Peak Signal to Noise ratio of the compared frames for the component
16772 specified by the suffix.
16773 max_avg, max_y, max_u, max_v
16775 Maximum allowed value for each channel, and average over all
16776 channels.
16777 Examples
16779 • For example:
16781 movie=ref_movie.mpg, setpts=PTS-STARTPTS [main];
16783 [main][ref] psnr="stats_file=stats.log" [out]
16784 On this example the input file being processed is compared with the
16786 reference file ref_movie.mpg. The PSNR of each individual frame is
16787 stored in stats.log.
16788 • Another example with different containers:
16790 ffmpeg -i main.mpg -i ref.mkv -lavfi "[0:v]settb=AVTB,setpts=PTS-STARTPTS[main];[1:v]settb=AVTB,setpts=PTS-STARTPTS[ref];[main][ref]psnr" -f null -
16792 pullup
16794 Pulldown reversal (inverse telecine) filter, capable of handling mixed
16795 hard-telecine, 24000/1001 fps progressive, and 30000/1001 fps
16796 progressive content.
16797 The pullup filter is designed to take advantage of future context in
16799 making its decisions. This filter is stateless in the sense that it
16800 does not lock onto a pattern to follow, but it instead looks forward to
16801 the following fields in order to identify matches and rebuild
16802 progressive frames.
16803 To produce content with an even framerate, insert the fps filter after
16805 pullup, use "fps=24000/1001" if the input frame rate is 29.97fps,
16806 "fps=24" for 30fps and the (rare) telecined 25fps input.
16807 The filter accepts the following options:
16809 jl
16811 jr
16812 jt
16813 jb These options set the amount of "junk" to ignore at the left,
16814 right, top, and bottom of the image, respectively. Left and right
16815 are in units of 8 pixels, while top and bottom are in units of 2
16816 lines. The default is 8 pixels on each side.
16817 sb Set the strict breaks. Setting this option to 1 will reduce the
16819 chances of filter generating an occasional mismatched frame, but it
16820 may also cause an excessive number of frames to be dropped during
16821 high motion sequences. Conversely, setting it to -1 will make
16822 filter match fields more easily. This may help processing of video
16823 where there is slight blurring between the fields, but may also
16824 cause there to be interlaced frames in the output. Default value
16825 is 0.
16826 mp Set the metric plane to use. It accepts the following values:
16828 l Use luma plane.
16830 u Use chroma blue plane.
16832 v Use chroma red plane.
16834 This option may be set to use chroma plane instead of the default
16836 luma plane for doing filter's computations. This may improve
16837 accuracy on very clean source material, but more likely will
16838 decrease accuracy, especially if there is chroma noise (rainbow
16839 effect) or any grayscale video. The main purpose of setting mp to
16840 a chroma plane is to reduce CPU load and make pullup usable in
16841 realtime on slow machines.
16842 For best results (without duplicated frames in the output file) it is
16844 necessary to change the output frame rate. For example, to inverse
16845 telecine NTSC input:
16846 ffmpeg -i input -vf pullup -r 24000/1001 ...
16848 qp
16850 Change video quantization parameters (QP).
16851 The filter accepts the following option:
16853 qp Set expression for quantization parameter.
16855 The expression is evaluated through the eval API and can contain, among
16857 others, the following constants:
16858 known
16860 1 if index is not 129, 0 otherwise.
16861 qp Sequential index starting from -129 to 128.
16863 Examples
16865 • Some equation like:
16867 qp=2+2*sin(PI*qp)
16869 random
16871 Flush video frames from internal cache of frames into a random order.
16872 No frame is discarded. Inspired by frei0r nervous filter.
16873 frames
16875 Set size in number of frames of internal cache, in range from 2 to
16876 512. Default is 30.
16877 seed
16879 Set seed for random number generator, must be an integer included
16880 between 0 and "UINT32_MAX". If not specified, or if explicitly set
16881 to less than 0, the filter will try to use a good random seed on a
16882 best effort basis.
16883 readeia608
16885 Read closed captioning (EIA-608) information from the top lines of a
16886 video frame.
16887 This filter adds frame metadata for "lavfi.readeia608.X.cc" and
16889 "lavfi.readeia608.X.line", where "X" is the number of the identified
16890 line with EIA-608 data (starting from 0). A description of each
16891 metadata value follows:
16892 lavfi.readeia608.X.cc
16894 The two bytes stored as EIA-608 data (printed in hexadecimal).
16895 lavfi.readeia608.X.line
16897 The number of the line on which the EIA-608 data was identified and
16898 read.
16899 This filter accepts the following options:
16901 scan_min
16903 Set the line to start scanning for EIA-608 data. Default is 0.
16904 scan_max
16906 Set the line to end scanning for EIA-608 data. Default is 29.
16907 spw Set the ratio of width reserved for sync code detection. Default
16909 is 0.27. Allowed range is "[0.1 - 0.7]".
16910 chp Enable checking the parity bit. In the event of a parity error, the
16912 filter will output 0x00 for that character. Default is false.
16913 lp Lowpass lines prior to further processing. Default is enabled.
16915 Commands
16917 This filter supports the all above options as commands.
16919 Examples
16921 • Output a csv with presentation time and the first two lines of
16923 identified EIA-608 captioning data.
16924 ffprobe -f lavfi -i movie=captioned_video.mov,readeia608 -show_entries frame=pts_time:frame_tags=lavfi.readeia608.0.cc,lavfi.readeia608.1.cc -of csv
16926 readvitc
16928 Read vertical interval timecode (VITC) information from the top lines
16929 of a video frame.
16930 The filter adds frame metadata key "lavfi.readvitc.tc_str" with the
16932 timecode value, if a valid timecode has been detected. Further metadata
16933 key "lavfi.readvitc.found" is set to 0/1 depending on whether timecode
16934 data has been found or not.
16935 This filter accepts the following options:
16937 scan_max
16939 Set the maximum number of lines to scan for VITC data. If the value
16940 is set to "-1" the full video frame is scanned. Default is 45.
16941 thr_b
16943 Set the luma threshold for black. Accepts float numbers in the
16944 range [0.0,1.0], default value is 0.2. The value must be equal or
16945 less than "thr_w".
16946 thr_w
16948 Set the luma threshold for white. Accepts float numbers in the
16949 range [0.0,1.0], default value is 0.6. The value must be equal or
16950 greater than "thr_b".
16951 Examples
16953 • Detect and draw VITC data onto the video frame; if no valid VITC is
16955 detected, draw "--:--:--:--" as a placeholder:
16956 ffmpeg -i input.avi -filter:v 'readvitc,drawtext=fontfile=FreeMono.ttf:text=%{metadata\\:lavfi.readvitc.tc_str\\:--\\\\\\:--\\\\\\:--\\\\\\:--}:x=(w-tw)/2:y=400-ascent'
16958 remap
16960 Remap pixels using 2nd: Xmap and 3rd: Ymap input video stream.
16961 Destination pixel at position (X, Y) will be picked from source (x, y)
16963 position where x = Xmap(X, Y) and y = Ymap(X, Y). If mapping values are
16964 out of range, zero value for pixel will be used for destination pixel.
16965 Xmap and Ymap input video streams must be of same dimensions. Output
16967 video stream will have Xmap/Ymap video stream dimensions. Xmap and
16968 Ymap input video streams are 16bit depth, single channel.
16969 format
16971 Specify pixel format of output from this filter. Can be "color" or
16972 "gray". Default is "color".
16973 fill
16975 Specify the color of the unmapped pixels. For the syntax of this
16976 option, check the "Color" section in the ffmpeg-utils manual.
16977 Default color is "black".
16978 removegrain
16980 The removegrain filter is a spatial denoiser for progressive video.
16981 m0 Set mode for the first plane.
16983 m1 Set mode for the second plane.
16985 m2 Set mode for the third plane.
16987 m3 Set mode for the fourth plane.
16989 Range of mode is from 0 to 24. Description of each mode follows:
16991 0 Leave input plane unchanged. Default.
16993 1 Clips the pixel with the minimum and maximum of the 8 neighbour
16995 pixels.
16996 2 Clips the pixel with the second minimum and maximum of the 8
16998 neighbour pixels.
16999 3 Clips the pixel with the third minimum and maximum of the 8
17001 neighbour pixels.
17002 4 Clips the pixel with the fourth minimum and maximum of the 8
17004 neighbour pixels. This is equivalent to a median filter.
17005 5 Line-sensitive clipping giving the minimal change.
17007 6 Line-sensitive clipping, intermediate.
17009 7 Line-sensitive clipping, intermediate.
17011 8 Line-sensitive clipping, intermediate.
17013 9 Line-sensitive clipping on a line where the neighbours pixels are
17015 the closest.
17016 10 Replaces the target pixel with the closest neighbour.
17018 11 [1 2 1] horizontal and vertical kernel blur.
17020 12 Same as mode 11.
17022 13 Bob mode, interpolates top field from the line where the neighbours
17024 pixels are the closest.
17025 14 Bob mode, interpolates bottom field from the line where the
17027 neighbours pixels are the closest.
17028 15 Bob mode, interpolates top field. Same as 13 but with a more
17030 complicated interpolation formula.
17031 16 Bob mode, interpolates bottom field. Same as 14 but with a more
17033 complicated interpolation formula.
17034 17 Clips the pixel with the minimum and maximum of respectively the
17036 maximum and minimum of each pair of opposite neighbour pixels.
17037 18 Line-sensitive clipping using opposite neighbours whose greatest
17039 distance from the current pixel is minimal.
17040 19 Replaces the pixel with the average of its 8 neighbours.
17042 20 Averages the 9 pixels ([1 1 1] horizontal and vertical blur).
17044 21 Clips pixels using the averages of opposite neighbour.
17046 22 Same as mode 21 but simpler and faster.
17048 23 Small edge and halo removal, but reputed useless.
17050 24 Similar as 23.
17052 removelogo
17054 Suppress a TV station logo, using an image file to determine which
17055 pixels comprise the logo. It works by filling in the pixels that
17056 comprise the logo with neighboring pixels.
17057 The filter accepts the following options:
17059 filename, f
17061 Set the filter bitmap file, which can be any image format supported
17062 by libavformat. The width and height of the image file must match
17063 those of the video stream being processed.
17064 Pixels in the provided bitmap image with a value of zero are not
17066 considered part of the logo, non-zero pixels are considered part of the
17067 logo. If you use white (255) for the logo and black (0) for the rest,
17068 you will be safe. For making the filter bitmap, it is recommended to
17069 take a screen capture of a black frame with the logo visible, and then
17070 using a threshold filter followed by the erode filter once or twice.
17071 If needed, little splotches can be fixed manually. Remember that if
17073 logo pixels are not covered, the filter quality will be much reduced.
17074 Marking too many pixels as part of the logo does not hurt as much, but
17075 it will increase the amount of blurring needed to cover over the image
17076 and will destroy more information than necessary, and extra pixels will
17077 slow things down on a large logo.
17078 repeatfields
17080 This filter uses the repeat_field flag from the Video ES headers and
17081 hard repeats fields based on its value.
17082 reverse
17084 Reverse a video clip.
17085 Warning: This filter requires memory to buffer the entire clip, so
17087 trimming is suggested.
17088 Examples
17090 • Take the first 5 seconds of a clip, and reverse it.
17092 trim=end=5,reverse
17094 rgbashift
17096 Shift R/G/B/A pixels horizontally and/or vertically.
17097 The filter accepts the following options:
17099 rh Set amount to shift red horizontally.
17101 rv Set amount to shift red vertically.
17103 gh Set amount to shift green horizontally.
17105 gv Set amount to shift green vertically.
17107 bh Set amount to shift blue horizontally.
17109 bv Set amount to shift blue vertically.
17111 ah Set amount to shift alpha horizontally.
17113 av Set amount to shift alpha vertically.
17115 edge
17117 Set edge mode, can be smear, default, or warp.
17118 Commands
17120 This filter supports the all above options as commands.
17122 roberts
17124 Apply roberts cross operator to input video stream.
17125 The filter accepts the following option:
17127 planes
17129 Set which planes will be processed, unprocessed planes will be
17130 copied. By default value 0xf, all planes will be processed.
17131 scale
17133 Set value which will be multiplied with filtered result.
17134 delta
17136 Set value which will be added to filtered result.
17137 Commands
17139 This filter supports the all above options as commands.
17141 rotate
17143 Rotate video by an arbitrary angle expressed in radians.
17144 The filter accepts the following options:
17146 A description of the optional parameters follows.
17148 angle, a
17150 Set an expression for the angle by which to rotate the input video
17151 clockwise, expressed as a number of radians. A negative value will
17152 result in a counter-clockwise rotation. By default it is set to
17153 "0".
17154 This expression is evaluated for each frame.
17156 out_w, ow
17158 Set the output width expression, default value is "iw". This
17159 expression is evaluated just once during configuration.
17160 out_h, oh
17162 Set the output height expression, default value is "ih". This
17163 expression is evaluated just once during configuration.
17164 bilinear
17166 Enable bilinear interpolation if set to 1, a value of 0 disables
17167 it. Default value is 1.
17168 fillcolor, c
17170 Set the color used to fill the output area not covered by the
17171 rotated image. For the general syntax of this option, check the
17172 "Color" section in the ffmpeg-utils manual. If the special value
17173 "none" is selected then no background is printed (useful for
17174 example if the background is never shown).
17175 Default value is "black".
17177 The expressions for the angle and the output size can contain the
17179 following constants and functions:
17180 n sequential number of the input frame, starting from 0. It is always
17182 NAN before the first frame is filtered.
17183 t time in seconds of the input frame, it is set to 0 when the filter
17185 is configured. It is always NAN before the first frame is filtered.
17186 hsub
17188 vsub
17189 horizontal and vertical chroma subsample values. For example for
17190 the pixel format "yuv422p" hsub is 2 and vsub is 1.
17191 in_w, iw
17193 in_h, ih
17194 the input video width and height
17195 out_w, ow
17197 out_h, oh
17198 the output width and height, that is the size of the padded area as
17199 specified by the width and height expressions
17200 rotw(a)
17202 roth(a)
17203 the minimal width/height required for completely containing the
17204 input video rotated by a radians.
17205 These are only available when computing the out_w and out_h
17207 expressions.
17208 Examples
17210 • Rotate the input by PI/6 radians clockwise:
17212 rotate=PI/6
17214 • Rotate the input by PI/6 radians counter-clockwise:
17216 rotate=-PI/6
17218 • Rotate the input by 45 degrees clockwise:
17220 rotate=45*PI/180
17222 • Apply a constant rotation with period T, starting from an angle of
17224 PI/3:
17225 rotate=PI/3+2*PI*t/T
17227 • Make the input video rotation oscillating with a period of T
17229 seconds and an amplitude of A radians:
17230 rotate=A*sin(2*PI/T*t)
17232 • Rotate the video, output size is chosen so that the whole rotating
17234 input video is always completely contained in the output:
17235 rotate='2*PI*t:ow=hypot(iw,ih):oh=ow'
17237 • Rotate the video, reduce the output size so that no background is
17239 ever shown:
17240 rotate=2*PI*t:ow='min(iw,ih)/sqrt(2)':oh=ow:c=none
17242 Commands
17244 The filter supports the following commands:
17246 a, angle
17248 Set the angle expression. The command accepts the same syntax of
17249 the corresponding option.
17250 If the specified expression is not valid, it is kept at its current
17252 value.
17253 sab
17255 Apply Shape Adaptive Blur.
17256 The filter accepts the following options:
17258 luma_radius, lr
17260 Set luma blur filter strength, must be a value in range 0.1-4.0,
17261 default value is 1.0. A greater value will result in a more blurred
17262 image, and in slower processing.
17263 luma_pre_filter_radius, lpfr
17265 Set luma pre-filter radius, must be a value in the 0.1-2.0 range,
17266 default value is 1.0.
17267 luma_strength, ls
17269 Set luma maximum difference between pixels to still be considered,
17270 must be a value in the 0.1-100.0 range, default value is 1.0.
17271 chroma_radius, cr
17273 Set chroma blur filter strength, must be a value in range -0.9-4.0.
17274 A greater value will result in a more blurred image, and in slower
17275 processing.
17276 chroma_pre_filter_radius, cpfr
17278 Set chroma pre-filter radius, must be a value in the -0.9-2.0
17279 range.
17280 chroma_strength, cs
17282 Set chroma maximum difference between pixels to still be
17283 considered, must be a value in the -0.9-100.0 range.
17284 Each chroma option value, if not explicitly specified, is set to the
17286 corresponding luma option value.
17287 scale
17289 Scale (resize) the input video, using the libswscale library.
17290 The scale filter forces the output display aspect ratio to be the same
17292 of the input, by changing the output sample aspect ratio.
17293 If the input image format is different from the format requested by the
17295 next filter, the scale filter will convert the input to the requested
17296 format.
17297 Options
17299 The filter accepts the following options, or any of the options
17301 supported by the libswscale scaler.
17302 See the ffmpeg-scaler manual for the complete list of scaler options.
17304 width, w
17306 height, h
17307 Set the output video dimension expression. Default value is the
17308 input dimension.
17309 If the width or w value is 0, the input width is used for the
17311 output. If the height or h value is 0, the input height is used for
17312 the output.
17313 If one and only one of the values is -n with n >= 1, the scale
17315 filter will use a value that maintains the aspect ratio of the
17316 input image, calculated from the other specified dimension. After
17317 that it will, however, make sure that the calculated dimension is
17318 divisible by n and adjust the value if necessary.
17319 If both values are -n with n >= 1, the behavior will be identical
17321 to both values being set to 0 as previously detailed.
17322 See below for the list of accepted constants for use in the
17324 dimension expression.
17325 eval
17327 Specify when to evaluate width and height expression. It accepts
17328 the following values:
17329 init
17331 Only evaluate expressions once during the filter initialization
17332 or when a command is processed.
17333 frame
17335 Evaluate expressions for each incoming frame.
17336 Default value is init.
17338 interl
17340 Set the interlacing mode. It accepts the following values:
17341 1 Force interlaced aware scaling.
17343 0 Do not apply interlaced scaling.
17345 -1 Select interlaced aware scaling depending on whether the source
17347 frames are flagged as interlaced or not.
17348 Default value is 0.
17350 flags
17352 Set libswscale scaling flags. See the ffmpeg-scaler manual for the
17353 complete list of values. If not explicitly specified the filter
17354 applies the default flags.
17355 param0, param1
17357 Set libswscale input parameters for scaling algorithms that need
17358 them. See the ffmpeg-scaler manual for the complete documentation.
17359 If not explicitly specified the filter applies empty parameters.
17360 size, s
17362 Set the video size. For the syntax of this option, check the "Video
17363 size" section in the ffmpeg-utils manual.
17364 in_color_matrix
17366 out_color_matrix
17367 Set in/output YCbCr color space type.
17368 This allows the autodetected value to be overridden as well as
17370 allows forcing a specific value used for the output and encoder.
17371 If not specified, the color space type depends on the pixel format.
17373 Possible values:
17375 auto
17377 Choose automatically.
17378 bt709
17380 Format conforming to International Telecommunication Union
17381 (ITU) Recommendation BT.709.
17382 fcc Set color space conforming to the United States Federal
17384 Communications Commission (FCC) Code of Federal Regulations
17385 (CFR) Title 47 (2003) 73.682 (a).
17386 bt601
17388 bt470
17389 smpte170m
17390 Set color space conforming to:
17391 • ITU Radiocommunication Sector (ITU-R) Recommendation BT.601
17393 • ITU-R Rec. BT.470-6 (1998) Systems B, B1, and G
17395 • Society of Motion Picture and Television Engineers (SMPTE)
17397 ST 170:2004
17398 smpte240m
17400 Set color space conforming to SMPTE ST 240:1999.
17401 bt2020
17403 Set color space conforming to ITU-R BT.2020 non-constant
17404 luminance system.
17405 in_range
17407 out_range
17408 Set in/output YCbCr sample range.
17409 This allows the autodetected value to be overridden as well as
17411 allows forcing a specific value used for the output and encoder. If
17412 not specified, the range depends on the pixel format. Possible
17413 values:
17414 auto/unknown
17416 Choose automatically.
17417 jpeg/full/pc
17419 Set full range (0-255 in case of 8-bit luma).
17420 mpeg/limited/tv
17422 Set "MPEG" range (16-235 in case of 8-bit luma).
17423 force_original_aspect_ratio
17425 Enable decreasing or increasing output video width or height if
17426 necessary to keep the original aspect ratio. Possible values:
17427 disable
17429 Scale the video as specified and disable this feature.
17430 decrease
17432 The output video dimensions will automatically be decreased if
17433 needed.
17434 increase
17436 The output video dimensions will automatically be increased if
17437 needed.
17438 One useful instance of this option is that when you know a specific
17440 device's maximum allowed resolution, you can use this to limit the
17441 output video to that, while retaining the aspect ratio. For
17442 example, device A allows 1280x720 playback, and your video is
17443 1920x800. Using this option (set it to decrease) and specifying
17444 1280x720 to the command line makes the output 1280x533.
17445 Please note that this is a different thing than specifying -1 for w
17447 or h, you still need to specify the output resolution for this
17448 option to work.
17449 force_divisible_by
17451 Ensures that both the output dimensions, width and height, are
17452 divisible by the given integer when used together with
17453 force_original_aspect_ratio. This works similar to using "-n" in
17454 the w and h options.
17455 This option respects the value set for force_original_aspect_ratio,
17457 increasing or decreasing the resolution accordingly. The video's
17458 aspect ratio may be slightly modified.
17459 This option can be handy if you need to have a video fit within or
17461 exceed a defined resolution using force_original_aspect_ratio but
17462 also have encoder restrictions on width or height divisibility.
17463 The values of the w and h options are expressions containing the
17465 following constants:
17466 in_w
17468 in_h
17469 The input width and height
17470 iw
17472 ih These are the same as in_w and in_h.
17473 out_w
17475 out_h
17476 The output (scaled) width and height
17477 ow
17479 oh These are the same as out_w and out_h
17480 a The same as iw / ih
17482 sar input sample aspect ratio
17484 dar The input display aspect ratio. Calculated from "(iw / ih) * sar".
17486 hsub
17488 vsub
17489 horizontal and vertical input chroma subsample values. For example
17490 for the pixel format "yuv422p" hsub is 2 and vsub is 1.
17491 ohsub
17493 ovsub
17494 horizontal and vertical output chroma subsample values. For example
17495 for the pixel format "yuv422p" hsub is 2 and vsub is 1.
17496 n The (sequential) number of the input frame, starting from 0. Only
17498 available with "eval=frame".
17499 t The presentation timestamp of the input frame, expressed as a
17501 number of seconds. Only available with "eval=frame".
17502 pos The position (byte offset) of the frame in the input stream, or NaN
17504 if this information is unavailable and/or meaningless (for example
17505 in case of synthetic video). Only available with "eval=frame".
17506 Examples
17508 • Scale the input video to a size of 200x100
17510 scale=w=200:h=100
17512 This is equivalent to:
17514 scale=200:100
17516 or:
17518 scale=200x100
17520 • Specify a size abbreviation for the output size:
17522 scale=qcif
17524 which can also be written as:
17526 scale=size=qcif
17528 • Scale the input to 2x:
17530 scale=w=2*iw:h=2*ih
17532 • The above is the same as:
17534 scale=2*in_w:2*in_h
17536 • Scale the input to 2x with forced interlaced scaling:
17538 scale=2*iw:2*ih:interl=1
17540 • Scale the input to half size:
17542 scale=w=iw/2:h=ih/2
17544 • Increase the width, and set the height to the same size:
17546 scale=3/2*iw:ow
17548 • Seek Greek harmony:
17550 scale=iw:1/PHI*iw
17552 scale=ih*PHI:ih
17553 • Increase the height, and set the width to 3/2 of the height:
17555 scale=w=3/2*oh:h=3/5*ih
17557 • Increase the size, making the size a multiple of the chroma
17559 subsample values:
17560 scale="trunc(3/2*iw/hsub)*hsub:trunc(3/2*ih/vsub)*vsub"
17562 • Increase the width to a maximum of 500 pixels, keeping the same
17564 aspect ratio as the input:
17565 scale=w='min(500\, iw*3/2):h=-1'
17567 • Make pixels square by combining scale and setsar:
17569 scale='trunc(ih*dar):ih',setsar=1/1
17571 • Make pixels square by combining scale and setsar, making sure the
17573 resulting resolution is even (required by some codecs):
17574 scale='trunc(ih*dar/2)*2:trunc(ih/2)*2',setsar=1/1
17576 Commands
17578 This filter supports the following commands:
17580 width, w
17582 height, h
17583 Set the output video dimension expression. The command accepts the
17584 same syntax of the corresponding option.
17585 If the specified expression is not valid, it is kept at its current
17587 value.
17588 scale_cuda
17590 Scale (resize) and convert (pixel format) the input video, using
17591 accelerated CUDA kernels. Setting the output width and height works in
17592 the same way as for the scale filter.
17593 The filter accepts the following options:
17595 w
17597 h Set the output video dimension expression. Default value is the
17598 input dimension.
17599 Allows for the same expressions as the scale filter.
17601 interp_algo
17603 Sets the algorithm used for scaling:
17604 nearest
17606 Nearest neighbour
17607 Used by default if input parameters match the desired output.
17609 bilinear
17611 Bilinear
17612 bicubic
17614 Bicubic
17615 This is the default.
17617 lanczos
17619 Lanczos
17620 format
17622 Controls the output pixel format. By default, or if none is
17623 specified, the input pixel format is used.
17624 The filter does not support converting between YUV and RGB pixel
17626 formats.
17627 passthrough
17629 If set to 0, every frame is processed, even if no conversion is
17630 neccesary. This mode can be useful to use the filter as a buffer
17631 for a downstream frame-consumer that exhausts the limited decoder
17632 frame pool.
17633 If set to 1, frames are passed through as-is if they match the
17635 desired output parameters. This is the default behaviour.
17636 param
17638 Algorithm-Specific parameter.
17639 Affects the curves of the bicubic algorithm.
17641 force_original_aspect_ratio
17643 force_divisible_by
17644 Work the same as the identical scale filter options.
17645 Examples
17647 • Scale input to 720p, keeping aspect ratio and ensuring the output
17649 is yuv420p.
17650 scale_cuda=-2:720:format=yuv420p
17652 • Upscale to 4K using nearest neighbour algorithm.
17654 scale_cuda=4096:2160:interp_algo=nearest
17656 • Don't do any conversion or scaling, but copy all input frames into
17658 newly allocated ones. This can be useful to deal with a filter and
17659 encode chain that otherwise exhausts the decoders frame pool.
17660 scale_cuda=passthrough=0
17662 scale_npp
17664 Use the NVIDIA Performance Primitives (libnpp) to perform scaling
17665 and/or pixel format conversion on CUDA video frames. Setting the output
17666 width and height works in the same way as for the scale filter.
17667 The following additional options are accepted:
17669 format
17671 The pixel format of the output CUDA frames. If set to the string
17672 "same" (the default), the input format will be kept. Note that
17673 automatic format negotiation and conversion is not yet supported
17674 for hardware frames
17675 interp_algo
17677 The interpolation algorithm used for resizing. One of the
17678 following:
17679 nn Nearest neighbour.
17681 linear
17683 cubic
17684 cubic2p_bspline
17685 2-parameter cubic (B=1, C=0)
17686 cubic2p_catmullrom
17688 2-parameter cubic (B=0, C=1/2)
17689 cubic2p_b05c03
17691 2-parameter cubic (B=1/2, C=3/10)
17692 super
17694 Supersampling
17695 lanczos
17697 force_original_aspect_ratio
17698 Enable decreasing or increasing output video width or height if
17699 necessary to keep the original aspect ratio. Possible values:
17700 disable
17702 Scale the video as specified and disable this feature.
17703 decrease
17705 The output video dimensions will automatically be decreased if
17706 needed.
17707 increase
17709 The output video dimensions will automatically be increased if
17710 needed.
17711 One useful instance of this option is that when you know a specific
17713 device's maximum allowed resolution, you can use this to limit the
17714 output video to that, while retaining the aspect ratio. For
17715 example, device A allows 1280x720 playback, and your video is
17716 1920x800. Using this option (set it to decrease) and specifying
17717 1280x720 to the command line makes the output 1280x533.
17718 Please note that this is a different thing than specifying -1 for w
17720 or h, you still need to specify the output resolution for this
17721 option to work.
17722 force_divisible_by
17724 Ensures that both the output dimensions, width and height, are
17725 divisible by the given integer when used together with
17726 force_original_aspect_ratio. This works similar to using "-n" in
17727 the w and h options.
17728 This option respects the value set for force_original_aspect_ratio,
17730 increasing or decreasing the resolution accordingly. The video's
17731 aspect ratio may be slightly modified.
17732 This option can be handy if you need to have a video fit within or
17734 exceed a defined resolution using force_original_aspect_ratio but
17735 also have encoder restrictions on width or height divisibility.
17736 eval
17738 Specify when to evaluate width and height expression. It accepts
17739 the following values:
17740 init
17742 Only evaluate expressions once during the filter initialization
17743 or when a command is processed.
17744 frame
17746 Evaluate expressions for each incoming frame.
17747 The values of the w and h options are expressions containing the
17749 following constants:
17750 in_w
17752 in_h
17753 The input width and height
17754 iw
17756 ih These are the same as in_w and in_h.
17757 out_w
17759 out_h
17760 The output (scaled) width and height
17761 ow
17763 oh These are the same as out_w and out_h
17764 a The same as iw / ih
17766 sar input sample aspect ratio
17768 dar The input display aspect ratio. Calculated from "(iw / ih) * sar".
17770 n The (sequential) number of the input frame, starting from 0. Only
17772 available with "eval=frame".
17773 t The presentation timestamp of the input frame, expressed as a
17775 number of seconds. Only available with "eval=frame".
17776 pos The position (byte offset) of the frame in the input stream, or NaN
17778 if this information is unavailable and/or meaningless (for example
17779 in case of synthetic video). Only available with "eval=frame".
17780 scale2ref
17782 Scale (resize) the input video, based on a reference video.
17783 See the scale filter for available options, scale2ref supports the same
17785 but uses the reference video instead of the main input as basis.
17786 scale2ref also supports the following additional constants for the w
17787 and h options:
17788 main_w
17790 main_h
17791 The main input video's width and height
17792 main_a
17794 The same as main_w / main_h
17795 main_sar
17797 The main input video's sample aspect ratio
17798 main_dar, mdar
17800 The main input video's display aspect ratio. Calculated from
17801 "(main_w / main_h) * main_sar".
17802 main_hsub
17804 main_vsub
17805 The main input video's horizontal and vertical chroma subsample
17806 values. For example for the pixel format "yuv422p" hsub is 2 and
17807 vsub is 1.
17808 main_n
17810 The (sequential) number of the main input frame, starting from 0.
17811 Only available with "eval=frame".
17812 main_t
17814 The presentation timestamp of the main input frame, expressed as a
17815 number of seconds. Only available with "eval=frame".
17816 main_pos
17818 The position (byte offset) of the frame in the main input stream,
17819 or NaN if this information is unavailable and/or meaningless (for
17820 example in case of synthetic video). Only available with
17821 "eval=frame".
17822 Examples
17824 • Scale a subtitle stream (b) to match the main video (a) in size
17826 before overlaying
17827 'scale2ref[b][a];[a][b]overlay'
17829 • Scale a logo to 1/10th the height of a video, while preserving its
17831 display aspect ratio.
17832 [logo-in][video-in]scale2ref=w=oh*mdar:h=ih/10[logo-out][video-out]
17834 Commands
17836 This filter supports the following commands:
17838 width, w
17840 height, h
17841 Set the output video dimension expression. The command accepts the
17842 same syntax of the corresponding option.
17843 If the specified expression is not valid, it is kept at its current
17845 value.
17846 scale2ref_npp
17848 Use the NVIDIA Performance Primitives (libnpp) to scale (resize) the
17849 input video, based on a reference video.
17850 See the scale_npp filter for available options, scale2ref_npp supports
17852 the same but uses the reference video instead of the main input as
17853 basis. scale2ref_npp also supports the following additional constants
17854 for the w and h options:
17855 main_w
17857 main_h
17858 The main input video's width and height
17859 main_a
17861 The same as main_w / main_h
17862 main_sar
17864 The main input video's sample aspect ratio
17865 main_dar, mdar
17867 The main input video's display aspect ratio. Calculated from
17868 "(main_w / main_h) * main_sar".
17869 main_n
17871 The (sequential) number of the main input frame, starting from 0.
17872 Only available with "eval=frame".
17873 main_t
17875 The presentation timestamp of the main input frame, expressed as a
17876 number of seconds. Only available with "eval=frame".
17877 main_pos
17879 The position (byte offset) of the frame in the main input stream,
17880 or NaN if this information is unavailable and/or meaningless (for
17881 example in case of synthetic video). Only available with
17882 "eval=frame".
17883 Examples
17885 • Scale a subtitle stream (b) to match the main video (a) in size
17887 before overlaying
17888 'scale2ref_npp[b][a];[a][b]overlay_cuda'
17890 • Scale a logo to 1/10th the height of a video, while preserving its
17892 display aspect ratio.
17893 [logo-in][video-in]scale2ref_npp=w=oh*mdar:h=ih/10[logo-out][video-out]
17895 scharr
17897 Apply scharr operator to input video stream.
17898 The filter accepts the following option:
17900 planes
17902 Set which planes will be processed, unprocessed planes will be
17903 copied. By default value 0xf, all planes will be processed.
17904 scale
17906 Set value which will be multiplied with filtered result.
17907 delta
17909 Set value which will be added to filtered result.
17910 Commands
17912 This filter supports the all above options as commands.
17914 scroll
17916 Scroll input video horizontally and/or vertically by constant speed.
17917 The filter accepts the following options:
17919 horizontal, h
17921 Set the horizontal scrolling speed. Default is 0. Allowed range is
17922 from -1 to 1. Negative values changes scrolling direction.
17923 vertical, v
17925 Set the vertical scrolling speed. Default is 0. Allowed range is
17926 from -1 to 1. Negative values changes scrolling direction.
17927 hpos
17929 Set the initial horizontal scrolling position. Default is 0.
17930 Allowed range is from 0 to 1.
17931 vpos
17933 Set the initial vertical scrolling position. Default is 0. Allowed
17934 range is from 0 to 1.
17935 Commands
17937 This filter supports the following commands:
17939 horizontal, h
17941 Set the horizontal scrolling speed.
17942 vertical, v
17944 Set the vertical scrolling speed.
17945 scdet
17947 Detect video scene change.
17948 This filter sets frame metadata with mafd between frame, the scene
17950 score, and forward the frame to the next filter, so they can use these
17951 metadata to detect scene change or others.
17952 In addition, this filter logs a message and sets frame metadata when it
17954 detects a scene change by threshold.
17955 "lavfi.scd.mafd" metadata keys are set with mafd for every frame.
17957 "lavfi.scd.score" metadata keys are set with scene change score for
17959 every frame to detect scene change.
17960 "lavfi.scd.time" metadata keys are set with current filtered frame time
17962 which detect scene change with threshold.
17963 The filter accepts the following options:
17965 threshold, t
17967 Set the scene change detection threshold as a percentage of maximum
17968 change. Good values are in the "[8.0, 14.0]" range. The range for
17969 threshold is "[0., 100.]".
17970 Default value is 10..
17972 sc_pass, s
17974 Set the flag to pass scene change frames to the next filter.
17975 Default value is 0 You can enable it if you want to get snapshot of
17976 scene change frames only.
17977 selectivecolor
17979 Adjust cyan, magenta, yellow and black (CMYK) to certain ranges of
17980 colors (such as "reds", "yellows", "greens", "cyans", ...). The
17981 adjustment range is defined by the "purity" of the color (that is, how
17982 saturated it already is).
17983 This filter is similar to the Adobe Photoshop Selective Color tool.
17985 The filter accepts the following options:
17987 correction_method
17989 Select color correction method.
17990 Available values are:
17992 absolute
17994 Specified adjustments are applied "as-is" (added/subtracted to
17995 original pixel component value).
17996 relative
17998 Specified adjustments are relative to the original component
17999 value.
18000 Default is "absolute".
18002 reds
18004 Adjustments for red pixels (pixels where the red component is the
18005 maximum)
18006 yellows
18008 Adjustments for yellow pixels (pixels where the blue component is
18009 the minimum)
18010 greens
18012 Adjustments for green pixels (pixels where the green component is
18013 the maximum)
18014 cyans
18016 Adjustments for cyan pixels (pixels where the red component is the
18017 minimum)
18018 blues
18020 Adjustments for blue pixels (pixels where the blue component is the
18021 maximum)
18022 magentas
18024 Adjustments for magenta pixels (pixels where the green component is
18025 the minimum)
18026 whites
18028 Adjustments for white pixels (pixels where all components are
18029 greater than 128)
18030 neutrals
18032 Adjustments for all pixels except pure black and pure white
18033 blacks
18035 Adjustments for black pixels (pixels where all components are
18036 lesser than 128)
18037 psfile
18039 Specify a Photoshop selective color file (".asv") to import the
18040 settings from.
18041 All the adjustment settings (reds, yellows, ...) accept up to 4 space
18043 separated floating point adjustment values in the [-1,1] range,
18044 respectively to adjust the amount of cyan, magenta, yellow and black
18045 for the pixels of its range.
18046 Examples
18048 • Increase cyan by 50% and reduce yellow by 33% in every green areas,
18050 and increase magenta by 27% in blue areas:
18051 selectivecolor=greens=.5 0 -.33 0:blues=0 .27
18053 • Use a Photoshop selective color preset:
18055 selectivecolor=psfile=MySelectiveColorPresets/Misty.asv
18057 separatefields
18059 The "separatefields" takes a frame-based video input and splits each
18060 frame into its components fields, producing a new half height clip with
18061 twice the frame rate and twice the frame count.
18062 This filter use field-dominance information in frame to decide which of
18064 each pair of fields to place first in the output. If it gets it wrong
18065 use setfield filter before "separatefields" filter.
18066 setdar, setsar
18068 The "setdar" filter sets the Display Aspect Ratio for the filter output
18069 video.
18070 This is done by changing the specified Sample (aka Pixel) Aspect Ratio,
18072 according to the following equation:
18073 <DAR> = <HORIZONTAL_RESOLUTION> / <VERTICAL_RESOLUTION> * <SAR>
18075 Keep in mind that the "setdar" filter does not modify the pixel
18077 dimensions of the video frame. Also, the display aspect ratio set by
18078 this filter may be changed by later filters in the filterchain, e.g. in
18079 case of scaling or if another "setdar" or a "setsar" filter is applied.
18080 The "setsar" filter sets the Sample (aka Pixel) Aspect Ratio for the
18082 filter output video.
18083 Note that as a consequence of the application of this filter, the
18085 output display aspect ratio will change according to the equation
18086 above.
18087 Keep in mind that the sample aspect ratio set by the "setsar" filter
18089 may be changed by later filters in the filterchain, e.g. if another
18090 "setsar" or a "setdar" filter is applied.
18091 It accepts the following parameters:
18093 r, ratio, dar ("setdar" only), sar ("setsar" only)
18095 Set the aspect ratio used by the filter.
18096 The parameter can be a floating point number string, an expression,
18098 or a string of the form num:den, where num and den are the
18099 numerator and denominator of the aspect ratio. If the parameter is
18100 not specified, it is assumed the value "0". In case the form
18101 "num:den" is used, the ":" character should be escaped.
18102 max Set the maximum integer value to use for expressing numerator and
18104 denominator when reducing the expressed aspect ratio to a rational.
18105 Default value is 100.
18106 The parameter sar is an expression containing the following constants:
18108 E, PI, PHI
18110 These are approximated values for the mathematical constants e
18111 (Euler's number), pi (Greek pi), and phi (the golden ratio).
18112 w, h
18114 The input width and height.
18115 a These are the same as w / h.
18117 sar The input sample aspect ratio.
18119 dar The input display aspect ratio. It is the same as (w / h) * sar.
18121 hsub, vsub
18123 Horizontal and vertical chroma subsample values. For example, for
18124 the pixel format "yuv422p" hsub is 2 and vsub is 1.
18125 Examples
18127 • To change the display aspect ratio to 16:9, specify one of the
18129 following:
18130 setdar=dar=1.77777
18132 setdar=dar=16/9
18133 • To change the sample aspect ratio to 10:11, specify:
18135 setsar=sar=10/11
18137 • To set a display aspect ratio of 16:9, and specify a maximum
18139 integer value of 1000 in the aspect ratio reduction, use the
18140 command:
18141 setdar=ratio=16/9:max=1000
18143 setfield
18145 Force field for the output video frame.
18146 The "setfield" filter marks the interlace type field for the output
18148 frames. It does not change the input frame, but only sets the
18149 corresponding property, which affects how the frame is treated by
18150 following filters (e.g. "fieldorder" or "yadif").
18151 The filter accepts the following options:
18153 mode
18155 Available values are:
18156 auto
18158 Keep the same field property.
18159 bff Mark the frame as bottom-field-first.
18161 tff Mark the frame as top-field-first.
18163 prog
18165 Mark the frame as progressive.
18166 setparams
18168 Force frame parameter for the output video frame.
18169 The "setparams" filter marks interlace and color range for the output
18171 frames. It does not change the input frame, but only sets the
18172 corresponding property, which affects how the frame is treated by
18173 filters/encoders.
18174 field_mode
18176 Available values are:
18177 auto
18179 Keep the same field property (default).
18180 bff Mark the frame as bottom-field-first.
18182 tff Mark the frame as top-field-first.
18184 prog
18186 Mark the frame as progressive.
18187 range
18189 Available values are:
18190 auto
18192 Keep the same color range property (default).
18193 unspecified, unknown
18195 Mark the frame as unspecified color range.
18196 limited, tv, mpeg
18198 Mark the frame as limited range.
18199 full, pc, jpeg
18201 Mark the frame as full range.
18202 color_primaries
18204 Set the color primaries. Available values are:
18205 auto
18207 Keep the same color primaries property (default).
18208 bt709
18210 unknown
18211 bt470m
18212 bt470bg
18213 smpte170m
18214 smpte240m
18215 film
18216 bt2020
18217 smpte428
18218 smpte431
18219 smpte432
18220 jedec-p22
18221 color_trc
18222 Set the color transfer. Available values are:
18223 auto
18225 Keep the same color trc property (default).
18226 bt709
18228 unknown
18229 bt470m
18230 bt470bg
18231 smpte170m
18232 smpte240m
18233 linear
18234 log100
18235 log316
18236 iec61966-2-4
18237 bt1361e
18238 iec61966-2-1
18239 bt2020-10
18240 bt2020-12
18241 smpte2084
18242 smpte428
18243 arib-std-b67
18244 colorspace
18245 Set the colorspace. Available values are:
18246 auto
18248 Keep the same colorspace property (default).
18249 gbr
18251 bt709
18252 unknown
18253 fcc
18254 bt470bg
18255 smpte170m
18256 smpte240m
18257 ycgco
18258 bt2020nc
18259 bt2020c
18260 smpte2085
18261 chroma-derived-nc
18262 chroma-derived-c
18263 ictcp
18264 sharpen_npp
18266 Use the NVIDIA Performance Primitives (libnpp) to perform image
18267 sharpening with border control.
18268 The following additional options are accepted:
18270 border_type
18272 Type of sampling to be used ad frame borders. One of the following:
18273 replicate
18275 Replicate pixel values.
18276 shear
18278 Apply shear transform to input video.
18279 This filter supports the following options:
18281 shx Shear factor in X-direction. Default value is 0. Allowed range is
18283 from -2 to 2.
18284 shy Shear factor in Y-direction. Default value is 0. Allowed range is
18286 from -2 to 2.
18287 fillcolor, c
18289 Set the color used to fill the output area not covered by the
18290 transformed video. For the general syntax of this option, check the
18291 "Color" section in the ffmpeg-utils manual. If the special value
18292 "none" is selected then no background is printed (useful for
18293 example if the background is never shown).
18294 Default value is "black".
18296 interp
18298 Set interpolation type. Can be "bilinear" or "nearest". Default is
18299 "bilinear".
18300 Commands
18302 This filter supports the all above options as commands.
18304 showinfo
18306 Show a line containing various information for each input video frame.
18307 The input video is not modified.
18308 This filter supports the following options:
18310 checksum
18312 Calculate checksums of each plane. By default enabled.
18313 The shown line contains a sequence of key/value pairs of the form
18315 key:value.
18316 The following values are shown in the output:
18318 n The (sequential) number of the input frame, starting from 0.
18320 pts The Presentation TimeStamp of the input frame, expressed as a
18322 number of time base units. The time base unit depends on the filter
18323 input pad.
18324 pts_time
18326 The Presentation TimeStamp of the input frame, expressed as a
18327 number of seconds.
18328 pos The position of the frame in the input stream, or -1 if this
18330 information is unavailable and/or meaningless (for example in case
18331 of synthetic video).
18332 fmt The pixel format name.
18334 sar The sample aspect ratio of the input frame, expressed in the form
18336 num/den.
18337 s The size of the input frame. For the syntax of this option, check
18339 the "Video size" section in the ffmpeg-utils manual.
18340 i The type of interlaced mode ("P" for "progressive", "T" for top
18342 field first, "B" for bottom field first).
18343 iskey
18345 This is 1 if the frame is a key frame, 0 otherwise.
18346 type
18348 The picture type of the input frame ("I" for an I-frame, "P" for a
18349 P-frame, "B" for a B-frame, or "?" for an unknown type). Also
18350 refer to the documentation of the "AVPictureType" enum and of the
18351 "av_get_picture_type_char" function defined in libavutil/avutil.h.
18352 checksum
18354 The Adler-32 checksum (printed in hexadecimal) of all the planes of
18355 the input frame.
18356 plane_checksum
18358 The Adler-32 checksum (printed in hexadecimal) of each plane of the
18359 input frame, expressed in the form "[c0 c1 c2 c3]".
18360 mean
18362 The mean value of pixels in each plane of the input frame,
18363 expressed in the form "[mean0 mean1 mean2 mean3]".
18364 stdev
18366 The standard deviation of pixel values in each plane of the input
18367 frame, expressed in the form "[stdev0 stdev1 stdev2 stdev3]".
18368 showpalette
18370 Displays the 256 colors palette of each frame. This filter is only
18371 relevant for pal8 pixel format frames.
18372 It accepts the following option:
18374 s Set the size of the box used to represent one palette color entry.
18376 Default is 30 (for a "30x30" pixel box).
18377 shuffleframes
18379 Reorder and/or duplicate and/or drop video frames.
18380 It accepts the following parameters:
18382 mapping
18384 Set the destination indexes of input frames. This is space or '|'
18385 separated list of indexes that maps input frames to output frames.
18386 Number of indexes also sets maximal value that each index may have.
18387 '-1' index have special meaning and that is to drop frame.
18388 The first frame has the index 0. The default is to keep the input
18390 unchanged.
18391 Examples
18393 • Swap second and third frame of every three frames of the input:
18395 ffmpeg -i INPUT -vf "shuffleframes=0 2 1" OUTPUT
18397 • Swap 10th and 1st frame of every ten frames of the input:
18399 ffmpeg -i INPUT -vf "shuffleframes=9 1 2 3 4 5 6 7 8 0" OUTPUT
18401 shufflepixels
18403 Reorder pixels in video frames.
18404 This filter accepts the following options:
18406 direction, d
18408 Set shuffle direction. Can be forward or inverse direction.
18409 Default direction is forward.
18410 mode, m
18412 Set shuffle mode. Can be horizontal, vertical or block mode.
18413 width, w
18415 height, h
18416 Set shuffle block_size. In case of horizontal shuffle mode only
18417 width part of size is used, and in case of vertical shuffle mode
18418 only height part of size is used.
18419 seed, s
18421 Set random seed used with shuffling pixels. Mainly useful to set to
18422 be able to reverse filtering process to get original input. For
18423 example, to reverse forward shuffle you need to use same parameters
18424 and exact same seed and to set direction to inverse.
18425 shuffleplanes
18427 Reorder and/or duplicate video planes.
18428 It accepts the following parameters:
18430 map0
18432 The index of the input plane to be used as the first output plane.
18433 map1
18435 The index of the input plane to be used as the second output plane.
18436 map2
18438 The index of the input plane to be used as the third output plane.
18439 map3
18441 The index of the input plane to be used as the fourth output plane.
18442 The first plane has the index 0. The default is to keep the input
18444 unchanged.
18445 Examples
18447 • Swap the second and third planes of the input:
18449 ffmpeg -i INPUT -vf shuffleplanes=0:2:1:3 OUTPUT
18451 signalstats
18453 Evaluate various visual metrics that assist in determining issues
18454 associated with the digitization of analog video media.
18455 By default the filter will log these metadata values:
18457 YMIN
18459 Display the minimal Y value contained within the input frame.
18460 Expressed in range of [0-255].
18461 YLOW
18463 Display the Y value at the 10% percentile within the input frame.
18464 Expressed in range of [0-255].
18465 YAVG
18467 Display the average Y value within the input frame. Expressed in
18468 range of [0-255].
18469 YHIGH
18471 Display the Y value at the 90% percentile within the input frame.
18472 Expressed in range of [0-255].
18473 YMAX
18475 Display the maximum Y value contained within the input frame.
18476 Expressed in range of [0-255].
18477 UMIN
18479 Display the minimal U value contained within the input frame.
18480 Expressed in range of [0-255].
18481 ULOW
18483 Display the U value at the 10% percentile within the input frame.
18484 Expressed in range of [0-255].
18485 UAVG
18487 Display the average U value within the input frame. Expressed in
18488 range of [0-255].
18489 UHIGH
18491 Display the U value at the 90% percentile within the input frame.
18492 Expressed in range of [0-255].
18493 UMAX
18495 Display the maximum U value contained within the input frame.
18496 Expressed in range of [0-255].
18497 VMIN
18499 Display the minimal V value contained within the input frame.
18500 Expressed in range of [0-255].
18501 VLOW
18503 Display the V value at the 10% percentile within the input frame.
18504 Expressed in range of [0-255].
18505 VAVG
18507 Display the average V value within the input frame. Expressed in
18508 range of [0-255].
18509 VHIGH
18511 Display the V value at the 90% percentile within the input frame.
18512 Expressed in range of [0-255].
18513 VMAX
18515 Display the maximum V value contained within the input frame.
18516 Expressed in range of [0-255].
18517 SATMIN
18519 Display the minimal saturation value contained within the input
18520 frame. Expressed in range of [0-~181.02].
18521 SATLOW
18523 Display the saturation value at the 10% percentile within the input
18524 frame. Expressed in range of [0-~181.02].
18525 SATAVG
18527 Display the average saturation value within the input frame.
18528 Expressed in range of [0-~181.02].
18529 SATHIGH
18531 Display the saturation value at the 90% percentile within the input
18532 frame. Expressed in range of [0-~181.02].
18533 SATMAX
18535 Display the maximum saturation value contained within the input
18536 frame. Expressed in range of [0-~181.02].
18537 HUEMED
18539 Display the median value for hue within the input frame. Expressed
18540 in range of [0-360].
18541 HUEAVG
18543 Display the average value for hue within the input frame. Expressed
18544 in range of [0-360].
18545 YDIF
18547 Display the average of sample value difference between all values
18548 of the Y plane in the current frame and corresponding values of the
18549 previous input frame. Expressed in range of [0-255].
18550 UDIF
18552 Display the average of sample value difference between all values
18553 of the U plane in the current frame and corresponding values of the
18554 previous input frame. Expressed in range of [0-255].
18555 VDIF
18557 Display the average of sample value difference between all values
18558 of the V plane in the current frame and corresponding values of the
18559 previous input frame. Expressed in range of [0-255].
18560 YBITDEPTH
18562 Display bit depth of Y plane in current frame. Expressed in range
18563 of [0-16].
18564 UBITDEPTH
18566 Display bit depth of U plane in current frame. Expressed in range
18567 of [0-16].
18568 VBITDEPTH
18570 Display bit depth of V plane in current frame. Expressed in range
18571 of [0-16].
18572 The filter accepts the following options:
18574 stat
18576 out stat specify an additional form of image analysis. out output
18577 video with the specified type of pixel highlighted.
18578 Both options accept the following values:
18580 tout
18582 Identify temporal outliers pixels. A temporal outlier is a
18583 pixel unlike the neighboring pixels of the same field. Examples
18584 of temporal outliers include the results of video dropouts,
18585 head clogs, or tape tracking issues.
18586 vrep
18588 Identify vertical line repetition. Vertical line repetition
18589 includes similar rows of pixels within a frame. In born-digital
18590 video vertical line repetition is common, but this pattern is
18591 uncommon in video digitized from an analog source. When it
18592 occurs in video that results from the digitization of an analog
18593 source it can indicate concealment from a dropout compensator.
18594 brng
18596 Identify pixels that fall outside of legal broadcast range.
18597 color, c
18599 Set the highlight color for the out option. The default color is
18600 yellow.
18601 Examples
18603 • Output data of various video metrics:
18605 ffprobe -f lavfi movie=example.mov,signalstats="stat=tout+vrep+brng" -show_frames
18607 • Output specific data about the minimum and maximum values of the Y
18609 plane per frame:
18610 ffprobe -f lavfi movie=example.mov,signalstats -show_entries frame_tags=lavfi.signalstats.YMAX,lavfi.signalstats.YMIN
18612 • Playback video while highlighting pixels that are outside of
18614 broadcast range in red.
18615 ffplay example.mov -vf signalstats="out=brng:color=red"
18617 • Playback video with signalstats metadata drawn over the frame.
18619 ffplay example.mov -vf signalstats=stat=brng+vrep+tout,drawtext=fontfile=FreeSerif.ttf:textfile=signalstat_drawtext.txt
18621 The contents of signalstat_drawtext.txt used in the command are:
18623 time %{pts:hms}
18625 Y (%{metadata:lavfi.signalstats.YMIN}-%{metadata:lavfi.signalstats.YMAX})
18626 U (%{metadata:lavfi.signalstats.UMIN}-%{metadata:lavfi.signalstats.UMAX})
18627 V (%{metadata:lavfi.signalstats.VMIN}-%{metadata:lavfi.signalstats.VMAX})
18628 saturation maximum: %{metadata:lavfi.signalstats.SATMAX}
18629 signature
18631 Calculates the MPEG-7 Video Signature. The filter can handle more than
18632 one input. In this case the matching between the inputs can be
18633 calculated additionally. The filter always passes through the first
18634 input. The signature of each stream can be written into a file.
18635 It accepts the following options:
18637 detectmode
18639 Enable or disable the matching process.
18640 Available values are:
18642 off Disable the calculation of a matching (default).
18644 full
18646 Calculate the matching for the whole video and output whether
18647 the whole video matches or only parts.
18648 fast
18650 Calculate only until a matching is found or the video ends.
18651 Should be faster in some cases.
18652 nb_inputs
18654 Set the number of inputs. The option value must be a non negative
18655 integer. Default value is 1.
18656 filename
18658 Set the path to which the output is written. If there is more than
18659 one input, the path must be a prototype, i.e. must contain %d or
18660 %0nd (where n is a positive integer), that will be replaced with
18661 the input number. If no filename is specified, no output will be
18662 written. This is the default.
18663 format
18665 Choose the output format.
18666 Available values are:
18668 binary
18670 Use the specified binary representation (default).
18671 xml Use the specified xml representation.
18673 th_d
18675 Set threshold to detect one word as similar. The option value must
18676 be an integer greater than zero. The default value is 9000.
18677 th_dc
18679 Set threshold to detect all words as similar. The option value must
18680 be an integer greater than zero. The default value is 60000.
18681 th_xh
18683 Set threshold to detect frames as similar. The option value must be
18684 an integer greater than zero. The default value is 116.
18685 th_di
18687 Set the minimum length of a sequence in frames to recognize it as
18688 matching sequence. The option value must be a non negative integer
18689 value. The default value is 0.
18690 th_it
18692 Set the minimum relation, that matching frames to all frames must
18693 have. The option value must be a double value between 0 and 1. The
18694 default value is 0.5.
18695 Examples
18697 • To calculate the signature of an input video and store it in
18699 signature.bin:
18700 ffmpeg -i input.mkv -vf signature=filename=signature.bin -map 0:v -f null -
18702 • To detect whether two videos match and store the signatures in XML
18704 format in signature0.xml and signature1.xml:
18705 ffmpeg -i input1.mkv -i input2.mkv -filter_complex "[0:v][1:v] signature=nb_inputs=2:detectmode=full:format=xml:filename=signature%d.xml" -map :v -f null -
18707 siti
18709 Calculate Spatial Info (SI) and Temporal Info (TI) scores for a video,
18710 as defined in ITU-T P.910: Subjective video quality assessment methods
18711 for multimedia applications. Available PDF at
18712 <https://www.itu.int/rec/T-REC-P.910-199909-S/en >.
18713 It accepts the following option:
18715 print_summary
18717 If set to 1, Summary statistics will be printed to the console.
18718 Default 0.
18719 Examples
18721 • To calculate SI/TI metrics and print summary:
18723 ffmpeg -i input.mp4 -vf siti=print_summary=1 -f null -
18725 smartblur
18727 Blur the input video without impacting the outlines.
18728 It accepts the following options:
18730 luma_radius, lr
18732 Set the luma radius. The option value must be a float number in the
18733 range [0.1,5.0] that specifies the variance of the gaussian filter
18734 used to blur the image (slower if larger). Default value is 1.0.
18735 luma_strength, ls
18737 Set the luma strength. The option value must be a float number in
18738 the range [-1.0,1.0] that configures the blurring. A value included
18739 in [0.0,1.0] will blur the image whereas a value included in
18740 [-1.0,0.0] will sharpen the image. Default value is 1.0.
18741 luma_threshold, lt
18743 Set the luma threshold used as a coefficient to determine whether a
18744 pixel should be blurred or not. The option value must be an integer
18745 in the range [-30,30]. A value of 0 will filter all the image, a
18746 value included in [0,30] will filter flat areas and a value
18747 included in [-30,0] will filter edges. Default value is 0.
18748 chroma_radius, cr
18750 Set the chroma radius. The option value must be a float number in
18751 the range [0.1,5.0] that specifies the variance of the gaussian
18752 filter used to blur the image (slower if larger). Default value is
18753 luma_radius.
18754 chroma_strength, cs
18756 Set the chroma strength. The option value must be a float number in
18757 the range [-1.0,1.0] that configures the blurring. A value included
18758 in [0.0,1.0] will blur the image whereas a value included in
18759 [-1.0,0.0] will sharpen the image. Default value is luma_strength.
18760 chroma_threshold, ct
18762 Set the chroma threshold used as a coefficient to determine whether
18763 a pixel should be blurred or not. The option value must be an
18764 integer in the range [-30,30]. A value of 0 will filter all the
18765 image, a value included in [0,30] will filter flat areas and a
18766 value included in [-30,0] will filter edges. Default value is
18767 luma_threshold.
18768 If a chroma option is not explicitly set, the corresponding luma value
18770 is set.
18771 sobel
18773 Apply sobel operator to input video stream.
18774 The filter accepts the following option:
18776 planes
18778 Set which planes will be processed, unprocessed planes will be
18779 copied. By default value 0xf, all planes will be processed.
18780 scale
18782 Set value which will be multiplied with filtered result.
18783 delta
18785 Set value which will be added to filtered result.
18786 Commands
18788 This filter supports the all above options as commands.
18790 spp
18792 Apply a simple postprocessing filter that compresses and decompresses
18793 the image at several (or - in the case of quality level 6 - all) shifts
18794 and average the results.
18795 The filter accepts the following options:
18797 quality
18799 Set quality. This option defines the number of levels for
18800 averaging. It accepts an integer in the range 0-6. If set to 0, the
18801 filter will have no effect. A value of 6 means the higher quality.
18802 For each increment of that value the speed drops by a factor of
18803 approximately 2. Default value is 3.
18804 qp Force a constant quantization parameter. If not set, the filter
18806 will use the QP from the video stream (if available).
18807 mode
18809 Set thresholding mode. Available modes are:
18810 hard
18812 Set hard thresholding (default).
18813 soft
18815 Set soft thresholding (better de-ringing effect, but likely
18816 blurrier).
18817 use_bframe_qp
18819 Enable the use of the QP from the B-Frames if set to 1. Using this
18820 option may cause flicker since the B-Frames have often larger QP.
18821 Default is 0 (not enabled).
18822 Commands
18824 This filter supports the following commands:
18826 quality, level
18828 Set quality level. The value "max" can be used to set the maximum
18829 level, currently 6.
18830 sr
18832 Scale the input by applying one of the super-resolution methods based
18833 on convolutional neural networks. Supported models:
18834 • Super-Resolution Convolutional Neural Network model (SRCNN). See
18836 <https://arxiv.org/abs/1501.00092>.
18837 • Efficient Sub-Pixel Convolutional Neural Network model (ESPCN).
18839 See <https://arxiv.org/abs/1609.05158>.
18840 Training scripts as well as scripts for model file (.pb) saving can be
18842 found at <https://github.com/XueweiMeng/sr/tree/sr_dnn_native>.
18843 Original repository is at
18844 <https://github.com/HighVoltageRocknRoll/sr.git>.
18845 Native model files (.model) can be generated from TensorFlow model
18847 files (.pb) by using tools/python/convert.py
18848 The filter accepts the following options:
18850 dnn_backend
18852 Specify which DNN backend to use for model loading and execution.
18853 This option accepts the following values:
18854 native
18856 Native implementation of DNN loading and execution.
18857 tensorflow
18859 TensorFlow backend. To enable this backend you need to install
18860 the TensorFlow for C library (see
18861 <https://www.tensorflow.org/install/lang_c>) and configure
18862 FFmpeg with "--enable-libtensorflow"
18863 Default value is native.
18865 model
18867 Set path to model file specifying network architecture and its
18868 parameters. Note that different backends use different file
18869 formats. TensorFlow backend can load files for both formats, while
18870 native backend can load files for only its format.
18871 scale_factor
18873 Set scale factor for SRCNN model. Allowed values are 2, 3 and 4.
18874 Default value is 2. Scale factor is necessary for SRCNN model,
18875 because it accepts input upscaled using bicubic upscaling with
18876 proper scale factor.
18877 To get full functionality (such as async execution), please use the
18879 dnn_processing filter.
18880 ssim
18882 Obtain the SSIM (Structural SImilarity Metric) between two input
18883 videos.
18884 This filter takes in input two input videos, the first input is
18886 considered the "main" source and is passed unchanged to the output. The
18887 second input is used as a "reference" video for computing the SSIM.
18888 Both video inputs must have the same resolution and pixel format for
18890 this filter to work correctly. Also it assumes that both inputs have
18891 the same number of frames, which are compared one by one.
18892 The filter stores the calculated SSIM of each frame.
18894 The description of the accepted parameters follows.
18896 stats_file, f
18898 If specified the filter will use the named file to save the SSIM of
18899 each individual frame. When filename equals "-" the data is sent to
18900 standard output.
18901 The file printed if stats_file is selected, contains a sequence of
18903 key/value pairs of the form key:value for each compared couple of
18904 frames.
18905 A description of each shown parameter follows:
18907 n sequential number of the input frame, starting from 1
18909 Y, U, V, R, G, B
18911 SSIM of the compared frames for the component specified by the
18912 suffix.
18913 All SSIM of the compared frames for the whole frame.
18915 dB Same as above but in dB representation.
18917 This filter also supports the framesync options.
18919 Examples
18921 • For example:
18923 movie=ref_movie.mpg, setpts=PTS-STARTPTS [main];
18925 [main][ref] ssim="stats_file=stats.log" [out]
18926 On this example the input file being processed is compared with the
18928 reference file ref_movie.mpg. The SSIM of each individual frame is
18929 stored in stats.log.
18930 • Another example with both psnr and ssim at same time:
18932 ffmpeg -i main.mpg -i ref.mpg -lavfi "ssim;[0:v][1:v]psnr" -f null -
18934 • Another example with different containers:
18936 ffmpeg -i main.mpg -i ref.mkv -lavfi "[0:v]settb=AVTB,setpts=PTS-STARTPTS[main];[1:v]settb=AVTB,setpts=PTS-STARTPTS[ref];[main][ref]ssim" -f null -
18938 stereo3d
18940 Convert between different stereoscopic image formats.
18941 The filters accept the following options:
18943 in Set stereoscopic image format of input.
18945 Available values for input image formats are:
18947 sbsl
18949 side by side parallel (left eye left, right eye right)
18950 sbsr
18952 side by side crosseye (right eye left, left eye right)
18953 sbs2l
18955 side by side parallel with half width resolution (left eye
18956 left, right eye right)
18957 sbs2r
18959 side by side crosseye with half width resolution (right eye
18960 left, left eye right)
18961 abl
18963 tbl above-below (left eye above, right eye below)
18964 abr
18966 tbr above-below (right eye above, left eye below)
18967 ab2l
18969 tb2l
18970 above-below with half height resolution (left eye above, right
18971 eye below)
18972 ab2r
18974 tb2r
18975 above-below with half height resolution (right eye above, left
18976 eye below)
18977 al alternating frames (left eye first, right eye second)
18979 ar alternating frames (right eye first, left eye second)
18981 irl interleaved rows (left eye has top row, right eye starts on
18983 next row)
18984 irr interleaved rows (right eye has top row, left eye starts on
18986 next row)
18987 icl interleaved columns, left eye first
18989 icr interleaved columns, right eye first
18991 Default value is sbsl.
18993 out Set stereoscopic image format of output.
18995 sbsl
18997 side by side parallel (left eye left, right eye right)
18998 sbsr
19000 side by side crosseye (right eye left, left eye right)
19001 sbs2l
19003 side by side parallel with half width resolution (left eye
19004 left, right eye right)
19005 sbs2r
19007 side by side crosseye with half width resolution (right eye
19008 left, left eye right)
19009 abl
19011 tbl above-below (left eye above, right eye below)
19012 abr
19014 tbr above-below (right eye above, left eye below)
19015 ab2l
19017 tb2l
19018 above-below with half height resolution (left eye above, right
19019 eye below)
19020 ab2r
19022 tb2r
19023 above-below with half height resolution (right eye above, left
19024 eye below)
19025 al alternating frames (left eye first, right eye second)
19027 ar alternating frames (right eye first, left eye second)
19029 irl interleaved rows (left eye has top row, right eye starts on
19031 next row)
19032 irr interleaved rows (right eye has top row, left eye starts on
19034 next row)
19035 arbg
19037 anaglyph red/blue gray (red filter on left eye, blue filter on
19038 right eye)
19039 argg
19041 anaglyph red/green gray (red filter on left eye, green filter
19042 on right eye)
19043 arcg
19045 anaglyph red/cyan gray (red filter on left eye, cyan filter on
19046 right eye)
19047 arch
19049 anaglyph red/cyan half colored (red filter on left eye, cyan
19050 filter on right eye)
19051 arcc
19053 anaglyph red/cyan color (red filter on left eye, cyan filter on
19054 right eye)
19055 arcd
19057 anaglyph red/cyan color optimized with the least squares
19058 projection of dubois (red filter on left eye, cyan filter on
19059 right eye)
19060 agmg
19062 anaglyph green/magenta gray (green filter on left eye, magenta
19063 filter on right eye)
19064 agmh
19066 anaglyph green/magenta half colored (green filter on left eye,
19067 magenta filter on right eye)
19068 agmc
19070 anaglyph green/magenta colored (green filter on left eye,
19071 magenta filter on right eye)
19072 agmd
19074 anaglyph green/magenta color optimized with the least squares
19075 projection of dubois (green filter on left eye, magenta filter
19076 on right eye)
19077 aybg
19079 anaglyph yellow/blue gray (yellow filter on left eye, blue
19080 filter on right eye)
19081 aybh
19083 anaglyph yellow/blue half colored (yellow filter on left eye,
19084 blue filter on right eye)
19085 aybc
19087 anaglyph yellow/blue colored (yellow filter on left eye, blue
19088 filter on right eye)
19089 aybd
19091 anaglyph yellow/blue color optimized with the least squares
19092 projection of dubois (yellow filter on left eye, blue filter on
19093 right eye)
19094 ml mono output (left eye only)
19096 mr mono output (right eye only)
19098 chl checkerboard, left eye first
19100 chr checkerboard, right eye first
19102 icl interleaved columns, left eye first
19104 icr interleaved columns, right eye first
19106 hdmi
19108 HDMI frame pack
19109 Default value is arcd.
19111 Examples
19113 • Convert input video from side by side parallel to anaglyph
19115 yellow/blue dubois:
19116 stereo3d=sbsl:aybd
19118 • Convert input video from above below (left eye above, right eye
19120 below) to side by side crosseye.
19121 stereo3d=abl:sbsr
19123 streamselect, astreamselect
19125 Select video or audio streams.
19126 The filter accepts the following options:
19128 inputs
19130 Set number of inputs. Default is 2.
19131 map Set input indexes to remap to outputs.
19133 Commands
19135 The "streamselect" and "astreamselect" filter supports the following
19137 commands:
19138 map Set input indexes to remap to outputs.
19140 Examples
19142 • Select first 5 seconds 1st stream and rest of time 2nd stream:
19144 sendcmd='5.0 streamselect map 1',streamselect=inputs=2:map=0
19146 • Same as above, but for audio:
19148 asendcmd='5.0 astreamselect map 1',astreamselect=inputs=2:map=0
19150 subtitles
19152 Draw subtitles on top of input video using the libass library.
19153 To enable compilation of this filter you need to configure FFmpeg with
19155 "--enable-libass". This filter also requires a build with libavcodec
19156 and libavformat to convert the passed subtitles file to ASS (Advanced
19157 Substation Alpha) subtitles format.
19158 The filter accepts the following options:
19160 filename, f
19162 Set the filename of the subtitle file to read. It must be
19163 specified.
19164 original_size
19166 Specify the size of the original video, the video for which the ASS
19167 file was composed. For the syntax of this option, check the "Video
19168 size" section in the ffmpeg-utils manual. Due to a misdesign in
19169 ASS aspect ratio arithmetic, this is necessary to correctly scale
19170 the fonts if the aspect ratio has been changed.
19171 fontsdir
19173 Set a directory path containing fonts that can be used by the
19174 filter. These fonts will be used in addition to whatever the font
19175 provider uses.
19176 alpha
19178 Process alpha channel, by default alpha channel is untouched.
19179 charenc
19181 Set subtitles input character encoding. "subtitles" filter only.
19182 Only useful if not UTF-8.
19183 stream_index, si
19185 Set subtitles stream index. "subtitles" filter only.
19186 force_style
19188 Override default style or script info parameters of the subtitles.
19189 It accepts a string containing ASS style format "KEY=VALUE" couples
19190 separated by ",".
19191 If the first key is not specified, it is assumed that the first value
19193 specifies the filename.
19194 For example, to render the file sub.srt on top of the input video, use
19196 the command:
19197 subtitles=sub.srt
19199 which is equivalent to:
19201 subtitles=filename=sub.srt
19203 To render the default subtitles stream from file video.mkv, use:
19205 subtitles=video.mkv
19207 To render the second subtitles stream from that file, use:
19209 subtitles=video.mkv:si=1
19211 To make the subtitles stream from sub.srt appear in 80% transparent
19213 blue "DejaVu Serif", use:
19214 subtitles=sub.srt:force_style='Fontname=DejaVu Serif,PrimaryColour=&HCCFF0000'
19216 super2xsai
19218 Scale the input by 2x and smooth using the Super2xSaI (Scale and
19219 Interpolate) pixel art scaling algorithm.
19220 Useful for enlarging pixel art images without reducing sharpness.
19222 swaprect
19224 Swap two rectangular objects in video.
19225 This filter accepts the following options:
19227 w Set object width.
19229 h Set object height.
19231 x1 Set 1st rect x coordinate.
19233 y1 Set 1st rect y coordinate.
19235 x2 Set 2nd rect x coordinate.
19237 y2 Set 2nd rect y coordinate.
19239 All expressions are evaluated once for each frame.
19241 The all options are expressions containing the following constants:
19243 w
19245 h The input width and height.
19246 a same as w / h
19248 sar input sample aspect ratio
19250 dar input display aspect ratio, it is the same as (w / h) * sar
19252 n The number of the input frame, starting from 0.
19254 t The timestamp expressed in seconds. It's NAN if the input timestamp
19256 is unknown.
19257 pos the position in the file of the input frame, NAN if unknown
19259 Commands
19261 This filter supports the all above options as commands.
19263 swapuv
19265 Swap U & V plane.
19266 tblend
19268 Blend successive video frames.
19269 See blend
19271 telecine
19273 Apply telecine process to the video.
19274 This filter accepts the following options:
19276 first_field
19278 top, t
19279 top field first
19280 bottom, b
19282 bottom field first The default value is "top".
19283 pattern
19285 A string of numbers representing the pulldown pattern you wish to
19286 apply. The default value is 23.
19287 Some typical patterns:
19289 NTSC output (30i):
19291 27.5p: 32222
19292 24p: 23 (classic)
19293 24p: 2332 (preferred)
19294 20p: 33
19295 18p: 334
19296 16p: 3444
19297 PAL output (25i):
19299 27.5p: 12222
19300 24p: 222222222223 ("Euro pulldown")
19301 16.67p: 33
19302 16p: 33333334
19303 thistogram
19305 Compute and draw a color distribution histogram for the input video
19306 across time.
19307 Unlike histogram video filter which only shows histogram of single
19309 input frame at certain time, this filter shows also past histograms of
19310 number of frames defined by "width" option.
19311 The computed histogram is a representation of the color component
19313 distribution in an image.
19314 The filter accepts the following options:
19316 width, w
19318 Set width of single color component output. Default value is 0.
19319 Value of 0 means width will be picked from input video. This also
19320 set number of passed histograms to keep. Allowed range is [0,
19321 8192].
19322 display_mode, d
19324 Set display mode. It accepts the following values:
19325 stack
19327 Per color component graphs are placed below each other.
19328 parade
19330 Per color component graphs are placed side by side.
19331 overlay
19333 Presents information identical to that in the "parade", except
19334 that the graphs representing color components are superimposed
19335 directly over one another.
19336 Default is "stack".
19338 levels_mode, m
19340 Set mode. Can be either "linear", or "logarithmic". Default is
19341 "linear".
19342 components, c
19344 Set what color components to display. Default is 7.
19345 bgopacity, b
19347 Set background opacity. Default is 0.9.
19348 envelope, e
19350 Show envelope. Default is disabled.
19351 ecolor, ec
19353 Set envelope color. Default is "gold".
19354 slide
19356 Set slide mode.
19357 Available values for slide is:
19359 frame
19361 Draw new frame when right border is reached.
19362 replace
19364 Replace old columns with new ones.
19365 scroll
19367 Scroll from right to left.
19368 rscroll
19370 Scroll from left to right.
19371 picture
19373 Draw single picture.
19374 Default is "replace".
19376 threshold
19378 Apply threshold effect to video stream.
19379 This filter needs four video streams to perform thresholding. First
19381 stream is stream we are filtering. Second stream is holding threshold
19382 values, third stream is holding min values, and last, fourth stream is
19383 holding max values.
19384 The filter accepts the following option:
19386 planes
19388 Set which planes will be processed, unprocessed planes will be
19389 copied. By default value 0xf, all planes will be processed.
19390 For example if first stream pixel's component value is less then
19392 threshold value of pixel component from 2nd threshold stream, third
19393 stream value will picked, otherwise fourth stream pixel component value
19394 will be picked.
19395 Using color source filter one can perform various types of
19397 thresholding:
19398 Commands
19400 This filter supports the all options as commands.
19402 Examples
19404 • Binary threshold, using gray color as threshold:
19406 ffmpeg -i 320x240.avi -f lavfi -i color=gray -f lavfi -i color=black -f lavfi -i color=white -lavfi threshold output.avi
19408 • Inverted binary threshold, using gray color as threshold:
19410 ffmpeg -i 320x240.avi -f lavfi -i color=gray -f lavfi -i color=white -f lavfi -i color=black -lavfi threshold output.avi
19412 • Truncate binary threshold, using gray color as threshold:
19414 ffmpeg -i 320x240.avi -f lavfi -i color=gray -i 320x240.avi -f lavfi -i color=gray -lavfi threshold output.avi
19416 • Threshold to zero, using gray color as threshold:
19418 ffmpeg -i 320x240.avi -f lavfi -i color=gray -f lavfi -i color=white -i 320x240.avi -lavfi threshold output.avi
19420 • Inverted threshold to zero, using gray color as threshold:
19422 ffmpeg -i 320x240.avi -f lavfi -i color=gray -i 320x240.avi -f lavfi -i color=white -lavfi threshold output.avi
19424 thumbnail
19426 Select the most representative frame in a given sequence of consecutive
19427 frames.
19428 The filter accepts the following options:
19430 n Set the frames batch size to analyze; in a set of n frames, the
19432 filter will pick one of them, and then handle the next batch of n
19433 frames until the end. Default is 100.
19434 Since the filter keeps track of the whole frames sequence, a bigger n
19436 value will result in a higher memory usage, so a high value is not
19437 recommended.
19438 Examples
19440 • Extract one picture each 50 frames:
19442 thumbnail=50
19444 • Complete example of a thumbnail creation with ffmpeg:
19446 ffmpeg -i in.avi -vf thumbnail,scale=300:200 -frames:v 1 out.png
19448 tile
19450 Tile several successive frames together.
19451 The untile filter can do the reverse.
19453 The filter accepts the following options:
19455 layout
19457 Set the grid size in the form "COLUMNSxROWS". Range is upto
19458 UINT_MAX cells. Default is "6x5".
19459 nb_frames
19461 Set the maximum number of frames to render in the given area. It
19462 must be less than or equal to wxh. The default value is 0, meaning
19463 all the area will be used.
19464 margin
19466 Set the outer border margin in pixels. Range is 0 to 1024. Default
19467 is 0.
19468 padding
19470 Set the inner border thickness (i.e. the number of pixels between
19471 frames). For more advanced padding options (such as having
19472 different values for the edges), refer to the pad video filter.
19473 Range is 0 to 1024. Default is 0.
19474 color
19476 Specify the color of the unused area. For the syntax of this
19477 option, check the "Color" section in the ffmpeg-utils manual. The
19478 default value of color is "black".
19479 overlap
19481 Set the number of frames to overlap when tiling several successive
19482 frames together. The value must be between 0 and nb_frames - 1.
19483 Default is 0.
19484 init_padding
19486 Set the number of frames to initially be empty before displaying
19487 first output frame. This controls how soon will one get first
19488 output frame. The value must be between 0 and nb_frames - 1.
19489 Default is 0.
19490 Examples
19492 • Produce 8x8 PNG tiles of all keyframes (-skip_frame nokey) in a
19494 movie:
19495 ffmpeg -skip_frame nokey -i file.avi -vf 'scale=128:72,tile=8x8' -an -vsync 0 keyframes%03d.png
19497 The -vsync 0 is necessary to prevent ffmpeg from duplicating each
19499 output frame to accommodate the originally detected frame rate.
19500 • Display 5 pictures in an area of "3x2" frames, with 7 pixels
19502 between them, and 2 pixels of initial margin, using mixed flat and
19503 named options:
19504 tile=3x2:nb_frames=5:padding=7:margin=2
19506 tinterlace
19508 Perform various types of temporal field interlacing.
19509 Frames are counted starting from 1, so the first input frame is
19511 considered odd.
19512 The filter accepts the following options:
19514 mode
19516 Specify the mode of the interlacing. This option can also be
19517 specified as a value alone. See below for a list of values for this
19518 option.
19519 Available values are:
19521 merge, 0
19523 Move odd frames into the upper field, even into the lower
19524 field, generating a double height frame at half frame rate.
19525 ------> time
19527 Input:
19528 Frame 1 Frame 2 Frame 3 Frame 4
19529 11111 22222 33333 44444
19531 11111 22222 33333 44444
19532 11111 22222 33333 44444
19533 11111 22222 33333 44444
19534 Output:
19536 11111 33333
19537 22222 44444
19538 11111 33333
19539 22222 44444
19540 11111 33333
19541 22222 44444
19542 11111 33333
19543 22222 44444
19544 drop_even, 1
19546 Only output odd frames, even frames are dropped, generating a
19547 frame with unchanged height at half frame rate.
19548 ------> time
19550 Input:
19551 Frame 1 Frame 2 Frame 3 Frame 4
19552 11111 22222 33333 44444
19554 11111 22222 33333 44444
19555 11111 22222 33333 44444
19556 11111 22222 33333 44444
19557 Output:
19559 11111 33333
19560 11111 33333
19561 11111 33333
19562 11111 33333
19563 drop_odd, 2
19565 Only output even frames, odd frames are dropped, generating a
19566 frame with unchanged height at half frame rate.
19567 ------> time
19569 Input:
19570 Frame 1 Frame 2 Frame 3 Frame 4
19571 11111 22222 33333 44444
19573 11111 22222 33333 44444
19574 11111 22222 33333 44444
19575 11111 22222 33333 44444
19576 Output:
19578 22222 44444
19579 22222 44444
19580 22222 44444
19581 22222 44444
19582 pad, 3
19584 Expand each frame to full height, but pad alternate lines with
19585 black, generating a frame with double height at the same input
19586 frame rate.
19587 ------> time
19589 Input:
19590 Frame 1 Frame 2 Frame 3 Frame 4
19591 11111 22222 33333 44444
19593 11111 22222 33333 44444
19594 11111 22222 33333 44444
19595 11111 22222 33333 44444
19596 Output:
19598 11111 ..... 33333 .....
19599 ..... 22222 ..... 44444
19600 11111 ..... 33333 .....
19601 ..... 22222 ..... 44444
19602 11111 ..... 33333 .....
19603 ..... 22222 ..... 44444
19604 11111 ..... 33333 .....
19605 ..... 22222 ..... 44444
19606 interleave_top, 4
19608 Interleave the upper field from odd frames with the lower field
19609 from even frames, generating a frame with unchanged height at
19610 half frame rate.
19611 ------> time
19613 Input:
19614 Frame 1 Frame 2 Frame 3 Frame 4
19615 11111<- 22222 33333<- 44444
19617 11111 22222<- 33333 44444<-
19618 11111<- 22222 33333<- 44444
19619 11111 22222<- 33333 44444<-
19620 Output:
19622 11111 33333
19623 22222 44444
19624 11111 33333
19625 22222 44444
19626 interleave_bottom, 5
19628 Interleave the lower field from odd frames with the upper field
19629 from even frames, generating a frame with unchanged height at
19630 half frame rate.
19631 ------> time
19633 Input:
19634 Frame 1 Frame 2 Frame 3 Frame 4
19635 11111 22222<- 33333 44444<-
19637 11111<- 22222 33333<- 44444
19638 11111 22222<- 33333 44444<-
19639 11111<- 22222 33333<- 44444
19640 Output:
19642 22222 44444
19643 11111 33333
19644 22222 44444
19645 11111 33333
19646 interlacex2, 6
19648 Double frame rate with unchanged height. Frames are inserted
19649 each containing the second temporal field from the previous
19650 input frame and the first temporal field from the next input
19651 frame. This mode relies on the top_field_first flag. Useful for
19652 interlaced video displays with no field synchronisation.
19653 ------> time
19655 Input:
19656 Frame 1 Frame 2 Frame 3 Frame 4
19657 11111 22222 33333 44444
19659 11111 22222 33333 44444
19660 11111 22222 33333 44444
19661 11111 22222 33333 44444
19662 Output:
19664 11111 22222 22222 33333 33333 44444 44444
19665 11111 11111 22222 22222 33333 33333 44444
19666 11111 22222 22222 33333 33333 44444 44444
19667 11111 11111 22222 22222 33333 33333 44444
19668 mergex2, 7
19670 Move odd frames into the upper field, even into the lower
19671 field, generating a double height frame at same frame rate.
19672 ------> time
19674 Input:
19675 Frame 1 Frame 2 Frame 3 Frame 4
19676 11111 22222 33333 44444
19678 11111 22222 33333 44444
19679 11111 22222 33333 44444
19680 11111 22222 33333 44444
19681 Output:
19683 11111 33333 33333 55555
19684 22222 22222 44444 44444
19685 11111 33333 33333 55555
19686 22222 22222 44444 44444
19687 11111 33333 33333 55555
19688 22222 22222 44444 44444
19689 11111 33333 33333 55555
19690 22222 22222 44444 44444
19691 Numeric values are deprecated but are accepted for backward
19693 compatibility reasons.
19694 Default mode is "merge".
19696 flags
19698 Specify flags influencing the filter process.
19699 Available value for flags is:
19701 low_pass_filter, vlpf
19703 Enable linear vertical low-pass filtering in the filter.
19704 Vertical low-pass filtering is required when creating an
19705 interlaced destination from a progressive source which contains
19706 high-frequency vertical detail. Filtering will reduce interlace
19707 'twitter' and Moire patterning.
19708 complex_filter, cvlpf
19710 Enable complex vertical low-pass filtering. This will slightly
19711 less reduce interlace 'twitter' and Moire patterning but better
19712 retain detail and subjective sharpness impression.
19713 bypass_il
19715 Bypass already interlaced frames, only adjust the frame rate.
19716 Vertical low-pass filtering and bypassing already interlaced frames
19718 can only be enabled for mode interleave_top and interleave_bottom.
19719 tmedian
19721 Pick median pixels from several successive input video frames.
19722 The filter accepts the following options:
19724 radius
19726 Set radius of median filter. Default is 1. Allowed range is from 1
19727 to 127.
19728 planes
19730 Set which planes to filter. Default value is 15, by which all
19731 planes are processed.
19732 percentile
19734 Set median percentile. Default value is 0.5. Default value of 0.5
19735 will pick always median values, while 0 will pick minimum values,
19736 and 1 maximum values.
19737 Commands
19739 This filter supports all above options as commands, excluding option
19741 "radius".
19742 tmidequalizer
19744 Apply Temporal Midway Video Equalization effect.
19745 Midway Video Equalization adjusts a sequence of video frames to have
19747 the same histograms, while maintaining their dynamics as much as
19748 possible. It's useful for e.g. matching exposures from a video frames
19749 sequence.
19750 This filter accepts the following option:
19752 radius
19754 Set filtering radius. Default is 5. Allowed range is from 1 to 127.
19755 sigma
19757 Set filtering sigma. Default is 0.5. This controls strength of
19758 filtering. Setting this option to 0 effectively does nothing.
19759 planes
19761 Set which planes to process. Default is 15, which is all available
19762 planes.
19763 tmix
19765 Mix successive video frames.
19766 A description of the accepted options follows.
19768 frames
19770 The number of successive frames to mix. If unspecified, it defaults
19771 to 3.
19772 weights
19774 Specify weight of each input video frame. Each weight is separated
19775 by space. If number of weights is smaller than number of frames
19776 last specified weight will be used for all remaining unset weights.
19777 scale
19779 Specify scale, if it is set it will be multiplied with sum of each
19780 weight multiplied with pixel values to give final destination pixel
19781 value. By default scale is auto scaled to sum of weights.
19782 planes
19784 Set which planes to filter. Default is all. Allowed range is from 0
19785 to 15.
19786 Examples
19788 • Average 7 successive frames:
19790 tmix=frames=7:weights="1 1 1 1 1 1 1"
19792 • Apply simple temporal convolution:
19794 tmix=frames=3:weights="-1 3 -1"
19796 • Similar as above but only showing temporal differences:
19798 tmix=frames=3:weights="-1 2 -1":scale=1
19800 Commands
19802 This filter supports the following commands:
19804 weights
19806 scale
19807 planes
19808 Syntax is same as option with same name.
19809 tonemap
19811 Tone map colors from different dynamic ranges.
19812 This filter expects data in single precision floating point, as it
19814 needs to operate on (and can output) out-of-range values. Another
19815 filter, such as zscale, is needed to convert the resulting frame to a
19816 usable format.
19817 The tonemapping algorithms implemented only work on linear light, so
19819 input data should be linearized beforehand (and possibly correctly
19820 tagged).
19821 ffmpeg -i INPUT -vf zscale=transfer=linear,tonemap=clip,zscale=transfer=bt709,format=yuv420p OUTPUT
19823 Options
19825 The filter accepts the following options.
19827 tonemap
19829 Set the tone map algorithm to use.
19830 Possible values are:
19832 none
19834 Do not apply any tone map, only desaturate overbright pixels.
19835 clip
19837 Hard-clip any out-of-range values. Use it for perfect color
19838 accuracy for in-range values, while distorting out-of-range
19839 values.
19840 linear
19842 Stretch the entire reference gamut to a linear multiple of the
19843 display.
19844 gamma
19846 Fit a logarithmic transfer between the tone curves.
19847 reinhard
19849 Preserve overall image brightness with a simple curve, using
19850 nonlinear contrast, which results in flattening details and
19851 degrading color accuracy.
19852 hable
19854 Preserve both dark and bright details better than reinhard, at
19855 the cost of slightly darkening everything. Use it when detail
19856 preservation is more important than color and brightness
19857 accuracy.
19858 mobius
19860 Smoothly map out-of-range values, while retaining contrast and
19861 colors for in-range material as much as possible. Use it when
19862 color accuracy is more important than detail preservation.
19863 Default is none.
19865 param
19867 Tune the tone mapping algorithm.
19868 This affects the following algorithms:
19870 none
19872 Ignored.
19873 linear
19875 Specifies the scale factor to use while stretching. Default to
19876 1.0.
19877 gamma
19879 Specifies the exponent of the function. Default to 1.8.
19880 clip
19882 Specify an extra linear coefficient to multiply into the signal
19883 before clipping. Default to 1.0.
19884 reinhard
19886 Specify the local contrast coefficient at the display peak.
19887 Default to 0.5, which means that in-gamut values will be about
19888 half as bright as when clipping.
19889 hable
19891 Ignored.
19892 mobius
19894 Specify the transition point from linear to mobius transform.
19895 Every value below this point is guaranteed to be mapped 1:1.
19896 The higher the value, the more accurate the result will be, at
19897 the cost of losing bright details. Default to 0.3, which due
19898 to the steep initial slope still preserves in-range colors
19899 fairly accurately.
19900 desat
19902 Apply desaturation for highlights that exceed this level of
19903 brightness. The higher the parameter, the more color information
19904 will be preserved. This setting helps prevent unnaturally blown-out
19905 colors for super-highlights, by (smoothly) turning into white
19906 instead. This makes images feel more natural, at the cost of
19907 reducing information about out-of-range colors.
19908 The default of 2.0 is somewhat conservative and will mostly just
19910 apply to skies or directly sunlit surfaces. A setting of 0.0
19911 disables this option.
19912 This option works only if the input frame has a supported color
19914 tag.
19915 peak
19917 Override signal/nominal/reference peak with this value. Useful when
19918 the embedded peak information in display metadata is not reliable
19919 or when tone mapping from a lower range to a higher range.
19920 tpad
19922 Temporarily pad video frames.
19923 The filter accepts the following options:
19925 start
19927 Specify number of delay frames before input video stream. Default
19928 is 0.
19929 stop
19931 Specify number of padding frames after input video stream. Set to
19932 -1 to pad indefinitely. Default is 0.
19933 start_mode
19935 Set kind of frames added to beginning of stream. Can be either add
19936 or clone. With add frames of solid-color are added. With clone
19937 frames are clones of first frame. Default is add.
19938 stop_mode
19940 Set kind of frames added to end of stream. Can be either add or
19941 clone. With add frames of solid-color are added. With clone
19942 frames are clones of last frame. Default is add.
19943 start_duration, stop_duration
19945 Specify the duration of the start/stop delay. See the Time duration
19946 section in the ffmpeg-utils(1) manual for the accepted syntax.
19947 These options override start and stop. Default is 0.
19948 color
19950 Specify the color of the padded area. For the syntax of this
19951 option, check the "Color" section in the ffmpeg-utils manual.
19952 The default value of color is "black".
19954 transpose
19956 Transpose rows with columns in the input video and optionally flip it.
19957 It accepts the following parameters:
19959 dir Specify the transposition direction.
19961 Can assume the following values:
19963 0, 4, cclock_flip
19965 Rotate by 90 degrees counterclockwise and vertically flip
19966 (default), that is:
19967 L.R L.l
19969 . . -> . .
19970 l.r R.r
19971 1, 5, clock
19973 Rotate by 90 degrees clockwise, that is:
19974 L.R l.L
19976 . . -> . .
19977 l.r r.R
19978 2, 6, cclock
19980 Rotate by 90 degrees counterclockwise, that is:
19981 L.R R.r
19983 . . -> . .
19984 l.r L.l
19985 3, 7, clock_flip
19987 Rotate by 90 degrees clockwise and vertically flip, that is:
19988 L.R r.R
19990 . . -> . .
19991 l.r l.L
19992 For values between 4-7, the transposition is only done if the input
19994 video geometry is portrait and not landscape. These values are
19995 deprecated, the "passthrough" option should be used instead.
19996 Numerical values are deprecated, and should be dropped in favor of
19998 symbolic constants.
19999 passthrough
20001 Do not apply the transposition if the input geometry matches the
20002 one specified by the specified value. It accepts the following
20003 values:
20004 none
20006 Always apply transposition.
20007 portrait
20009 Preserve portrait geometry (when height >= width).
20010 landscape
20012 Preserve landscape geometry (when width >= height).
20013 Default value is "none".
20015 For example to rotate by 90 degrees clockwise and preserve portrait
20017 layout:
20018 transpose=dir=1:passthrough=portrait
20020 The command above can also be specified as:
20022 transpose=1:portrait
20024 transpose_npp
20026 Transpose rows with columns in the input video and optionally flip it.
20027 For more in depth examples see the transpose video filter, which shares
20028 mostly the same options.
20029 It accepts the following parameters:
20031 dir Specify the transposition direction.
20033 Can assume the following values:
20035 cclock_flip
20037 Rotate by 90 degrees counterclockwise and vertically flip.
20038 (default)
20039 clock
20041 Rotate by 90 degrees clockwise.
20042 cclock
20044 Rotate by 90 degrees counterclockwise.
20045 clock_flip
20047 Rotate by 90 degrees clockwise and vertically flip.
20048 passthrough
20050 Do not apply the transposition if the input geometry matches the
20051 one specified by the specified value. It accepts the following
20052 values:
20053 none
20055 Always apply transposition. (default)
20056 portrait
20058 Preserve portrait geometry (when height >= width).
20059 landscape
20061 Preserve landscape geometry (when width >= height).
20062 trim
20064 Trim the input so that the output contains one continuous subpart of
20065 the input.
20066 It accepts the following parameters:
20068 start
20070 Specify the time of the start of the kept section, i.e. the frame
20071 with the timestamp start will be the first frame in the output.
20072 end Specify the time of the first frame that will be dropped, i.e. the
20074 frame immediately preceding the one with the timestamp end will be
20075 the last frame in the output.
20076 start_pts
20078 This is the same as start, except this option sets the start
20079 timestamp in timebase units instead of seconds.
20080 end_pts
20082 This is the same as end, except this option sets the end timestamp
20083 in timebase units instead of seconds.
20084 duration
20086 The maximum duration of the output in seconds.
20087 start_frame
20089 The number of the first frame that should be passed to the output.
20090 end_frame
20092 The number of the first frame that should be dropped.
20093 start, end, and duration are expressed as time duration specifications;
20095 see the Time duration section in the ffmpeg-utils(1) manual for the
20096 accepted syntax.
20097 Note that the first two sets of the start/end options and the duration
20099 option look at the frame timestamp, while the _frame variants simply
20100 count the frames that pass through the filter. Also note that this
20101 filter does not modify the timestamps. If you wish for the output
20102 timestamps to start at zero, insert a setpts filter after the trim
20103 filter.
20104 If multiple start or end options are set, this filter tries to be
20106 greedy and keep all the frames that match at least one of the specified
20107 constraints. To keep only the part that matches all the constraints at
20108 once, chain multiple trim filters.
20109 The defaults are such that all the input is kept. So it is possible to
20111 set e.g. just the end values to keep everything before the specified
20112 time.
20113 Examples:
20115 • Drop everything except the second minute of input:
20117 ffmpeg -i INPUT -vf trim=60:120
20119 • Keep only the first second:
20121 ffmpeg -i INPUT -vf trim=duration=1
20123 unpremultiply
20125 Apply alpha unpremultiply effect to input video stream using first
20126 plane of second stream as alpha.
20127 Both streams must have same dimensions and same pixel format.
20129 The filter accepts the following option:
20131 planes
20133 Set which planes will be processed, unprocessed planes will be
20134 copied. By default value 0xf, all planes will be processed.
20135 If the format has 1 or 2 components, then luma is bit 0. If the
20137 format has 3 or 4 components: for RGB formats bit 0 is green, bit 1
20138 is blue and bit 2 is red; for YUV formats bit 0 is luma, bit 1 is
20139 chroma-U and bit 2 is chroma-V. If present, the alpha channel is
20140 always the last bit.
20141 inplace
20143 Do not require 2nd input for processing, instead use alpha plane
20144 from input stream.
20145 unsharp
20147 Sharpen or blur the input video.
20148 It accepts the following parameters:
20150 luma_msize_x, lx
20152 Set the luma matrix horizontal size. It must be an odd integer
20153 between 3 and 23. The default value is 5.
20154 luma_msize_y, ly
20156 Set the luma matrix vertical size. It must be an odd integer
20157 between 3 and 23. The default value is 5.
20158 luma_amount, la
20160 Set the luma effect strength. It must be a floating point number,
20161 reasonable values lay between -1.5 and 1.5.
20162 Negative values will blur the input video, while positive values
20164 will sharpen it, a value of zero will disable the effect.
20165 Default value is 1.0.
20167 chroma_msize_x, cx
20169 Set the chroma matrix horizontal size. It must be an odd integer
20170 between 3 and 23. The default value is 5.
20171 chroma_msize_y, cy
20173 Set the chroma matrix vertical size. It must be an odd integer
20174 between 3 and 23. The default value is 5.
20175 chroma_amount, ca
20177 Set the chroma effect strength. It must be a floating point number,
20178 reasonable values lay between -1.5 and 1.5.
20179 Negative values will blur the input video, while positive values
20181 will sharpen it, a value of zero will disable the effect.
20182 Default value is 0.0.
20184 alpha_msize_x, ax
20186 Set the alpha matrix horizontal size. It must be an odd integer
20187 between 3 and 23. The default value is 5.
20188 alpha_msize_y, ay
20190 Set the alpha matrix vertical size. It must be an odd integer
20191 between 3 and 23. The default value is 5.
20192 alpha_amount, aa
20194 Set the alpha effect strength. It must be a floating point number,
20195 reasonable values lay between -1.5 and 1.5.
20196 Negative values will blur the input video, while positive values
20198 will sharpen it, a value of zero will disable the effect.
20199 Default value is 0.0.
20201 All parameters are optional and default to the equivalent of the string
20203 '5:5:1.0:5:5:0.0'.
20204 Examples
20206 • Apply strong luma sharpen effect:
20208 unsharp=luma_msize_x=7:luma_msize_y=7:luma_amount=2.5
20210 • Apply a strong blur of both luma and chroma parameters:
20212 unsharp=7:7:-2:7:7:-2
20214 untile
20216 Decompose a video made of tiled images into the individual images.
20217 The frame rate of the output video is the frame rate of the input video
20219 multiplied by the number of tiles.
20220 This filter does the reverse of tile.
20222 The filter accepts the following options:
20224 layout
20226 Set the grid size (i.e. the number of lines and columns). For the
20227 syntax of this option, check the "Video size" section in the
20228 ffmpeg-utils manual.
20229 Examples
20231 • Produce a 1-second video from a still image file made of 25 frames
20233 stacked vertically, like an analogic film reel:
20234 ffmpeg -r 1 -i image.jpg -vf untile=1x25 movie.mkv
20236 uspp
20238 Apply ultra slow/simple postprocessing filter that compresses and
20239 decompresses the image at several (or - in the case of quality level 8
20240 - all) shifts and average the results.
20241 The way this differs from the behavior of spp is that uspp actually
20243 encodes & decodes each case with libavcodec Snow, whereas spp uses a
20244 simplified intra only 8x8 DCT similar to MJPEG.
20245 This filter is only available in ffmpeg version 4.4 or earlier.
20247 The filter accepts the following options:
20249 quality
20251 Set quality. This option defines the number of levels for
20252 averaging. It accepts an integer in the range 0-8. If set to 0, the
20253 filter will have no effect. A value of 8 means the higher quality.
20254 For each increment of that value the speed drops by a factor of
20255 approximately 2. Default value is 3.
20256 qp Force a constant quantization parameter. If not set, the filter
20258 will use the QP from the video stream (if available).
20259 v360
20261 Convert 360 videos between various formats.
20262 The filter accepts the following options:
20264 input
20266 output
20267 Set format of the input/output video.
20268 Available formats:
20270 e
20272 equirect
20273 Equirectangular projection.
20274 c3x2
20276 c6x1
20277 c1x6
20278 Cubemap with 3x2/6x1/1x6 layout.
20279 Format specific options:
20281 in_pad
20283 out_pad
20284 Set padding proportion for the input/output cubemap. Values
20285 in decimals.
20286 Example values:
20288 0 No padding.
20290 0.01
20292 1% of face is padding. For example, with 1920x1280
20293 resolution face size would be 640x640 and padding would
20294 be 3 pixels from each side. (640 * 0.01 = 6 pixels)
20295 Default value is @samp{0}. Maximum value is @samp{0.1}.
20297 fin_pad
20299 fout_pad
20300 Set fixed padding for the input/output cubemap. Values in
20301 pixels.
20302 Default value is @samp{0}. If greater than zero it
20304 overrides other padding options.
20305 in_forder
20307 out_forder
20308 Set order of faces for the input/output cubemap. Choose one
20309 direction for each position.
20310 Designation of directions:
20312 r right
20314 l left
20316 u up
20318 d down
20320 f forward
20322 b back
20324 Default value is @samp{rludfb}.
20326 in_frot
20328 out_frot
20329 Set rotation of faces for the input/output cubemap. Choose
20330 one angle for each position.
20331 Designation of angles:
20333 0 0 degrees clockwise
20335 1 90 degrees clockwise
20337 2 180 degrees clockwise
20339 3 270 degrees clockwise
20341 Default value is @samp{000000}.
20343 eac Equi-Angular Cubemap.
20345 flat
20347 gnomonic
20348 rectilinear
20349 Regular video.
20350 Format specific options:
20352 h_fov
20354 v_fov
20355 d_fov
20356 Set output horizontal/vertical/diagonal field of view.
20357 Values in degrees.
20358 If diagonal field of view is set it overrides horizontal
20360 and vertical field of view.
20361 ih_fov
20363 iv_fov
20364 id_fov
20365 Set input horizontal/vertical/diagonal field of view.
20366 Values in degrees.
20367 If diagonal field of view is set it overrides horizontal
20369 and vertical field of view.
20370 dfisheye
20372 Dual fisheye.
20373 Format specific options:
20375 h_fov
20377 v_fov
20378 d_fov
20379 Set output horizontal/vertical/diagonal field of view.
20380 Values in degrees.
20381 If diagonal field of view is set it overrides horizontal
20383 and vertical field of view.
20384 ih_fov
20386 iv_fov
20387 id_fov
20388 Set input horizontal/vertical/diagonal field of view.
20389 Values in degrees.
20390 If diagonal field of view is set it overrides horizontal
20392 and vertical field of view.
20393 barrel
20395 fb
20396 barrelsplit
20397 Facebook's 360 formats.
20398 sg Stereographic format.
20400 Format specific options:
20402 h_fov
20404 v_fov
20405 d_fov
20406 Set output horizontal/vertical/diagonal field of view.
20407 Values in degrees.
20408 If diagonal field of view is set it overrides horizontal
20410 and vertical field of view.
20411 ih_fov
20413 iv_fov
20414 id_fov
20415 Set input horizontal/vertical/diagonal field of view.
20416 Values in degrees.
20417 If diagonal field of view is set it overrides horizontal
20419 and vertical field of view.
20420 mercator
20422 Mercator format.
20423 ball
20425 Ball format, gives significant distortion toward the back.
20426 hammer
20428 Hammer-Aitoff map projection format.
20429 sinusoidal
20431 Sinusoidal map projection format.
20432 fisheye
20434 Fisheye projection.
20435 Format specific options:
20437 h_fov
20439 v_fov
20440 d_fov
20441 Set output horizontal/vertical/diagonal field of view.
20442 Values in degrees.
20443 If diagonal field of view is set it overrides horizontal
20445 and vertical field of view.
20446 ih_fov
20448 iv_fov
20449 id_fov
20450 Set input horizontal/vertical/diagonal field of view.
20451 Values in degrees.
20452 If diagonal field of view is set it overrides horizontal
20454 and vertical field of view.
20455 pannini
20457 Pannini projection.
20458 Format specific options:
20460 h_fov
20462 Set output pannini parameter.
20463 ih_fov
20465 Set input pannini parameter.
20466 cylindrical
20468 Cylindrical projection.
20469 Format specific options:
20471 h_fov
20473 v_fov
20474 d_fov
20475 Set output horizontal/vertical/diagonal field of view.
20476 Values in degrees.
20477 If diagonal field of view is set it overrides horizontal
20479 and vertical field of view.
20480 ih_fov
20482 iv_fov
20483 id_fov
20484 Set input horizontal/vertical/diagonal field of view.
20485 Values in degrees.
20486 If diagonal field of view is set it overrides horizontal
20488 and vertical field of view.
20489 perspective
20491 Perspective projection. (output only)
20492 Format specific options:
20494 v_fov
20496 Set perspective parameter.
20497 tetrahedron
20499 Tetrahedron projection.
20500 tsp Truncated square pyramid projection.
20502 he
20504 hequirect
20505 Half equirectangular projection.
20506 equisolid
20508 Equisolid format.
20509 Format specific options:
20511 h_fov
20513 v_fov
20514 d_fov
20515 Set output horizontal/vertical/diagonal field of view.
20516 Values in degrees.
20517 If diagonal field of view is set it overrides horizontal
20519 and vertical field of view.
20520 ih_fov
20522 iv_fov
20523 id_fov
20524 Set input horizontal/vertical/diagonal field of view.
20525 Values in degrees.
20526 If diagonal field of view is set it overrides horizontal
20528 and vertical field of view.
20529 og Orthographic format.
20531 Format specific options:
20533 h_fov
20535 v_fov
20536 d_fov
20537 Set output horizontal/vertical/diagonal field of view.
20538 Values in degrees.
20539 If diagonal field of view is set it overrides horizontal
20541 and vertical field of view.
20542 ih_fov
20544 iv_fov
20545 id_fov
20546 Set input horizontal/vertical/diagonal field of view.
20547 Values in degrees.
20548 If diagonal field of view is set it overrides horizontal
20550 and vertical field of view.
20551 octahedron
20553 Octahedron projection.
20554 cylindricalea
20556 Cylindrical Equal Area projection.
20557 interp
20559 Set interpolation method.Note: more complex interpolation methods
20560 require much more memory to run.
20561 Available methods:
20563 near
20565 nearest
20566 Nearest neighbour.
20567 line
20569 linear
20570 Bilinear interpolation.
20571 lagrange9
20573 Lagrange9 interpolation.
20574 cube
20576 cubic
20577 Bicubic interpolation.
20578 lanc
20580 lanczos
20581 Lanczos interpolation.
20582 sp16
20584 spline16
20585 Spline16 interpolation.
20586 gauss
20588 gaussian
20589 Gaussian interpolation.
20590 mitchell
20592 Mitchell interpolation.
20593 Default value is @samp{line}.
20595 w
20597 h Set the output video resolution.
20598 Default resolution depends on formats.
20600 in_stereo
20602 out_stereo
20603 Set the input/output stereo format.
20604 2d 2D mono
20606 sbs Side by side
20608 tb Top bottom
20610 Default value is @samp{2d} for input and output format.
20612 yaw
20614 pitch
20615 roll
20616 Set rotation for the output video. Values in degrees.
20617 rorder
20619 Set rotation order for the output video. Choose one item for each
20620 position.
20621 y, Y
20623 yaw
20624 p, P
20626 pitch
20627 r, R
20629 roll
20630 Default value is @samp{ypr}.
20632 h_flip
20634 v_flip
20635 d_flip
20636 Flip the output video horizontally(swaps
20637 left-right)/vertically(swaps up-down)/in-depth(swaps back-forward).
20638 Boolean values.
20639 ih_flip
20641 iv_flip
20642 Set if input video is flipped horizontally/vertically. Boolean
20643 values.
20644 in_trans
20646 Set if input video is transposed. Boolean value, by default
20647 disabled.
20648 out_trans
20650 Set if output video needs to be transposed. Boolean value, by
20651 default disabled.
20652 h_offset
20654 v_offset
20655 Set output horizontal/vertical off-axis offset. Default is set to
20656 0. Allowed range is from -1 to 1.
20657 alpha_mask
20659 Build mask in alpha plane for all unmapped pixels by marking them
20660 fully transparent. Boolean value, by default disabled.
20661 reset_rot
20663 Reset rotation of output video. Boolean value, by default disabled.
20664 Examples
20666 • Convert equirectangular video to cubemap with 3x2 layout and 1%
20668 padding using bicubic interpolation:
20669 ffmpeg -i input.mkv -vf v360=e:c3x2:cubic:out_pad=0.01 output.mkv
20671 • Extract back view of Equi-Angular Cubemap:
20673 ffmpeg -i input.mkv -vf v360=eac:flat:yaw=180 output.mkv
20675 • Convert transposed and horizontally flipped Equi-Angular Cubemap in
20677 side-by-side stereo format to equirectangular top-bottom stereo
20678 format:
20679 v360=eac:equirect:in_stereo=sbs:in_trans=1:ih_flip=1:out_stereo=tb
20681 Commands
20683 This filter supports subset of above options as commands.
20685 vaguedenoiser
20687 Apply a wavelet based denoiser.
20688 It transforms each frame from the video input into the wavelet domain,
20690 using Cohen-Daubechies-Feauveau 9/7. Then it applies some filtering to
20691 the obtained coefficients. It does an inverse wavelet transform after.
20692 Due to wavelet properties, it should give a nice smoothed result, and
20693 reduced noise, without blurring picture features.
20694 This filter accepts the following options:
20696 threshold
20698 The filtering strength. The higher, the more filtered the video
20699 will be. Hard thresholding can use a higher threshold than soft
20700 thresholding before the video looks overfiltered. Default value is
20701 2.
20702 method
20704 The filtering method the filter will use.
20705 It accepts the following values:
20707 hard
20709 All values under the threshold will be zeroed.
20710 soft
20712 All values under the threshold will be zeroed. All values above
20713 will be reduced by the threshold.
20714 garrote
20716 Scales or nullifies coefficients - intermediary between (more)
20717 soft and (less) hard thresholding.
20718 Default is garrote.
20720 nsteps
20722 Number of times, the wavelet will decompose the picture. Picture
20723 can't be decomposed beyond a particular point (typically, 8 for a
20724 640x480 frame - as 2^9 = 512 > 480). Valid values are integers
20725 between 1 and 32. Default value is 6.
20726 percent
20728 Partial of full denoising (limited coefficients shrinking), from 0
20729 to 100. Default value is 85.
20730 planes
20732 A list of the planes to process. By default all planes are
20733 processed.
20734 type
20736 The threshold type the filter will use.
20737 It accepts the following values:
20739 universal
20741 Threshold used is same for all decompositions.
20742 bayes
20744 Threshold used depends also on each decomposition coefficients.
20745 Default is universal.
20747 varblur
20749 Apply variable blur filter by using 2nd video stream to set blur
20750 radius. The 2nd stream must have the same dimensions.
20751 This filter accepts the following options:
20753 min_r
20755 Set min allowed radius. Allowed range is from 0 to 254. Default is
20756 0.
20757 max_r
20759 Set max allowed radius. Allowed range is from 1 to 255. Default is
20760 8.
20761 planes
20763 Set which planes to process. By default, all are used.
20764 The "varblur" filter also supports the framesync options.
20766 Commands
20768 This filter supports all the above options as commands.
20770 vectorscope
20772 Display 2 color component values in the two dimensional graph (which is
20773 called a vectorscope).
20774 This filter accepts the following options:
20776 mode, m
20778 Set vectorscope mode.
20779 It accepts the following values:
20781 gray
20783 tint
20784 Gray values are displayed on graph, higher brightness means
20785 more pixels have same component color value on location in
20786 graph. This is the default mode.
20787 color
20789 Gray values are displayed on graph. Surrounding pixels values
20790 which are not present in video frame are drawn in gradient of 2
20791 color components which are set by option "x" and "y". The 3rd
20792 color component is static.
20793 color2
20795 Actual color components values present in video frame are
20796 displayed on graph.
20797 color3
20799 Similar as color2 but higher frequency of same values "x" and
20800 "y" on graph increases value of another color component, which
20801 is luminance by default values of "x" and "y".
20802 color4
20804 Actual colors present in video frame are displayed on graph. If
20805 two different colors map to same position on graph then color
20806 with higher value of component not present in graph is picked.
20807 color5
20809 Gray values are displayed on graph. Similar to "color" but with
20810 3rd color component picked from radial gradient.
20811 x Set which color component will be represented on X-axis. Default is
20813 1.
20814 y Set which color component will be represented on Y-axis. Default is
20816 2.
20817 intensity, i
20819 Set intensity, used by modes: gray, color, color3 and color5 for
20820 increasing brightness of color component which represents frequency
20821 of (X, Y) location in graph.
20822 envelope, e
20824 none
20825 No envelope, this is default.
20826 instant
20828 Instant envelope, even darkest single pixel will be clearly
20829 highlighted.
20830 peak
20832 Hold maximum and minimum values presented in graph over time.
20833 This way you can still spot out of range values without
20834 constantly looking at vectorscope.
20835 peak+instant
20837 Peak and instant envelope combined together.
20838 graticule, g
20840 Set what kind of graticule to draw.
20841 none
20843 green
20844 color
20845 invert
20846 opacity, o
20847 Set graticule opacity.
20848 flags, f
20850 Set graticule flags.
20851 white
20853 Draw graticule for white point.
20854 black
20856 Draw graticule for black point.
20857 name
20859 Draw color points short names.
20860 bgopacity, b
20862 Set background opacity.
20863 lthreshold, l
20865 Set low threshold for color component not represented on X or Y
20866 axis. Values lower than this value will be ignored. Default is 0.
20867 Note this value is multiplied with actual max possible value one
20868 pixel component can have. So for 8-bit input and low threshold
20869 value of 0.1 actual threshold is 0.1 * 255 = 25.
20870 hthreshold, h
20872 Set high threshold for color component not represented on X or Y
20873 axis. Values higher than this value will be ignored. Default is 1.
20874 Note this value is multiplied with actual max possible value one
20875 pixel component can have. So for 8-bit input and high threshold
20876 value of 0.9 actual threshold is 0.9 * 255 = 230.
20877 colorspace, c
20879 Set what kind of colorspace to use when drawing graticule.
20880 auto
20882 601
20883 709
20884 Default is auto.
20886 tint0, t0
20888 tint1, t1
20889 Set color tint for gray/tint vectorscope mode. By default both
20890 options are zero. This means no tint, and output will remain gray.
20891 vidstabdetect
20893 Analyze video stabilization/deshaking. Perform pass 1 of 2, see
20894 vidstabtransform for pass 2.
20895 This filter generates a file with relative translation and rotation
20897 transform information about subsequent frames, which is then used by
20898 the vidstabtransform filter.
20899 To enable compilation of this filter you need to configure FFmpeg with
20901 "--enable-libvidstab".
20902 This filter accepts the following options:
20904 result
20906 Set the path to the file used to write the transforms information.
20907 Default value is transforms.trf.
20908 shakiness
20910 Set how shaky the video is and how quick the camera is. It accepts
20911 an integer in the range 1-10, a value of 1 means little shakiness,
20912 a value of 10 means strong shakiness. Default value is 5.
20913 accuracy
20915 Set the accuracy of the detection process. It must be a value in
20916 the range 1-15. A value of 1 means low accuracy, a value of 15
20917 means high accuracy. Default value is 15.
20918 stepsize
20920 Set stepsize of the search process. The region around minimum is
20921 scanned with 1 pixel resolution. Default value is 6.
20922 mincontrast
20924 Set minimum contrast. Below this value a local measurement field is
20925 discarded. Must be a floating point value in the range 0-1. Default
20926 value is 0.3.
20927 tripod
20929 Set reference frame number for tripod mode.
20930 If enabled, the motion of the frames is compared to a reference
20932 frame in the filtered stream, identified by the specified number.
20933 The idea is to compensate all movements in a more-or-less static
20934 scene and keep the camera view absolutely still.
20935 If set to 0, it is disabled. The frames are counted starting from
20937 1.
20938 show
20940 Show fields and transforms in the resulting frames. It accepts an
20941 integer in the range 0-2. Default value is 0, which disables any
20942 visualization.
20943 Examples
20945 • Use default values:
20947 vidstabdetect
20949 • Analyze strongly shaky movie and put the results in file
20951 mytransforms.trf:
20952 vidstabdetect=shakiness=10:accuracy=15:result="mytransforms.trf"
20954 • Visualize the result of internal transformations in the resulting
20956 video:
20957 vidstabdetect=show=1
20959 • Analyze a video with medium shakiness using ffmpeg:
20961 ffmpeg -i input -vf vidstabdetect=shakiness=5:show=1 dummy.avi
20963 vidstabtransform
20965 Video stabilization/deshaking: pass 2 of 2, see vidstabdetect for pass
20966 1.
20967 Read a file with transform information for each frame and
20969 apply/compensate them. Together with the vidstabdetect filter this can
20970 be used to deshake videos. See also
20971 <http://public.hronopik.de/vid.stab>. It is important to also use the
20972 unsharp filter, see below.
20973 To enable compilation of this filter you need to configure FFmpeg with
20975 "--enable-libvidstab".
20976 Options
20978 input
20980 Set path to the file used to read the transforms. Default value is
20981 transforms.trf.
20982 smoothing
20984 Set the number of frames (value*2 + 1) used for lowpass filtering
20985 the camera movements. Default value is 10.
20986 For example a number of 10 means that 21 frames are used (10 in the
20988 past and 10 in the future) to smoothen the motion in the video. A
20989 larger value leads to a smoother video, but limits the acceleration
20990 of the camera (pan/tilt movements). 0 is a special case where a
20991 static camera is simulated.
20992 optalgo
20994 Set the camera path optimization algorithm.
20995 Accepted values are:
20997 gauss
20999 gaussian kernel low-pass filter on camera motion (default)
21000 avg averaging on transformations
21002 maxshift
21004 Set maximal number of pixels to translate frames. Default value is
21005 -1, meaning no limit.
21006 maxangle
21008 Set maximal angle in radians (degree*PI/180) to rotate frames.
21009 Default value is -1, meaning no limit.
21010 crop
21012 Specify how to deal with borders that may be visible due to
21013 movement compensation.
21014 Available values are:
21016 keep
21018 keep image information from previous frame (default)
21019 black
21021 fill the border black
21022 invert
21024 Invert transforms if set to 1. Default value is 0.
21025 relative
21027 Consider transforms as relative to previous frame if set to 1,
21028 absolute if set to 0. Default value is 0.
21029 zoom
21031 Set percentage to zoom. A positive value will result in a zoom-in
21032 effect, a negative value in a zoom-out effect. Default value is 0
21033 (no zoom).
21034 optzoom
21036 Set optimal zooming to avoid borders.
21037 Accepted values are:
21039 0 disabled
21041 1 optimal static zoom value is determined (only very strong
21043 movements will lead to visible borders) (default)
21044 2 optimal adaptive zoom value is determined (no borders will be
21046 visible), see zoomspeed
21047 Note that the value given at zoom is added to the one calculated
21049 here.
21050 zoomspeed
21052 Set percent to zoom maximally each frame (enabled when optzoom is
21053 set to 2). Range is from 0 to 5, default value is 0.25.
21054 interpol
21056 Specify type of interpolation.
21057 Available values are:
21059 no no interpolation
21061 linear
21063 linear only horizontal
21064 bilinear
21066 linear in both directions (default)
21067 bicubic
21069 cubic in both directions (slow)
21070 tripod
21072 Enable virtual tripod mode if set to 1, which is equivalent to
21073 "relative=0:smoothing=0". Default value is 0.
21074 Use also "tripod" option of vidstabdetect.
21076 debug
21078 Increase log verbosity if set to 1. Also the detected global
21079 motions are written to the temporary file global_motions.trf.
21080 Default value is 0.
21081 Examples
21083 • Use ffmpeg for a typical stabilization with default values:
21085 ffmpeg -i inp.mpeg -vf vidstabtransform,unsharp=5:5:0.8:3:3:0.4 inp_stabilized.mpeg
21087 Note the use of the unsharp filter which is always recommended.
21089 • Zoom in a bit more and load transform data from a given file:
21091 vidstabtransform=zoom=5:input="mytransforms.trf"
21093 • Smoothen the video even more:
21095 vidstabtransform=smoothing=30
21097 vflip
21099 Flip the input video vertically.
21100 For example, to vertically flip a video with ffmpeg:
21102 ffmpeg -i in.avi -vf "vflip" out.avi
21104 vfrdet
21106 Detect variable frame rate video.
21107 This filter tries to detect if the input is variable or constant frame
21109 rate.
21110 At end it will output number of frames detected as having variable
21112 delta pts, and ones with constant delta pts. If there was frames with
21113 variable delta, than it will also show min, max and average delta
21114 encountered.
21115 vibrance
21117 Boost or alter saturation.
21118 The filter accepts the following options:
21120 intensity
21122 Set strength of boost if positive value or strength of alter if
21123 negative value. Default is 0. Allowed range is from -2 to 2.
21124 rbal
21126 Set the red balance. Default is 1. Allowed range is from -10 to 10.
21127 gbal
21129 Set the green balance. Default is 1. Allowed range is from -10 to
21130 10.
21131 bbal
21133 Set the blue balance. Default is 1. Allowed range is from -10 to
21134 10.
21135 rlum
21137 Set the red luma coefficient.
21138 glum
21140 Set the green luma coefficient.
21141 blum
21143 Set the blue luma coefficient.
21144 alternate
21146 If "intensity" is negative and this is set to 1, colors will
21147 change, otherwise colors will be less saturated, more towards gray.
21148 Commands
21150 This filter supports the all above options as commands.
21152 vif
21154 Obtain the average VIF (Visual Information Fidelity) between two input
21155 videos.
21156 This filter takes two input videos.
21158 Both input videos must have the same resolution and pixel format for
21160 this filter to work correctly. Also it assumes that both inputs have
21161 the same number of frames, which are compared one by one.
21162 The obtained average VIF score is printed through the logging system.
21164 The filter stores the calculated VIF score of each frame.
21166 In the below example the input file main.mpg being processed is
21168 compared with the reference file ref.mpg.
21169 ffmpeg -i main.mpg -i ref.mpg -lavfi vif -f null -
21171 vignette
21173 Make or reverse a natural vignetting effect.
21174 The filter accepts the following options:
21176 angle, a
21178 Set lens angle expression as a number of radians.
21179 The value is clipped in the "[0,PI/2]" range.
21181 Default value: "PI/5"
21183 x0
21185 y0 Set center coordinates expressions. Respectively "w/2" and "h/2" by
21186 default.
21187 mode
21189 Set forward/backward mode.
21190 Available modes are:
21192 forward
21194 The larger the distance from the central point, the darker the
21195 image becomes.
21196 backward
21198 The larger the distance from the central point, the brighter
21199 the image becomes. This can be used to reverse a vignette
21200 effect, though there is no automatic detection to extract the
21201 lens angle and other settings (yet). It can also be used to
21202 create a burning effect.
21203 Default value is forward.
21205 eval
21207 Set evaluation mode for the expressions (angle, x0, y0).
21208 It accepts the following values:
21210 init
21212 Evaluate expressions only once during the filter
21213 initialization.
21214 frame
21216 Evaluate expressions for each incoming frame. This is way
21217 slower than the init mode since it requires all the scalers to
21218 be re-computed, but it allows advanced dynamic expressions.
21219 Default value is init.
21221 dither
21223 Set dithering to reduce the circular banding effects. Default is 1
21224 (enabled).
21225 aspect
21227 Set vignette aspect. This setting allows one to adjust the shape of
21228 the vignette. Setting this value to the SAR of the input will make
21229 a rectangular vignetting following the dimensions of the video.
21230 Default is "1/1".
21232 Expressions
21234 The alpha, x0 and y0 expressions can contain the following parameters.
21236 w
21238 h input width and height
21239 n the number of input frame, starting from 0
21241 pts the PTS (Presentation TimeStamp) time of the filtered video frame,
21243 expressed in TB units, NAN if undefined
21244 r frame rate of the input video, NAN if the input frame rate is
21246 unknown
21247 t the PTS (Presentation TimeStamp) of the filtered video frame,
21249 expressed in seconds, NAN if undefined
21250 tb time base of the input video
21252 Examples
21254 • Apply simple strong vignetting effect:
21256 vignette=PI/4
21258 • Make a flickering vignetting:
21260 vignette='PI/4+random(1)*PI/50':eval=frame
21262 vmafmotion
21264 Obtain the average VMAF motion score of a video. It is one of the
21265 component metrics of VMAF.
21266 The obtained average motion score is printed through the logging
21268 system.
21269 The filter accepts the following options:
21271 stats_file
21273 If specified, the filter will use the named file to save the motion
21274 score of each frame with respect to the previous frame. When
21275 filename equals "-" the data is sent to standard output.
21276 Example:
21278 ffmpeg -i ref.mpg -vf vmafmotion -f null -
21280 vstack
21282 Stack input videos vertically.
21283 All streams must be of same pixel format and of same width.
21285 Note that this filter is faster than using overlay and pad filter to
21287 create same output.
21288 The filter accepts the following options:
21290 inputs
21292 Set number of input streams. Default is 2.
21293 shortest
21295 If set to 1, force the output to terminate when the shortest input
21296 terminates. Default value is 0.
21297 w3fdif
21299 Deinterlace the input video ("w3fdif" stands for "Weston 3 Field
21300 Deinterlacing Filter").
21301 Based on the process described by Martin Weston for BBC R&D, and
21303 implemented based on the de-interlace algorithm written by Jim
21304 Easterbrook for BBC R&D, the Weston 3 field deinterlacing filter uses
21305 filter coefficients calculated by BBC R&D.
21306 This filter uses field-dominance information in frame to decide which
21308 of each pair of fields to place first in the output. If it gets it
21309 wrong use setfield filter before "w3fdif" filter.
21310 There are two sets of filter coefficients, so called "simple" and
21312 "complex". Which set of filter coefficients is used can be set by
21313 passing an optional parameter:
21314 filter
21316 Set the interlacing filter coefficients. Accepts one of the
21317 following values:
21318 simple
21320 Simple filter coefficient set.
21321 complex
21323 More-complex filter coefficient set.
21324 Default value is complex.
21326 mode
21328 The interlacing mode to adopt. It accepts one of the following
21329 values:
21330 frame
21332 Output one frame for each frame.
21333 field
21335 Output one frame for each field.
21336 The default value is "field".
21338 parity
21340 The picture field parity assumed for the input interlaced video. It
21341 accepts one of the following values:
21342 tff Assume the top field is first.
21344 bff Assume the bottom field is first.
21346 auto
21348 Enable automatic detection of field parity.
21349 The default value is "auto". If the interlacing is unknown or the
21351 decoder does not export this information, top field first will be
21352 assumed.
21353 deint
21355 Specify which frames to deinterlace. Accepts one of the following
21356 values:
21357 all Deinterlace all frames,
21359 interlaced
21361 Only deinterlace frames marked as interlaced.
21362 Default value is all.
21364 Commands
21366 This filter supports same commands as options.
21368 waveform
21370 Video waveform monitor.
21371 The waveform monitor plots color component intensity. By default
21373 luminance only. Each column of the waveform corresponds to a column of
21374 pixels in the source video.
21375 It accepts the following options:
21377 mode, m
21379 Can be either "row", or "column". Default is "column". In row
21380 mode, the graph on the left side represents color component value 0
21381 and the right side represents value = 255. In column mode, the top
21382 side represents color component value = 0 and bottom side
21383 represents value = 255.
21384 intensity, i
21386 Set intensity. Smaller values are useful to find out how many
21387 values of the same luminance are distributed across input
21388 rows/columns. Default value is 0.04. Allowed range is [0, 1].
21389 mirror, r
21391 Set mirroring mode. 0 means unmirrored, 1 means mirrored. In
21392 mirrored mode, higher values will be represented on the left side
21393 for "row" mode and at the top for "column" mode. Default is 1
21394 (mirrored).
21395 display, d
21397 Set display mode. It accepts the following values:
21398 overlay
21400 Presents information identical to that in the "parade", except
21401 that the graphs representing color components are superimposed
21402 directly over one another.
21403 This display mode makes it easier to spot relative differences
21405 or similarities in overlapping areas of the color components
21406 that are supposed to be identical, such as neutral whites,
21407 grays, or blacks.
21408 stack
21410 Display separate graph for the color components side by side in
21411 "row" mode or one below the other in "column" mode.
21412 parade
21414 Display separate graph for the color components side by side in
21415 "column" mode or one below the other in "row" mode.
21416 Using this display mode makes it easy to spot color casts in
21418 the highlights and shadows of an image, by comparing the
21419 contours of the top and the bottom graphs of each waveform.
21420 Since whites, grays, and blacks are characterized by exactly
21421 equal amounts of red, green, and blue, neutral areas of the
21422 picture should display three waveforms of roughly equal
21423 width/height. If not, the correction is easy to perform by
21424 making level adjustments the three waveforms.
21425 Default is "stack".
21427 components, c
21429 Set which color components to display. Default is 1, which means
21430 only luminance or red color component if input is in RGB
21431 colorspace. If is set for example to 7 it will display all 3 (if)
21432 available color components.
21433 envelope, e
21435 none
21436 No envelope, this is default.
21437 instant
21439 Instant envelope, minimum and maximum values presented in graph
21440 will be easily visible even with small "step" value.
21441 peak
21443 Hold minimum and maximum values presented in graph across time.
21444 This way you can still spot out of range values without
21445 constantly looking at waveforms.
21446 peak+instant
21448 Peak and instant envelope combined together.
21449 filter, f
21451 lowpass
21452 No filtering, this is default.
21453 flat
21455 Luma and chroma combined together.
21456 aflat
21458 Similar as above, but shows difference between blue and red
21459 chroma.
21460 xflat
21462 Similar as above, but use different colors.
21463 yflat
21465 Similar as above, but again with different colors.
21466 chroma
21468 Displays only chroma.
21469 color
21471 Displays actual color value on waveform.
21472 acolor
21474 Similar as above, but with luma showing frequency of chroma
21475 values.
21476 graticule, g
21478 Set which graticule to display.
21479 none
21481 Do not display graticule.
21482 green
21484 Display green graticule showing legal broadcast ranges.
21485 orange
21487 Display orange graticule showing legal broadcast ranges.
21488 invert
21490 Display invert graticule showing legal broadcast ranges.
21491 opacity, o
21493 Set graticule opacity.
21494 flags, fl
21496 Set graticule flags.
21497 numbers
21499 Draw numbers above lines. By default enabled.
21500 dots
21502 Draw dots instead of lines.
21503 scale, s
21505 Set scale used for displaying graticule.
21506 digital
21508 millivolts
21509 ire
21510 Default is digital.
21512 bgopacity, b
21514 Set background opacity.
21515 tint0, t0
21517 tint1, t1
21518 Set tint for output. Only used with lowpass filter and when
21519 display is not overlay and input pixel formats are not RGB.
21520 fitmode, fm
21522 Set sample aspect ratio of video output frames. Can be used to
21523 configure waveform so it is not streched too much in one of
21524 directions.
21525 none
21527 Set sample aspect ration to 1/1.
21528 size
21530 Set sample aspect ratio to match input size of video
21531 Default is none.
21533 weave, doubleweave
21535 The "weave" takes a field-based video input and join each two
21536 sequential fields into single frame, producing a new double height clip
21537 with half the frame rate and half the frame count.
21538 The "doubleweave" works same as "weave" but without halving frame rate
21540 and frame count.
21541 It accepts the following option:
21543 first_field
21545 Set first field. Available values are:
21546 top, t
21548 Set the frame as top-field-first.
21549 bottom, b
21551 Set the frame as bottom-field-first.
21552 Examples
21554 • Interlace video using select and separatefields filter:
21556 separatefields,select=eq(mod(n,4),0)+eq(mod(n,4),3),weave
21558 xbr
21560 Apply the xBR high-quality magnification filter which is designed for
21561 pixel art. It follows a set of edge-detection rules, see
21562 <https://forums.libretro.com/t/xbr-algorithm-tutorial/123>.
21563 It accepts the following option:
21565 n Set the scaling dimension: 2 for "2xBR", 3 for "3xBR" and 4 for
21567 "4xBR". Default is 3.
21568 xcorrelate
21570 Apply normalized cross-correlation between first and second input video
21571 stream.
21572 Second input video stream dimensions must be lower than first input
21574 video stream.
21575 The filter accepts the following options:
21577 planes
21579 Set which planes to process.
21580 secondary
21582 Set which secondary video frames will be processed from second
21583 input video stream, can be first or all. Default is all.
21584 The "xcorrelate" filter also supports the framesync options.
21586 xfade
21588 Apply cross fade from one input video stream to another input video
21589 stream. The cross fade is applied for specified duration.
21590 Both inputs must be constant frame-rate and have the same resolution,
21592 pixel format, frame rate and timebase.
21593 The filter accepts the following options:
21595 transition
21597 Set one of available transition effects:
21598 custom
21600 fade
21601 wipeleft
21602 wiperight
21603 wipeup
21604 wipedown
21605 slideleft
21606 slideright
21607 slideup
21608 slidedown
21609 circlecrop
21610 rectcrop
21611 distance
21612 fadeblack
21613 fadewhite
21614 radial
21615 smoothleft
21616 smoothright
21617 smoothup
21618 smoothdown
21619 circleopen
21620 circleclose
21621 vertopen
21622 vertclose
21623 horzopen
21624 horzclose
21625 dissolve
21626 pixelize
21627 diagtl
21628 diagtr
21629 diagbl
21630 diagbr
21631 hlslice
21632 hrslice
21633 vuslice
21634 vdslice
21635 hblur
21636 fadegrays
21637 wipetl
21638 wipetr
21639 wipebl
21640 wipebr
21641 squeezeh
21642 squeezev
21643 zoomin
21644 fadefast
21645 fadeslow
21646 Default transition effect is fade.
21648 duration
21650 Set cross fade duration in seconds. Range is 0 to 60 seconds.
21651 Default duration is 1 second.
21652 offset
21654 Set cross fade start relative to first input stream in seconds.
21655 Default offset is 0.
21656 expr
21658 Set expression for custom transition effect.
21659 The expressions can use the following variables and functions:
21661 X
21663 Y The coordinates of the current sample.
21664 W
21666 H The width and height of the image.
21667 P Progress of transition effect.
21669 PLANE
21671 Currently processed plane.
21672 A Return value of first input at current location and plane.
21674 B Return value of second input at current location and plane.
21676 a0(x, y)
21678 a1(x, y)
21679 a2(x, y)
21680 a3(x, y)
21681 Return the value of the pixel at location (x,y) of the
21682 first/second/third/fourth component of first input.
21683 b0(x, y)
21685 b1(x, y)
21686 b2(x, y)
21687 b3(x, y)
21688 Return the value of the pixel at location (x,y) of the
21689 first/second/third/fourth component of second input.
21690 Examples
21692 • Cross fade from one input video to another input video, with fade
21694 transition and duration of transition of 2 seconds starting at
21695 offset of 5 seconds:
21696 ffmpeg -i first.mp4 -i second.mp4 -filter_complex xfade=transition=fade:duration=2:offset=5 output.mp4
21698 xmedian
21700 Pick median pixels from several input videos.
21701 The filter accepts the following options:
21703 inputs
21705 Set number of inputs. Default is 3. Allowed range is from 3 to
21706 255. If number of inputs is even number, than result will be mean
21707 value between two median values.
21708 planes
21710 Set which planes to filter. Default value is 15, by which all
21711 planes are processed.
21712 percentile
21714 Set median percentile. Default value is 0.5. Default value of 0.5
21715 will pick always median values, while 0 will pick minimum values,
21716 and 1 maximum values.
21717 Commands
21719 This filter supports all above options as commands, excluding option
21721 "inputs".
21722 xstack
21724 Stack video inputs into custom layout.
21725 All streams must be of same pixel format.
21727 The filter accepts the following options:
21729 inputs
21731 Set number of input streams. Default is 2.
21732 layout
21734 Specify layout of inputs. This option requires the desired layout
21735 configuration to be explicitly set by the user. This sets position
21736 of each video input in output. Each input is separated by '|'. The
21737 first number represents the column, and the second number
21738 represents the row. Numbers start at 0 and are separated by '_'.
21739 Optionally one can use wX and hX, where X is video input from which
21740 to take width or height. Multiple values can be used when
21741 separated by '+'. In such case values are summed together.
21742 Note that if inputs are of different sizes gaps may appear, as not
21744 all of the output video frame will be filled. Similarly, videos can
21745 overlap each other if their position doesn't leave enough space for
21746 the full frame of adjoining videos.
21747 For 2 inputs, a default layout of "0_0|w0_0" (equivalent to
21749 "grid=2x1") is set. In all other cases, a layout or a grid must be
21750 set by the user. Either "grid" or "layout" can be specified at a
21751 time. Specifying both will result in an error.
21752 grid
21754 Specify a fixed size grid of inputs. This option is used to create
21755 a fixed size grid of the input streams. Set the grid size in the
21756 form "COLUMNSxROWS". There must be "ROWS * COLUMNS" input streams
21757 and they will be arranged as a grid with "ROWS" rows and "COLUMNS"
21758 columns. When using this option, each input stream within a row
21759 must have the same height and all the rows must have the same
21760 width.
21761 If "grid" is set, then "inputs" option is ignored and is implicitly
21763 set to "ROWS * COLUMNS".
21764 For 2 inputs, a default grid of "2x1" (equivalent to
21766 "layout=0_0|w0_0") is set. In all other cases, a layout or a grid
21767 must be set by the user. Either "grid" or "layout" can be specified
21768 at a time. Specifying both will result in an error.
21769 shortest
21771 If set to 1, force the output to terminate when the shortest input
21772 terminates. Default value is 0.
21773 fill
21775 If set to valid color, all unused pixels will be filled with that
21776 color. By default fill is set to none, so it is disabled.
21777 Examples
21779 • Display 4 inputs into 2x2 grid.
21781 Layout:
21783 input1(0, 0) | input3(w0, 0)
21785 input2(0, h0) | input4(w0, h0)
21786 xstack=inputs=4:layout=0_0|0_h0|w0_0|w0_h0
21790 Note that if inputs are of different sizes, gaps or overlaps may
21792 occur.
21793 • Display 4 inputs into 1x4 grid.
21795 Layout:
21797 input1(0, 0)
21799 input2(0, h0)
21800 input3(0, h0+h1)
21801 input4(0, h0+h1+h2)
21802 xstack=inputs=4:layout=0_0|0_h0|0_h0+h1|0_h0+h1+h2
21806 Note that if inputs are of different widths, unused space will
21808 appear.
21809 • Display 9 inputs into 3x3 grid.
21811 Layout:
21813 input1(0, 0) | input4(w0, 0) | input7(w0+w3, 0)
21815 input2(0, h0) | input5(w0, h0) | input8(w0+w3, h0)
21816 input3(0, h0+h1) | input6(w0, h0+h1) | input9(w0+w3, h0+h1)
21817 xstack=inputs=9:layout=0_0|0_h0|0_h0+h1|w0_0|w0_h0|w0_h0+h1|w0+w3_0|w0+w3_h0|w0+w3_h0+h1
21821 Note that if inputs are of different sizes, gaps or overlaps may
21823 occur.
21824 • Display 16 inputs into 4x4 grid.
21826 Layout:
21828 input1(0, 0) | input5(w0, 0) | input9 (w0+w4, 0) | input13(w0+w4+w8, 0)
21830 input2(0, h0) | input6(w0, h0) | input10(w0+w4, h0) | input14(w0+w4+w8, h0)
21831 input3(0, h0+h1) | input7(w0, h0+h1) | input11(w0+w4, h0+h1) | input15(w0+w4+w8, h0+h1)
21832 input4(0, h0+h1+h2)| input8(w0, h0+h1+h2)| input12(w0+w4, h0+h1+h2)| input16(w0+w4+w8, h0+h1+h2)
21833 xstack=inputs=16:layout=0_0|0_h0|0_h0+h1|0_h0+h1+h2|w0_0|w0_h0|w0_h0+h1|w0_h0+h1+h2|w0+w4_0|
21837 w0+w4_h0|w0+w4_h0+h1|w0+w4_h0+h1+h2|w0+w4+w8_0|w0+w4+w8_h0|w0+w4+w8_h0+h1|w0+w4+w8_h0+h1+h2
21838 Note that if inputs are of different sizes, gaps or overlaps may
21840 occur.
21841 yadif
21843 Deinterlace the input video ("yadif" means "yet another deinterlacing
21844 filter").
21845 It accepts the following parameters:
21847 mode
21849 The interlacing mode to adopt. It accepts one of the following
21850 values:
21851 0, send_frame
21853 Output one frame for each frame.
21854 1, send_field
21856 Output one frame for each field.
21857 2, send_frame_nospatial
21859 Like "send_frame", but it skips the spatial interlacing check.
21860 3, send_field_nospatial
21862 Like "send_field", but it skips the spatial interlacing check.
21863 The default value is "send_frame".
21865 parity
21867 The picture field parity assumed for the input interlaced video. It
21868 accepts one of the following values:
21869 0, tff
21871 Assume the top field is first.
21872 1, bff
21874 Assume the bottom field is first.
21875 -1, auto
21877 Enable automatic detection of field parity.
21878 The default value is "auto". If the interlacing is unknown or the
21880 decoder does not export this information, top field first will be
21881 assumed.
21882 deint
21884 Specify which frames to deinterlace. Accepts one of the following
21885 values:
21886 0, all
21888 Deinterlace all frames.
21889 1, interlaced
21891 Only deinterlace frames marked as interlaced.
21892 The default value is "all".
21894 yadif_cuda
21896 Deinterlace the input video using the yadif algorithm, but implemented
21897 in CUDA so that it can work as part of a GPU accelerated pipeline with
21898 nvdec and/or nvenc.
21899 It accepts the following parameters:
21901 mode
21903 The interlacing mode to adopt. It accepts one of the following
21904 values:
21905 0, send_frame
21907 Output one frame for each frame.
21908 1, send_field
21910 Output one frame for each field.
21911 2, send_frame_nospatial
21913 Like "send_frame", but it skips the spatial interlacing check.
21914 3, send_field_nospatial
21916 Like "send_field", but it skips the spatial interlacing check.
21917 The default value is "send_frame".
21919 parity
21921 The picture field parity assumed for the input interlaced video. It
21922 accepts one of the following values:
21923 0, tff
21925 Assume the top field is first.
21926 1, bff
21928 Assume the bottom field is first.
21929 -1, auto
21931 Enable automatic detection of field parity.
21932 The default value is "auto". If the interlacing is unknown or the
21934 decoder does not export this information, top field first will be
21935 assumed.
21936 deint
21938 Specify which frames to deinterlace. Accepts one of the following
21939 values:
21940 0, all
21942 Deinterlace all frames.
21943 1, interlaced
21945 Only deinterlace frames marked as interlaced.
21946 The default value is "all".
21948 yaepblur
21950 Apply blur filter while preserving edges ("yaepblur" means "yet another
21951 edge preserving blur filter"). The algorithm is described in "J. S.
21952 Lee, Digital image enhancement and noise filtering by use of local
21953 statistics, IEEE Trans. Pattern Anal. Mach. Intell. PAMI-2, 1980."
21954 It accepts the following parameters:
21956 radius, r
21958 Set the window radius. Default value is 3.
21959 planes, p
21961 Set which planes to filter. Default is only the first plane.
21962 sigma, s
21964 Set blur strength. Default value is 128.
21965 Commands
21967 This filter supports same commands as options.
21969 zoompan
21971 Apply Zoom & Pan effect.
21972 This filter accepts the following options:
21974 zoom, z
21976 Set the zoom expression. Range is 1-10. Default is 1.
21977 x
21979 y Set the x and y expression. Default is 0.
21980 d Set the duration expression in number of frames. This sets for how
21982 many number of frames effect will last for single input image.
21983 Default is 90.
21984 s Set the output image size, default is 'hd720'.
21986 fps Set the output frame rate, default is '25'.
21988 Each expression can contain the following constants:
21990 in_w, iw
21992 Input width.
21993 in_h, ih
21995 Input height.
21996 out_w, ow
21998 Output width.
21999 out_h, oh
22001 Output height.
22002 in Input frame count.
22004 on Output frame count.
22006 in_time, it
22008 The input timestamp expressed in seconds. It's NAN if the input
22009 timestamp is unknown.
22010 out_time, time, ot
22012 The output timestamp expressed in seconds.
22013 x
22015 y Last calculated 'x' and 'y' position from 'x' and 'y' expression
22016 for current input frame.
22017 px
22019 py 'x' and 'y' of last output frame of previous input frame or 0 when
22020 there was not yet such frame (first input frame).
22021 zoom
22023 Last calculated zoom from 'z' expression for current input frame.
22024 pzoom
22026 Last calculated zoom of last output frame of previous input frame.
22027 duration
22029 Number of output frames for current input frame. Calculated from
22030 'd' expression for each input frame.
22031 pduration
22033 number of output frames created for previous input frame
22034 a Rational number: input width / input height
22036 sar sample aspect ratio
22038 dar display aspect ratio
22040 Examples
22042 • Zoom in up to 1.5x and pan at same time to some spot near center of
22044 picture:
22045 zoompan=z='min(zoom+0.0015,1.5)':d=700:x='if(gte(zoom,1.5),x,x+1/a)':y='if(gte(zoom,1.5),y,y+1)':s=640x360
22047 • Zoom in up to 1.5x and pan always at center of picture:
22049 zoompan=z='min(zoom+0.0015,1.5)':d=700:x='iw/2-(iw/zoom/2)':y='ih/2-(ih/zoom/2)'
22051 • Same as above but without pausing:
22053 zoompan=z='min(max(zoom,pzoom)+0.0015,1.5)':d=1:x='iw/2-(iw/zoom/2)':y='ih/2-(ih/zoom/2)'
22055 • Zoom in 2x into center of picture only for the first second of the
22057 input video:
22058 zoompan=z='if(between(in_time,0,1),2,1)':d=1:x='iw/2-(iw/zoom/2)':y='ih/2-(ih/zoom/2)'
22060 zscale
22062 Scale (resize) the input video, using the z.lib library:
22063 <https://github.com/sekrit-twc/zimg>. To enable compilation of this
22064 filter, you need to configure FFmpeg with "--enable-libzimg".
22065 The zscale filter forces the output display aspect ratio to be the same
22067 as the input, by changing the output sample aspect ratio.
22068 If the input image format is different from the format requested by the
22070 next filter, the zscale filter will convert the input to the requested
22071 format.
22072 Options
22074 The filter accepts the following options.
22076 width, w
22078 height, h
22079 Set the output video dimension expression. Default value is the
22080 input dimension.
22081 If the width or w value is 0, the input width is used for the
22083 output. If the height or h value is 0, the input height is used for
22084 the output.
22085 If one and only one of the values is -n with n >= 1, the zscale
22087 filter will use a value that maintains the aspect ratio of the
22088 input image, calculated from the other specified dimension. After
22089 that it will, however, make sure that the calculated dimension is
22090 divisible by n and adjust the value if necessary.
22091 If both values are -n with n >= 1, the behavior will be identical
22093 to both values being set to 0 as previously detailed.
22094 See below for the list of accepted constants for use in the
22096 dimension expression.
22097 size, s
22099 Set the video size. For the syntax of this option, check the "Video
22100 size" section in the ffmpeg-utils manual.
22101 dither, d
22103 Set the dither type.
22104 Possible values are:
22106 none
22108 ordered
22109 random
22110 error_diffusion
22111 Default is none.
22113 filter, f
22115 Set the resize filter type.
22116 Possible values are:
22118 point
22120 bilinear
22121 bicubic
22122 spline16
22123 spline36
22124 lanczos
22125 Default is bilinear.
22127 range, r
22129 Set the color range.
22130 Possible values are:
22132 input
22134 limited
22135 full
22136 Default is same as input.
22138 primaries, p
22140 Set the color primaries.
22141 Possible values are:
22143 input
22145 709
22146 unspecified
22147 170m
22148 240m
22149 2020
22150 Default is same as input.
22152 transfer, t
22154 Set the transfer characteristics.
22155 Possible values are:
22157 input
22159 709
22160 unspecified
22161 601
22162 linear
22163 2020_10
22164 2020_12
22165 smpte2084
22166 iec61966-2-1
22167 arib-std-b67
22168 Default is same as input.
22170 matrix, m
22172 Set the colorspace matrix.
22173 Possible value are:
22175 input
22177 709
22178 unspecified
22179 470bg
22180 170m
22181 2020_ncl
22182 2020_cl
22183 Default is same as input.
22185 rangein, rin
22187 Set the input color range.
22188 Possible values are:
22190 input
22192 limited
22193 full
22194 Default is same as input.
22196 primariesin, pin
22198 Set the input color primaries.
22199 Possible values are:
22201 input
22203 709
22204 unspecified
22205 170m
22206 240m
22207 2020
22208 Default is same as input.
22210 transferin, tin
22212 Set the input transfer characteristics.
22213 Possible values are:
22215 input
22217 709
22218 unspecified
22219 601
22220 linear
22221 2020_10
22222 2020_12
22223 Default is same as input.
22225 matrixin, min
22227 Set the input colorspace matrix.
22228 Possible value are:
22230 input
22232 709
22233 unspecified
22234 470bg
22235 170m
22236 2020_ncl
22237 2020_cl
22238 chromal, c
22239 Set the output chroma location.
22240 Possible values are:
22242 input
22244 left
22245 center
22246 topleft
22247 top
22248 bottomleft
22249 bottom
22250 chromalin, cin
22251 Set the input chroma location.
22252 Possible values are:
22254 input
22256 left
22257 center
22258 topleft
22259 top
22260 bottomleft
22261 bottom
22262 npl Set the nominal peak luminance.
22263 param_a
22265 Parameter A for scaling filters. Parameter "b" for bicubic, and the
22266 number of filter taps for lanczos.
22267 param_b
22269 Parameter B for scaling filters. Parameter "c" for bicubic.
22270 The values of the w and h options are expressions containing the
22272 following constants:
22273 in_w
22275 in_h
22276 The input width and height
22277 iw
22279 ih These are the same as in_w and in_h.
22280 out_w
22282 out_h
22283 The output (scaled) width and height
22284 ow
22286 oh These are the same as out_w and out_h
22287 a The same as iw / ih
22289 sar input sample aspect ratio
22291 dar The input display aspect ratio. Calculated from "(iw / ih) * sar".
22293 hsub
22295 vsub
22296 horizontal and vertical input chroma subsample values. For example
22297 for the pixel format "yuv422p" hsub is 2 and vsub is 1.
22298 ohsub
22300 ovsub
22301 horizontal and vertical output chroma subsample values. For example
22302 for the pixel format "yuv422p" hsub is 2 and vsub is 1.
22303 Commands
22305 This filter supports the following commands:
22307 width, w
22309 height, h
22310 Set the output video dimension expression. The command accepts the
22311 same syntax of the corresponding option.
22312 If the specified expression is not valid, it is kept at its current
22314 value.
22315
22317 Below is a description of the currently available OpenCL video filters.
22318 To enable compilation of these filters you need to configure FFmpeg
22320 with "--enable-opencl".
22321 Running OpenCL filters requires you to initialize a hardware device and
22323 to pass that device to all filters in any filter graph.
22324 -init_hw_device opencl[=name][:device[,key=value...]]
22326 Initialise a new hardware device of type opencl called name, using
22327 the given device parameters.
22328 -filter_hw_device name
22330 Pass the hardware device called name to all filters in any filter
22331 graph.
22332 For more detailed information see
22334 <https://www.ffmpeg.org/ffmpeg.html#Advanced-Video-options>
22335 • Example of choosing the first device on the second platform and
22337 running avgblur_opencl filter with default parameters on it.
22338 -init_hw_device opencl=gpu:1.0 -filter_hw_device gpu -i INPUT -vf "hwupload, avgblur_opencl, hwdownload" OUTPUT
22340 Since OpenCL filters are not able to access frame data in normal
22342 memory, all frame data needs to be uploaded(hwupload) to hardware
22343 surfaces connected to the appropriate device before being used and then
22344 downloaded(hwdownload) back to normal memory. Note that hwupload will
22345 upload to a surface with the same layout as the software frame, so it
22346 may be necessary to add a format filter immediately before to get the
22347 input into the right format and hwdownload does not support all formats
22348 on the output - it may be necessary to insert an additional format
22349 filter immediately following in the graph to get the output in a
22350 supported format.
22351 avgblur_opencl
22353 Apply average blur filter.
22354 The filter accepts the following options:
22356 sizeX
22358 Set horizontal radius size. Range is "[1, 1024]" and default value
22359 is 1.
22360 planes
22362 Set which planes to filter. Default value is 0xf, by which all
22363 planes are processed.
22364 sizeY
22366 Set vertical radius size. Range is "[1, 1024]" and default value is
22367 0. If zero, "sizeX" value will be used.
22368 Example
22370 • Apply average blur filter with horizontal and vertical size of 3,
22372 setting each pixel of the output to the average value of the 7x7
22373 region centered on it in the input. For pixels on the edges of the
22374 image, the region does not extend beyond the image boundaries, and
22375 so out-of-range coordinates are not used in the calculations.
22376 -i INPUT -vf "hwupload, avgblur_opencl=3, hwdownload" OUTPUT
22378 boxblur_opencl
22380 Apply a boxblur algorithm to the input video.
22381 It accepts the following parameters:
22383 luma_radius, lr
22385 luma_power, lp
22386 chroma_radius, cr
22387 chroma_power, cp
22388 alpha_radius, ar
22389 alpha_power, ap
22390 A description of the accepted options follows.
22392 luma_radius, lr
22394 chroma_radius, cr
22395 alpha_radius, ar
22396 Set an expression for the box radius in pixels used for blurring
22397 the corresponding input plane.
22398 The radius value must be a non-negative number, and must not be
22400 greater than the value of the expression "min(w,h)/2" for the luma
22401 and alpha planes, and of "min(cw,ch)/2" for the chroma planes.
22402 Default value for luma_radius is "2". If not specified,
22404 chroma_radius and alpha_radius default to the corresponding value
22405 set for luma_radius.
22406 The expressions can contain the following constants:
22408 w
22410 h The input width and height in pixels.
22411 cw
22413 ch The input chroma image width and height in pixels.
22414 hsub
22416 vsub
22417 The horizontal and vertical chroma subsample values. For
22418 example, for the pixel format "yuv422p", hsub is 2 and vsub is
22419 1.
22420 luma_power, lp
22422 chroma_power, cp
22423 alpha_power, ap
22424 Specify how many times the boxblur filter is applied to the
22425 corresponding plane.
22426 Default value for luma_power is 2. If not specified, chroma_power
22428 and alpha_power default to the corresponding value set for
22429 luma_power.
22430 A value of 0 will disable the effect.
22432 Examples
22434 Apply boxblur filter, setting each pixel of the output to the average
22436 value of box-radiuses luma_radius, chroma_radius, alpha_radius for each
22437 plane respectively. The filter will apply luma_power, chroma_power,
22438 alpha_power times onto the corresponding plane. For pixels on the edges
22439 of the image, the radius does not extend beyond the image boundaries,
22440 and so out-of-range coordinates are not used in the calculations.
22441 • Apply a boxblur filter with the luma, chroma, and alpha radius set
22443 to 2 and luma, chroma, and alpha power set to 3. The filter will
22444 run 3 times with box-radius set to 2 for every plane of the image.
22445 -i INPUT -vf "hwupload, boxblur_opencl=luma_radius=2:luma_power=3, hwdownload" OUTPUT
22447 -i INPUT -vf "hwupload, boxblur_opencl=2:3, hwdownload" OUTPUT
22448 • Apply a boxblur filter with luma radius set to 2, luma_power to 1,
22450 chroma_radius to 4, chroma_power to 5, alpha_radius to 3 and
22451 alpha_power to 7.
22452 For the luma plane, a 2x2 box radius will be run once.
22454 For the chroma plane, a 4x4 box radius will be run 5 times.
22456 For the alpha plane, a 3x3 box radius will be run 7 times.
22458 -i INPUT -vf "hwupload, boxblur_opencl=2:1:4:5:3:7, hwdownload" OUTPUT
22460 colorkey_opencl
22462 RGB colorspace color keying.
22463 The filter accepts the following options:
22465 color
22467 The color which will be replaced with transparency.
22468 similarity
22470 Similarity percentage with the key color.
22471 0.01 matches only the exact key color, while 1.0 matches
22473 everything.
22474 blend
22476 Blend percentage.
22477 0.0 makes pixels either fully transparent, or not transparent at
22479 all.
22480 Higher values result in semi-transparent pixels, with a higher
22482 transparency the more similar the pixels color is to the key color.
22483 Examples
22485 • Make every semi-green pixel in the input transparent with some
22487 slight blending:
22488 -i INPUT -vf "hwupload, colorkey_opencl=green:0.3:0.1, hwdownload" OUTPUT
22490 convolution_opencl
22492 Apply convolution of 3x3, 5x5, 7x7 matrix.
22493 The filter accepts the following options:
22495 0m
22497 1m
22498 2m
22499 3m Set matrix for each plane. Matrix is sequence of 9, 25 or 49
22500 signed numbers. Default value for each plane is "0 0 0 0 1 0 0 0
22501 0".
22502 0rdiv
22504 1rdiv
22505 2rdiv
22506 3rdiv
22507 Set multiplier for calculated value for each plane. If unset or 0,
22508 it will be sum of all matrix elements. The option value must be a
22509 float number greater or equal to 0.0. Default value is 1.0.
22510 0bias
22512 1bias
22513 2bias
22514 3bias
22515 Set bias for each plane. This value is added to the result of the
22516 multiplication. Useful for making the overall image brighter or
22517 darker. The option value must be a float number greater or equal
22518 to 0.0. Default value is 0.0.
22519 Examples
22521 • Apply sharpen:
22523 -i INPUT -vf "hwupload, convolution_opencl=0 -1 0 -1 5 -1 0 -1 0:0 -1 0 -1 5 -1 0 -1 0:0 -1 0 -1 5 -1 0 -1 0:0 -1 0 -1 5 -1 0 -1 0, hwdownload" OUTPUT
22525 • Apply blur:
22527 -i INPUT -vf "hwupload, convolution_opencl=1 1 1 1 1 1 1 1 1:1 1 1 1 1 1 1 1 1:1 1 1 1 1 1 1 1 1:1 1 1 1 1 1 1 1 1:1/9:1/9:1/9:1/9, hwdownload" OUTPUT
22529 • Apply edge enhance:
22531 -i INPUT -vf "hwupload, convolution_opencl=0 0 0 -1 1 0 0 0 0:0 0 0 -1 1 0 0 0 0:0 0 0 -1 1 0 0 0 0:0 0 0 -1 1 0 0 0 0:5:1:1:1:0:128:128:128, hwdownload" OUTPUT
22533 • Apply edge detect:
22535 -i INPUT -vf "hwupload, convolution_opencl=0 1 0 1 -4 1 0 1 0:0 1 0 1 -4 1 0 1 0:0 1 0 1 -4 1 0 1 0:0 1 0 1 -4 1 0 1 0:5:5:5:1:0:128:128:128, hwdownload" OUTPUT
22537 • Apply laplacian edge detector which includes diagonals:
22539 -i INPUT -vf "hwupload, convolution_opencl=1 1 1 1 -8 1 1 1 1:1 1 1 1 -8 1 1 1 1:1 1 1 1 -8 1 1 1 1:1 1 1 1 -8 1 1 1 1:5:5:5:1:0:128:128:0, hwdownload" OUTPUT
22541 • Apply emboss:
22543 -i INPUT -vf "hwupload, convolution_opencl=-2 -1 0 -1 1 1 0 1 2:-2 -1 0 -1 1 1 0 1 2:-2 -1 0 -1 1 1 0 1 2:-2 -1 0 -1 1 1 0 1 2, hwdownload" OUTPUT
22545 erosion_opencl
22547 Apply erosion effect to the video.
22548 This filter replaces the pixel by the local(3x3) minimum.
22550 It accepts the following options:
22552 threshold0
22554 threshold1
22555 threshold2
22556 threshold3
22557 Limit the maximum change for each plane. Range is "[0, 65535]" and
22558 default value is 65535. If 0, plane will remain unchanged.
22559 coordinates
22561 Flag which specifies the pixel to refer to. Range is "[0, 255]"
22562 and default value is 255, i.e. all eight pixels are used.
22563 Flags to local 3x3 coordinates region centered on "x":
22565 1 2 3
22567 4 x 5
22569 6 7 8
22571 Example
22573 • Apply erosion filter with threshold0 set to 30, threshold1 set 40,
22575 threshold2 set to 50 and coordinates set to 231, setting each pixel
22576 of the output to the local minimum between pixels: 1, 2, 3, 6, 7, 8
22577 of the 3x3 region centered on it in the input. If the difference
22578 between input pixel and local minimum is more then threshold of the
22579 corresponding plane, output pixel will be set to input pixel -
22580 threshold of corresponding plane.
22581 -i INPUT -vf "hwupload, erosion_opencl=30:40:50:coordinates=231, hwdownload" OUTPUT
22583 deshake_opencl
22585 Feature-point based video stabilization filter.
22586 The filter accepts the following options:
22588 tripod
22590 Simulates a tripod by preventing any camera movement whatsoever
22591 from the original frame. Defaults to 0.
22592 debug
22594 Whether or not additional debug info should be displayed, both in
22595 the processed output and in the console.
22596 Note that in order to see console debug output you will also need
22598 to pass "-v verbose" to ffmpeg.
22599 Viewing point matches in the output video is only supported for RGB
22601 input.
22602 Defaults to 0.
22604 adaptive_crop
22606 Whether or not to do a tiny bit of cropping at the borders to cut
22607 down on the amount of mirrored pixels.
22608 Defaults to 1.
22610 refine_features
22612 Whether or not feature points should be refined at a sub-pixel
22613 level.
22614 This can be turned off for a slight performance gain at the cost of
22616 precision.
22617 Defaults to 1.
22619 smooth_strength
22621 The strength of the smoothing applied to the camera path from 0.0
22622 to 1.0.
22623 1.0 is the maximum smoothing strength while values less than that
22625 result in less smoothing.
22626 0.0 causes the filter to adaptively choose a smoothing strength on
22628 a per-frame basis.
22629 Defaults to 0.0.
22631 smooth_window_multiplier
22633 Controls the size of the smoothing window (the number of frames
22634 buffered to determine motion information from).
22635 The size of the smoothing window is determined by multiplying the
22637 framerate of the video by this number.
22638 Acceptable values range from 0.1 to 10.0.
22640 Larger values increase the amount of motion data available for
22642 determining how to smooth the camera path, potentially improving
22643 smoothness, but also increase latency and memory usage.
22644 Defaults to 2.0.
22646 Examples
22648 • Stabilize a video with a fixed, medium smoothing strength:
22650 -i INPUT -vf "hwupload, deshake_opencl=smooth_strength=0.5, hwdownload" OUTPUT
22652 • Stabilize a video with debugging (both in console and in rendered
22654 video):
22655 -i INPUT -filter_complex "[0:v]format=rgba, hwupload, deshake_opencl=debug=1, hwdownload, format=rgba, format=yuv420p" -v verbose OUTPUT
22657 dilation_opencl
22659 Apply dilation effect to the video.
22660 This filter replaces the pixel by the local(3x3) maximum.
22662 It accepts the following options:
22664 threshold0
22666 threshold1
22667 threshold2
22668 threshold3
22669 Limit the maximum change for each plane. Range is "[0, 65535]" and
22670 default value is 65535. If 0, plane will remain unchanged.
22671 coordinates
22673 Flag which specifies the pixel to refer to. Range is "[0, 255]"
22674 and default value is 255, i.e. all eight pixels are used.
22675 Flags to local 3x3 coordinates region centered on "x":
22677 1 2 3
22679 4 x 5
22681 6 7 8
22683 Example
22685 • Apply dilation filter with threshold0 set to 30, threshold1 set 40,
22687 threshold2 set to 50 and coordinates set to 231, setting each pixel
22688 of the output to the local maximum between pixels: 1, 2, 3, 6, 7, 8
22689 of the 3x3 region centered on it in the input. If the difference
22690 between input pixel and local maximum is more then threshold of the
22691 corresponding plane, output pixel will be set to input pixel +
22692 threshold of corresponding plane.
22693 -i INPUT -vf "hwupload, dilation_opencl=30:40:50:coordinates=231, hwdownload" OUTPUT
22695 nlmeans_opencl
22697 Non-local Means denoise filter through OpenCL, this filter accepts same
22698 options as nlmeans.
22699 overlay_opencl
22701 Overlay one video on top of another.
22702 It takes two inputs and has one output. The first input is the "main"
22704 video on which the second input is overlaid. This filter requires same
22705 memory layout for all the inputs. So, format conversion may be needed.
22706 The filter accepts the following options:
22708 x Set the x coordinate of the overlaid video on the main video.
22710 Default value is 0.
22711 y Set the y coordinate of the overlaid video on the main video.
22713 Default value is 0.
22714 Examples
22716 • Overlay an image LOGO at the top-left corner of the INPUT video.
22718 Both inputs are yuv420p format.
22719 -i INPUT -i LOGO -filter_complex "[0:v]hwupload[a], [1:v]format=yuv420p, hwupload[b], [a][b]overlay_opencl, hwdownload" OUTPUT
22721 • The inputs have same memory layout for color channels , the overlay
22723 has additional alpha plane, like INPUT is yuv420p, and the LOGO is
22724 yuva420p.
22725 -i INPUT -i LOGO -filter_complex "[0:v]hwupload[a], [1:v]format=yuva420p, hwupload[b], [a][b]overlay_opencl, hwdownload" OUTPUT
22727 pad_opencl
22729 Add paddings to the input image, and place the original input at the
22730 provided x, y coordinates.
22731 It accepts the following options:
22733 width, w
22735 height, h
22736 Specify an expression for the size of the output image with the
22737 paddings added. If the value for width or height is 0, the
22738 corresponding input size is used for the output.
22739 The width expression can reference the value set by the height
22741 expression, and vice versa.
22742 The default value of width and height is 0.
22744 x
22746 y Specify the offsets to place the input image at within the padded
22747 area, with respect to the top/left border of the output image.
22748 The x expression can reference the value set by the y expression,
22750 and vice versa.
22751 The default value of x and y is 0.
22753 If x or y evaluate to a negative number, they'll be changed so the
22755 input image is centered on the padded area.
22756 color
22758 Specify the color of the padded area. For the syntax of this
22759 option, check the "Color" section in the ffmpeg-utils manual.
22760 aspect
22762 Pad to an aspect instead to a resolution.
22763 The value for the width, height, x, and y options are expressions
22765 containing the following constants:
22766 in_w
22768 in_h
22769 The input video width and height.
22770 iw
22772 ih These are the same as in_w and in_h.
22773 out_w
22775 out_h
22776 The output width and height (the size of the padded area), as
22777 specified by the width and height expressions.
22778 ow
22780 oh These are the same as out_w and out_h.
22781 x
22783 y The x and y offsets as specified by the x and y expressions, or NAN
22784 if not yet specified.
22785 a same as iw / ih
22787 sar input sample aspect ratio
22789 dar input display aspect ratio, it is the same as (iw / ih) * sar
22791 prewitt_opencl
22793 Apply the Prewitt operator
22794 (<https://en.wikipedia.org/wiki/Prewitt_operator>) to input video
22795 stream.
22796 The filter accepts the following option:
22798 planes
22800 Set which planes to filter. Default value is 0xf, by which all
22801 planes are processed.
22802 scale
22804 Set value which will be multiplied with filtered result. Range is
22805 "[0.0, 65535]" and default value is 1.0.
22806 delta
22808 Set value which will be added to filtered result. Range is
22809 "[-65535, 65535]" and default value is 0.0.
22810 Example
22812 • Apply the Prewitt operator with scale set to 2 and delta set to 10.
22814 -i INPUT -vf "hwupload, prewitt_opencl=scale=2:delta=10, hwdownload" OUTPUT
22816 program_opencl
22818 Filter video using an OpenCL program.
22819 source
22821 OpenCL program source file.
22822 kernel
22824 Kernel name in program.
22825 inputs
22827 Number of inputs to the filter. Defaults to 1.
22828 size, s
22830 Size of output frames. Defaults to the same as the first input.
22831 The "program_opencl" filter also supports the framesync options.
22833 The program source file must contain a kernel function with the given
22835 name, which will be run once for each plane of the output. Each run on
22836 a plane gets enqueued as a separate 2D global NDRange with one work-
22837 item for each pixel to be generated. The global ID offset for each
22838 work-item is therefore the coordinates of a pixel in the destination
22839 image.
22840 The kernel function needs to take the following arguments:
22842 • Destination image, __write_only image2d_t.
22844 This image will become the output; the kernel should write all of
22846 it.
22847 • Frame index, unsigned int.
22849 This is a counter starting from zero and increasing by one for each
22851 frame.
22852 • Source images, __read_only image2d_t.
22854 These are the most recent images on each input. The kernel may
22856 read from them to generate the output, but they can't be written
22857 to.
22858 Example programs:
22860 • Copy the input to the output (output must be the same size as the
22862 input).
22863 __kernel void copy(__write_only image2d_t destination,
22865 unsigned int index,
22866 __read_only image2d_t source)
22867 {
22868 const sampler_t sampler = CLK_NORMALIZED_COORDS_FALSE;
22869 int2 location = (int2)(get_global_id(0), get_global_id(1));
22871 float4 value = read_imagef(source, sampler, location);
22873 write_imagef(destination, location, value);
22875 }
22876 • Apply a simple transformation, rotating the input by an amount
22878 increasing with the index counter. Pixel values are linearly
22879 interpolated by the sampler, and the output need not have the same
22880 dimensions as the input.
22881 __kernel void rotate_image(__write_only image2d_t dst,
22883 unsigned int index,
22884 __read_only image2d_t src)
22885 {
22886 const sampler_t sampler = (CLK_NORMALIZED_COORDS_FALSE |
22887 CLK_FILTER_LINEAR);
22888 float angle = (float)index / 100.0f;
22890 float2 dst_dim = convert_float2(get_image_dim(dst));
22892 float2 src_dim = convert_float2(get_image_dim(src));
22893 float2 dst_cen = dst_dim / 2.0f;
22895 float2 src_cen = src_dim / 2.0f;
22896 int2 dst_loc = (int2)(get_global_id(0), get_global_id(1));
22898 float2 dst_pos = convert_float2(dst_loc) - dst_cen;
22900 float2 src_pos = {
22901 cos(angle) * dst_pos.x - sin(angle) * dst_pos.y,
22902 sin(angle) * dst_pos.x + cos(angle) * dst_pos.y
22903 };
22904 src_pos = src_pos * src_dim / dst_dim;
22905 float2 src_loc = src_pos + src_cen;
22907 if (src_loc.x < 0.0f || src_loc.y < 0.0f ||
22909 src_loc.x > src_dim.x || src_loc.y > src_dim.y)
22910 write_imagef(dst, dst_loc, 0.5f);
22911 else
22912 write_imagef(dst, dst_loc, read_imagef(src, sampler, src_loc));
22913 }
22914 • Blend two inputs together, with the amount of each input used
22916 varying with the index counter.
22917 __kernel void blend_images(__write_only image2d_t dst,
22919 unsigned int index,
22920 __read_only image2d_t src1,
22921 __read_only image2d_t src2)
22922 {
22923 const sampler_t sampler = (CLK_NORMALIZED_COORDS_FALSE |
22924 CLK_FILTER_LINEAR);
22925 float blend = (cos((float)index / 50.0f) + 1.0f) / 2.0f;
22927 int2 dst_loc = (int2)(get_global_id(0), get_global_id(1));
22929 int2 src1_loc = dst_loc * get_image_dim(src1) / get_image_dim(dst);
22930 int2 src2_loc = dst_loc * get_image_dim(src2) / get_image_dim(dst);
22931 float4 val1 = read_imagef(src1, sampler, src1_loc);
22933 float4 val2 = read_imagef(src2, sampler, src2_loc);
22934 write_imagef(dst, dst_loc, val1 * blend + val2 * (1.0f - blend));
22936 }
22937 remap_opencl
22939 Remap pixels using 2nd: Xmap and 3rd: Ymap input video stream.
22940 Destination pixel at position (X, Y) will be picked from source (x, y)
22942 position where x = Xmap(X, Y) and y = Ymap(X, Y). If mapping values are
22943 out of range, zero value for pixel will be used for destination pixel.
22944 Xmap and Ymap input video streams must be of same dimensions. Output
22946 video stream will have Xmap/Ymap video stream dimensions. Xmap and
22947 Ymap input video streams are 32bit float pixel format, single channel.
22948 interp
22950 Specify interpolation used for remapping of pixels. Allowed values
22951 are "near" and "linear". Default value is "linear".
22952 fill
22954 Specify the color of the unmapped pixels. For the syntax of this
22955 option, check the "Color" section in the ffmpeg-utils manual.
22956 Default color is "black".
22957 roberts_opencl
22959 Apply the Roberts cross operator
22960 (<https://en.wikipedia.org/wiki/Roberts_cross>) to input video stream.
22961 The filter accepts the following option:
22963 planes
22965 Set which planes to filter. Default value is 0xf, by which all
22966 planes are processed.
22967 scale
22969 Set value which will be multiplied with filtered result. Range is
22970 "[0.0, 65535]" and default value is 1.0.
22971 delta
22973 Set value which will be added to filtered result. Range is
22974 "[-65535, 65535]" and default value is 0.0.
22975 Example
22977 • Apply the Roberts cross operator with scale set to 2 and delta set
22979 to 10
22980 -i INPUT -vf "hwupload, roberts_opencl=scale=2:delta=10, hwdownload" OUTPUT
22982 sobel_opencl
22984 Apply the Sobel operator
22985 (<https://en.wikipedia.org/wiki/Sobel_operator>) to input video stream.
22986 The filter accepts the following option:
22988 planes
22990 Set which planes to filter. Default value is 0xf, by which all
22991 planes are processed.
22992 scale
22994 Set value which will be multiplied with filtered result. Range is
22995 "[0.0, 65535]" and default value is 1.0.
22996 delta
22998 Set value which will be added to filtered result. Range is
22999 "[-65535, 65535]" and default value is 0.0.
23000 Example
23002 • Apply sobel operator with scale set to 2 and delta set to 10
23004 -i INPUT -vf "hwupload, sobel_opencl=scale=2:delta=10, hwdownload" OUTPUT
23006 tonemap_opencl
23008 Perform HDR(PQ/HLG) to SDR conversion with tone-mapping.
23009 It accepts the following parameters:
23011 tonemap
23013 Specify the tone-mapping operator to be used. Same as tonemap
23014 option in tonemap.
23015 param
23017 Tune the tone mapping algorithm. same as param option in tonemap.
23018 desat
23020 Apply desaturation for highlights that exceed this level of
23021 brightness. The higher the parameter, the more color information
23022 will be preserved. This setting helps prevent unnaturally blown-out
23023 colors for super-highlights, by (smoothly) turning into white
23024 instead. This makes images feel more natural, at the cost of
23025 reducing information about out-of-range colors.
23026 The default value is 0.5, and the algorithm here is a little
23028 different from the cpu version tonemap currently. A setting of 0.0
23029 disables this option.
23030 threshold
23032 The tonemapping algorithm parameters is fine-tuned per each scene.
23033 And a threshold is used to detect whether the scene has changed or
23034 not. If the distance between the current frame average brightness
23035 and the current running average exceeds a threshold value, we would
23036 re-calculate scene average and peak brightness. The default value
23037 is 0.2.
23038 format
23040 Specify the output pixel format.
23041 Currently supported formats are:
23043 p010
23045 nv12
23046 range, r
23047 Set the output color range.
23048 Possible values are:
23050 tv/mpeg
23052 pc/jpeg
23053 Default is same as input.
23055 primaries, p
23057 Set the output color primaries.
23058 Possible values are:
23060 bt709
23062 bt2020
23063 Default is same as input.
23065 transfer, t
23067 Set the output transfer characteristics.
23068 Possible values are:
23070 bt709
23072 bt2020
23073 Default is bt709.
23075 matrix, m
23077 Set the output colorspace matrix.
23078 Possible value are:
23080 bt709
23082 bt2020
23083 Default is same as input.
23085 Example
23087 • Convert HDR(PQ/HLG) video to bt2020-transfer-characteristic p010
23089 format using linear operator.
23090 -i INPUT -vf "format=p010,hwupload,tonemap_opencl=t=bt2020:tonemap=linear:format=p010,hwdownload,format=p010" OUTPUT
23092 unsharp_opencl
23094 Sharpen or blur the input video.
23095 It accepts the following parameters:
23097 luma_msize_x, lx
23099 Set the luma matrix horizontal size. Range is "[1, 23]" and
23100 default value is 5.
23101 luma_msize_y, ly
23103 Set the luma matrix vertical size. Range is "[1, 23]" and default
23104 value is 5.
23105 luma_amount, la
23107 Set the luma effect strength. Range is "[-10, 10]" and default
23108 value is 1.0.
23109 Negative values will blur the input video, while positive values
23111 will sharpen it, a value of zero will disable the effect.
23112 chroma_msize_x, cx
23114 Set the chroma matrix horizontal size. Range is "[1, 23]" and
23115 default value is 5.
23116 chroma_msize_y, cy
23118 Set the chroma matrix vertical size. Range is "[1, 23]" and
23119 default value is 5.
23120 chroma_amount, ca
23122 Set the chroma effect strength. Range is "[-10, 10]" and default
23123 value is 0.0.
23124 Negative values will blur the input video, while positive values
23126 will sharpen it, a value of zero will disable the effect.
23127 All parameters are optional and default to the equivalent of the string
23129 '5:5:1.0:5:5:0.0'.
23130 Examples
23132 • Apply strong luma sharpen effect:
23134 -i INPUT -vf "hwupload, unsharp_opencl=luma_msize_x=7:luma_msize_y=7:luma_amount=2.5, hwdownload" OUTPUT
23136 • Apply a strong blur of both luma and chroma parameters:
23138 -i INPUT -vf "hwupload, unsharp_opencl=7:7:-2:7:7:-2, hwdownload" OUTPUT
23140 xfade_opencl
23142 Cross fade two videos with custom transition effect by using OpenCL.
23143 It accepts the following options:
23145 transition
23147 Set one of possible transition effects.
23148 custom
23150 Select custom transition effect, the actual transition
23151 description will be picked from source and kernel options.
23152 fade
23154 wipeleft
23155 wiperight
23156 wipeup
23157 wipedown
23158 slideleft
23159 slideright
23160 slideup
23161 slidedown
23162 Default transition is fade.
23163 source
23165 OpenCL program source file for custom transition.
23166 kernel
23168 Set name of kernel to use for custom transition from program source
23169 file.
23170 duration
23172 Set duration of video transition.
23173 offset
23175 Set time of start of transition relative to first video.
23176 The program source file must contain a kernel function with the given
23178 name, which will be run once for each plane of the output. Each run on
23179 a plane gets enqueued as a separate 2D global NDRange with one work-
23180 item for each pixel to be generated. The global ID offset for each
23181 work-item is therefore the coordinates of a pixel in the destination
23182 image.
23183 The kernel function needs to take the following arguments:
23185 • Destination image, __write_only image2d_t.
23187 This image will become the output; the kernel should write all of
23189 it.
23190 • First Source image, __read_only image2d_t. Second Source image,
23192 __read_only image2d_t.
23193 These are the most recent images on each input. The kernel may
23195 read from them to generate the output, but they can't be written
23196 to.
23197 • Transition progress, float. This value is always between 0 and 1
23199 inclusive.
23200 Example programs:
23202 • Apply dots curtain transition effect:
23204 __kernel void blend_images(__write_only image2d_t dst,
23206 __read_only image2d_t src1,
23207 __read_only image2d_t src2,
23208 float progress)
23209 {
23210 const sampler_t sampler = (CLK_NORMALIZED_COORDS_FALSE |
23211 CLK_FILTER_LINEAR);
23212 int2 p = (int2)(get_global_id(0), get_global_id(1));
23213 float2 rp = (float2)(get_global_id(0), get_global_id(1));
23214 float2 dim = (float2)(get_image_dim(src1).x, get_image_dim(src1).y);
23215 rp = rp / dim;
23216 float2 dots = (float2)(20.0, 20.0);
23218 float2 center = (float2)(0,0);
23219 float2 unused;
23220 float4 val1 = read_imagef(src1, sampler, p);
23222 float4 val2 = read_imagef(src2, sampler, p);
23223 bool next = distance(fract(rp * dots, &unused), (float2)(0.5, 0.5)) < (progress / distance(rp, center));
23224 write_imagef(dst, p, next ? val1 : val2);
23226 }
23227
23229 VAAPI Video filters are usually used with VAAPI decoder and VAAPI
23230 encoder. Below is a description of VAAPI video filters.
23231 To enable compilation of these filters you need to configure FFmpeg
23233 with "--enable-vaapi".
23234 To use vaapi filters, you need to setup the vaapi device correctly. For
23236 more information, please read
23237 <https://trac.ffmpeg.org/wiki/Hardware/VAAPI>
23238 overlay_vaapi
23240 Overlay one video on the top of another.
23241 It takes two inputs and has one output. The first input is the "main"
23243 video on which the second input is overlaid. This filter requires same
23244 memory layout for all the inputs. So, format conversion may be needed.
23245 The filter accepts the following options:
23247 x Set the x coordinate of the overlaid video on the main video.
23249 Default value is 0.
23250 y Set the y coordinate of the overlaid video on the main video.
23252 Default value is 0.
23253 w Set the width of the overlaid video on the main video. Default
23255 value is the width of input overlay video.
23256 h Set the height of the overlaid video on the main video. Default
23258 value is the height of input overlay video.
23259 alpha
23261 Set blocking detection thresholds. Allowed range is 0.0 to 1.0, it
23262 requires an input video with alpha channel. Default value is 0.0.
23263 Examples
23265 • Overlay an image LOGO at the top-left corner of the INPUT video.
23267 Both inputs for this filter are yuv420p format.
23268 -i INPUT -i LOGO -filter_complex "[0:v]hwupload[a], [1:v]format=yuv420p, hwupload[b], [a][b]overlay_vaapi" OUTPUT
23270 • Overlay an image LOGO at the offset (200, 100) from the top-left
23272 corner of the INPUT video. The inputs have same memory layout for
23273 color channels, the overlay has additional alpha plane, like INPUT
23274 is yuv420p, and the LOGO is yuva420p.
23275 -i INPUT -i LOGO -filter_complex "[0:v]hwupload[a], [1:v]format=yuva420p, hwupload[b], [a][b]overlay_vaapi=x=200:y=100:w=400:h=300:alpha=1.0, hwdownload, format=nv12" OUTPUT
23277 tonemap_vaapi
23279 Perform HDR(High Dynamic Range) to SDR(Standard Dynamic Range)
23280 conversion with tone-mapping. It maps the dynamic range of HDR10
23281 content to the SDR content. It currently only accepts HDR10 as input.
23282 It accepts the following parameters:
23284 format
23286 Specify the output pixel format.
23287 Currently supported formats are:
23289 p010
23291 nv12
23292 Default is nv12.
23294 primaries, p
23296 Set the output color primaries.
23297 Default is same as input.
23299 transfer, t
23301 Set the output transfer characteristics.
23302 Default is bt709.
23304 matrix, m
23306 Set the output colorspace matrix.
23307 Default is same as input.
23309 Example
23311 • Convert HDR(HDR10) video to bt2020-transfer-characteristic p010
23313 format
23314 tonemap_vaapi=format=p010:t=bt2020-10
23316
23318 Below is a description of the currently available video sources.
23319 buffer
23321 Buffer video frames, and make them available to the filter chain.
23322 This source is mainly intended for a programmatic use, in particular
23324 through the interface defined in libavfilter/buffersrc.h.
23325 It accepts the following parameters:
23327 video_size
23329 Specify the size (width and height) of the buffered video frames.
23330 For the syntax of this option, check the "Video size" section in
23331 the ffmpeg-utils manual.
23332 width
23334 The input video width.
23335 height
23337 The input video height.
23338 pix_fmt
23340 A string representing the pixel format of the buffered video
23341 frames. It may be a number corresponding to a pixel format, or a
23342 pixel format name.
23343 time_base
23345 Specify the timebase assumed by the timestamps of the buffered
23346 frames.
23347 frame_rate
23349 Specify the frame rate expected for the video stream.
23350 pixel_aspect, sar
23352 The sample (pixel) aspect ratio of the input video.
23353 sws_param
23355 This option is deprecated and ignored. Prepend "sws_flags=flags;"
23356 to the filtergraph description to specify swscale flags for
23357 automatically inserted scalers. See Filtergraph syntax.
23358 hw_frames_ctx
23360 When using a hardware pixel format, this should be a reference to
23361 an AVHWFramesContext describing input frames.
23362 For example:
23364 buffer=width=320:height=240:pix_fmt=yuv410p:time_base=1/24:sar=1
23366 will instruct the source to accept video frames with size 320x240 and
23368 with format "yuv410p", assuming 1/24 as the timestamps timebase and
23369 square pixels (1:1 sample aspect ratio). Since the pixel format with
23370 name "yuv410p" corresponds to the number 6 (check the enum
23371 AVPixelFormat definition in libavutil/pixfmt.h), this example
23372 corresponds to:
23373 buffer=size=320x240:pixfmt=6:time_base=1/24:pixel_aspect=1/1
23375 Alternatively, the options can be specified as a flat string, but this
23377 syntax is deprecated:
23378 width:height:pix_fmt:time_base.num:time_base.den:pixel_aspect.num:pixel_aspect.den
23380 cellauto
23382 Create a pattern generated by an elementary cellular automaton.
23383 The initial state of the cellular automaton can be defined through the
23385 filename and pattern options. If such options are not specified an
23386 initial state is created randomly.
23387 At each new frame a new row in the video is filled with the result of
23389 the cellular automaton next generation. The behavior when the whole
23390 frame is filled is defined by the scroll option.
23391 This source accepts the following options:
23393 filename, f
23395 Read the initial cellular automaton state, i.e. the starting row,
23396 from the specified file. In the file, each non-whitespace
23397 character is considered an alive cell, a newline will terminate the
23398 row, and further characters in the file will be ignored.
23399 pattern, p
23401 Read the initial cellular automaton state, i.e. the starting row,
23402 from the specified string.
23403 Each non-whitespace character in the string is considered an alive
23405 cell, a newline will terminate the row, and further characters in
23406 the string will be ignored.
23407 rate, r
23409 Set the video rate, that is the number of frames generated per
23410 second. Default is 25.
23411 random_fill_ratio, ratio
23413 Set the random fill ratio for the initial cellular automaton row.
23414 It is a floating point number value ranging from 0 to 1, defaults
23415 to 1/PHI.
23416 This option is ignored when a file or a pattern is specified.
23418 random_seed, seed
23420 Set the seed for filling randomly the initial row, must be an
23421 integer included between 0 and UINT32_MAX. If not specified, or if
23422 explicitly set to -1, the filter will try to use a good random seed
23423 on a best effort basis.
23424 rule
23426 Set the cellular automaton rule, it is a number ranging from 0 to
23427 255. Default value is 110.
23428 size, s
23430 Set the size of the output video. For the syntax of this option,
23431 check the "Video size" section in the ffmpeg-utils manual.
23432 If filename or pattern is specified, the size is set by default to
23434 the width of the specified initial state row, and the height is set
23435 to width * PHI.
23436 If size is set, it must contain the width of the specified pattern
23438 string, and the specified pattern will be centered in the larger
23439 row.
23440 If a filename or a pattern string is not specified, the size value
23442 defaults to "320x518" (used for a randomly generated initial
23443 state).
23444 scroll
23446 If set to 1, scroll the output upward when all the rows in the
23447 output have been already filled. If set to 0, the new generated row
23448 will be written over the top row just after the bottom row is
23449 filled. Defaults to 1.
23450 start_full, full
23452 If set to 1, completely fill the output with generated rows before
23453 outputting the first frame. This is the default behavior, for
23454 disabling set the value to 0.
23455 stitch
23457 If set to 1, stitch the left and right row edges together. This is
23458 the default behavior, for disabling set the value to 0.
23459 Examples
23461 • Read the initial state from pattern, and specify an output of size
23463 200x400.
23464 cellauto=f=pattern:s=200x400
23466 • Generate a random initial row with a width of 200 cells, with a
23468 fill ratio of 2/3:
23469 cellauto=ratio=2/3:s=200x200
23471 • Create a pattern generated by rule 18 starting by a single alive
23473 cell centered on an initial row with width 100:
23474 cellauto=p=@s=100x400:full=0:rule=18
23476 • Specify a more elaborated initial pattern:
23478 cellauto=p='@@ @ @@':s=100x400:full=0:rule=18
23480 coreimagesrc
23482 Video source generated on GPU using Apple's CoreImage API on OSX.
23483 This video source is a specialized version of the coreimage video
23485 filter. Use a core image generator at the beginning of the applied
23486 filterchain to generate the content.
23487 The coreimagesrc video source accepts the following options:
23489 list_generators
23491 List all available generators along with all their respective
23492 options as well as possible minimum and maximum values along with
23493 the default values.
23494 list_generators=true
23496 size, s
23498 Specify the size of the sourced video. For the syntax of this
23499 option, check the "Video size" section in the ffmpeg-utils manual.
23500 The default value is "320x240".
23501 rate, r
23503 Specify the frame rate of the sourced video, as the number of
23504 frames generated per second. It has to be a string in the format
23505 frame_rate_num/frame_rate_den, an integer number, a floating point
23506 number or a valid video frame rate abbreviation. The default value
23507 is "25".
23508 sar Set the sample aspect ratio of the sourced video.
23510 duration, d
23512 Set the duration of the sourced video. See the Time duration
23513 section in the ffmpeg-utils(1) manual for the accepted syntax.
23514 If not specified, or the expressed duration is negative, the video
23516 is supposed to be generated forever.
23517 Additionally, all options of the coreimage video filter are accepted.
23519 A complete filterchain can be used for further processing of the
23520 generated input without CPU-HOST transfer. See coreimage documentation
23521 and examples for details.
23522 Examples
23524 • Use CIQRCodeGenerator to create a QR code for the FFmpeg homepage,
23526 given as complete and escaped command-line for Apple's standard
23527 bash shell:
23528 ffmpeg -f lavfi -i coreimagesrc=s=100x100:filter=CIQRCodeGenerator@inputMessage=https\\\\\://FFmpeg.org/@inputCorrectionLevel=H -frames:v 1 QRCode.png
23530 This example is equivalent to the QRCode example of coreimage
23532 without the need for a nullsrc video source.
23533 gradients
23535 Generate several gradients.
23536 size, s
23538 Set frame size. For the syntax of this option, check the "Video
23539 size" section in the ffmpeg-utils manual. Default value is
23540 "640x480".
23541 rate, r
23543 Set frame rate, expressed as number of frames per second. Default
23544 value is "25".
23545 c0, c1, c2, c3, c4, c5, c6, c7
23547 Set 8 colors. Default values for colors is to pick random one.
23548 x0, y0, y0, y1
23550 Set gradient line source and destination points. If negative or out
23551 of range, random ones are picked.
23552 nb_colors, n
23554 Set number of colors to use at once. Allowed range is from 2 to 8.
23555 Default value is 2.
23556 seed
23558 Set seed for picking gradient line points.
23559 duration, d
23561 Set the duration of the sourced video. See the Time duration
23562 section in the ffmpeg-utils(1) manual for the accepted syntax.
23563 If not specified, or the expressed duration is negative, the video
23565 is supposed to be generated forever.
23566 speed
23568 Set speed of gradients rotation.
23569 type, t
23571 Set type of gradients, can be "linear" or "radial" or "circular" or
23572 "spiral".
23573 mandelbrot
23575 Generate a Mandelbrot set fractal, and progressively zoom towards the
23576 point specified with start_x and start_y.
23577 This source accepts the following options:
23579 end_pts
23581 Set the terminal pts value. Default value is 400.
23582 end_scale
23584 Set the terminal scale value. Must be a floating point value.
23585 Default value is 0.3.
23586 inner
23588 Set the inner coloring mode, that is the algorithm used to draw the
23589 Mandelbrot fractal internal region.
23590 It shall assume one of the following values:
23592 black
23594 Set black mode.
23595 convergence
23597 Show time until convergence.
23598 mincol
23600 Set color based on point closest to the origin of the
23601 iterations.
23602 period
23604 Set period mode.
23605 Default value is mincol.
23607 bailout
23609 Set the bailout value. Default value is 10.0.
23610 maxiter
23612 Set the maximum of iterations performed by the rendering algorithm.
23613 Default value is 7189.
23614 outer
23616 Set outer coloring mode. It shall assume one of following values:
23617 iteration_count
23619 Set iteration count mode.
23620 normalized_iteration_count
23622 set normalized iteration count mode.
23623 Default value is normalized_iteration_count.
23625 rate, r
23627 Set frame rate, expressed as number of frames per second. Default
23628 value is "25".
23629 size, s
23631 Set frame size. For the syntax of this option, check the "Video
23632 size" section in the ffmpeg-utils manual. Default value is
23633 "640x480".
23634 start_scale
23636 Set the initial scale value. Default value is 3.0.
23637 start_x
23639 Set the initial x position. Must be a floating point value between
23640 -100 and 100. Default value is
23641 -0.743643887037158704752191506114774.
23642 start_y
23644 Set the initial y position. Must be a floating point value between
23645 -100 and 100. Default value is
23646 -0.131825904205311970493132056385139.
23647 mptestsrc
23649 Generate various test patterns, as generated by the MPlayer test
23650 filter.
23651 The size of the generated video is fixed, and is 256x256. This source
23653 is useful in particular for testing encoding features.
23654 This source accepts the following options:
23656 rate, r
23658 Specify the frame rate of the sourced video, as the number of
23659 frames generated per second. It has to be a string in the format
23660 frame_rate_num/frame_rate_den, an integer number, a floating point
23661 number or a valid video frame rate abbreviation. The default value
23662 is "25".
23663 duration, d
23665 Set the duration of the sourced video. See the Time duration
23666 section in the ffmpeg-utils(1) manual for the accepted syntax.
23667 If not specified, or the expressed duration is negative, the video
23669 is supposed to be generated forever.
23670 test, t
23672 Set the number or the name of the test to perform. Supported tests
23673 are:
23674 dc_luma
23676 dc_chroma
23677 freq_luma
23678 freq_chroma
23679 amp_luma
23680 amp_chroma
23681 cbp
23682 mv
23683 ring1
23684 ring2
23685 all
23686 max_frames, m
23687 Set the maximum number of frames generated for each test,
23688 default value is 30.
23689 Default value is "all", which will cycle through the list of all
23691 tests.
23692 Some examples:
23694 mptestsrc=t=dc_luma
23696 will generate a "dc_luma" test pattern.
23698 frei0r_src
23700 Provide a frei0r source.
23701 To enable compilation of this filter you need to install the frei0r
23703 header and configure FFmpeg with "--enable-frei0r".
23704 This source accepts the following parameters:
23706 size
23708 The size of the video to generate. For the syntax of this option,
23709 check the "Video size" section in the ffmpeg-utils manual.
23710 framerate
23712 The framerate of the generated video. It may be a string of the
23713 form num/den or a frame rate abbreviation.
23714 filter_name
23716 The name to the frei0r source to load. For more information
23717 regarding frei0r and how to set the parameters, read the frei0r
23718 section in the video filters documentation.
23719 filter_params
23721 A '|'-separated list of parameters to pass to the frei0r source.
23722 For example, to generate a frei0r partik0l source with size 200x200 and
23724 frame rate 10 which is overlaid on the overlay filter main input:
23725 frei0r_src=size=200x200:framerate=10:filter_name=partik0l:filter_params=1234 [overlay]; [in][overlay] overlay
23727 life
23729 Generate a life pattern.
23730 This source is based on a generalization of John Conway's life game.
23732 The sourced input represents a life grid, each pixel represents a cell
23734 which can be in one of two possible states, alive or dead. Every cell
23735 interacts with its eight neighbours, which are the cells that are
23736 horizontally, vertically, or diagonally adjacent.
23737 At each interaction the grid evolves according to the adopted rule,
23739 which specifies the number of neighbor alive cells which will make a
23740 cell stay alive or born. The rule option allows one to specify the rule
23741 to adopt.
23742 This source accepts the following options:
23744 filename, f
23746 Set the file from which to read the initial grid state. In the
23747 file, each non-whitespace character is considered an alive cell,
23748 and newline is used to delimit the end of each row.
23749 If this option is not specified, the initial grid is generated
23751 randomly.
23752 rate, r
23754 Set the video rate, that is the number of frames generated per
23755 second. Default is 25.
23756 random_fill_ratio, ratio
23758 Set the random fill ratio for the initial random grid. It is a
23759 floating point number value ranging from 0 to 1, defaults to 1/PHI.
23760 It is ignored when a file is specified.
23761 random_seed, seed
23763 Set the seed for filling the initial random grid, must be an
23764 integer included between 0 and UINT32_MAX. If not specified, or if
23765 explicitly set to -1, the filter will try to use a good random seed
23766 on a best effort basis.
23767 rule
23769 Set the life rule.
23770 A rule can be specified with a code of the kind "SNS/BNB", where NS
23772 and NB are sequences of numbers in the range 0-8, NS specifies the
23773 number of alive neighbor cells which make a live cell stay alive,
23774 and NB the number of alive neighbor cells which make a dead cell to
23775 become alive (i.e. to "born"). "s" and "b" can be used in place of
23776 "S" and "B", respectively.
23777 Alternatively a rule can be specified by an 18-bits integer. The 9
23779 high order bits are used to encode the next cell state if it is
23780 alive for each number of neighbor alive cells, the low order bits
23781 specify the rule for "borning" new cells. Higher order bits encode
23782 for an higher number of neighbor cells. For example the number
23783 6153 = "(12<<9)+9" specifies a stay alive rule of 12 and a born
23784 rule of 9, which corresponds to "S23/B03".
23785 Default value is "S23/B3", which is the original Conway's game of
23787 life rule, and will keep a cell alive if it has 2 or 3 neighbor
23788 alive cells, and will born a new cell if there are three alive
23789 cells around a dead cell.
23790 size, s
23792 Set the size of the output video. For the syntax of this option,
23793 check the "Video size" section in the ffmpeg-utils manual.
23794 If filename is specified, the size is set by default to the same
23796 size of the input file. If size is set, it must contain the size
23797 specified in the input file, and the initial grid defined in that
23798 file is centered in the larger resulting area.
23799 If a filename is not specified, the size value defaults to
23801 "320x240" (used for a randomly generated initial grid).
23802 stitch
23804 If set to 1, stitch the left and right grid edges together, and the
23805 top and bottom edges also. Defaults to 1.
23806 mold
23808 Set cell mold speed. If set, a dead cell will go from death_color
23809 to mold_color with a step of mold. mold can have a value from 0 to
23810 255.
23811 life_color
23813 Set the color of living (or new born) cells.
23814 death_color
23816 Set the color of dead cells. If mold is set, this is the first
23817 color used to represent a dead cell.
23818 mold_color
23820 Set mold color, for definitely dead and moldy cells.
23821 For the syntax of these 3 color options, check the "Color" section
23823 in the ffmpeg-utils manual.
23824 Examples
23826 • Read a grid from pattern, and center it on a grid of size 300x300
23828 pixels:
23829 life=f=pattern:s=300x300
23831 • Generate a random grid of size 200x200, with a fill ratio of 2/3:
23833 life=ratio=2/3:s=200x200
23835 • Specify a custom rule for evolving a randomly generated grid:
23837 life=rule=S14/B34
23839 • Full example with slow death effect (mold) using ffplay:
23841 ffplay -f lavfi life=s=300x200:mold=10:r=60:ratio=0.1:death_color=#C83232:life_color=#00ff00,scale=1200:800:flags=16
23843 allrgb, allyuv, color, colorchart, colorspectrum, haldclutsrc, nullsrc,
23845 pal75bars, pal100bars, rgbtestsrc, smptebars, smptehdbars, testsrc,
23846 testsrc2, yuvtestsrc
23847 The "allrgb" source returns frames of size 4096x4096 of all rgb colors.
23848 The "allyuv" source returns frames of size 4096x4096 of all yuv colors.
23850 The "color" source provides an uniformly colored input.
23852 The "colorchart" source provides a colors checker chart.
23854 The "colorspectrum" source provides a color spectrum input.
23856 The "haldclutsrc" source provides an identity Hald CLUT. See also
23858 haldclut filter.
23859 The "nullsrc" source returns unprocessed video frames. It is mainly
23861 useful to be employed in analysis / debugging tools, or as the source
23862 for filters which ignore the input data.
23863 The "pal75bars" source generates a color bars pattern, based on EBU PAL
23865 recommendations with 75% color levels.
23866 The "pal100bars" source generates a color bars pattern, based on EBU
23868 PAL recommendations with 100% color levels.
23869 The "rgbtestsrc" source generates an RGB test pattern useful for
23871 detecting RGB vs BGR issues. You should see a red, green and blue
23872 stripe from top to bottom.
23873 The "smptebars" source generates a color bars pattern, based on the
23875 SMPTE Engineering Guideline EG 1-1990.
23876 The "smptehdbars" source generates a color bars pattern, based on the
23878 SMPTE RP 219-2002.
23879 The "testsrc" source generates a test video pattern, showing a color
23881 pattern, a scrolling gradient and a timestamp. This is mainly intended
23882 for testing purposes.
23883 The "testsrc2" source is similar to testsrc, but supports more pixel
23885 formats instead of just "rgb24". This allows using it as an input for
23886 other tests without requiring a format conversion.
23887 The "yuvtestsrc" source generates an YUV test pattern. You should see a
23889 y, cb and cr stripe from top to bottom.
23890 The sources accept the following parameters:
23892 level
23894 Specify the level of the Hald CLUT, only available in the
23895 "haldclutsrc" source. A level of "N" generates a picture of "N*N*N"
23896 by "N*N*N" pixels to be used as identity matrix for 3D lookup
23897 tables. Each component is coded on a "1/(N*N)" scale.
23898 color, c
23900 Specify the color of the source, only available in the "color"
23901 source. For the syntax of this option, check the "Color" section in
23902 the ffmpeg-utils manual.
23903 size, s
23905 Specify the size of the sourced video. For the syntax of this
23906 option, check the "Video size" section in the ffmpeg-utils manual.
23907 The default value is "320x240".
23908 This option is not available with the "allrgb", "allyuv", and
23910 "haldclutsrc" filters.
23911 rate, r
23913 Specify the frame rate of the sourced video, as the number of
23914 frames generated per second. It has to be a string in the format
23915 frame_rate_num/frame_rate_den, an integer number, a floating point
23916 number or a valid video frame rate abbreviation. The default value
23917 is "25".
23918 duration, d
23920 Set the duration of the sourced video. See the Time duration
23921 section in the ffmpeg-utils(1) manual for the accepted syntax.
23922 If not specified, or the expressed duration is negative, the video
23924 is supposed to be generated forever.
23925 Since the frame rate is used as time base, all frames including the
23927 last one will have their full duration. If the specified duration
23928 is not a multiple of the frame duration, it will be rounded up.
23929 sar Set the sample aspect ratio of the sourced video.
23931 alpha
23933 Specify the alpha (opacity) of the background, only available in
23934 the "testsrc2" source. The value must be between 0 (fully
23935 transparent) and 255 (fully opaque, the default).
23936 decimals, n
23938 Set the number of decimals to show in the timestamp, only available
23939 in the "testsrc" source.
23940 The displayed timestamp value will correspond to the original
23942 timestamp value multiplied by the power of 10 of the specified
23943 value. Default value is 0.
23944 type
23946 Set the type of the color spectrum, only available in the
23947 "colorspectrum" source. Can be one of the following:
23948 black
23950 white
23951 all
23952 patch_size
23953 Set patch size of single color patch, only available in the
23954 "colorchart" source. Default is "64x64".
23955 preset
23957 Set colorchecker colors preset, only available in the "colorchart"
23958 source.
23959 Available values are:
23961 reference
23963 skintones
23964 Default value is "reference".
23966 Examples
23968 • Generate a video with a duration of 5.3 seconds, with size 176x144
23970 and a frame rate of 10 frames per second:
23971 testsrc=duration=5.3:size=qcif:rate=10
23973 • The following graph description will generate a red source with an
23975 opacity of 0.2, with size "qcif" and a frame rate of 10 frames per
23976 second:
23977 color=c=red@0.2:s=qcif:r=10
23979 • If the input content is to be ignored, "nullsrc" can be used. The
23981 following command generates noise in the luminance plane by
23982 employing the "geq" filter:
23983 nullsrc=s=256x256, geq=random(1)*255:128:128
23985 Commands
23987 The "color" source supports the following commands:
23989 c, color
23991 Set the color of the created image. Accepts the same syntax of the
23992 corresponding color option.
23993 openclsrc
23995 Generate video using an OpenCL program.
23996 source
23998 OpenCL program source file.
23999 kernel
24001 Kernel name in program.
24002 size, s
24004 Size of frames to generate. This must be set.
24005 format
24007 Pixel format to use for the generated frames. This must be set.
24008 rate, r
24010 Number of frames generated every second. Default value is '25'.
24011 For details of how the program loading works, see the program_opencl
24013 filter.
24014 Example programs:
24016 • Generate a colour ramp by setting pixel values from the position of
24018 the pixel in the output image. (Note that this will work with all
24019 pixel formats, but the generated output will not be the same.)
24020 __kernel void ramp(__write_only image2d_t dst,
24022 unsigned int index)
24023 {
24024 int2 loc = (int2)(get_global_id(0), get_global_id(1));
24025 float4 val;
24027 val.xy = val.zw = convert_float2(loc) / convert_float2(get_image_dim(dst));
24028 write_imagef(dst, loc, val);
24030 }
24031 • Generate a Sierpinski carpet pattern, panning by a single pixel
24033 each frame.
24034 __kernel void sierpinski_carpet(__write_only image2d_t dst,
24036 unsigned int index)
24037 {
24038 int2 loc = (int2)(get_global_id(0), get_global_id(1));
24039 float4 value = 0.0f;
24041 int x = loc.x + index;
24042 int y = loc.y + index;
24043 while (x > 0 || y > 0) {
24044 if (x % 3 == 1 && y % 3 == 1) {
24045 value = 1.0f;
24046 break;
24047 }
24048 x /= 3;
24049 y /= 3;
24050 }
24051 write_imagef(dst, loc, value);
24053 }
24054 sierpinski
24056 Generate a Sierpinski carpet/triangle fractal, and randomly pan around.
24057 This source accepts the following options:
24059 size, s
24061 Set frame size. For the syntax of this option, check the "Video
24062 size" section in the ffmpeg-utils manual. Default value is
24063 "640x480".
24064 rate, r
24066 Set frame rate, expressed as number of frames per second. Default
24067 value is "25".
24068 seed
24070 Set seed which is used for random panning.
24071 jump
24073 Set max jump for single pan destination. Allowed range is from 1 to
24074 10000.
24075 type
24077 Set fractal type, can be default "carpet" or "triangle".
24078
24080 Below is a description of the currently available video sinks.
24081 buffersink
24083 Buffer video frames, and make them available to the end of the filter
24084 graph.
24085 This sink is mainly intended for programmatic use, in particular
24087 through the interface defined in libavfilter/buffersink.h or the
24088 options system.
24089 It accepts a pointer to an AVBufferSinkContext structure, which defines
24091 the incoming buffers' formats, to be passed as the opaque parameter to
24092 "avfilter_init_filter" for initialization.
24093 nullsink
24095 Null video sink: do absolutely nothing with the input video. It is
24096 mainly useful as a template and for use in analysis / debugging tools.
24097
24099 Below is a description of the currently available multimedia filters.
24100 abitscope
24102 Convert input audio to a video output, displaying the audio bit scope.
24103 The filter accepts the following options:
24105 rate, r
24107 Set frame rate, expressed as number of frames per second. Default
24108 value is "25".
24109 size, s
24111 Specify the video size for the output. For the syntax of this
24112 option, check the "Video size" section in the ffmpeg-utils manual.
24113 Default value is "1024x256".
24114 colors
24116 Specify list of colors separated by space or by '|' which will be
24117 used to draw channels. Unrecognized or missing colors will be
24118 replaced by white color.
24119 mode, m
24121 Set output mode. Can be "bars" or "trace". Default is "bars".
24122 adrawgraph
24124 Draw a graph using input audio metadata.
24125 See drawgraph
24127 agraphmonitor
24129 See graphmonitor.
24130 ahistogram
24132 Convert input audio to a video output, displaying the volume histogram.
24133 The filter accepts the following options:
24135 dmode
24137 Specify how histogram is calculated.
24138 It accepts the following values:
24140 single
24142 Use single histogram for all channels.
24143 separate
24145 Use separate histogram for each channel.
24146 Default is "single".
24148 rate, r
24150 Set frame rate, expressed as number of frames per second. Default
24151 value is "25".
24152 size, s
24154 Specify the video size for the output. For the syntax of this
24155 option, check the "Video size" section in the ffmpeg-utils manual.
24156 Default value is "hd720".
24157 scale
24159 Set display scale.
24160 It accepts the following values:
24162 log logarithmic
24164 sqrt
24166 square root
24167 cbrt
24169 cubic root
24170 lin linear
24172 rlog
24174 reverse logarithmic
24175 Default is "log".
24177 ascale
24179 Set amplitude scale.
24180 It accepts the following values:
24182 log logarithmic
24184 lin linear
24186 Default is "log".
24188 acount
24190 Set how much frames to accumulate in histogram. Default is 1.
24191 Setting this to -1 accumulates all frames.
24192 rheight
24194 Set histogram ratio of window height.
24195 slide
24197 Set sonogram sliding.
24198 It accepts the following values:
24200 replace
24202 replace old rows with new ones.
24203 scroll
24205 scroll from top to bottom.
24206 Default is "replace".
24208 hmode
24210 Set histogram mode.
24211 It accepts the following values:
24213 abs Use absolute values of samples.
24215 sign
24217 Use untouched values of samples.
24218 Default is "abs".
24220 aphasemeter
24222 Measures phase of input audio, which is exported as metadata
24223 "lavfi.aphasemeter.phase", representing mean phase of current audio
24224 frame. A video output can also be produced and is enabled by default.
24225 The audio is passed through as first output.
24226 Audio will be rematrixed to stereo if it has a different channel
24228 layout. Phase value is in range "[-1, 1]" where "-1" means left and
24229 right channels are completely out of phase and 1 means channels are in
24230 phase.
24231 The filter accepts the following options, all related to its video
24233 output:
24234 rate, r
24236 Set the output frame rate. Default value is 25.
24237 size, s
24239 Set the video size for the output. For the syntax of this option,
24240 check the "Video size" section in the ffmpeg-utils manual. Default
24241 value is "800x400".
24242 rc
24244 gc
24245 bc Specify the red, green, blue contrast. Default values are 2, 7 and
24246 1. Allowed range is "[0, 255]".
24247 mpc Set color which will be used for drawing median phase. If color is
24249 "none" which is default, no median phase value will be drawn.
24250 video
24252 Enable video output. Default is enabled.
24253 phasing detection
24255 The filter also detects out of phase and mono sequences in stereo
24257 streams. It logs the sequence start, end and duration when it lasts
24258 longer or as long as the minimum set.
24259 The filter accepts the following options for this detection:
24261 phasing
24263 Enable mono and out of phase detection. Default is disabled.
24264 tolerance, t
24266 Set phase tolerance for mono detection, in amplitude ratio. Default
24267 is 0. Allowed range is "[0, 1]".
24268 angle, a
24270 Set angle threshold for out of phase detection, in degree. Default
24271 is 170. Allowed range is "[90, 180]".
24272 duration, d
24274 Set mono or out of phase duration until notification, expressed in
24275 seconds. Default is 2.
24276 Examples
24278 • Complete example with ffmpeg to detect 1 second of mono with 0.001
24280 phase tolerance:
24281 ffmpeg -i stereo.wav -af aphasemeter=video=0:phasing=1:duration=1:tolerance=0.001 -f null -
24283 avectorscope
24285 Convert input audio to a video output, representing the audio vector
24286 scope.
24287 The filter is used to measure the difference between channels of stereo
24289 audio stream. A monaural signal, consisting of identical left and right
24290 signal, results in straight vertical line. Any stereo separation is
24291 visible as a deviation from this line, creating a Lissajous figure. If
24292 the straight (or deviation from it) but horizontal line appears this
24293 indicates that the left and right channels are out of phase.
24294 The filter accepts the following options:
24296 mode, m
24298 Set the vectorscope mode.
24299 Available values are:
24301 lissajous
24303 Lissajous rotated by 45 degrees.
24304 lissajous_xy
24306 Same as above but not rotated.
24307 polar
24309 Shape resembling half of circle.
24310 Default value is lissajous.
24312 size, s
24314 Set the video size for the output. For the syntax of this option,
24315 check the "Video size" section in the ffmpeg-utils manual. Default
24316 value is "400x400".
24317 rate, r
24319 Set the output frame rate. Default value is 25.
24320 rc
24322 gc
24323 bc
24324 ac Specify the red, green, blue and alpha contrast. Default values are
24325 40, 160, 80 and 255. Allowed range is "[0, 255]".
24326 rf
24328 gf
24329 bf
24330 af Specify the red, green, blue and alpha fade. Default values are 15,
24331 10, 5 and 5. Allowed range is "[0, 255]".
24332 zoom
24334 Set the zoom factor. Default value is 1. Allowed range is "[0,
24335 10]". Values lower than 1 will auto adjust zoom factor to maximal
24336 possible value.
24337 draw
24339 Set the vectorscope drawing mode.
24340 Available values are:
24342 dot Draw dot for each sample.
24344 line
24346 Draw line between previous and current sample.
24347 Default value is dot.
24349 scale
24351 Specify amplitude scale of audio samples.
24352 Available values are:
24354 lin Linear.
24356 sqrt
24358 Square root.
24359 cbrt
24361 Cubic root.
24362 log Logarithmic.
24364 swap
24366 Swap left channel axis with right channel axis.
24367 mirror
24369 Mirror axis.
24370 none
24372 No mirror.
24373 x Mirror only x axis.
24375 y Mirror only y axis.
24377 xy Mirror both axis.
24379 Examples
24381 • Complete example using ffplay:
24383 ffplay -f lavfi 'amovie=input.mp3, asplit [a][out1];
24385 [a] avectorscope=zoom=1.3:rc=2:gc=200:bc=10:rf=1:gf=8:bf=7 [out0]'
24386 Commands
24388 This filter supports the all above options as commands except options
24390 "size" and "rate".
24391 bench, abench
24393 Benchmark part of a filtergraph.
24394 The filter accepts the following options:
24396 action
24398 Start or stop a timer.
24399 Available values are:
24401 start
24403 Get the current time, set it as frame metadata (using the key
24404 "lavfi.bench.start_time"), and forward the frame to the next
24405 filter.
24406 stop
24408 Get the current time and fetch the "lavfi.bench.start_time"
24409 metadata from the input frame metadata to get the time
24410 difference. Time difference, average, maximum and minimum time
24411 (respectively "t", "avg", "max" and "min") are then printed.
24412 The timestamps are expressed in seconds.
24413 Examples
24415 • Benchmark selectivecolor filter:
24417 bench=start,selectivecolor=reds=-.2 .12 -.49,bench=stop
24419 concat
24421 Concatenate audio and video streams, joining them together one after
24422 the other.
24423 The filter works on segments of synchronized video and audio streams.
24425 All segments must have the same number of streams of each type, and
24426 that will also be the number of streams at output.
24427 The filter accepts the following options:
24429 n Set the number of segments. Default is 2.
24431 v Set the number of output video streams, that is also the number of
24433 video streams in each segment. Default is 1.
24434 a Set the number of output audio streams, that is also the number of
24436 audio streams in each segment. Default is 0.
24437 unsafe
24439 Activate unsafe mode: do not fail if segments have a different
24440 format.
24441 The filter has v+a outputs: first v video outputs, then a audio
24443 outputs.
24444 There are nx(v+a) inputs: first the inputs for the first segment, in
24446 the same order as the outputs, then the inputs for the second segment,
24447 etc.
24448 Related streams do not always have exactly the same duration, for
24450 various reasons including codec frame size or sloppy authoring. For
24451 that reason, related synchronized streams (e.g. a video and its audio
24452 track) should be concatenated at once. The concat filter will use the
24453 duration of the longest stream in each segment (except the last one),
24454 and if necessary pad shorter audio streams with silence.
24455 For this filter to work correctly, all segments must start at timestamp
24457 0.
24458 All corresponding streams must have the same parameters in all
24460 segments; the filtering system will automatically select a common pixel
24461 format for video streams, and a common sample format, sample rate and
24462 channel layout for audio streams, but other settings, such as
24463 resolution, must be converted explicitly by the user.
24464 Different frame rates are acceptable but will result in variable frame
24466 rate at output; be sure to configure the output file to handle it.
24467 Examples
24469 • Concatenate an opening, an episode and an ending, all in bilingual
24471 version (video in stream 0, audio in streams 1 and 2):
24472 ffmpeg -i opening.mkv -i episode.mkv -i ending.mkv -filter_complex \
24474 '[0:0] [0:1] [0:2] [1:0] [1:1] [1:2] [2:0] [2:1] [2:2]
24475 concat=n=3:v=1:a=2 [v] [a1] [a2]' \
24476 -map '[v]' -map '[a1]' -map '[a2]' output.mkv
24477 • Concatenate two parts, handling audio and video separately, using
24479 the (a)movie sources, and adjusting the resolution:
24480 movie=part1.mp4, scale=512:288 [v1] ; amovie=part1.mp4 [a1] ;
24482 movie=part2.mp4, scale=512:288 [v2] ; amovie=part2.mp4 [a2] ;
24483 [v1] [v2] concat [outv] ; [a1] [a2] concat=v=0:a=1 [outa]
24484 Note that a desync will happen at the stitch if the audio and video
24486 streams do not have exactly the same duration in the first file.
24487 Commands
24489 This filter supports the following commands:
24491 next
24493 Close the current segment and step to the next one
24494 ebur128
24496 EBU R128 scanner filter. This filter takes an audio stream and analyzes
24497 its loudness level. By default, it logs a message at a frequency of
24498 10Hz with the Momentary loudness (identified by "M"), Short-term
24499 loudness ("S"), Integrated loudness ("I") and Loudness Range ("LRA").
24500 The filter can only analyze streams which have sample format is double-
24502 precision floating point. The input stream will be converted to this
24503 specification, if needed. Users may need to insert aformat and/or
24504 aresample filters after this filter to obtain the original parameters.
24505 The filter also has a video output (see the video option) with a real
24507 time graph to observe the loudness evolution. The graphic contains the
24508 logged message mentioned above, so it is not printed anymore when this
24509 option is set, unless the verbose logging is set. The main graphing
24510 area contains the short-term loudness (3 seconds of analysis), and the
24511 gauge on the right is for the momentary loudness (400 milliseconds),
24512 but can optionally be configured to instead display short-term loudness
24513 (see gauge).
24514 The green area marks a +/- 1LU target range around the target loudness
24516 (-23LUFS by default, unless modified through target).
24517 More information about the Loudness Recommendation EBU R128 on
24519 <http://tech.ebu.ch/loudness>.
24520 The filter accepts the following options:
24522 video
24524 Activate the video output. The audio stream is passed unchanged
24525 whether this option is set or no. The video stream will be the
24526 first output stream if activated. Default is 0.
24527 size
24529 Set the video size. This option is for video only. For the syntax
24530 of this option, check the "Video size" section in the ffmpeg-utils
24531 manual. Default and minimum resolution is "640x480".
24532 meter
24534 Set the EBU scale meter. Default is 9. Common values are 9 and 18,
24535 respectively for EBU scale meter +9 and EBU scale meter +18. Any
24536 other integer value between this range is allowed.
24537 metadata
24539 Set metadata injection. If set to 1, the audio input will be
24540 segmented into 100ms output frames, each of them containing various
24541 loudness information in metadata. All the metadata keys are
24542 prefixed with "lavfi.r128.".
24543 Default is 0.
24545 framelog
24547 Force the frame logging level.
24548 Available values are:
24550 info
24552 information logging level
24553 verbose
24555 verbose logging level
24556 By default, the logging level is set to info. If the video or the
24558 metadata options are set, it switches to verbose.
24559 peak
24561 Set peak mode(s).
24562 Available modes can be cumulated (the option is a "flag" type).
24564 Possible values are:
24565 none
24567 Disable any peak mode (default).
24568 sample
24570 Enable sample-peak mode.
24571 Simple peak mode looking for the higher sample value. It logs a
24573 message for sample-peak (identified by "SPK").
24574 true
24576 Enable true-peak mode.
24577 If enabled, the peak lookup is done on an over-sampled version
24579 of the input stream for better peak accuracy. It logs a message
24580 for true-peak. (identified by "TPK") and true-peak per frame
24581 (identified by "FTPK"). This mode requires a build with
24582 "libswresample".
24583 dualmono
24585 Treat mono input files as "dual mono". If a mono file is intended
24586 for playback on a stereo system, its EBU R128 measurement will be
24587 perceptually incorrect. If set to "true", this option will
24588 compensate for this effect. Multi-channel input files are not
24589 affected by this option.
24590 panlaw
24592 Set a specific pan law to be used for the measurement of dual mono
24593 files. This parameter is optional, and has a default value of
24594 -3.01dB.
24595 target
24597 Set a specific target level (in LUFS) used as relative zero in the
24598 visualization. This parameter is optional and has a default value
24599 of -23LUFS as specified by EBU R128. However, material published
24600 online may prefer a level of -16LUFS (e.g. for use with podcasts or
24601 video platforms).
24602 gauge
24604 Set the value displayed by the gauge. Valid values are "momentary"
24605 and s "shortterm". By default the momentary value will be used, but
24606 in certain scenarios it may be more useful to observe the short
24607 term value instead (e.g. live mixing).
24608 scale
24610 Sets the display scale for the loudness. Valid parameters are
24611 "absolute" (in LUFS) or "relative" (LU) relative to the target.
24612 This only affects the video output, not the summary or continuous
24613 log output.
24614 Examples
24616 • Real-time graph using ffplay, with a EBU scale meter +18:
24618 ffplay -f lavfi -i "amovie=input.mp3,ebur128=video=1:meter=18 [out0][out1]"
24620 • Run an analysis with ffmpeg:
24622 ffmpeg -nostats -i input.mp3 -filter_complex ebur128 -f null -
24624 interleave, ainterleave
24626 Temporally interleave frames from several inputs.
24627 "interleave" works with video inputs, "ainterleave" with audio.
24629 These filters read frames from several inputs and send the oldest
24631 queued frame to the output.
24632 Input streams must have well defined, monotonically increasing frame
24634 timestamp values.
24635 In order to submit one frame to output, these filters need to enqueue
24637 at least one frame for each input, so they cannot work in case one
24638 input is not yet terminated and will not receive incoming frames.
24639 For example consider the case when one input is a "select" filter which
24641 always drops input frames. The "interleave" filter will keep reading
24642 from that input, but it will never be able to send new frames to output
24643 until the input sends an end-of-stream signal.
24644 Also, depending on inputs synchronization, the filters will drop frames
24646 in case one input receives more frames than the other ones, and the
24647 queue is already filled.
24648 These filters accept the following options:
24650 nb_inputs, n
24652 Set the number of different inputs, it is 2 by default.
24653 duration
24655 How to determine the end-of-stream.
24656 longest
24658 The duration of the longest input. (default)
24659 shortest
24661 The duration of the shortest input.
24662 first
24664 The duration of the first input.
24665 Examples
24667 • Interleave frames belonging to different streams using ffmpeg:
24669 ffmpeg -i bambi.avi -i pr0n.mkv -filter_complex "[0:v][1:v] interleave" out.avi
24671 • Add flickering blur effect:
24673 select='if(gt(random(0), 0.2), 1, 2)':n=2 [tmp], boxblur=2:2, [tmp] interleave
24675 latency, alatency
24677 Measure filtering latency.
24678 Report previous filter filtering latency, delay in number of audio
24680 samples for audio filters or number of video frames for video filters.
24681 On end of input stream, filter will report min and max measured latency
24683 for previous running filter in filtergraph.
24684 metadata, ametadata
24686 Manipulate frame metadata.
24687 This filter accepts the following options:
24689 mode
24691 Set mode of operation of the filter.
24692 Can be one of the following:
24694 select
24696 If both "value" and "key" is set, select frames which have such
24697 metadata. If only "key" is set, select every frame that has
24698 such key in metadata.
24699 add Add new metadata "key" and "value". If key is already available
24701 do nothing.
24702 modify
24704 Modify value of already present key.
24705 delete
24707 If "value" is set, delete only keys that have such value.
24708 Otherwise, delete key. If "key" is not set, delete all metadata
24709 values in the frame.
24710 print
24712 Print key and its value if metadata was found. If "key" is not
24713 set print all metadata values available in frame.
24714 key Set key used with all modes. Must be set for all modes except
24716 "print" and "delete".
24717 value
24719 Set metadata value which will be used. This option is mandatory for
24720 "modify" and "add" mode.
24721 function
24723 Which function to use when comparing metadata value and "value".
24724 Can be one of following:
24726 same_str
24728 Values are interpreted as strings, returns true if metadata
24729 value is same as "value".
24730 starts_with
24732 Values are interpreted as strings, returns true if metadata
24733 value starts with the "value" option string.
24734 less
24736 Values are interpreted as floats, returns true if metadata
24737 value is less than "value".
24738 equal
24740 Values are interpreted as floats, returns true if "value" is
24741 equal with metadata value.
24742 greater
24744 Values are interpreted as floats, returns true if metadata
24745 value is greater than "value".
24746 expr
24748 Values are interpreted as floats, returns true if expression
24749 from option "expr" evaluates to true.
24750 ends_with
24752 Values are interpreted as strings, returns true if metadata
24753 value ends with the "value" option string.
24754 expr
24756 Set expression which is used when "function" is set to "expr". The
24757 expression is evaluated through the eval API and can contain the
24758 following constants:
24759 VALUE1, FRAMEVAL
24761 Float representation of "value" from metadata key.
24762 VALUE2, USERVAL
24764 Float representation of "value" as supplied by user in "value"
24765 option.
24766 file
24768 If specified in "print" mode, output is written to the named file.
24769 Instead of plain filename any writable url can be specified.
24770 Filename ``-'' is a shorthand for standard output. If "file" option
24771 is not set, output is written to the log with AV_LOG_INFO loglevel.
24772 direct
24774 Reduces buffering in print mode when output is written to a URL set
24775 using file.
24776 Examples
24778 • Print all metadata values for frames with key
24780 "lavfi.signalstats.YDIF" with values between 0 and 1.
24781 signalstats,metadata=print:key=lavfi.signalstats.YDIF:value=0:function=expr:expr='between(VALUE1,0,1)'
24783 • Print silencedetect output to file metadata.txt.
24785 silencedetect,ametadata=mode=print:file=metadata.txt
24787 • Direct all metadata to a pipe with file descriptor 4.
24789 metadata=mode=print:file='pipe\:4'
24791 perms, aperms
24793 Set read/write permissions for the output frames.
24794 These filters are mainly aimed at developers to test direct path in the
24796 following filter in the filtergraph.
24797 The filters accept the following options:
24799 mode
24801 Select the permissions mode.
24802 It accepts the following values:
24804 none
24806 Do nothing. This is the default.
24807 ro Set all the output frames read-only.
24809 rw Set all the output frames directly writable.
24811 toggle
24813 Make the frame read-only if writable, and writable if read-
24814 only.
24815 random
24817 Set each output frame read-only or writable randomly.
24818 seed
24820 Set the seed for the random mode, must be an integer included
24821 between 0 and "UINT32_MAX". If not specified, or if explicitly set
24822 to "-1", the filter will try to use a good random seed on a best
24823 effort basis.
24824 Note: in case of auto-inserted filter between the permission filter and
24826 the following one, the permission might not be received as expected in
24827 that following filter. Inserting a format or aformat filter before the
24828 perms/aperms filter can avoid this problem.
24829 realtime, arealtime
24831 Slow down filtering to match real time approximately.
24832 These filters will pause the filtering for a variable amount of time to
24834 match the output rate with the input timestamps. They are similar to
24835 the re option to "ffmpeg".
24836 They accept the following options:
24838 limit
24840 Time limit for the pauses. Any pause longer than that will be
24841 considered a timestamp discontinuity and reset the timer. Default
24842 is 2 seconds.
24843 speed
24845 Speed factor for processing. The value must be a float larger than
24846 zero. Values larger than 1.0 will result in faster than realtime
24847 processing, smaller will slow processing down. The limit is
24848 automatically adapted accordingly. Default is 1.0.
24849 A processing speed faster than what is possible without these
24851 filters cannot be achieved.
24852 Commands
24854 Both filters supports the all above options as commands.
24856 segment, asegment
24858 Split single input stream into multiple streams.
24859 This filter does opposite of concat filters.
24861 "segment" works on video frames, "asegment" on audio samples.
24863 This filter accepts the following options:
24865 timestamps
24867 Timestamps of output segments separated by '|'. The first segment
24868 will run from the beginning of the input stream. The last segment
24869 will run until the end of the input stream
24870 frames, samples
24872 Exact frame/sample count to split the segments.
24873 In all cases, prefixing an each segment with '+' will make it relative
24875 to the previous segment.
24876 Examples
24878 • Split input audio stream into three output audio streams, starting
24880 at start of input audio stream and storing that in 1st output audio
24881 stream, then following at 60th second and storing than in 2nd
24882 output audio stream, and last after 150th second of input audio
24883 stream store in 3rd output audio stream:
24884 asegment=timestamps="60|150"
24886 select, aselect
24888 Select frames to pass in output.
24889 This filter accepts the following options:
24891 expr, e
24893 Set expression, which is evaluated for each input frame.
24894 If the expression is evaluated to zero, the frame is discarded.
24896 If the evaluation result is negative or NaN, the frame is sent to
24898 the first output; otherwise it is sent to the output with index
24899 "ceil(val)-1", assuming that the input index starts from 0.
24900 For example a value of 1.2 corresponds to the output with index
24902 "ceil(1.2)-1 = 2-1 = 1", that is the second output.
24903 outputs, n
24905 Set the number of outputs. The output to which to send the selected
24906 frame is based on the result of the evaluation. Default value is 1.
24907 The expression can contain the following constants:
24909 n The (sequential) number of the filtered frame, starting from 0.
24911 selected_n
24913 The (sequential) number of the selected frame, starting from 0.
24914 prev_selected_n
24916 The sequential number of the last selected frame. It's NAN if
24917 undefined.
24918 TB The timebase of the input timestamps.
24920 pts The PTS (Presentation TimeStamp) of the filtered frame, expressed
24922 in TB units. It's NAN if undefined.
24923 t The PTS of the filtered frame, expressed in seconds. It's NAN if
24925 undefined.
24926 prev_pts
24928 The PTS of the previously filtered frame. It's NAN if undefined.
24929 prev_selected_pts
24931 The PTS of the last previously filtered frame. It's NAN if
24932 undefined.
24933 prev_selected_t
24935 The PTS of the last previously selected frame, expressed in
24936 seconds. It's NAN if undefined.
24937 start_pts
24939 The first PTS in the stream which is not NAN. It remains NAN if not
24940 found.
24941 start_t
24943 The first PTS, in seconds, in the stream which is not NAN. It
24944 remains NAN if not found.
24945 pict_type (video only)
24947 The type of the filtered frame. It can assume one of the following
24948 values:
24949 I
24951 P
24952 B
24953 S
24954 SI
24955 SP
24956 BI
24957 interlace_type (video only)
24958 The frame interlace type. It can assume one of the following
24959 values:
24960 PROGRESSIVE
24962 The frame is progressive (not interlaced).
24963 TOPFIRST
24965 The frame is top-field-first.
24966 BOTTOMFIRST
24968 The frame is bottom-field-first.
24969 consumed_sample_n (audio only)
24971 the number of selected samples before the current frame
24972 samples_n (audio only)
24974 the number of samples in the current frame
24975 sample_rate (audio only)
24977 the input sample rate
24978 key This is 1 if the filtered frame is a key-frame, 0 otherwise.
24980 pos the position in the file of the filtered frame, -1 if the
24982 information is not available (e.g. for synthetic video)
24983 scene (video only)
24985 value between 0 and 1 to indicate a new scene; a low value reflects
24986 a low probability for the current frame to introduce a new scene,
24987 while a higher value means the current frame is more likely to be
24988 one (see the example below)
24989 concatdec_select
24991 The concat demuxer can select only part of a concat input file by
24992 setting an inpoint and an outpoint, but the output packets may not
24993 be entirely contained in the selected interval. By using this
24994 variable, it is possible to skip frames generated by the concat
24995 demuxer which are not exactly contained in the selected interval.
24996 This works by comparing the frame pts against the
24998 lavf.concat.start_time and the lavf.concat.duration packet metadata
24999 values which are also present in the decoded frames.
25000 The concatdec_select variable is -1 if the frame pts is at least
25002 start_time and either the duration metadata is missing or the frame
25003 pts is less than start_time + duration, 0 otherwise, and NaN if the
25004 start_time metadata is missing.
25005 That basically means that an input frame is selected if its pts is
25007 within the interval set by the concat demuxer.
25008 The default value of the select expression is "1".
25010 Examples
25012 • Select all frames in input:
25014 select
25016 The example above is the same as:
25018 select=1
25020 • Skip all frames:
25022 select=0
25024 • Select only I-frames:
25026 select='eq(pict_type\,I)'
25028 • Select one frame every 100:
25030 select='not(mod(n\,100))'
25032 • Select only frames contained in the 10-20 time interval:
25034 select=between(t\,10\,20)
25036 • Select only I-frames contained in the 10-20 time interval:
25038 select=between(t\,10\,20)*eq(pict_type\,I)
25040 • Select frames with a minimum distance of 10 seconds:
25042 select='isnan(prev_selected_t)+gte(t-prev_selected_t\,10)'
25044 • Use aselect to select only audio frames with samples number > 100:
25046 aselect='gt(samples_n\,100)'
25048 • Create a mosaic of the first scenes:
25050 ffmpeg -i video.avi -vf select='gt(scene\,0.4)',scale=160:120,tile -frames:v 1 preview.png
25052 Comparing scene against a value between 0.3 and 0.5 is generally a
25054 sane choice.
25055 • Send even and odd frames to separate outputs, and compose them:
25057 select=n=2:e='mod(n, 2)+1' [odd][even]; [odd] pad=h=2*ih [tmp]; [tmp][even] overlay=y=h
25059 • Select useful frames from an ffconcat file which is using inpoints
25061 and outpoints but where the source files are not intra frame only.
25062 ffmpeg -copyts -vsync 0 -segment_time_metadata 1 -i input.ffconcat -vf select=concatdec_select -af aselect=concatdec_select output.avi
25064 sendcmd, asendcmd
25066 Send commands to filters in the filtergraph.
25067 These filters read commands to be sent to other filters in the
25069 filtergraph.
25070 "sendcmd" must be inserted between two video filters, "asendcmd" must
25072 be inserted between two audio filters, but apart from that they act the
25073 same way.
25074 The specification of commands can be provided in the filter arguments
25076 with the commands option, or in a file specified by the filename
25077 option.
25078 These filters accept the following options:
25080 commands, c
25082 Set the commands to be read and sent to the other filters.
25083 filename, f
25085 Set the filename of the commands to be read and sent to the other
25086 filters.
25087 Commands syntax
25089 A commands description consists of a sequence of interval
25091 specifications, comprising a list of commands to be executed when a
25092 particular event related to that interval occurs. The occurring event
25093 is typically the current frame time entering or leaving a given time
25094 interval.
25095 An interval is specified by the following syntax:
25097 <START>[-<END>] <COMMANDS>;
25099 The time interval is specified by the START and END times. END is
25101 optional and defaults to the maximum time.
25102 The current frame time is considered within the specified interval if
25104 it is included in the interval [START, END), that is when the time is
25105 greater or equal to START and is lesser than END.
25106 COMMANDS consists of a sequence of one or more command specifications,
25108 separated by ",", relating to that interval. The syntax of a command
25109 specification is given by:
25110 [<FLAGS>] <TARGET> <COMMAND> <ARG>
25112 FLAGS is optional and specifies the type of events relating to the time
25114 interval which enable sending the specified command, and must be a non-
25115 null sequence of identifier flags separated by "+" or "|" and enclosed
25116 between "[" and "]".
25117 The following flags are recognized:
25119 enter
25121 The command is sent when the current frame timestamp enters the
25122 specified interval. In other words, the command is sent when the
25123 previous frame timestamp was not in the given interval, and the
25124 current is.
25125 leave
25127 The command is sent when the current frame timestamp leaves the
25128 specified interval. In other words, the command is sent when the
25129 previous frame timestamp was in the given interval, and the current
25130 is not.
25131 expr
25133 The command ARG is interpreted as expression and result of
25134 expression is passed as ARG.
25135 The expression is evaluated through the eval API and can contain
25137 the following constants:
25138 POS Original position in the file of the frame, or undefined if
25140 undefined for the current frame.
25141 PTS The presentation timestamp in input.
25143 N The count of the input frame for video or audio, starting from
25145 0.
25146 T The time in seconds of the current frame.
25148 TS The start time in seconds of the current command interval.
25150 TE The end time in seconds of the current command interval.
25152 TI The interpolated time of the current command interval, TI = (T
25154 - TS) / (TE - TS).
25155 W The video frame width.
25157 H The video frame height.
25159 If FLAGS is not specified, a default value of "[enter]" is assumed.
25161 TARGET specifies the target of the command, usually the name of the
25163 filter class or a specific filter instance name.
25164 COMMAND specifies the name of the command for the target filter.
25166 ARG is optional and specifies the optional list of argument for the
25168 given COMMAND.
25169 Between one interval specification and another, whitespaces, or
25171 sequences of characters starting with "#" until the end of line, are
25172 ignored and can be used to annotate comments.
25173 A simplified BNF description of the commands specification syntax
25175 follows:
25176 <COMMAND_FLAG> ::= "enter" | "leave"
25178 <COMMAND_FLAGS> ::= <COMMAND_FLAG> [(+|"|")<COMMAND_FLAG>]
25179 <COMMAND> ::= ["[" <COMMAND_FLAGS> "]"] <TARGET> <COMMAND> [<ARG>]
25180 <COMMANDS> ::= <COMMAND> [,<COMMANDS>]
25181 <INTERVAL> ::= <START>[-<END>] <COMMANDS>
25182 <INTERVALS> ::= <INTERVAL>[;<INTERVALS>]
25183 Examples
25185 • Specify audio tempo change at second 4:
25187 asendcmd=c='4.0 atempo tempo 1.5',atempo
25189 • Target a specific filter instance:
25191 asendcmd=c='4.0 atempo@my tempo 1.5',atempo@my
25193 • Specify a list of drawtext and hue commands in a file.
25195 # show text in the interval 5-10
25197 5.0-10.0 [enter] drawtext reinit 'fontfile=FreeSerif.ttf:text=hello world',
25198 [leave] drawtext reinit 'fontfile=FreeSerif.ttf:text=';
25199 # desaturate the image in the interval 15-20
25201 15.0-20.0 [enter] hue s 0,
25202 [enter] drawtext reinit 'fontfile=FreeSerif.ttf:text=nocolor',
25203 [leave] hue s 1,
25204 [leave] drawtext reinit 'fontfile=FreeSerif.ttf:text=color';
25205 # apply an exponential saturation fade-out effect, starting from time 25
25207 25 [enter] hue s exp(25-t)
25208 A filtergraph allowing to read and process the above command list
25210 stored in a file test.cmd, can be specified with:
25211 sendcmd=f=test.cmd,drawtext=fontfile=FreeSerif.ttf:text='',hue
25213 setpts, asetpts
25215 Change the PTS (presentation timestamp) of the input frames.
25216 "setpts" works on video frames, "asetpts" on audio frames.
25218 This filter accepts the following options:
25220 expr
25222 The expression which is evaluated for each frame to construct its
25223 timestamp.
25224 The expression is evaluated through the eval API and can contain the
25226 following constants:
25227 FRAME_RATE, FR
25229 frame rate, only defined for constant frame-rate video
25230 PTS The presentation timestamp in input
25232 N The count of the input frame for video or the number of consumed
25234 samples, not including the current frame for audio, starting from
25235 0.
25236 NB_CONSUMED_SAMPLES
25238 The number of consumed samples, not including the current frame
25239 (only audio)
25240 NB_SAMPLES, S
25242 The number of samples in the current frame (only audio)
25243 SAMPLE_RATE, SR
25245 The audio sample rate.
25246 STARTPTS
25248 The PTS of the first frame.
25249 STARTT
25251 the time in seconds of the first frame
25252 INTERLACED
25254 State whether the current frame is interlaced.
25255 T the time in seconds of the current frame
25257 POS original position in the file of the frame, or undefined if
25259 undefined for the current frame
25260 PREV_INPTS
25262 The previous input PTS.
25263 PREV_INT
25265 previous input time in seconds
25266 PREV_OUTPTS
25268 The previous output PTS.
25269 PREV_OUTT
25271 previous output time in seconds
25272 RTCTIME
25274 The wallclock (RTC) time in microseconds. This is deprecated, use
25275 time(0) instead.
25276 RTCSTART
25278 The wallclock (RTC) time at the start of the movie in microseconds.
25279 TB The timebase of the input timestamps.
25281 Examples
25283 • Start counting PTS from zero
25285 setpts=PTS-STARTPTS
25287 • Apply fast motion effect:
25289 setpts=0.5*PTS
25291 • Apply slow motion effect:
25293 setpts=2.0*PTS
25295 • Set fixed rate of 25 frames per second:
25297 setpts=N/(25*TB)
25299 • Set fixed rate 25 fps with some jitter:
25301 setpts='1/(25*TB) * (N + 0.05 * sin(N*2*PI/25))'
25303 • Apply an offset of 10 seconds to the input PTS:
25305 setpts=PTS+10/TB
25307 • Generate timestamps from a "live source" and rebase onto the
25309 current timebase:
25310 setpts='(RTCTIME - RTCSTART) / (TB * 1000000)'
25312 • Generate timestamps by counting samples:
25314 asetpts=N/SR/TB
25316 setrange
25318 Force color range for the output video frame.
25319 The "setrange" filter marks the color range property for the output
25321 frames. It does not change the input frame, but only sets the
25322 corresponding property, which affects how the frame is treated by
25323 following filters.
25324 The filter accepts the following options:
25326 range
25328 Available values are:
25329 auto
25331 Keep the same color range property.
25332 unspecified, unknown
25334 Set the color range as unspecified.
25335 limited, tv, mpeg
25337 Set the color range as limited.
25338 full, pc, jpeg
25340 Set the color range as full.
25341 settb, asettb
25343 Set the timebase to use for the output frames timestamps. It is mainly
25344 useful for testing timebase configuration.
25345 It accepts the following parameters:
25347 expr, tb
25349 The expression which is evaluated into the output timebase.
25350 The value for tb is an arithmetic expression representing a rational.
25352 The expression can contain the constants "AVTB" (the default timebase),
25353 "intb" (the input timebase) and "sr" (the sample rate, audio only).
25354 Default value is "intb".
25355 Examples
25357 • Set the timebase to 1/25:
25359 settb=expr=1/25
25361 • Set the timebase to 1/10:
25363 settb=expr=0.1
25365 • Set the timebase to 1001/1000:
25367 settb=1+0.001
25369 • Set the timebase to 2*intb:
25371 settb=2*intb
25373 • Set the default timebase value:
25375 settb=AVTB
25377 showcqt
25379 Convert input audio to a video output representing frequency spectrum
25380 logarithmically using Brown-Puckette constant Q transform algorithm
25381 with direct frequency domain coefficient calculation (but the transform
25382 itself is not really constant Q, instead the Q factor is actually
25383 variable/clamped), with musical tone scale, from E0 to D#10.
25384 The filter accepts the following options:
25386 size, s
25388 Specify the video size for the output. It must be even. For the
25389 syntax of this option, check the "Video size" section in the
25390 ffmpeg-utils manual. Default value is "1920x1080".
25391 fps, rate, r
25393 Set the output frame rate. Default value is 25.
25394 bar_h
25396 Set the bargraph height. It must be even. Default value is "-1"
25397 which computes the bargraph height automatically.
25398 axis_h
25400 Set the axis height. It must be even. Default value is "-1" which
25401 computes the axis height automatically.
25402 sono_h
25404 Set the sonogram height. It must be even. Default value is "-1"
25405 which computes the sonogram height automatically.
25406 fullhd
25408 Set the fullhd resolution. This option is deprecated, use size, s
25409 instead. Default value is 1.
25410 sono_v, volume
25412 Specify the sonogram volume expression. It can contain variables:
25413 bar_v
25415 the bar_v evaluated expression
25416 frequency, freq, f
25418 the frequency where it is evaluated
25419 timeclamp, tc
25421 the value of timeclamp option
25422 and functions:
25424 a_weighting(f)
25426 A-weighting of equal loudness
25427 b_weighting(f)
25429 B-weighting of equal loudness
25430 c_weighting(f)
25432 C-weighting of equal loudness.
25433 Default value is 16.
25435 bar_v, volume2
25437 Specify the bargraph volume expression. It can contain variables:
25438 sono_v
25440 the sono_v evaluated expression
25441 frequency, freq, f
25443 the frequency where it is evaluated
25444 timeclamp, tc
25446 the value of timeclamp option
25447 and functions:
25449 a_weighting(f)
25451 A-weighting of equal loudness
25452 b_weighting(f)
25454 B-weighting of equal loudness
25455 c_weighting(f)
25457 C-weighting of equal loudness.
25458 Default value is "sono_v".
25460 sono_g, gamma
25462 Specify the sonogram gamma. Lower gamma makes the spectrum more
25463 contrast, higher gamma makes the spectrum having more range.
25464 Default value is 3. Acceptable range is "[1, 7]".
25465 bar_g, gamma2
25467 Specify the bargraph gamma. Default value is 1. Acceptable range is
25468 "[1, 7]".
25469 bar_t
25471 Specify the bargraph transparency level. Lower value makes the
25472 bargraph sharper. Default value is 1. Acceptable range is "[0,
25473 1]".
25474 timeclamp, tc
25476 Specify the transform timeclamp. At low frequency, there is trade-
25477 off between accuracy in time domain and frequency domain. If
25478 timeclamp is lower, event in time domain is represented more
25479 accurately (such as fast bass drum), otherwise event in frequency
25480 domain is represented more accurately (such as bass guitar).
25481 Acceptable range is "[0.002, 1]". Default value is 0.17.
25482 attack
25484 Set attack time in seconds. The default is 0 (disabled). Otherwise,
25485 it limits future samples by applying asymmetric windowing in time
25486 domain, useful when low latency is required. Accepted range is "[0,
25487 1]".
25488 basefreq
25490 Specify the transform base frequency. Default value is
25491 20.01523126408007475, which is frequency 50 cents below E0.
25492 Acceptable range is "[10, 100000]".
25493 endfreq
25495 Specify the transform end frequency. Default value is
25496 20495.59681441799654, which is frequency 50 cents above D#10.
25497 Acceptable range is "[10, 100000]".
25498 coeffclamp
25500 This option is deprecated and ignored.
25501 tlength
25503 Specify the transform length in time domain. Use this option to
25504 control accuracy trade-off between time domain and frequency domain
25505 at every frequency sample. It can contain variables:
25506 frequency, freq, f
25508 the frequency where it is evaluated
25509 timeclamp, tc
25511 the value of timeclamp option.
25512 Default value is "384*tc/(384+tc*f)".
25514 count
25516 Specify the transform count for every video frame. Default value is
25517 6. Acceptable range is "[1, 30]".
25518 fcount
25520 Specify the transform count for every single pixel. Default value
25521 is 0, which makes it computed automatically. Acceptable range is
25522 "[0, 10]".
25523 fontfile
25525 Specify font file for use with freetype to draw the axis. If not
25526 specified, use embedded font. Note that drawing with font file or
25527 embedded font is not implemented with custom basefreq and endfreq,
25528 use axisfile option instead.
25529 font
25531 Specify fontconfig pattern. This has lower priority than fontfile.
25532 The ":" in the pattern may be replaced by "|" to avoid unnecessary
25533 escaping.
25534 fontcolor
25536 Specify font color expression. This is arithmetic expression that
25537 should return integer value 0xRRGGBB. It can contain variables:
25538 frequency, freq, f
25540 the frequency where it is evaluated
25541 timeclamp, tc
25543 the value of timeclamp option
25544 and functions:
25546 midi(f)
25548 midi number of frequency f, some midi numbers: E0(16), C1(24),
25549 C2(36), A4(69)
25550 r(x), g(x), b(x)
25552 red, green, and blue value of intensity x.
25553 Default value is "st(0, (midi(f)-59.5)/12); st(1,
25555 if(between(ld(0),0,1), 0.5-0.5*cos(2*PI*ld(0)), 0)); r(1-ld(1)) +
25556 b(ld(1))".
25557 axisfile
25559 Specify image file to draw the axis. This option override fontfile
25560 and fontcolor option.
25561 axis, text
25563 Enable/disable drawing text to the axis. If it is set to 0, drawing
25564 to the axis is disabled, ignoring fontfile and axisfile option.
25565 Default value is 1.
25566 csp Set colorspace. The accepted values are:
25568 unspecified
25570 Unspecified (default)
25571 bt709
25573 BT.709
25574 fcc FCC
25576 bt470bg
25578 BT.470BG or BT.601-6 625
25579 smpte170m
25581 SMPTE-170M or BT.601-6 525
25582 smpte240m
25584 SMPTE-240M
25585 bt2020ncl
25587 BT.2020 with non-constant luminance
25588 cscheme
25590 Set spectrogram color scheme. This is list of floating point values
25591 with format "left_r|left_g|left_b|right_r|right_g|right_b". The
25592 default is "1|0.5|0|0|0.5|1".
25593 Examples
25595 • Playing audio while showing the spectrum:
25597 ffplay -f lavfi 'amovie=a.mp3, asplit [a][out1]; [a] showcqt [out0]'
25599 • Same as above, but with frame rate 30 fps:
25601 ffplay -f lavfi 'amovie=a.mp3, asplit [a][out1]; [a] showcqt=fps=30:count=5 [out0]'
25603 • Playing at 1280x720:
25605 ffplay -f lavfi 'amovie=a.mp3, asplit [a][out1]; [a] showcqt=s=1280x720:count=4 [out0]'
25607 • Disable sonogram display:
25609 sono_h=0
25611 • A1 and its harmonics: A1, A2, (near)E3, A3:
25613 ffplay -f lavfi 'aevalsrc=0.1*sin(2*PI*55*t)+0.1*sin(4*PI*55*t)+0.1*sin(6*PI*55*t)+0.1*sin(8*PI*55*t),
25615 asplit[a][out1]; [a] showcqt [out0]'
25616 • Same as above, but with more accuracy in frequency domain:
25618 ffplay -f lavfi 'aevalsrc=0.1*sin(2*PI*55*t)+0.1*sin(4*PI*55*t)+0.1*sin(6*PI*55*t)+0.1*sin(8*PI*55*t),
25620 asplit[a][out1]; [a] showcqt=timeclamp=0.5 [out0]'
25621 • Custom volume:
25623 bar_v=10:sono_v=bar_v*a_weighting(f)
25625 • Custom gamma, now spectrum is linear to the amplitude.
25627 bar_g=2:sono_g=2
25629 • Custom tlength equation:
25631 tc=0.33:tlength='st(0,0.17); 384*tc / (384 / ld(0) + tc*f /(1-ld(0))) + 384*tc / (tc*f / ld(0) + 384 /(1-ld(0)))'
25633 • Custom fontcolor and fontfile, C-note is colored green, others are
25635 colored blue:
25636 fontcolor='if(mod(floor(midi(f)+0.5),12), 0x0000FF, g(1))':fontfile=myfont.ttf
25638 • Custom font using fontconfig:
25640 font='Courier New,Monospace,mono|bold'
25642 • Custom frequency range with custom axis using image file:
25644 axisfile=myaxis.png:basefreq=40:endfreq=10000
25646 showfreqs
25648 Convert input audio to video output representing the audio power
25649 spectrum. Audio amplitude is on Y-axis while frequency is on X-axis.
25650 The filter accepts the following options:
25652 size, s
25654 Specify size of video. For the syntax of this option, check the
25655 "Video size" section in the ffmpeg-utils manual. Default is
25656 "1024x512".
25657 rate, r
25659 Set video rate. Default is 25.
25660 mode
25662 Set display mode. This set how each frequency bin will be
25663 represented.
25664 It accepts the following values:
25666 line
25668 bar
25669 dot
25670 Default is "bar".
25672 ascale
25674 Set amplitude scale.
25675 It accepts the following values:
25677 lin Linear scale.
25679 sqrt
25681 Square root scale.
25682 cbrt
25684 Cubic root scale.
25685 log Logarithmic scale.
25687 Default is "log".
25689 fscale
25691 Set frequency scale.
25692 It accepts the following values:
25694 lin Linear scale.
25696 log Logarithmic scale.
25698 rlog
25700 Reverse logarithmic scale.
25701 Default is "lin".
25703 win_size
25705 Set window size. Allowed range is from 16 to 65536.
25706 Default is 2048
25708 win_func
25710 Set windowing function.
25711 It accepts the following values:
25713 rect
25715 bartlett
25716 hanning
25717 hamming
25718 blackman
25719 welch
25720 flattop
25721 bharris
25722 bnuttall
25723 bhann
25724 sine
25725 nuttall
25726 lanczos
25727 gauss
25728 tukey
25729 dolph
25730 cauchy
25731 parzen
25732 poisson
25733 bohman
25734 Default is "hanning".
25736 overlap
25738 Set window overlap. In range "[0, 1]". Default is 1, which means
25739 optimal overlap for selected window function will be picked.
25740 averaging
25742 Set time averaging. Setting this to 0 will display current maximal
25743 peaks. Default is 1, which means time averaging is disabled.
25744 colors
25746 Specify list of colors separated by space or by '|' which will be
25747 used to draw channel frequencies. Unrecognized or missing colors
25748 will be replaced by white color.
25749 cmode
25751 Set channel display mode.
25752 It accepts the following values:
25754 combined
25756 separate
25757 Default is "combined".
25759 minamp
25761 Set minimum amplitude used in "log" amplitude scaler.
25762 data
25764 Set data display mode.
25765 It accepts the following values:
25767 magnitude
25769 phase
25770 delay
25771 Default is "magnitude".
25773 channels
25775 Set channels to use when processing audio. By default all are
25776 processed.
25777 showspatial
25779 Convert stereo input audio to a video output, representing the spatial
25780 relationship between two channels.
25781 The filter accepts the following options:
25783 size, s
25785 Specify the video size for the output. For the syntax of this
25786 option, check the "Video size" section in the ffmpeg-utils manual.
25787 Default value is "512x512".
25788 win_size
25790 Set window size. Allowed range is from 1024 to 65536. Default size
25791 is 4096.
25792 win_func
25794 Set window function.
25795 It accepts the following values:
25797 rect
25799 bartlett
25800 hann
25801 hanning
25802 hamming
25803 blackman
25804 welch
25805 flattop
25806 bharris
25807 bnuttall
25808 bhann
25809 sine
25810 nuttall
25811 lanczos
25812 gauss
25813 tukey
25814 dolph
25815 cauchy
25816 parzen
25817 poisson
25818 bohman
25819 Default value is "hann".
25821 overlap
25823 Set ratio of overlap window. Default value is 0.5. When value is 1
25824 overlap is set to recommended size for specific window function
25825 currently used.
25826 showspectrum
25828 Convert input audio to a video output, representing the audio frequency
25829 spectrum.
25830 The filter accepts the following options:
25832 size, s
25834 Specify the video size for the output. For the syntax of this
25835 option, check the "Video size" section in the ffmpeg-utils manual.
25836 Default value is "640x512".
25837 slide
25839 Specify how the spectrum should slide along the window.
25840 It accepts the following values:
25842 replace
25844 the samples start again on the left when they reach the right
25845 scroll
25847 the samples scroll from right to left
25848 fullframe
25850 frames are only produced when the samples reach the right
25851 rscroll
25853 the samples scroll from left to right
25854 lreplace
25856 the samples start again on the right when they reach the left
25857 Default value is "replace".
25859 mode
25861 Specify display mode.
25862 It accepts the following values:
25864 combined
25866 all channels are displayed in the same row
25867 separate
25869 all channels are displayed in separate rows
25870 Default value is combined.
25872 color
25874 Specify display color mode.
25875 It accepts the following values:
25877 channel
25879 each channel is displayed in a separate color
25880 intensity
25882 each channel is displayed using the same color scheme
25883 rainbow
25885 each channel is displayed using the rainbow color scheme
25886 moreland
25888 each channel is displayed using the moreland color scheme
25889 nebulae
25891 each channel is displayed using the nebulae color scheme
25892 fire
25894 each channel is displayed using the fire color scheme
25895 fiery
25897 each channel is displayed using the fiery color scheme
25898 fruit
25900 each channel is displayed using the fruit color scheme
25901 cool
25903 each channel is displayed using the cool color scheme
25904 magma
25906 each channel is displayed using the magma color scheme
25907 green
25909 each channel is displayed using the green color scheme
25910 viridis
25912 each channel is displayed using the viridis color scheme
25913 plasma
25915 each channel is displayed using the plasma color scheme
25916 cividis
25918 each channel is displayed using the cividis color scheme
25919 terrain
25921 each channel is displayed using the terrain color scheme
25922 Default value is channel.
25924 scale
25926 Specify scale used for calculating intensity color values.
25927 It accepts the following values:
25929 lin linear
25931 sqrt
25933 square root, default
25934 cbrt
25936 cubic root
25937 log logarithmic
25939 4thrt
25941 4th root
25942 5thrt
25944 5th root
25945 Default value is sqrt.
25947 fscale
25949 Specify frequency scale.
25950 It accepts the following values:
25952 lin linear
25954 log logarithmic
25956 Default value is lin.
25958 saturation
25960 Set saturation modifier for displayed colors. Negative values
25961 provide alternative color scheme. 0 is no saturation at all.
25962 Saturation must be in [-10.0, 10.0] range. Default value is 1.
25963 win_func
25965 Set window function.
25966 It accepts the following values:
25968 rect
25970 bartlett
25971 hann
25972 hanning
25973 hamming
25974 blackman
25975 welch
25976 flattop
25977 bharris
25978 bnuttall
25979 bhann
25980 sine
25981 nuttall
25982 lanczos
25983 gauss
25984 tukey
25985 dolph
25986 cauchy
25987 parzen
25988 poisson
25989 bohman
25990 Default value is "hann".
25992 orientation
25994 Set orientation of time vs frequency axis. Can be "vertical" or
25995 "horizontal". Default is "vertical".
25996 overlap
25998 Set ratio of overlap window. Default value is 0. When value is 1
25999 overlap is set to recommended size for specific window function
26000 currently used.
26001 gain
26003 Set scale gain for calculating intensity color values. Default
26004 value is 1.
26005 data
26007 Set which data to display. Can be "magnitude", default or "phase",
26008 or unwrapped phase: "uphase".
26009 rotation
26011 Set color rotation, must be in [-1.0, 1.0] range. Default value is
26012 0.
26013 start
26015 Set start frequency from which to display spectrogram. Default is
26016 0.
26017 stop
26019 Set stop frequency to which to display spectrogram. Default is 0.
26020 fps Set upper frame rate limit. Default is "auto", unlimited.
26022 legend
26024 Draw time and frequency axes and legends. Default is disabled.
26025 drange
26027 Set dynamic range used to calculate intensity color values. Default
26028 is 120 dBFS. Allowed range is from 10 to 200.
26029 limit
26031 Set upper limit of input audio samples volume in dBFS. Default is 0
26032 dBFS. Allowed range is from -100 to 100.
26033 opacity
26035 Set opacity strength when using pixel format output with alpha
26036 component.
26037 The usage is very similar to the showwaves filter; see the examples in
26039 that section.
26040 Examples
26042 • Large window with logarithmic color scaling:
26044 showspectrum=s=1280x480:scale=log
26046 • Complete example for a colored and sliding spectrum per channel
26048 using ffplay:
26049 ffplay -f lavfi 'amovie=input.mp3, asplit [a][out1];
26051 [a] showspectrum=mode=separate:color=intensity:slide=1:scale=cbrt [out0]'
26052 showspectrumpic
26054 Convert input audio to a single video frame, representing the audio
26055 frequency spectrum.
26056 The filter accepts the following options:
26058 size, s
26060 Specify the video size for the output. For the syntax of this
26061 option, check the "Video size" section in the ffmpeg-utils manual.
26062 Default value is "4096x2048".
26063 mode
26065 Specify display mode.
26066 It accepts the following values:
26068 combined
26070 all channels are displayed in the same row
26071 separate
26073 all channels are displayed in separate rows
26074 Default value is combined.
26076 color
26078 Specify display color mode.
26079 It accepts the following values:
26081 channel
26083 each channel is displayed in a separate color
26084 intensity
26086 each channel is displayed using the same color scheme
26087 rainbow
26089 each channel is displayed using the rainbow color scheme
26090 moreland
26092 each channel is displayed using the moreland color scheme
26093 nebulae
26095 each channel is displayed using the nebulae color scheme
26096 fire
26098 each channel is displayed using the fire color scheme
26099 fiery
26101 each channel is displayed using the fiery color scheme
26102 fruit
26104 each channel is displayed using the fruit color scheme
26105 cool
26107 each channel is displayed using the cool color scheme
26108 magma
26110 each channel is displayed using the magma color scheme
26111 green
26113 each channel is displayed using the green color scheme
26114 viridis
26116 each channel is displayed using the viridis color scheme
26117 plasma
26119 each channel is displayed using the plasma color scheme
26120 cividis
26122 each channel is displayed using the cividis color scheme
26123 terrain
26125 each channel is displayed using the terrain color scheme
26126 Default value is intensity.
26128 scale
26130 Specify scale used for calculating intensity color values.
26131 It accepts the following values:
26133 lin linear
26135 sqrt
26137 square root, default
26138 cbrt
26140 cubic root
26141 log logarithmic
26143 4thrt
26145 4th root
26146 5thrt
26148 5th root
26149 Default value is log.
26151 fscale
26153 Specify frequency scale.
26154 It accepts the following values:
26156 lin linear
26158 log logarithmic
26160 Default value is lin.
26162 saturation
26164 Set saturation modifier for displayed colors. Negative values
26165 provide alternative color scheme. 0 is no saturation at all.
26166 Saturation must be in [-10.0, 10.0] range. Default value is 1.
26167 win_func
26169 Set window function.
26170 It accepts the following values:
26172 rect
26174 bartlett
26175 hann
26176 hanning
26177 hamming
26178 blackman
26179 welch
26180 flattop
26181 bharris
26182 bnuttall
26183 bhann
26184 sine
26185 nuttall
26186 lanczos
26187 gauss
26188 tukey
26189 dolph
26190 cauchy
26191 parzen
26192 poisson
26193 bohman
26194 Default value is "hann".
26196 orientation
26198 Set orientation of time vs frequency axis. Can be "vertical" or
26199 "horizontal". Default is "vertical".
26200 gain
26202 Set scale gain for calculating intensity color values. Default
26203 value is 1.
26204 legend
26206 Draw time and frequency axes and legends. Default is enabled.
26207 rotation
26209 Set color rotation, must be in [-1.0, 1.0] range. Default value is
26210 0.
26211 start
26213 Set start frequency from which to display spectrogram. Default is
26214 0.
26215 stop
26217 Set stop frequency to which to display spectrogram. Default is 0.
26218 drange
26220 Set dynamic range used to calculate intensity color values. Default
26221 is 120 dBFS. Allowed range is from 10 to 200.
26222 limit
26224 Set upper limit of input audio samples volume in dBFS. Default is 0
26225 dBFS. Allowed range is from -100 to 100.
26226 opacity
26228 Set opacity strength when using pixel format output with alpha
26229 component.
26230 Examples
26232 • Extract an audio spectrogram of a whole audio track in a 1024x1024
26234 picture using ffmpeg:
26235 ffmpeg -i audio.flac -lavfi showspectrumpic=s=1024x1024 spectrogram.png
26237 showvolume
26239 Convert input audio volume to a video output.
26240 The filter accepts the following options:
26242 rate, r
26244 Set video rate.
26245 b Set border width, allowed range is [0, 5]. Default is 1.
26247 w Set channel width, allowed range is [80, 8192]. Default is 400.
26249 h Set channel height, allowed range is [1, 900]. Default is 20.
26251 f Set fade, allowed range is [0, 1]. Default is 0.95.
26253 c Set volume color expression.
26255 The expression can use the following variables:
26257 VOLUME
26259 Current max volume of channel in dB.
26260 PEAK
26262 Current peak.
26263 CHANNEL
26265 Current channel number, starting from 0.
26266 t If set, displays channel names. Default is enabled.
26268 v If set, displays volume values. Default is enabled.
26270 o Set orientation, can be horizontal: "h" or vertical: "v", default
26272 is "h".
26273 s Set step size, allowed range is [0, 5]. Default is 0, which means
26275 step is disabled.
26276 p Set background opacity, allowed range is [0, 1]. Default is 0.
26278 m Set metering mode, can be peak: "p" or rms: "r", default is "p".
26280 ds Set display scale, can be linear: "lin" or log: "log", default is
26282 "lin".
26283 dm In second. If set to > 0., display a line for the max level in the
26285 previous seconds. default is disabled: 0.
26286 dmc The color of the max line. Use when "dm" option is set to > 0.
26288 default is: "orange"
26289 showwaves
26291 Convert input audio to a video output, representing the samples waves.
26292 The filter accepts the following options:
26294 size, s
26296 Specify the video size for the output. For the syntax of this
26297 option, check the "Video size" section in the ffmpeg-utils manual.
26298 Default value is "600x240".
26299 mode
26301 Set display mode.
26302 Available values are:
26304 point
26306 Draw a point for each sample.
26307 line
26309 Draw a vertical line for each sample.
26310 p2p Draw a point for each sample and a line between them.
26312 cline
26314 Draw a centered vertical line for each sample.
26315 Default value is "point".
26317 n Set the number of samples which are printed on the same column. A
26319 larger value will decrease the frame rate. Must be a positive
26320 integer. This option can be set only if the value for rate is not
26321 explicitly specified.
26322 rate, r
26324 Set the (approximate) output frame rate. This is done by setting
26325 the option n. Default value is "25".
26326 split_channels
26328 Set if channels should be drawn separately or overlap. Default
26329 value is 0.
26330 colors
26332 Set colors separated by '|' which are going to be used for drawing
26333 of each channel.
26334 scale
26336 Set amplitude scale.
26337 Available values are:
26339 lin Linear.
26341 log Logarithmic.
26343 sqrt
26345 Square root.
26346 cbrt
26348 Cubic root.
26349 Default is linear.
26351 draw
26353 Set the draw mode. This is mostly useful to set for high n.
26354 Available values are:
26356 scale
26358 Scale pixel values for each drawn sample.
26359 full
26361 Draw every sample directly.
26362 Default value is "scale".
26364 Examples
26366 • Output the input file audio and the corresponding video
26368 representation at the same time:
26369 amovie=a.mp3,asplit[out0],showwaves[out1]
26371 • Create a synthetic signal and show it with showwaves, forcing a
26373 frame rate of 30 frames per second:
26374 aevalsrc=sin(1*2*PI*t)*sin(880*2*PI*t):cos(2*PI*200*t),asplit[out0],showwaves=r=30[out1]
26376 showwavespic
26378 Convert input audio to a single video frame, representing the samples
26379 waves.
26380 The filter accepts the following options:
26382 size, s
26384 Specify the video size for the output. For the syntax of this
26385 option, check the "Video size" section in the ffmpeg-utils manual.
26386 Default value is "600x240".
26387 split_channels
26389 Set if channels should be drawn separately or overlap. Default
26390 value is 0.
26391 colors
26393 Set colors separated by '|' which are going to be used for drawing
26394 of each channel.
26395 scale
26397 Set amplitude scale.
26398 Available values are:
26400 lin Linear.
26402 log Logarithmic.
26404 sqrt
26406 Square root.
26407 cbrt
26409 Cubic root.
26410 Default is linear.
26412 draw
26414 Set the draw mode.
26415 Available values are:
26417 scale
26419 Scale pixel values for each drawn sample.
26420 full
26422 Draw every sample directly.
26423 Default value is "scale".
26425 filter
26427 Set the filter mode.
26428 Available values are:
26430 average
26432 Use average samples values for each drawn sample.
26433 peak
26435 Use peak samples values for each drawn sample.
26436 Default value is "average".
26438 Examples
26440 • Extract a channel split representation of the wave form of a whole
26442 audio track in a 1024x800 picture using ffmpeg:
26443 ffmpeg -i audio.flac -lavfi showwavespic=split_channels=1:s=1024x800 waveform.png
26445 sidedata, asidedata
26447 Delete frame side data, or select frames based on it.
26448 This filter accepts the following options:
26450 mode
26452 Set mode of operation of the filter.
26453 Can be one of the following:
26455 select
26457 Select every frame with side data of "type".
26458 delete
26460 Delete side data of "type". If "type" is not set, delete all
26461 side data in the frame.
26462 type
26464 Set side data type used with all modes. Must be set for "select"
26465 mode. For the list of frame side data types, refer to the
26466 "AVFrameSideDataType" enum in libavutil/frame.h. For example, to
26467 choose "AV_FRAME_DATA_PANSCAN" side data, you must specify
26468 "PANSCAN".
26469 spectrumsynth
26471 Synthesize audio from 2 input video spectrums, first input stream
26472 represents magnitude across time and second represents phase across
26473 time. The filter will transform from frequency domain as displayed in
26474 videos back to time domain as presented in audio output.
26475 This filter is primarily created for reversing processed showspectrum
26477 filter outputs, but can synthesize sound from other spectrograms too.
26478 But in such case results are going to be poor if the phase data is not
26479 available, because in such cases phase data need to be recreated,
26480 usually it's just recreated from random noise. For best results use
26481 gray only output ("channel" color mode in showspectrum filter) and
26482 "log" scale for magnitude video and "lin" scale for phase video. To
26483 produce phase, for 2nd video, use "data" option. Inputs videos should
26484 generally use "fullframe" slide mode as that saves resources needed for
26485 decoding video.
26486 The filter accepts the following options:
26488 sample_rate
26490 Specify sample rate of output audio, the sample rate of audio from
26491 which spectrum was generated may differ.
26492 channels
26494 Set number of channels represented in input video spectrums.
26495 scale
26497 Set scale which was used when generating magnitude input spectrum.
26498 Can be "lin" or "log". Default is "log".
26499 slide
26501 Set slide which was used when generating inputs spectrums. Can be
26502 "replace", "scroll", "fullframe" or "rscroll". Default is
26503 "fullframe".
26504 win_func
26506 Set window function used for resynthesis.
26507 overlap
26509 Set window overlap. In range "[0, 1]". Default is 1, which means
26510 optimal overlap for selected window function will be picked.
26511 orientation
26513 Set orientation of input videos. Can be "vertical" or "horizontal".
26514 Default is "vertical".
26515 Examples
26517 • First create magnitude and phase videos from audio, assuming audio
26519 is stereo with 44100 sample rate, then resynthesize videos back to
26520 audio with spectrumsynth:
26521 ffmpeg -i input.flac -lavfi showspectrum=mode=separate:scale=log:overlap=0.875:color=channel:slide=fullframe:data=magnitude -an -c:v rawvideo magnitude.nut
26523 ffmpeg -i input.flac -lavfi showspectrum=mode=separate:scale=lin:overlap=0.875:color=channel:slide=fullframe:data=phase -an -c:v rawvideo phase.nut
26524 ffmpeg -i magnitude.nut -i phase.nut -lavfi spectrumsynth=channels=2:sample_rate=44100:win_func=hann:overlap=0.875:slide=fullframe output.flac
26525 split, asplit
26527 Split input into several identical outputs.
26528 "asplit" works with audio input, "split" with video.
26530 The filter accepts a single parameter which specifies the number of
26532 outputs. If unspecified, it defaults to 2.
26533 Examples
26535 • Create two separate outputs from the same input:
26537 [in] split [out0][out1]
26539 • To create 3 or more outputs, you need to specify the number of
26541 outputs, like in:
26542 [in] asplit=3 [out0][out1][out2]
26544 • Create two separate outputs from the same input, one cropped and
26546 one padded:
26547 [in] split [splitout1][splitout2];
26549 [splitout1] crop=100:100:0:0 [cropout];
26550 [splitout2] pad=200:200:100:100 [padout];
26551 • Create 5 copies of the input audio with ffmpeg:
26553 ffmpeg -i INPUT -filter_complex asplit=5 OUTPUT
26555 zmq, azmq
26557 Receive commands sent through a libzmq client, and forward them to
26558 filters in the filtergraph.
26559 "zmq" and "azmq" work as a pass-through filters. "zmq" must be inserted
26561 between two video filters, "azmq" between two audio filters. Both are
26562 capable to send messages to any filter type.
26563 To enable these filters you need to install the libzmq library and
26565 headers and configure FFmpeg with "--enable-libzmq".
26566 For more information about libzmq see: <http://www.zeromq.org/>
26568 The "zmq" and "azmq" filters work as a libzmq server, which receives
26570 messages sent through a network interface defined by the bind_address
26571 (or the abbreviation "b") option. Default value of this option is
26572 tcp://localhost:5555. You may want to alter this value to your needs,
26573 but do not forget to escape any ':' signs (see filtergraph escaping).
26574 The received message must be in the form:
26576 <TARGET> <COMMAND> [<ARG>]
26578 TARGET specifies the target of the command, usually the name of the
26580 filter class or a specific filter instance name. The default filter
26581 instance name uses the pattern Parsed_<filter_name>_<index>, but you
26582 can override this by using the filter_name@id syntax (see Filtergraph
26583 syntax).
26584 COMMAND specifies the name of the command for the target filter.
26586 ARG is optional and specifies the optional argument list for the given
26588 COMMAND.
26589 Upon reception, the message is processed and the corresponding command
26591 is injected into the filtergraph. Depending on the result, the filter
26592 will send a reply to the client, adopting the format:
26593 <ERROR_CODE> <ERROR_REASON>
26595 <MESSAGE>
26596 MESSAGE is optional.
26598 Examples
26600 Look at tools/zmqsend for an example of a zmq client which can be used
26602 to send commands processed by these filters.
26603 Consider the following filtergraph generated by ffplay. In this
26605 example the last overlay filter has an instance name. All other filters
26606 will have default instance names.
26607 ffplay -dumpgraph 1 -f lavfi "
26609 color=s=100x100:c=red [l];
26610 color=s=100x100:c=blue [r];
26611 nullsrc=s=200x100, zmq [bg];
26612 [bg][l] overlay [bg+l];
26613 [bg+l][r] overlay@my=x=100 "
26614 To change the color of the left side of the video, the following
26616 command can be used:
26617 echo Parsed_color_0 c yellow | tools/zmqsend
26619 To change the right side:
26621 echo Parsed_color_1 c pink | tools/zmqsend
26623 To change the position of the right side:
26625 echo overlay@my x 150 | tools/zmqsend
26627
26629 Below is a description of the currently available multimedia sources.
26630 amovie
26632 This is the same as movie source, except it selects an audio stream by
26633 default.
26634 avsynctest
26636 Generate an Audio/Video Sync Test.
26637 Generated stream periodically shows flash video frame and emits beep in
26639 audio. Useful to inspect A/V sync issues.
26640 It accepts the following options:
26642 size, s
26644 Set output video size. Default value is "hd720".
26645 framerate, fr
26647 Set output video frame rate. Default value is 30.
26648 samplerate, sr
26650 Set output audio sample rate. Default value is 44100.
26651 amplitude, a
26653 Set output audio beep amplitude. Default value is 0.7.
26654 period, p
26656 Set output audio beep period in seconds. Default value is 3.
26657 delay, dl
26659 Set output video flash delay in number of frames. Default value is
26660 0.
26661 cycle, c
26663 Enable cycling of video delays, by default is disabled.
26664 duration, d
26666 Set stream output duration. By default duration is unlimited.
26667 fg, bg, ag
26669 Set foreground/background/additional color.
26670 movie
26672 Read audio and/or video stream(s) from a movie container.
26673 It accepts the following parameters:
26675 filename
26677 The name of the resource to read (not necessarily a file; it can
26678 also be a device or a stream accessed through some protocol).
26679 format_name, f
26681 Specifies the format assumed for the movie to read, and can be
26682 either the name of a container or an input device. If not
26683 specified, the format is guessed from movie_name or by probing.
26684 seek_point, sp
26686 Specifies the seek point in seconds. The frames will be output
26687 starting from this seek point. The parameter is evaluated with
26688 "av_strtod", so the numerical value may be suffixed by an IS
26689 postfix. The default value is "0".
26690 streams, s
26692 Specifies the streams to read. Several streams can be specified,
26693 separated by "+". The source will then have as many outputs, in the
26694 same order. The syntax is explained in the "Stream specifiers"
26695 section in the ffmpeg manual. Two special names, "dv" and "da"
26696 specify respectively the default (best suited) video and audio
26697 stream. Default is "dv", or "da" if the filter is called as
26698 "amovie".
26699 stream_index, si
26701 Specifies the index of the video stream to read. If the value is
26702 -1, the most suitable video stream will be automatically selected.
26703 The default value is "-1". Deprecated. If the filter is called
26704 "amovie", it will select audio instead of video.
26705 loop
26707 Specifies how many times to read the stream in sequence. If the
26708 value is 0, the stream will be looped infinitely. Default value is
26709 "1".
26710 Note that when the movie is looped the source timestamps are not
26712 changed, so it will generate non monotonically increasing
26713 timestamps.
26714 discontinuity
26716 Specifies the time difference between frames above which the point
26717 is considered a timestamp discontinuity which is removed by
26718 adjusting the later timestamps.
26719 dec_threads
26721 Specifies the number of threads for decoding
26722 format_opts
26724 Specify format options for the opened file. Format options can be
26725 specified as a list of key=value pairs separated by ':'. The
26726 following example shows how to add protocol_whitelist and
26727 protocol_blacklist options:
26728 ffplay -f lavfi
26730 "movie=filename='1.sdp':format_opts='protocol_whitelist=file,rtp,udp\:protocol_blacklist=http'"
26731 It allows overlaying a second video on top of the main input of a
26733 filtergraph, as shown in this graph:
26734 input -----------> deltapts0 --> overlay --> output
26736 ^
26737 |
26738 movie --> scale--> deltapts1 -------+
26739 Examples
26741 • Skip 3.2 seconds from the start of the AVI file in.avi, and overlay
26743 it on top of the input labelled "in":
26744 movie=in.avi:seek_point=3.2, scale=180:-1, setpts=PTS-STARTPTS [over];
26746 [in] setpts=PTS-STARTPTS [main];
26747 [main][over] overlay=16:16 [out]
26748 • Read from a video4linux2 device, and overlay it on top of the input
26750 labelled "in":
26751 movie=/dev/video0:f=video4linux2, scale=180:-1, setpts=PTS-STARTPTS [over];
26753 [in] setpts=PTS-STARTPTS [main];
26754 [main][over] overlay=16:16 [out]
26755 • Read the first video stream and the audio stream with id 0x81 from
26757 dvd.vob; the video is connected to the pad named "video" and the
26758 audio is connected to the pad named "audio":
26759 movie=dvd.vob:s=v:0+#0x81 [video] [audio]
26761 Commands
26763 Both movie and amovie support the following commands:
26765 seek
26767 Perform seek using "av_seek_frame". The syntax is: seek
26768 stream_index|timestamp|flags
26769 • stream_index: If stream_index is -1, a default stream is
26771 selected, and timestamp is automatically converted from
26772 AV_TIME_BASE units to the stream specific time_base.
26773 • timestamp: Timestamp in AVStream.time_base units or, if no
26775 stream is specified, in AV_TIME_BASE units.
26776 • flags: Flags which select direction and seeking mode.
26778 get_duration
26780 Get movie duration in AV_TIME_BASE units.
26781
26783 ffmpeg(1), ffplay(1), ffprobe(1), libavfilter(3)
26784
26786 The FFmpeg developers.
26787 For details about the authorship, see the Git history of the project
26789 (git://source.ffmpeg.org/ffmpeg), e.g. by typing the command git log in
26790 the FFmpeg source directory, or browsing the online repository at
26791 <http://source.ffmpeg.org>.
26792 Maintainers for the specific components are listed in the file
26794 MAINTAINERS in the source code tree.
26795 FFMPEG-FILTERS(1)