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 Examples
504 • Split input audio stream into two bands (low and high) with split
506 frequency of 1500 Hz, each band will be in separate stream:
507 ffmpeg -i in.flac -filter_complex 'acrossover=split=1500[LOW][HIGH]' -map '[LOW]' low.wav -map '[HIGH]' high.wav
509 • Same as above, but with higher filter order:
511 ffmpeg -i in.flac -filter_complex 'acrossover=split=1500:order=8th[LOW][HIGH]' -map '[LOW]' low.wav -map '[HIGH]' high.wav
513 • Same as above, but also with additional middle band (frequencies
515 between 1500 and 8000):
516 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
518 acrusher
520 Reduce audio bit resolution.
521 This filter is bit crusher with enhanced functionality. A bit crusher
523 is used to audibly reduce number of bits an audio signal is sampled
524 with. This doesn't change the bit depth at all, it just produces the
525 effect. Material reduced in bit depth sounds more harsh and "digital".
526 This filter is able to even round to continuous values instead of
527 discrete bit depths. Additionally it has a D/C offset which results in
528 different crushing of the lower and the upper half of the signal. An
529 Anti-Aliasing setting is able to produce "softer" crushing sounds.
530 Another feature of this filter is the logarithmic mode. This setting
532 switches from linear distances between bits to logarithmic ones. The
533 result is a much more "natural" sounding crusher which doesn't gate low
534 signals for example. The human ear has a logarithmic perception, so
535 this kind of crushing is much more pleasant. Logarithmic crushing is
536 also able to get anti-aliased.
537 The filter accepts the following options:
539 level_in
541 Set level in.
542 level_out
544 Set level out.
545 bits
547 Set bit reduction.
548 mix Set mixing amount.
550 mode
552 Can be linear: "lin" or logarithmic: "log".
553 dc Set DC.
555 aa Set anti-aliasing.
557 samples
559 Set sample reduction.
560 lfo Enable LFO. By default disabled.
562 lforange
564 Set LFO range.
565 lforate
567 Set LFO rate.
568 Commands
570 This filter supports the all above options as commands.
572 acue
574 Delay audio filtering until a given wallclock timestamp. See the cue
575 filter.
576 adeclick
578 Remove impulsive noise from input audio.
579 Samples detected as impulsive noise are replaced by interpolated
581 samples using autoregressive modelling.
582 window, w
584 Set window size, in milliseconds. Allowed range is from 10 to 100.
585 Default value is 55 milliseconds. This sets size of window which
586 will be processed at once.
587 overlap, o
589 Set window overlap, in percentage of window size. Allowed range is
590 from 50 to 95. Default value is 75 percent. Setting this to a very
591 high value increases impulsive noise removal but makes whole
592 process much slower.
593 arorder, a
595 Set autoregression order, in percentage of window size. Allowed
596 range is from 0 to 25. Default value is 2 percent. This option also
597 controls quality of interpolated samples using neighbour good
598 samples.
599 threshold, t
601 Set threshold value. Allowed range is from 1 to 100. Default value
602 is 2. This controls the strength of impulsive noise which is going
603 to be removed. The lower value, the more samples will be detected
604 as impulsive noise.
605 burst, b
607 Set burst fusion, in percentage of window size. Allowed range is 0
608 to 10. Default value is 2. If any two samples detected as noise
609 are spaced less than this value then any sample between those two
610 samples will be also detected as noise.
611 method, m
613 Set overlap method.
614 It accepts the following values:
616 add, a
618 Select overlap-add method. Even not interpolated samples are
619 slightly changed with this method.
620 save, s
622 Select overlap-save method. Not interpolated samples remain
623 unchanged.
624 Default value is "a".
626 adeclip
628 Remove clipped samples from input audio.
629 Samples detected as clipped are replaced by interpolated samples using
631 autoregressive modelling.
632 window, w
634 Set window size, in milliseconds. Allowed range is from 10 to 100.
635 Default value is 55 milliseconds. This sets size of window which
636 will be processed at once.
637 overlap, o
639 Set window overlap, in percentage of window size. Allowed range is
640 from 50 to 95. Default value is 75 percent.
641 arorder, a
643 Set autoregression order, in percentage of window size. Allowed
644 range is from 0 to 25. Default value is 8 percent. This option also
645 controls quality of interpolated samples using neighbour good
646 samples.
647 threshold, t
649 Set threshold value. Allowed range is from 1 to 100. Default value
650 is 10. Higher values make clip detection less aggressive.
651 hsize, n
653 Set size of histogram used to detect clips. Allowed range is from
654 100 to 9999. Default value is 1000. Higher values make clip
655 detection less aggressive.
656 method, m
658 Set overlap method.
659 It accepts the following values:
661 add, a
663 Select overlap-add method. Even not interpolated samples are
664 slightly changed with this method.
665 save, s
667 Select overlap-save method. Not interpolated samples remain
668 unchanged.
669 Default value is "a".
671 adecorrelate
673 Apply decorrelation to input audio stream.
674 The filter accepts the following options:
676 stages
678 Set decorrelation stages of filtering. Allowed range is from 1 to
679 16. Default value is 6.
680 seed
682 Set random seed used for setting delay in samples across channels.
683 adelay
685 Delay one or more audio channels.
686 Samples in delayed channel are filled with silence.
688 The filter accepts the following option:
690 delays
692 Set list of delays in milliseconds for each channel separated by
693 '|'. Unused delays will be silently ignored. If number of given
694 delays is smaller than number of channels all remaining channels
695 will not be delayed. If you want to delay exact number of samples,
696 append 'S' to number. If you want instead to delay in seconds,
697 append 's' to number.
698 all Use last set delay for all remaining channels. By default is
700 disabled. This option if enabled changes how option "delays" is
701 interpreted.
702 Examples
704 • Delay first channel by 1.5 seconds, the third channel by 0.5
706 seconds and leave the second channel (and any other channels that
707 may be present) unchanged.
708 adelay=1500|0|500
710 • Delay second channel by 500 samples, the third channel by 700
712 samples and leave the first channel (and any other channels that
713 may be present) unchanged.
714 adelay=0|500S|700S
716 • Delay all channels by same number of samples:
718 adelay=delays=64S:all=1
720 adenorm
722 Remedy denormals in audio by adding extremely low-level noise.
723 This filter shall be placed before any filter that can produce
725 denormals.
726 A description of the accepted parameters follows.
728 level
730 Set level of added noise in dB. Default is "-351". Allowed range
731 is from -451 to -90.
732 type
734 Set type of added noise.
735 dc Add DC signal.
737 ac Add AC signal.
739 square
741 Add square signal.
742 pulse
744 Add pulse signal.
745 Default is "dc".
747 Commands
749 This filter supports the all above options as commands.
751 aderivative, aintegral
753 Compute derivative/integral of audio stream.
754 Applying both filters one after another produces original audio.
756 adynamicequalizer
758 Apply dynamic equalization to input audio stream.
759 A description of the accepted options follows.
761 threshold
763 Set the detection threshold used to trigger equalization.
764 Threshold detection is using bandpass filter. Default value is 0.
765 Allowed range is from 0 to 100.
766 dfrequency
768 Set the detection frequency in Hz used for bandpass filter used to
769 trigger equalization. Default value is 1000 Hz. Allowed range is
770 between 2 and 1000000 Hz.
771 dqfactor
773 Set the detection resonance factor for bandpass filter used to
774 trigger equalization. Default value is 1. Allowed range is from
775 0.001 to 1000.
776 tfrequency
778 Set the target frequency of equalization filter. Default value is
779 1000 Hz. Allowed range is between 2 and 1000000 Hz.
780 tqfactor
782 Set the target resonance factor for target equalization filter.
783 Default value is 1. Allowed range is from 0.001 to 1000.
784 attack
786 Set the amount of milliseconds the signal from detection has to
787 rise above the detection threshold before equalization starts.
788 Default is 20. Allowed range is between 1 and 2000.
789 release
791 Set the amount of milliseconds the signal from detection has to
792 fall below the detection threshold before equalization ends.
793 Default is 200. Allowed range is between 1 and 2000.
794 knee
796 Curve the sharp knee around the detection threshold to calculate
797 equalization gain more softly. Default is 1. Allowed range is
798 between 0 and 8.
799 ratio
801 Set the ratio by which the equalization gain is raised. Default is
802 1. Allowed range is between 1 and 20.
803 makeup
805 Set the makeup offset in dB by which the equalization gain is
806 raised. Default is 0. Allowed range is between 0 and 30.
807 range
809 Set the max allowed cut/boost amount in dB. Default is 0. Allowed
810 range is from 0 to 200.
811 slew
813 Set the slew factor. Default is 1. Allowed range is from 1 to 200.
814 mode
816 Set the mode of filter operation, can be one of the following:
817 listen
819 Output only isolated bandpass signal.
820 cut Cut frequencies above detection threshold.
822 boost
824 Boost frequencies bellow detection threshold.
825 Default mode is cut.
827 Commands
829 This filter supports the all above options as commands.
831 adynamicsmooth
833 Apply dynamic smoothing to input audio stream.
834 A description of the accepted options follows.
836 sensitivity
838 Set an amount of sensitivity to frequency fluctations. Default is
839 2. Allowed range is from 0 to 1e+06.
840 basefreq
842 Set a base frequency for smoothing. Default value is 22050.
843 Allowed range is from 2 to 1e+06.
844 Commands
846 This filter supports the all above options as commands.
848 aecho
850 Apply echoing to the input audio.
851 Echoes are reflected sound and can occur naturally amongst mountains
853 (and sometimes large buildings) when talking or shouting; digital echo
854 effects emulate this behaviour and are often used to help fill out the
855 sound of a single instrument or vocal. The time difference between the
856 original signal and the reflection is the "delay", and the loudness of
857 the reflected signal is the "decay". Multiple echoes can have
858 different delays and decays.
859 A description of the accepted parameters follows.
861 in_gain
863 Set input gain of reflected signal. Default is 0.6.
864 out_gain
866 Set output gain of reflected signal. Default is 0.3.
867 delays
869 Set list of time intervals in milliseconds between original signal
870 and reflections separated by '|'. Allowed range for each "delay" is
871 "(0 - 90000.0]". Default is 1000.
872 decays
874 Set list of loudness of reflected signals separated by '|'.
875 Allowed range for each "decay" is "(0 - 1.0]". Default is 0.5.
876 Examples
878 • Make it sound as if there are twice as many instruments as are
880 actually playing:
881 aecho=0.8:0.88:60:0.4
883 • If delay is very short, then it sounds like a (metallic) robot
885 playing music:
886 aecho=0.8:0.88:6:0.4
888 • A longer delay will sound like an open air concert in the
890 mountains:
891 aecho=0.8:0.9:1000:0.3
893 • Same as above but with one more mountain:
895 aecho=0.8:0.9:1000|1800:0.3|0.25
897 aemphasis
899 Audio emphasis filter creates or restores material directly taken from
900 LPs or emphased CDs with different filter curves. E.g. to store music
901 on vinyl the signal has to be altered by a filter first to even out the
902 disadvantages of this recording medium. Once the material is played
903 back the inverse filter has to be applied to restore the distortion of
904 the frequency response.
905 The filter accepts the following options:
907 level_in
909 Set input gain.
910 level_out
912 Set output gain.
913 mode
915 Set filter mode. For restoring material use "reproduction" mode,
916 otherwise use "production" mode. Default is "reproduction" mode.
917 type
919 Set filter type. Selects medium. Can be one of the following:
920 col select Columbia.
922 emi select EMI.
924 bsi select BSI (78RPM).
926 riaa
928 select RIAA.
929 cd select Compact Disc (CD).
931 50fm
933 select 50Xs (FM).
934 75fm
936 select 75Xs (FM).
937 50kf
939 select 50Xs (FM-KF).
940 75kf
942 select 75Xs (FM-KF).
943 Commands
945 This filter supports the all above options as commands.
947 aeval
949 Modify an audio signal according to the specified expressions.
950 This filter accepts one or more expressions (one for each channel),
952 which are evaluated and used to modify a corresponding audio signal.
953 It accepts the following parameters:
955 exprs
957 Set the '|'-separated expressions list for each separate channel.
958 If the number of input channels is greater than the number of
959 expressions, the last specified expression is used for the
960 remaining output channels.
961 channel_layout, c
963 Set output channel layout. If not specified, the channel layout is
964 specified by the number of expressions. If set to same, it will use
965 by default the same input channel layout.
966 Each expression in exprs can contain the following constants and
968 functions:
969 ch channel number of the current expression
971 n number of the evaluated sample, starting from 0
973 s sample rate
975 t time of the evaluated sample expressed in seconds
977 nb_in_channels
979 nb_out_channels
980 input and output number of channels
981 val(CH)
983 the value of input channel with number CH
984 Note: this filter is slow. For faster processing you should use a
986 dedicated filter.
987 Examples
989 • Half volume:
991 aeval=val(ch)/2:c=same
993 • Invert phase of the second channel:
995 aeval=val(0)|-val(1)
997 aexciter
999 An exciter is used to produce high sound that is not present in the
1000 original signal. This is done by creating harmonic distortions of the
1001 signal which are restricted in range and added to the original signal.
1002 An Exciter raises the upper end of an audio signal without simply
1003 raising the higher frequencies like an equalizer would do to create a
1004 more "crisp" or "brilliant" sound.
1005 The filter accepts the following options:
1007 level_in
1009 Set input level prior processing of signal. Allowed range is from
1010 0 to 64. Default value is 1.
1011 level_out
1013 Set output level after processing of signal. Allowed range is from
1014 0 to 64. Default value is 1.
1015 amount
1017 Set the amount of harmonics added to original signal. Allowed
1018 range is from 0 to 64. Default value is 1.
1019 drive
1021 Set the amount of newly created harmonics. Allowed range is from
1022 0.1 to 10. Default value is 8.5.
1023 blend
1025 Set the octave of newly created harmonics. Allowed range is from
1026 -10 to 10. Default value is 0.
1027 freq
1029 Set the lower frequency limit of producing harmonics in Hz.
1030 Allowed range is from 2000 to 12000 Hz. Default is 7500 Hz.
1031 ceil
1033 Set the upper frequency limit of producing harmonics. Allowed
1034 range is from 9999 to 20000 Hz. If value is lower than 10000 Hz no
1035 limit is applied.
1036 listen
1038 Mute the original signal and output only added harmonics. By
1039 default is disabled.
1040 Commands
1042 This filter supports the all above options as commands.
1044 afade
1046 Apply fade-in/out effect to input audio.
1047 A description of the accepted parameters follows.
1049 type, t
1051 Specify the effect type, can be either "in" for fade-in, or "out"
1052 for a fade-out effect. Default is "in".
1053 start_sample, ss
1055 Specify the number of the start sample for starting to apply the
1056 fade effect. Default is 0.
1057 nb_samples, ns
1059 Specify the number of samples for which the fade effect has to
1060 last. At the end of the fade-in effect the output audio will have
1061 the same volume as the input audio, at the end of the fade-out
1062 transition the output audio will be silence. Default is 44100.
1063 start_time, st
1065 Specify the start time of the fade effect. Default is 0. The value
1066 must be specified as a time duration; see the Time duration section
1067 in the ffmpeg-utils(1) manual for the accepted syntax. If set this
1068 option is used instead of start_sample.
1069 duration, d
1071 Specify the duration of the fade effect. See the Time duration
1072 section in the ffmpeg-utils(1) manual for the accepted syntax. At
1073 the end of the fade-in effect the output audio will have the same
1074 volume as the input audio, at the end of the fade-out transition
1075 the output audio will be silence. By default the duration is
1076 determined by nb_samples. If set this option is used instead of
1077 nb_samples.
1078 curve
1080 Set curve for fade transition.
1081 It accepts the following values:
1083 tri select triangular, linear slope (default)
1085 qsin
1087 select quarter of sine wave
1088 hsin
1090 select half of sine wave
1091 esin
1093 select exponential sine wave
1094 log select logarithmic
1096 ipar
1098 select inverted parabola
1099 qua select quadratic
1101 cub select cubic
1103 squ select square root
1105 cbr select cubic root
1107 par select parabola
1109 exp select exponential
1111 iqsin
1113 select inverted quarter of sine wave
1114 ihsin
1116 select inverted half of sine wave
1117 dese
1119 select double-exponential seat
1120 desi
1122 select double-exponential sigmoid
1123 losi
1125 select logistic sigmoid
1126 sinc
1128 select sine cardinal function
1129 isinc
1131 select inverted sine cardinal function
1132 nofade
1134 no fade applied
1135 Commands
1137 This filter supports the all above options as commands.
1139 Examples
1141 • Fade in first 15 seconds of audio:
1143 afade=t=in:ss=0:d=15
1145 • Fade out last 25 seconds of a 900 seconds audio:
1147 afade=t=out:st=875:d=25
1149 afftdn
1151 Denoise audio samples with FFT.
1152 A description of the accepted parameters follows.
1154 nr Set the noise reduction in dB, allowed range is 0.01 to 97.
1156 Default value is 12 dB.
1157 nf Set the noise floor in dB, allowed range is -80 to -20. Default
1159 value is -50 dB.
1160 nt Set the noise type.
1162 It accepts the following values:
1164 w Select white noise.
1166 v Select vinyl noise.
1168 s Select shellac noise.
1170 c Select custom noise, defined in "bn" option.
1172 Default value is white noise.
1174 bn Set custom band noise for every one of 15 bands. Bands are
1176 separated by ' ' or '|'.
1177 rf Set the residual floor in dB, allowed range is -80 to -20. Default
1179 value is -38 dB.
1180 tn Enable noise tracking. By default is disabled. With this enabled,
1182 noise floor is automatically adjusted.
1183 tr Enable residual tracking. By default is disabled.
1185 om Set the output mode.
1187 It accepts the following values:
1189 i Pass input unchanged.
1191 o Pass noise filtered out.
1193 n Pass only noise.
1195 Default value is o.
1197 Commands
1199 This filter supports the following commands:
1201 sample_noise, sn
1203 Start or stop measuring noise profile. Syntax for the command is :
1204 "start" or "stop" string. After measuring noise profile is stopped
1205 it will be automatically applied in filtering.
1206 noise_reduction, nr
1208 Change noise reduction. Argument is single float number. Syntax
1209 for the command is : "noise_reduction"
1210 noise_floor, nf
1212 Change noise floor. Argument is single float number. Syntax for
1213 the command is : "noise_floor"
1214 output_mode, om
1216 Change output mode operation. Syntax for the command is : "i", "o"
1217 or "n" string.
1218 afftfilt
1220 Apply arbitrary expressions to samples in frequency domain.
1221 real
1223 Set frequency domain real expression for each separate channel
1224 separated by '|'. Default is "re". If the number of input channels
1225 is greater than the number of expressions, the last specified
1226 expression is used for the remaining output channels.
1227 imag
1229 Set frequency domain imaginary expression for each separate channel
1230 separated by '|'. Default is "im".
1231 Each expression in real and imag can contain the following
1233 constants and functions:
1234 sr sample rate
1236 b current frequency bin number
1238 nb number of available bins
1240 ch channel number of the current expression
1242 chs number of channels
1244 pts current frame pts
1246 re current real part of frequency bin of current channel
1248 im current imaginary part of frequency bin of current channel
1250 real(b, ch)
1252 Return the value of real part of frequency bin at location
1253 (bin,channel)
1254 imag(b, ch)
1256 Return the value of imaginary part of frequency bin at location
1257 (bin,channel)
1258 win_size
1260 Set window size. Allowed range is from 16 to 131072. Default is
1261 4096
1262 win_func
1264 Set window function.
1265 It accepts the following values:
1267 rect
1269 bartlett
1270 hann, hanning
1271 hamming
1272 blackman
1273 welch
1274 flattop
1275 bharris
1276 bnuttall
1277 bhann
1278 sine
1279 nuttall
1280 lanczos
1281 gauss
1282 tukey
1283 dolph
1284 cauchy
1285 parzen
1286 poisson
1287 bohman
1288 Default is "hann".
1290 overlap
1292 Set window overlap. If set to 1, the recommended overlap for
1293 selected window function will be picked. Default is 0.75.
1294 Examples
1296 • Leave almost only low frequencies in audio:
1298 afftfilt="'real=re * (1-clip((b/nb)*b,0,1))':imag='im * (1-clip((b/nb)*b,0,1))'"
1300 • Apply robotize effect:
1302 afftfilt="real='hypot(re,im)*sin(0)':imag='hypot(re,im)*cos(0)':win_size=512:overlap=0.75"
1304 • Apply whisper effect:
1306 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"
1308 afir
1310 Apply an arbitrary Finite Impulse Response filter.
1311 This filter is designed for applying long FIR filters, up to 60 seconds
1313 long.
1314 It can be used as component for digital crossover filters, room
1316 equalization, cross talk cancellation, wavefield synthesis,
1317 auralization, ambiophonics, ambisonics and spatialization.
1318 This filter uses the streams higher than first one as FIR coefficients.
1320 If the non-first stream holds a single channel, it will be used for all
1321 input channels in the first stream, otherwise the number of channels in
1322 the non-first stream must be same as the number of channels in the
1323 first stream.
1324 It accepts the following parameters:
1326 dry Set dry gain. This sets input gain.
1328 wet Set wet gain. This sets final output gain.
1330 length
1332 Set Impulse Response filter length. Default is 1, which means whole
1333 IR is processed.
1334 gtype
1336 Enable applying gain measured from power of IR.
1337 Set which approach to use for auto gain measurement.
1339 none
1341 Do not apply any gain.
1342 peak
1344 select peak gain, very conservative approach. This is default
1345 value.
1346 dc select DC gain, limited application.
1348 gn select gain to noise approach, this is most popular one.
1350 irgain
1352 Set gain to be applied to IR coefficients before filtering.
1353 Allowed range is 0 to 1. This gain is applied after any gain
1354 applied with gtype option.
1355 irfmt
1357 Set format of IR stream. Can be "mono" or "input". Default is
1358 "input".
1359 maxir
1361 Set max allowed Impulse Response filter duration in seconds.
1362 Default is 30 seconds. Allowed range is 0.1 to 60 seconds.
1363 response
1365 Show IR frequency response, magnitude(magenta), phase(green) and
1366 group delay(yellow) in additional video stream. By default it is
1367 disabled.
1368 channel
1370 Set for which IR channel to display frequency response. By default
1371 is first channel displayed. This option is used only when response
1372 is enabled.
1373 size
1375 Set video stream size. This option is used only when response is
1376 enabled.
1377 rate
1379 Set video stream frame rate. This option is used only when response
1380 is enabled.
1381 minp
1383 Set minimal partition size used for convolution. Default is 8192.
1384 Allowed range is from 1 to 32768. Lower values decreases latency
1385 at cost of higher CPU usage.
1386 maxp
1388 Set maximal partition size used for convolution. Default is 8192.
1389 Allowed range is from 8 to 32768. Lower values may increase CPU
1390 usage.
1391 nbirs
1393 Set number of input impulse responses streams which will be
1394 switchable at runtime. Allowed range is from 1 to 32. Default is
1395 1.
1396 ir Set IR stream which will be used for convolution, starting from 0,
1398 should always be lower than supplied value by "nbirs" option.
1399 Default is 0. This option can be changed at runtime via commands.
1400 Examples
1402 • Apply reverb to stream using mono IR file as second input, complete
1404 command using ffmpeg:
1405 ffmpeg -i input.wav -i middle_tunnel_1way_mono.wav -lavfi afir output.wav
1407 aformat
1409 Set output format constraints for the input audio. The framework will
1410 negotiate the most appropriate format to minimize conversions.
1411 It accepts the following parameters:
1413 sample_fmts, f
1415 A '|'-separated list of requested sample formats.
1416 sample_rates, r
1418 A '|'-separated list of requested sample rates.
1419 channel_layouts, cl
1421 A '|'-separated list of requested channel layouts.
1422 See the Channel Layout section in the ffmpeg-utils(1) manual for
1424 the required syntax.
1425 If a parameter is omitted, all values are allowed.
1427 Force the output to either unsigned 8-bit or signed 16-bit stereo
1429 aformat=sample_fmts=u8|s16:channel_layouts=stereo
1431 afreqshift
1433 Apply frequency shift to input audio samples.
1434 The filter accepts the following options:
1436 shift
1438 Specify frequency shift. Allowed range is -INT_MAX to INT_MAX.
1439 Default value is 0.0.
1440 level
1442 Set output gain applied to final output. Allowed range is from 0.0
1443 to 1.0. Default value is 1.0.
1444 order
1446 Set filter order used for filtering. Allowed range is from 1 to 16.
1447 Default value is 8.
1448 Commands
1450 This filter supports the all above options as commands.
1452 afwtdn
1454 Reduce broadband noise from input samples using Wavelets.
1455 A description of the accepted options follows.
1457 sigma
1459 Set the noise sigma, allowed range is from 0 to 1. Default value
1460 is 0. This option controls strength of denoising applied to input
1461 samples. Most useful way to set this option is via decibels, eg.
1462 -45dB.
1463 levels
1465 Set the number of wavelet levels of decomposition. Allowed range
1466 is from 1 to 12. Default value is 10. Setting this too low make
1467 denoising performance very poor.
1468 wavet
1470 Set wavelet type for decomposition of input frame. They are sorted
1471 by number of coefficients, from lowest to highest. More
1472 coefficients means worse filtering speed, but overall better
1473 quality. Available wavelets are:
1474 sym2
1476 sym4
1477 rbior68
1478 deb10
1479 sym10
1480 coif5
1481 bl3
1482 percent
1483 Set percent of full denoising. Allowed range is from 0 to 100
1484 percent. Default value is 85 percent or partial denoising.
1485 profile
1487 If enabled, first input frame will be used as noise profile. If
1488 first frame samples contain non-noise performance will be very
1489 poor.
1490 adaptive
1492 If enabled, input frames are analyzed for presence of noise. If
1493 noise is detected with high possibility then input frame profile
1494 will be used for processing following frames, until new noise frame
1495 is detected.
1496 samples
1498 Set size of single frame in number of samples. Allowed range is
1499 from 512 to 65536. Default frame size is 8192 samples.
1500 softness
1502 Set softness applied inside thresholding function. Allowed range is
1503 from 0 to 10. Default softness is 1.
1504 Commands
1506 This filter supports the all above options as commands.
1508 agate
1510 A gate is mainly used to reduce lower parts of a signal. This kind of
1511 signal processing reduces disturbing noise between useful signals.
1512 Gating is done by detecting the volume below a chosen level threshold
1514 and dividing it by the factor set with ratio. The bottom of the noise
1515 floor is set via range. Because an exact manipulation of the signal
1516 would cause distortion of the waveform the reduction can be levelled
1517 over time. This is done by setting attack and release.
1518 attack determines how long the signal has to fall below the threshold
1520 before any reduction will occur and release sets the time the signal
1521 has to rise above the threshold to reduce the reduction again. Shorter
1522 signals than the chosen attack time will be left untouched.
1523 level_in
1525 Set input level before filtering. Default is 1. Allowed range is
1526 from 0.015625 to 64.
1527 mode
1529 Set the mode of operation. Can be "upward" or "downward". Default
1530 is "downward". If set to "upward" mode, higher parts of signal will
1531 be amplified, expanding dynamic range in upward direction.
1532 Otherwise, in case of "downward" lower parts of signal will be
1533 reduced.
1534 range
1536 Set the level of gain reduction when the signal is below the
1537 threshold. Default is 0.06125. Allowed range is from 0 to 1.
1538 Setting this to 0 disables reduction and then filter behaves like
1539 expander.
1540 threshold
1542 If a signal rises above this level the gain reduction is released.
1543 Default is 0.125. Allowed range is from 0 to 1.
1544 ratio
1546 Set a ratio by which the signal is reduced. Default is 2. Allowed
1547 range is from 1 to 9000.
1548 attack
1550 Amount of milliseconds the signal has to rise above the threshold
1551 before gain reduction stops. Default is 20 milliseconds. Allowed
1552 range is from 0.01 to 9000.
1553 release
1555 Amount of milliseconds the signal has to fall below the threshold
1556 before the reduction is increased again. Default is 250
1557 milliseconds. Allowed range is from 0.01 to 9000.
1558 makeup
1560 Set amount of amplification of signal after processing. Default is
1561 1. Allowed range is from 1 to 64.
1562 knee
1564 Curve the sharp knee around the threshold to enter gain reduction
1565 more softly. Default is 2.828427125. Allowed range is from 1 to 8.
1566 detection
1568 Choose if exact signal should be taken for detection or an RMS like
1569 one. Default is "rms". Can be "peak" or "rms".
1570 link
1572 Choose if the average level between all channels or the louder
1573 channel affects the reduction. Default is "average". Can be
1574 "average" or "maximum".
1575 Commands
1577 This filter supports the all above options as commands.
1579 aiir
1581 Apply an arbitrary Infinite Impulse Response filter.
1582 It accepts the following parameters:
1584 zeros, z
1586 Set B/numerator/zeros/reflection coefficients.
1587 poles, p
1589 Set A/denominator/poles/ladder coefficients.
1590 gains, k
1592 Set channels gains.
1593 dry_gain
1595 Set input gain.
1596 wet_gain
1598 Set output gain.
1599 format, f
1601 Set coefficients format.
1602 ll lattice-ladder function
1604 sf analog transfer function
1606 tf digital transfer function
1608 zp Z-plane zeros/poles, cartesian (default)
1610 pr Z-plane zeros/poles, polar radians
1612 pd Z-plane zeros/poles, polar degrees
1614 sp S-plane zeros/poles
1616 process, r
1618 Set type of processing.
1619 d direct processing
1621 s serial processing
1623 p parallel processing
1625 precision, e
1627 Set filtering precision.
1628 dbl double-precision floating-point (default)
1630 flt single-precision floating-point
1632 i32 32-bit integers
1634 i16 16-bit integers
1636 normalize, n
1638 Normalize filter coefficients, by default is enabled. Enabling it
1639 will normalize magnitude response at DC to 0dB.
1640 mix How much to use filtered signal in output. Default is 1. Range is
1642 between 0 and 1.
1643 response
1645 Show IR frequency response, magnitude(magenta), phase(green) and
1646 group delay(yellow) in additional video stream. By default it is
1647 disabled.
1648 channel
1650 Set for which IR channel to display frequency response. By default
1651 is first channel displayed. This option is used only when response
1652 is enabled.
1653 size
1655 Set video stream size. This option is used only when response is
1656 enabled.
1657 Coefficients in "tf" and "sf" format are separated by spaces and are in
1659 ascending order.
1660 Coefficients in "zp" format are separated by spaces and order of
1662 coefficients doesn't matter. Coefficients in "zp" format are complex
1663 numbers with i imaginary unit.
1664 Different coefficients and gains can be provided for every channel, in
1666 such case use '|' to separate coefficients or gains. Last provided
1667 coefficients will be used for all remaining channels.
1668 Examples
1670 • Apply 2 pole elliptic notch at around 5000Hz for 48000 Hz sample
1672 rate:
1673 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
1675 • Same as above but in "zp" format:
1677 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
1679 • Apply 3-rd order analog normalized Butterworth low-pass filter,
1681 using analog transfer function format:
1682 aiir=z=1.3057 0 0 0:p=1.3057 2.3892 2.1860 1:f=sf:r=d
1684 alimiter
1686 The limiter prevents an input signal from rising over a desired
1687 threshold. This limiter uses lookahead technology to prevent your
1688 signal from distorting. It means that there is a small delay after the
1689 signal is processed. Keep in mind that the delay it produces is the
1690 attack time you set.
1691 The filter accepts the following options:
1693 level_in
1695 Set input gain. Default is 1.
1696 level_out
1698 Set output gain. Default is 1.
1699 limit
1701 Don't let signals above this level pass the limiter. Default is 1.
1702 attack
1704 The limiter will reach its attenuation level in this amount of time
1705 in milliseconds. Default is 5 milliseconds.
1706 release
1708 Come back from limiting to attenuation 1.0 in this amount of
1709 milliseconds. Default is 50 milliseconds.
1710 asc When gain reduction is always needed ASC takes care of releasing to
1712 an average reduction level rather than reaching a reduction of 0 in
1713 the release time.
1714 asc_level
1716 Select how much the release time is affected by ASC, 0 means nearly
1717 no changes in release time while 1 produces higher release times.
1718 level
1720 Auto level output signal. Default is enabled. This normalizes
1721 audio back to 0dB if enabled.
1722 Depending on picked setting it is recommended to upsample input 2x or
1724 4x times with aresample before applying this filter.
1725 allpass
1727 Apply a two-pole all-pass filter with central frequency (in Hz)
1728 frequency, and filter-width width. An all-pass filter changes the
1729 audio's frequency to phase relationship without changing its frequency
1730 to amplitude relationship.
1731 The filter accepts the following options:
1733 frequency, f
1735 Set frequency in Hz.
1736 width_type, t
1738 Set method to specify band-width of filter.
1739 h Hz
1741 q Q-Factor
1743 o octave
1745 s slope
1747 k kHz
1749 width, w
1751 Specify the band-width of a filter in width_type units.
1752 mix, m
1754 How much to use filtered signal in output. Default is 1. Range is
1755 between 0 and 1.
1756 channels, c
1758 Specify which channels to filter, by default all available are
1759 filtered.
1760 normalize, n
1762 Normalize biquad coefficients, by default is disabled. Enabling it
1763 will normalize magnitude response at DC to 0dB.
1764 order, o
1766 Set the filter order, can be 1 or 2. Default is 2.
1767 transform, a
1769 Set transform type of IIR filter.
1770 di
1772 dii
1773 tdii
1774 latt
1775 svf
1776 precision, r
1777 Set precison of filtering.
1778 auto
1780 Pick automatic sample format depending on surround filters.
1781 s16 Always use signed 16-bit.
1783 s32 Always use signed 32-bit.
1785 f32 Always use float 32-bit.
1787 f64 Always use float 64-bit.
1789 Commands
1791 This filter supports the following commands:
1793 frequency, f
1795 Change allpass frequency. Syntax for the command is : "frequency"
1796 width_type, t
1798 Change allpass width_type. Syntax for the command is :
1799 "width_type"
1800 width, w
1802 Change allpass width. Syntax for the command is : "width"
1803 mix, m
1805 Change allpass mix. Syntax for the command is : "mix"
1806 aloop
1808 Loop audio samples.
1809 The filter accepts the following options:
1811 loop
1813 Set the number of loops. Setting this value to -1 will result in
1814 infinite loops. Default is 0.
1815 size
1817 Set maximal number of samples. Default is 0.
1818 start
1820 Set first sample of loop. Default is 0.
1821 amerge
1823 Merge two or more audio streams into a single multi-channel stream.
1824 The filter accepts the following options:
1826 inputs
1828 Set the number of inputs. Default is 2.
1829 If the channel layouts of the inputs are disjoint, and therefore
1831 compatible, the channel layout of the output will be set accordingly
1832 and the channels will be reordered as necessary. If the channel layouts
1833 of the inputs are not disjoint, the output will have all the channels
1834 of the first input then all the channels of the second input, in that
1835 order, and the channel layout of the output will be the default value
1836 corresponding to the total number of channels.
1837 For example, if the first input is in 2.1 (FL+FR+LF) and the second
1839 input is FC+BL+BR, then the output will be in 5.1, with the channels in
1840 the following order: a1, a2, b1, a3, b2, b3 (a1 is the first channel of
1841 the first input, b1 is the first channel of the second input).
1842 On the other hand, if both input are in stereo, the output channels
1844 will be in the default order: a1, a2, b1, b2, and the channel layout
1845 will be arbitrarily set to 4.0, which may or may not be the expected
1846 value.
1847 All inputs must have the same sample rate, and format.
1849 If inputs do not have the same duration, the output will stop with the
1851 shortest.
1852 Examples
1854 • Merge two mono files into a stereo stream:
1856 amovie=left.wav [l] ; amovie=right.mp3 [r] ; [l] [r] amerge
1858 • Multiple merges assuming 1 video stream and 6 audio streams in
1860 input.mkv:
1861 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
1863 amix
1865 Mixes multiple audio inputs into a single output.
1866 Note that this filter only supports float samples (the amerge and pan
1868 audio filters support many formats). If the amix input has integer
1869 samples then aresample will be automatically inserted to perform the
1870 conversion to float samples.
1871 For example
1873 ffmpeg -i INPUT1 -i INPUT2 -i INPUT3 -filter_complex amix=inputs=3:duration=first:dropout_transition=3 OUTPUT
1875 will mix 3 input audio streams to a single output with the same
1877 duration as the first input and a dropout transition time of 3 seconds.
1878 It accepts the following parameters:
1880 inputs
1882 The number of inputs. If unspecified, it defaults to 2.
1883 duration
1885 How to determine the end-of-stream.
1886 longest
1888 The duration of the longest input. (default)
1889 shortest
1891 The duration of the shortest input.
1892 first
1894 The duration of the first input.
1895 dropout_transition
1897 The transition time, in seconds, for volume renormalization when an
1898 input stream ends. The default value is 2 seconds.
1899 weights
1901 Specify weight of each input audio stream as sequence. Each weight
1902 is separated by space. By default all inputs have same weight.
1903 normalize
1905 Always scale inputs instead of only doing summation of samples.
1906 Beware of heavy clipping if inputs are not normalized prior or
1907 after filtering by this filter if this option is disabled. By
1908 default is enabled.
1909 Commands
1911 This filter supports the following commands:
1913 weights
1915 normalize
1916 Syntax is same as option with same name.
1917 amultiply
1919 Multiply first audio stream with second audio stream and store result
1920 in output audio stream. Multiplication is done by multiplying each
1921 sample from first stream with sample at same position from second
1922 stream.
1923 With this element-wise multiplication one can create amplitude fades
1925 and amplitude modulations.
1926 anequalizer
1928 High-order parametric multiband equalizer for each channel.
1929 It accepts the following parameters:
1931 params
1933 This option string is in format: "cchn f=cf w=w g=g t=f | ..."
1934 Each equalizer band is separated by '|'.
1935 chn Set channel number to which equalization will be applied. If
1937 input doesn't have that channel the entry is ignored.
1938 f Set central frequency for band. If input doesn't have that
1940 frequency the entry is ignored.
1941 w Set band width in Hertz.
1943 g Set band gain in dB.
1945 t Set filter type for band, optional, can be:
1947 0 Butterworth, this is default.
1949 1 Chebyshev type 1.
1951 2 Chebyshev type 2.
1953 curves
1955 With this option activated frequency response of anequalizer is
1956 displayed in video stream.
1957 size
1959 Set video stream size. Only useful if curves option is activated.
1960 mgain
1962 Set max gain that will be displayed. Only useful if curves option
1963 is activated. Setting this to a reasonable value makes it possible
1964 to display gain which is derived from neighbour bands which are too
1965 close to each other and thus produce higher gain when both are
1966 activated.
1967 fscale
1969 Set frequency scale used to draw frequency response in video
1970 output. Can be linear or logarithmic. Default is logarithmic.
1971 colors
1973 Set color for each channel curve which is going to be displayed in
1974 video stream. This is list of color names separated by space or by
1975 '|'. Unrecognised or missing colors will be replaced by white
1976 color.
1977 Examples
1979 • Lower gain by 10 of central frequency 200Hz and width 100 Hz for
1981 first 2 channels using Chebyshev type 1 filter:
1982 anequalizer=c0 f=200 w=100 g=-10 t=1|c1 f=200 w=100 g=-10 t=1
1984 Commands
1986 This filter supports the following commands:
1988 change
1990 Alter existing filter parameters. Syntax for the commands is :
1991 "fN|f=freq|w=width|g=gain"
1992 fN is existing filter number, starting from 0, if no such filter is
1994 available error is returned. freq set new frequency parameter.
1995 width set new width parameter in Hertz. gain set new gain
1996 parameter in dB.
1997 Full filter invocation with asendcmd may look like this:
1999 asendcmd=c='4.0 anequalizer change
2000 0|f=200|w=50|g=1',anequalizer=...
2001 anlmdn
2003 Reduce broadband noise in audio samples using Non-Local Means
2004 algorithm.
2005 Each sample is adjusted by looking for other samples with similar
2007 contexts. This context similarity is defined by comparing their
2008 surrounding patches of size p. Patches are searched in an area of r
2009 around the sample.
2010 The filter accepts the following options:
2012 strength, s
2014 Set denoising strength. Allowed range is from 0.00001 to 10.
2015 Default value is 0.00001.
2016 patch, p
2018 Set patch radius duration. Allowed range is from 1 to 100
2019 milliseconds. Default value is 2 milliseconds.
2020 research, r
2022 Set research radius duration. Allowed range is from 2 to 300
2023 milliseconds. Default value is 6 milliseconds.
2024 output, o
2026 Set the output mode.
2027 It accepts the following values:
2029 i Pass input unchanged.
2031 o Pass noise filtered out.
2033 n Pass only noise.
2035 Default value is o.
2037 smooth, m
2039 Set smooth factor. Default value is 11. Allowed range is from 1 to
2040 15.
2041 Commands
2043 This filter supports the all above options as commands.
2045 anlmf, anlms
2047 Apply Normalized Least-Mean-(Squares|Fourth) algorithm to the first
2048 audio stream using the second audio stream.
2049 This adaptive filter is used to mimic a desired filter by finding the
2051 filter coefficients that relate to producing the least mean square of
2052 the error signal (difference between the desired, 2nd input audio
2053 stream and the actual signal, the 1st input audio stream).
2054 A description of the accepted options follows.
2056 order
2058 Set filter order.
2059 mu Set filter mu.
2061 eps Set the filter eps.
2063 leakage
2065 Set the filter leakage.
2066 out_mode
2068 It accepts the following values:
2069 i Pass the 1st input.
2071 d Pass the 2nd input.
2073 o Pass filtered samples.
2075 n Pass difference between desired and filtered samples.
2077 Default value is o.
2079 Examples
2081 • One of many usages of this filter is noise reduction, input audio
2083 is filtered with same samples that are delayed by fixed amount, one
2084 such example for stereo audio is:
2085 asplit[a][b],[a]adelay=32S|32S[a],[b][a]anlms=order=128:leakage=0.0005:mu=.5:out_mode=o
2087 Commands
2089 This filter supports the same commands as options, excluding option
2091 "order".
2092 anull
2094 Pass the audio source unchanged to the output.
2095 apad
2097 Pad the end of an audio stream with silence.
2098 This can be used together with ffmpeg -shortest to extend audio streams
2100 to the same length as the video stream.
2101 A description of the accepted options follows.
2103 packet_size
2105 Set silence packet size. Default value is 4096.
2106 pad_len
2108 Set the number of samples of silence to add to the end. After the
2109 value is reached, the stream is terminated. This option is mutually
2110 exclusive with whole_len.
2111 whole_len
2113 Set the minimum total number of samples in the output audio stream.
2114 If the value is longer than the input audio length, silence is
2115 added to the end, until the value is reached. This option is
2116 mutually exclusive with pad_len.
2117 pad_dur
2119 Specify the duration of samples of silence to add. See the Time
2120 duration section in the ffmpeg-utils(1) manual for the accepted
2121 syntax. Used only if set to non-negative value.
2122 whole_dur
2124 Specify the minimum total duration in the output audio stream. See
2125 the Time duration section in the ffmpeg-utils(1) manual for the
2126 accepted syntax. Used only if set to non-negative value. If the
2127 value is longer than the input audio length, silence is added to
2128 the end, until the value is reached. This option is mutually
2129 exclusive with pad_dur
2130 If neither the pad_len nor the whole_len nor pad_dur nor whole_dur
2132 option is set, the filter will add silence to the end of the input
2133 stream indefinitely.
2134 Note that for ffmpeg 4.4 and earlier a zero pad_dur or whole_dur also
2136 caused the filter to add silence indefinitely.
2137 Examples
2139 • Add 1024 samples of silence to the end of the input:
2141 apad=pad_len=1024
2143 • Make sure the audio output will contain at least 10000 samples, pad
2145 the input with silence if required:
2146 apad=whole_len=10000
2148 • Use ffmpeg to pad the audio input with silence, so that the video
2150 stream will always result the shortest and will be converted until
2151 the end in the output file when using the shortest option:
2152 ffmpeg -i VIDEO -i AUDIO -filter_complex "[1:0]apad" -shortest OUTPUT
2154 aphaser
2156 Add a phasing effect to the input audio.
2157 A phaser filter creates series of peaks and troughs in the frequency
2159 spectrum. The position of the peaks and troughs are modulated so that
2160 they vary over time, creating a sweeping effect.
2161 A description of the accepted parameters follows.
2163 in_gain
2165 Set input gain. Default is 0.4.
2166 out_gain
2168 Set output gain. Default is 0.74
2169 delay
2171 Set delay in milliseconds. Default is 3.0.
2172 decay
2174 Set decay. Default is 0.4.
2175 speed
2177 Set modulation speed in Hz. Default is 0.5.
2178 type
2180 Set modulation type. Default is triangular.
2181 It accepts the following values:
2183 triangular, t
2185 sinusoidal, s
2186 aphaseshift
2188 Apply phase shift to input audio samples.
2189 The filter accepts the following options:
2191 shift
2193 Specify phase shift. Allowed range is from -1.0 to 1.0. Default
2194 value is 0.0.
2195 level
2197 Set output gain applied to final output. Allowed range is from 0.0
2198 to 1.0. Default value is 1.0.
2199 order
2201 Set filter order used for filtering. Allowed range is from 1 to 16.
2202 Default value is 8.
2203 Commands
2205 This filter supports the all above options as commands.
2207 apsyclip
2209 Apply Psychoacoustic clipper to input audio stream.
2210 The filter accepts the following options:
2212 level_in
2214 Set input gain. By default it is 1. Range is [0.015625 - 64].
2215 level_out
2217 Set output gain. By default it is 1. Range is [0.015625 - 64].
2218 clip
2220 Set the clipping start value. Default value is 0dBFS or 1.
2221 diff
2223 Output only difference samples, useful to hear introduced
2224 distortions. By default is disabled.
2225 adaptive
2227 Set strenght of adaptive distortion applied. Default value is 0.5.
2228 Allowed range is from 0 to 1.
2229 iterations
2231 Set number of iterations of psychoacoustic clipper. Allowed range
2232 is from 1 to 20. Default value is 10.
2233 level
2235 Auto level output signal. Default is disabled. This normalizes
2236 audio back to 0dBFS if enabled.
2237 Commands
2239 This filter supports the all above options as commands.
2241 apulsator
2243 Audio pulsator is something between an autopanner and a tremolo. But
2244 it can produce funny stereo effects as well. Pulsator changes the
2245 volume of the left and right channel based on a LFO (low frequency
2246 oscillator) with different waveforms and shifted phases. This filter
2247 have the ability to define an offset between left and right channel. An
2248 offset of 0 means that both LFO shapes match each other. The left and
2249 right channel are altered equally - a conventional tremolo. An offset
2250 of 50% means that the shape of the right channel is exactly shifted in
2251 phase (or moved backwards about half of the frequency) - pulsator acts
2252 as an autopanner. At 1 both curves match again. Every setting in
2253 between moves the phase shift gapless between all stages and produces
2254 some "bypassing" sounds with sine and triangle waveforms. The more you
2255 set the offset near 1 (starting from the 0.5) the faster the signal
2256 passes from the left to the right speaker.
2257 The filter accepts the following options:
2259 level_in
2261 Set input gain. By default it is 1. Range is [0.015625 - 64].
2262 level_out
2264 Set output gain. By default it is 1. Range is [0.015625 - 64].
2265 mode
2267 Set waveform shape the LFO will use. Can be one of: sine, triangle,
2268 square, sawup or sawdown. Default is sine.
2269 amount
2271 Set modulation. Define how much of original signal is affected by
2272 the LFO.
2273 offset_l
2275 Set left channel offset. Default is 0. Allowed range is [0 - 1].
2276 offset_r
2278 Set right channel offset. Default is 0.5. Allowed range is [0 - 1].
2279 width
2281 Set pulse width. Default is 1. Allowed range is [0 - 2].
2282 timing
2284 Set possible timing mode. Can be one of: bpm, ms or hz. Default is
2285 hz.
2286 bpm Set bpm. Default is 120. Allowed range is [30 - 300]. Only used if
2288 timing is set to bpm.
2289 ms Set ms. Default is 500. Allowed range is [10 - 2000]. Only used if
2291 timing is set to ms.
2292 hz Set frequency in Hz. Default is 2. Allowed range is [0.01 - 100].
2294 Only used if timing is set to hz.
2295 aresample
2297 Resample the input audio to the specified parameters, using the
2298 libswresample library. If none are specified then the filter will
2299 automatically convert between its input and output.
2300 This filter is also able to stretch/squeeze the audio data to make it
2302 match the timestamps or to inject silence / cut out audio to make it
2303 match the timestamps, do a combination of both or do neither.
2304 The filter accepts the syntax [sample_rate:]resampler_options, where
2306 sample_rate expresses a sample rate and resampler_options is a list of
2307 key=value pairs, separated by ":". See the "Resampler Options" section
2308 in the ffmpeg-resampler(1) manual for the complete list of supported
2309 options.
2310 Examples
2312 • Resample the input audio to 44100Hz:
2314 aresample=44100
2316 • Stretch/squeeze samples to the given timestamps, with a maximum of
2318 1000 samples per second compensation:
2319 aresample=async=1000
2321 areverse
2323 Reverse an audio clip.
2324 Warning: This filter requires memory to buffer the entire clip, so
2326 trimming is suggested.
2327 Examples
2329 • Take the first 5 seconds of a clip, and reverse it.
2331 atrim=end=5,areverse
2333 arnndn
2335 Reduce noise from speech using Recurrent Neural Networks.
2336 This filter accepts the following options:
2338 model, m
2340 Set train model file to load. This option is always required.
2341 mix Set how much to mix filtered samples into final output. Allowed
2343 range is from -1 to 1. Default value is 1. Negative values are
2344 special, they set how much to keep filtered noise in the final
2345 filter output. Set this option to -1 to hear actual noise removed
2346 from input signal.
2347 Commands
2349 This filter supports the all above options as commands.
2351 asdr
2353 Measure Audio Signal-to-Distortion Ratio.
2354 This filter takes two audio streams for input, and outputs first audio
2356 stream. Results are in dB per channel at end of either input.
2357 asetnsamples
2359 Set the number of samples per each output audio frame.
2360 The last output packet may contain a different number of samples, as
2362 the filter will flush all the remaining samples when the input audio
2363 signals its end.
2364 The filter accepts the following options:
2366 nb_out_samples, n
2368 Set the number of frames per each output audio frame. The number is
2369 intended as the number of samples per each channel. Default value
2370 is 1024.
2371 pad, p
2373 If set to 1, the filter will pad the last audio frame with zeroes,
2374 so that the last frame will contain the same number of samples as
2375 the previous ones. Default value is 1.
2376 For example, to set the number of per-frame samples to 1234 and disable
2378 padding for the last frame, use:
2379 asetnsamples=n=1234:p=0
2381 asetrate
2383 Set the sample rate without altering the PCM data. This will result in
2384 a change of speed and pitch.
2385 The filter accepts the following options:
2387 sample_rate, r
2389 Set the output sample rate. Default is 44100 Hz.
2390 ashowinfo
2392 Show a line containing various information for each input audio frame.
2393 The input audio is not modified.
2394 The shown line contains a sequence of key/value pairs of the form
2396 key:value.
2397 The following values are shown in the output:
2399 n The (sequential) number of the input frame, starting from 0.
2401 pts The presentation timestamp of the input frame, in time base units;
2403 the time base depends on the filter input pad, and is usually
2404 1/sample_rate.
2405 pts_time
2407 The presentation timestamp of the input frame in seconds.
2408 pos position of the frame in the input stream, -1 if this information
2410 in unavailable and/or meaningless (for example in case of synthetic
2411 audio)
2412 fmt The sample format.
2414 chlayout
2416 The channel layout.
2417 rate
2419 The sample rate for the audio frame.
2420 nb_samples
2422 The number of samples (per channel) in the frame.
2423 checksum
2425 The Adler-32 checksum (printed in hexadecimal) of the audio data.
2426 For planar audio, the data is treated as if all the planes were
2427 concatenated.
2428 plane_checksums
2430 A list of Adler-32 checksums for each data plane.
2431 asoftclip
2433 Apply audio soft clipping.
2434 Soft clipping is a type of distortion effect where the amplitude of a
2436 signal is saturated along a smooth curve, rather than the abrupt shape
2437 of hard-clipping.
2438 This filter accepts the following options:
2440 type
2442 Set type of soft-clipping.
2443 It accepts the following values:
2445 hard
2447 tanh
2448 atan
2449 cubic
2450 exp
2451 alg
2452 quintic
2453 sin
2454 erf
2455 threshold
2456 Set threshold from where to start clipping. Default value is 0dB or
2457 1.
2458 output
2460 Set gain applied to output. Default value is 0dB or 1.
2461 param
2463 Set additional parameter which controls sigmoid function.
2464 oversample
2466 Set oversampling factor.
2467 Commands
2469 This filter supports the all above options as commands.
2471 aspectralstats
2473 Display frequency domain statistical information about the audio
2474 channels. Statistics are calculated and stored as metadata for each
2475 audio channel and for each audio frame.
2476 It accepts the following option:
2478 win_size
2480 Set the window length in samples. Default value is 2048. Allowed
2481 range is from 32 to 65536.
2482 win_func
2484 Set window function.
2485 It accepts the following values:
2487 rect
2489 bartlett
2490 hann, hanning
2491 hamming
2492 blackman
2493 welch
2494 flattop
2495 bharris
2496 bnuttall
2497 bhann
2498 sine
2499 nuttall
2500 lanczos
2501 gauss
2502 tukey
2503 dolph
2504 cauchy
2505 parzen
2506 poisson
2507 bohman
2508 Default is "hann".
2510 overlap
2512 Set window overlap. Allowed range is from 0 to 1. Default value is
2513 0.5.
2514 A list of each metadata key follows:
2516 mean
2518 variance
2519 centroid
2520 spread
2521 skewness
2522 kurtosis
2523 entropy
2524 flatness
2525 crest
2526 flux
2527 slope
2528 decrease
2529 rolloff
2530 asr
2532 Automatic Speech Recognition
2533 This filter uses PocketSphinx for speech recognition. To enable
2535 compilation of this filter, you need to configure FFmpeg with
2536 "--enable-pocketsphinx".
2537 It accepts the following options:
2539 rate
2541 Set sampling rate of input audio. Defaults is 16000. This need to
2542 match speech models, otherwise one will get poor results.
2543 hmm Set dictionary containing acoustic model files.
2545 dict
2547 Set pronunciation dictionary.
2548 lm Set language model file.
2550 lmctl
2552 Set language model set.
2553 lmname
2555 Set which language model to use.
2556 logfn
2558 Set output for log messages.
2559 The filter exports recognized speech as the frame metadata
2561 "lavfi.asr.text".
2562 astats
2564 Display time domain statistical information about the audio channels.
2565 Statistics are calculated and displayed for each audio channel and,
2566 where applicable, an overall figure is also given.
2567 It accepts the following option:
2569 length
2571 Short window length in seconds, used for peak and trough RMS
2572 measurement. Default is 0.05 (50 milliseconds). Allowed range is
2573 "[0 - 10]".
2574 metadata
2576 Set metadata injection. All the metadata keys are prefixed with
2577 "lavfi.astats.X", where "X" is channel number starting from 1 or
2578 string "Overall". Default is disabled.
2579 Available keys for each channel are: DC_offset Min_level Max_level
2581 Min_difference Max_difference Mean_difference RMS_difference
2582 Peak_level RMS_peak RMS_trough Crest_factor Flat_factor Peak_count
2583 Noise_floor Noise_floor_count Entropy Bit_depth Dynamic_range
2584 Zero_crossings Zero_crossings_rate Number_of_NaNs Number_of_Infs
2585 Number_of_denormals
2586 and for Overall: DC_offset Min_level Max_level Min_difference
2588 Max_difference Mean_difference RMS_difference Peak_level RMS_level
2589 RMS_peak RMS_trough Flat_factor Peak_count Noise_floor
2590 Noise_floor_count Entropy Bit_depth Number_of_samples
2591 Number_of_NaNs Number_of_Infs Number_of_denormals
2592 For example full key look like this "lavfi.astats.1.DC_offset" or
2594 this "lavfi.astats.Overall.Peak_count".
2595 For description what each key means read below.
2597 reset
2599 Set the number of frames over which cumulative stats are calculated
2600 before being reset Default is disabled.
2601 measure_perchannel
2603 Select the parameters which are measured per channel. The metadata
2604 keys can be used as flags, default is all which measures
2605 everything. none disables all per channel measurement.
2606 measure_overall
2608 Select the parameters which are measured overall. The metadata keys
2609 can be used as flags, default is all which measures everything.
2610 none disables all overall measurement.
2611 A description of each shown parameter follows:
2613 DC offset
2615 Mean amplitude displacement from zero.
2616 Min level
2618 Minimal sample level.
2619 Max level
2621 Maximal sample level.
2622 Min difference
2624 Minimal difference between two consecutive samples.
2625 Max difference
2627 Maximal difference between two consecutive samples.
2628 Mean difference
2630 Mean difference between two consecutive samples. The average of
2631 each difference between two consecutive samples.
2632 RMS difference
2634 Root Mean Square difference between two consecutive samples.
2635 Peak level dB
2637 RMS level dB
2638 Standard peak and RMS level measured in dBFS.
2639 RMS peak dB
2641 RMS trough dB
2642 Peak and trough values for RMS level measured over a short window.
2643 Crest factor
2645 Standard ratio of peak to RMS level (note: not in dB).
2646 Flat factor
2648 Flatness (i.e. consecutive samples with the same value) of the
2649 signal at its peak levels (i.e. either Min level or Max level).
2650 Peak count
2652 Number of occasions (not the number of samples) that the signal
2653 attained either Min level or Max level.
2654 Noise floor dB
2656 Minimum local peak measured in dBFS over a short window.
2657 Noise floor count
2659 Number of occasions (not the number of samples) that the signal
2660 attained Noise floor.
2661 Entropy
2663 Entropy measured across whole audio. Entropy of value near 1.0 is
2664 typically measured for white noise.
2665 Bit depth
2667 Overall bit depth of audio. Number of bits used for each sample.
2668 Dynamic range
2670 Measured dynamic range of audio in dB.
2671 Zero crossings
2673 Number of points where the waveform crosses the zero level axis.
2674 Zero crossings rate
2676 Rate of Zero crossings and number of audio samples.
2677 asubboost
2679 Boost subwoofer frequencies.
2680 The filter accepts the following options:
2682 dry Set dry gain, how much of original signal is kept. Allowed range is
2684 from 0 to 1. Default value is 0.7.
2685 wet Set wet gain, how much of filtered signal is kept. Allowed range is
2687 from 0 to 1. Default value is 0.7.
2688 decay
2690 Set delay line decay gain value. Allowed range is from 0 to 1.
2691 Default value is 0.7.
2692 feedback
2694 Set delay line feedback gain value. Allowed range is from 0 to 1.
2695 Default value is 0.9.
2696 cutoff
2698 Set cutoff frequency in Hertz. Allowed range is 50 to 900. Default
2699 value is 100.
2700 slope
2702 Set slope amount for cutoff frequency. Allowed range is 0.0001 to
2703 1. Default value is 0.5.
2704 delay
2706 Set delay. Allowed range is from 1 to 100. Default value is 20.
2707 Commands
2709 This filter supports the all above options as commands.
2711 asubcut
2713 Cut subwoofer frequencies.
2714 This filter allows to set custom, steeper roll off than highpass
2716 filter, and thus is able to more attenuate frequency content in stop-
2717 band.
2718 The filter accepts the following options:
2720 cutoff
2722 Set cutoff frequency in Hertz. Allowed range is 2 to 200. Default
2723 value is 20.
2724 order
2726 Set filter order. Available values are from 3 to 20. Default value
2727 is 10.
2728 level
2730 Set input gain level. Allowed range is from 0 to 1. Default value
2731 is 1.
2732 Commands
2734 This filter supports the all above options as commands.
2736 asupercut
2738 Cut super frequencies.
2739 The filter accepts the following options:
2741 cutoff
2743 Set cutoff frequency in Hertz. Allowed range is 20000 to 192000.
2744 Default value is 20000.
2745 order
2747 Set filter order. Available values are from 3 to 20. Default value
2748 is 10.
2749 level
2751 Set input gain level. Allowed range is from 0 to 1. Default value
2752 is 1.
2753 Commands
2755 This filter supports the all above options as commands.
2757 asuperpass
2759 Apply high order Butterworth band-pass filter.
2760 The filter accepts the following options:
2762 centerf
2764 Set center frequency in Hertz. Allowed range is 2 to 999999.
2765 Default value is 1000.
2766 order
2768 Set filter order. Available values are from 4 to 20. Default value
2769 is 4.
2770 qfactor
2772 Set Q-factor. Allowed range is from 0.01 to 100. Default value is
2773 1.
2774 level
2776 Set input gain level. Allowed range is from 0 to 2. Default value
2777 is 1.
2778 Commands
2780 This filter supports the all above options as commands.
2782 asuperstop
2784 Apply high order Butterworth band-stop filter.
2785 The filter accepts the following options:
2787 centerf
2789 Set center frequency in Hertz. Allowed range is 2 to 999999.
2790 Default value is 1000.
2791 order
2793 Set filter order. Available values are from 4 to 20. Default value
2794 is 4.
2795 qfactor
2797 Set Q-factor. Allowed range is from 0.01 to 100. Default value is
2798 1.
2799 level
2801 Set input gain level. Allowed range is from 0 to 2. Default value
2802 is 1.
2803 Commands
2805 This filter supports the all above options as commands.
2807 atempo
2809 Adjust audio tempo.
2810 The filter accepts exactly one parameter, the audio tempo. If not
2812 specified then the filter will assume nominal 1.0 tempo. Tempo must be
2813 in the [0.5, 100.0] range.
2814 Note that tempo greater than 2 will skip some samples rather than blend
2816 them in. If for any reason this is a concern it is always possible to
2817 daisy-chain several instances of atempo to achieve the desired product
2818 tempo.
2819 Examples
2821 • Slow down audio to 80% tempo:
2823 atempo=0.8
2825 • To speed up audio to 300% tempo:
2827 atempo=3
2829 • To speed up audio to 300% tempo by daisy-chaining two atempo
2831 instances:
2832 atempo=sqrt(3),atempo=sqrt(3)
2834 Commands
2836 This filter supports the following commands:
2838 tempo
2840 Change filter tempo scale factor. Syntax for the command is :
2841 "tempo"
2842 atilt
2844 Apply spectral tilt filter to audio stream.
2845 This filter apply any spectral roll-off slope over any specified
2847 frequency band.
2848 The filter accepts the following options:
2850 freq
2852 Set central frequency of tilt in Hz. Default is 10000 Hz.
2853 slope
2855 Set slope direction of tilt. Default is 0. Allowed range is from -1
2856 to 1.
2857 width
2859 Set width of tilt. Default is 1000. Allowed range is from 100 to
2860 10000.
2861 order
2863 Set order of tilt filter.
2864 level
2866 Set input volume level. Allowed range is from 0 to 4. Defalt is 1.
2867 Commands
2869 This filter supports the all above options as commands.
2871 atrim
2873 Trim the input so that the output contains one continuous subpart of
2874 the input.
2875 It accepts the following parameters:
2877 start
2879 Timestamp (in seconds) of the start of the section to keep. I.e.
2880 the audio sample with the timestamp start will be the first sample
2881 in the output.
2882 end Specify time of the first audio sample that will be dropped, i.e.
2884 the audio sample immediately preceding the one with the timestamp
2885 end will be the last sample in the output.
2886 start_pts
2888 Same as start, except this option sets the start timestamp in
2889 samples instead of seconds.
2890 end_pts
2892 Same as end, except this option sets the end timestamp in samples
2893 instead of seconds.
2894 duration
2896 The maximum duration of the output in seconds.
2897 start_sample
2899 The number of the first sample that should be output.
2900 end_sample
2902 The number of the first sample that should be dropped.
2903 start, end, and duration are expressed as time duration specifications;
2905 see the Time duration section in the ffmpeg-utils(1) manual.
2906 Note that the first two sets of the start/end options and the duration
2908 option look at the frame timestamp, while the _sample options simply
2909 count the samples that pass through the filter. So start/end_pts and
2910 start/end_sample will give different results when the timestamps are
2911 wrong, inexact or do not start at zero. Also note that this filter does
2912 not modify the timestamps. If you wish to have the output timestamps
2913 start at zero, insert the asetpts filter after the atrim filter.
2914 If multiple start or end options are set, this filter tries to be
2916 greedy and keep all samples that match at least one of the specified
2917 constraints. To keep only the part that matches all the constraints at
2918 once, chain multiple atrim filters.
2919 The defaults are such that all the input is kept. So it is possible to
2921 set e.g. just the end values to keep everything before the specified
2922 time.
2923 Examples:
2925 • Drop everything except the second minute of input:
2927 ffmpeg -i INPUT -af atrim=60:120
2929 • Keep only the first 1000 samples:
2931 ffmpeg -i INPUT -af atrim=end_sample=1000
2933 axcorrelate
2935 Calculate normalized windowed cross-correlation between two input audio
2936 streams.
2937 Resulted samples are always between -1 and 1 inclusive. If result is 1
2939 it means two input samples are highly correlated in that selected
2940 segment. Result 0 means they are not correlated at all. If result is
2941 -1 it means two input samples are out of phase, which means they cancel
2942 each other.
2943 The filter accepts the following options:
2945 size
2947 Set size of segment over which cross-correlation is calculated.
2948 Default is 256. Allowed range is from 2 to 131072.
2949 algo
2951 Set algorithm for cross-correlation. Can be "slow" or "fast".
2952 Default is "slow". Fast algorithm assumes mean values over any
2953 given segment are always zero and thus need much less calculations
2954 to make. This is generally not true, but is valid for typical
2955 audio streams.
2956 Examples
2958 • Calculate correlation between channels in stereo audio stream:
2960 ffmpeg -i stereo.wav -af channelsplit,axcorrelate=size=1024:algo=fast correlation.wav
2962 bandpass
2964 Apply a two-pole Butterworth band-pass filter with central frequency
2965 frequency, and (3dB-point) band-width width. The csg option selects a
2966 constant skirt gain (peak gain = Q) instead of the default: constant
2967 0dB peak gain. The filter roll off at 6dB per octave (20dB per
2968 decade).
2969 The filter accepts the following options:
2971 frequency, f
2973 Set the filter's central frequency. Default is 3000.
2974 csg Constant skirt gain if set to 1. Defaults to 0.
2976 width_type, t
2978 Set method to specify band-width of filter.
2979 h Hz
2981 q Q-Factor
2983 o octave
2985 s slope
2987 k kHz
2989 width, w
2991 Specify the band-width of a filter in width_type units.
2992 mix, m
2994 How much to use filtered signal in output. Default is 1. Range is
2995 between 0 and 1.
2996 channels, c
2998 Specify which channels to filter, by default all available are
2999 filtered.
3000 normalize, n
3002 Normalize biquad coefficients, by default is disabled. Enabling it
3003 will normalize magnitude response at DC to 0dB.
3004 transform, a
3006 Set transform type of IIR filter.
3007 di
3009 dii
3010 tdii
3011 latt
3012 svf
3013 precision, r
3014 Set precison of filtering.
3015 auto
3017 Pick automatic sample format depending on surround filters.
3018 s16 Always use signed 16-bit.
3020 s32 Always use signed 32-bit.
3022 f32 Always use float 32-bit.
3024 f64 Always use float 64-bit.
3026 Commands
3028 This filter supports the following commands:
3030 frequency, f
3032 Change bandpass frequency. Syntax for the command is : "frequency"
3033 width_type, t
3035 Change bandpass width_type. Syntax for the command is :
3036 "width_type"
3037 width, w
3039 Change bandpass width. Syntax for the command is : "width"
3040 mix, m
3042 Change bandpass mix. Syntax for the command is : "mix"
3043 bandreject
3045 Apply a two-pole Butterworth band-reject filter with central frequency
3046 frequency, and (3dB-point) band-width width. The filter roll off at
3047 6dB per octave (20dB per decade).
3048 The filter accepts the following options:
3050 frequency, f
3052 Set the filter's central frequency. Default is 3000.
3053 width_type, t
3055 Set method to specify band-width of filter.
3056 h Hz
3058 q Q-Factor
3060 o octave
3062 s slope
3064 k kHz
3066 width, w
3068 Specify the band-width of a filter in width_type units.
3069 mix, m
3071 How much to use filtered signal in output. Default is 1. Range is
3072 between 0 and 1.
3073 channels, c
3075 Specify which channels to filter, by default all available are
3076 filtered.
3077 normalize, n
3079 Normalize biquad coefficients, by default is disabled. Enabling it
3080 will normalize magnitude response at DC to 0dB.
3081 transform, a
3083 Set transform type of IIR filter.
3084 di
3086 dii
3087 tdii
3088 latt
3089 svf
3090 precision, r
3091 Set precison of filtering.
3092 auto
3094 Pick automatic sample format depending on surround filters.
3095 s16 Always use signed 16-bit.
3097 s32 Always use signed 32-bit.
3099 f32 Always use float 32-bit.
3101 f64 Always use float 64-bit.
3103 Commands
3105 This filter supports the following commands:
3107 frequency, f
3109 Change bandreject frequency. Syntax for the command is :
3110 "frequency"
3111 width_type, t
3113 Change bandreject width_type. Syntax for the command is :
3114 "width_type"
3115 width, w
3117 Change bandreject width. Syntax for the command is : "width"
3118 mix, m
3120 Change bandreject mix. Syntax for the command is : "mix"
3121 bass, lowshelf
3123 Boost or cut the bass (lower) frequencies of the audio using a two-pole
3124 shelving filter with a response similar to that of a standard hi-fi's
3125 tone-controls. This is also known as shelving equalisation (EQ).
3126 The filter accepts the following options:
3128 gain, g
3130 Give the gain at 0 Hz. Its useful range is about -20 (for a large
3131 cut) to +20 (for a large boost). Beware of clipping when using a
3132 positive gain.
3133 frequency, f
3135 Set the filter's central frequency and so can be used to extend or
3136 reduce the frequency range to be boosted or cut. The default value
3137 is 100 Hz.
3138 width_type, t
3140 Set method to specify band-width of filter.
3141 h Hz
3143 q Q-Factor
3145 o octave
3147 s slope
3149 k kHz
3151 width, w
3153 Determine how steep is the filter's shelf transition.
3154 poles, p
3156 Set number of poles. Default is 2.
3157 mix, m
3159 How much to use filtered signal in output. Default is 1. Range is
3160 between 0 and 1.
3161 channels, c
3163 Specify which channels to filter, by default all available are
3164 filtered.
3165 normalize, n
3167 Normalize biquad coefficients, by default is disabled. Enabling it
3168 will normalize magnitude response at DC to 0dB.
3169 transform, a
3171 Set transform type of IIR filter.
3172 di
3174 dii
3175 tdii
3176 latt
3177 svf
3178 precision, r
3179 Set precison of filtering.
3180 auto
3182 Pick automatic sample format depending on surround filters.
3183 s16 Always use signed 16-bit.
3185 s32 Always use signed 32-bit.
3187 f32 Always use float 32-bit.
3189 f64 Always use float 64-bit.
3191 Commands
3193 This filter supports the following commands:
3195 frequency, f
3197 Change bass frequency. Syntax for the command is : "frequency"
3198 width_type, t
3200 Change bass width_type. Syntax for the command is : "width_type"
3201 width, w
3203 Change bass width. Syntax for the command is : "width"
3204 gain, g
3206 Change bass gain. Syntax for the command is : "gain"
3207 mix, m
3209 Change bass mix. Syntax for the command is : "mix"
3210 biquad
3212 Apply a biquad IIR filter with the given coefficients. Where b0, b1,
3213 b2 and a0, a1, a2 are the numerator and denominator coefficients
3214 respectively. and channels, c specify which channels to filter, by
3215 default all available are filtered.
3216 Commands
3218 This filter supports the following commands:
3220 a0
3222 a1
3223 a2
3224 b0
3225 b1
3226 b2 Change biquad parameter. Syntax for the command is : "value"
3227 mix, m
3229 How much to use filtered signal in output. Default is 1. Range is
3230 between 0 and 1.
3231 channels, c
3233 Specify which channels to filter, by default all available are
3234 filtered.
3235 normalize, n
3237 Normalize biquad coefficients, by default is disabled. Enabling it
3238 will normalize magnitude response at DC to 0dB.
3239 transform, a
3241 Set transform type of IIR filter.
3242 di
3244 dii
3245 tdii
3246 latt
3247 svf
3248 precision, r
3249 Set precison of filtering.
3250 auto
3252 Pick automatic sample format depending on surround filters.
3253 s16 Always use signed 16-bit.
3255 s32 Always use signed 32-bit.
3257 f32 Always use float 32-bit.
3259 f64 Always use float 64-bit.
3261 bs2b
3263 Bauer stereo to binaural transformation, which improves headphone
3264 listening of stereo audio records.
3265 To enable compilation of this filter you need to configure FFmpeg with
3267 "--enable-libbs2b".
3268 It accepts the following parameters:
3270 profile
3272 Pre-defined crossfeed level.
3273 default
3275 Default level (fcut=700, feed=50).
3276 cmoy
3278 Chu Moy circuit (fcut=700, feed=60).
3279 jmeier
3281 Jan Meier circuit (fcut=650, feed=95).
3282 fcut
3284 Cut frequency (in Hz).
3285 feed
3287 Feed level (in Hz).
3288 channelmap
3290 Remap input channels to new locations.
3291 It accepts the following parameters:
3293 map Map channels from input to output. The argument is a '|'-separated
3295 list of mappings, each in the "in_channel-out_channel" or
3296 in_channel form. in_channel can be either the name of the input
3297 channel (e.g. FL for front left) or its index in the input channel
3298 layout. out_channel is the name of the output channel or its index
3299 in the output channel layout. If out_channel is not given then it
3300 is implicitly an index, starting with zero and increasing by one
3301 for each mapping.
3302 channel_layout
3304 The channel layout of the output stream.
3305 If no mapping is present, the filter will implicitly map input channels
3307 to output channels, preserving indices.
3308 Examples
3310 • For example, assuming a 5.1+downmix input MOV file,
3312 ffmpeg -i in.mov -filter 'channelmap=map=DL-FL|DR-FR' out.wav
3314 will create an output WAV file tagged as stereo from the downmix
3316 channels of the input.
3317 • To fix a 5.1 WAV improperly encoded in AAC's native channel order
3319 ffmpeg -i in.wav -filter 'channelmap=1|2|0|5|3|4:5.1' out.wav
3321 channelsplit
3323 Split each channel from an input audio stream into a separate output
3324 stream.
3325 It accepts the following parameters:
3327 channel_layout
3329 The channel layout of the input stream. The default is "stereo".
3330 channels
3332 A channel layout describing the channels to be extracted as
3333 separate output streams or "all" to extract each input channel as a
3334 separate stream. The default is "all".
3335 Choosing channels not present in channel layout in the input will
3337 result in an error.
3338 Examples
3340 • For example, assuming a stereo input MP3 file,
3342 ffmpeg -i in.mp3 -filter_complex channelsplit out.mkv
3344 will create an output Matroska file with two audio streams, one
3346 containing only the left channel and the other the right channel.
3347 • Split a 5.1 WAV file into per-channel files:
3349 ffmpeg -i in.wav -filter_complex
3351 'channelsplit=channel_layout=5.1[FL][FR][FC][LFE][SL][SR]'
3352 -map '[FL]' front_left.wav -map '[FR]' front_right.wav -map '[FC]'
3353 front_center.wav -map '[LFE]' lfe.wav -map '[SL]' side_left.wav -map '[SR]'
3354 side_right.wav
3355 • Extract only LFE from a 5.1 WAV file:
3357 ffmpeg -i in.wav -filter_complex 'channelsplit=channel_layout=5.1:channels=LFE[LFE]'
3359 -map '[LFE]' lfe.wav
3360 chorus
3362 Add a chorus effect to the audio.
3363 Can make a single vocal sound like a chorus, but can also be applied to
3365 instrumentation.
3366 Chorus resembles an echo effect with a short delay, but whereas with
3368 echo the delay is constant, with chorus, it is varied using using
3369 sinusoidal or triangular modulation. The modulation depth defines the
3370 range the modulated delay is played before or after the delay. Hence
3371 the delayed sound will sound slower or faster, that is the delayed
3372 sound tuned around the original one, like in a chorus where some vocals
3373 are slightly off key.
3374 It accepts the following parameters:
3376 in_gain
3378 Set input gain. Default is 0.4.
3379 out_gain
3381 Set output gain. Default is 0.4.
3382 delays
3384 Set delays. A typical delay is around 40ms to 60ms.
3385 decays
3387 Set decays.
3388 speeds
3390 Set speeds.
3391 depths
3393 Set depths.
3394 Examples
3396 • A single delay:
3398 chorus=0.7:0.9:55:0.4:0.25:2
3400 • Two delays:
3402 chorus=0.6:0.9:50|60:0.4|0.32:0.25|0.4:2|1.3
3404 • Fuller sounding chorus with three delays:
3406 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
3408 compand
3410 Compress or expand the audio's dynamic range.
3411 It accepts the following parameters:
3413 attacks
3415 decays
3416 A list of times in seconds for each channel over which the
3417 instantaneous level of the input signal is averaged to determine
3418 its volume. attacks refers to increase of volume and decays refers
3419 to decrease of volume. For most situations, the attack time
3420 (response to the audio getting louder) should be shorter than the
3421 decay time, because the human ear is more sensitive to sudden loud
3422 audio than sudden soft audio. A typical value for attack is 0.3
3423 seconds and a typical value for decay is 0.8 seconds. If specified
3424 number of attacks & decays is lower than number of channels, the
3425 last set attack/decay will be used for all remaining channels.
3426 points
3428 A list of points for the transfer function, specified in dB
3429 relative to the maximum possible signal amplitude. Each key points
3430 list must be defined using the following syntax:
3431 "x0/y0|x1/y1|x2/y2|...." or "x0/y0 x1/y1 x2/y2 ...."
3432 The input values must be in strictly increasing order but the
3434 transfer function does not have to be monotonically rising. The
3435 point "0/0" is assumed but may be overridden (by "0/out-dBn").
3436 Typical values for the transfer function are "-70/-70|-60/-20|1/0".
3437 soft-knee
3439 Set the curve radius in dB for all joints. It defaults to 0.01.
3440 gain
3442 Set the additional gain in dB to be applied at all points on the
3443 transfer function. This allows for easy adjustment of the overall
3444 gain. It defaults to 0.
3445 volume
3447 Set an initial volume, in dB, to be assumed for each channel when
3448 filtering starts. This permits the user to supply a nominal level
3449 initially, so that, for example, a very large gain is not applied
3450 to initial signal levels before the companding has begun to
3451 operate. A typical value for audio which is initially quiet is -90
3452 dB. It defaults to 0.
3453 delay
3455 Set a delay, in seconds. The input audio is analyzed immediately,
3456 but audio is delayed before being fed to the volume adjuster.
3457 Specifying a delay approximately equal to the attack/decay times
3458 allows the filter to effectively operate in predictive rather than
3459 reactive mode. It defaults to 0.
3460 Examples
3462 • Make music with both quiet and loud passages suitable for listening
3464 to in a noisy environment:
3465 compand=.3|.3:1|1:-90/-60|-60/-40|-40/-30|-20/-20:6:0:-90:0.2
3467 Another example for audio with whisper and explosion parts:
3469 compand=0|0:1|1:-90/-900|-70/-70|-30/-9|0/-3:6:0:0:0
3471 • A noise gate for when the noise is at a lower level than the
3473 signal:
3474 compand=.1|.1:.2|.2:-900/-900|-50.1/-900|-50/-50:.01:0:-90:.1
3476 • Here is another noise gate, this time for when the noise is at a
3478 higher level than the signal (making it, in some ways, similar to
3479 squelch):
3480 compand=.1|.1:.1|.1:-45.1/-45.1|-45/-900|0/-900:.01:45:-90:.1
3482 • 2:1 compression starting at -6dB:
3484 compand=points=-80/-80|-6/-6|0/-3.8|20/3.5
3486 • 2:1 compression starting at -9dB:
3488 compand=points=-80/-80|-9/-9|0/-5.3|20/2.9
3490 • 2:1 compression starting at -12dB:
3492 compand=points=-80/-80|-12/-12|0/-6.8|20/1.9
3494 • 2:1 compression starting at -18dB:
3496 compand=points=-80/-80|-18/-18|0/-9.8|20/0.7
3498 • 3:1 compression starting at -15dB:
3500 compand=points=-80/-80|-15/-15|0/-10.8|20/-5.2
3502 • Compressor/Gate:
3504 compand=points=-80/-105|-62/-80|-15.4/-15.4|0/-12|20/-7.6
3506 • Expander:
3508 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
3510 • Hard limiter at -6dB:
3512 compand=attacks=0:points=-80/-80|-6/-6|20/-6
3514 • Hard limiter at -12dB:
3516 compand=attacks=0:points=-80/-80|-12/-12|20/-12
3518 • Hard noise gate at -35 dB:
3520 compand=attacks=0:points=-80/-115|-35.1/-80|-35/-35|20/20
3522 • Soft limiter:
3524 compand=attacks=0:points=-80/-80|-12.4/-12.4|-6/-8|0/-6.8|20/-2.8
3526 compensationdelay
3528 Compensation Delay Line is a metric based delay to compensate differing
3529 positions of microphones or speakers.
3530 For example, you have recorded guitar with two microphones placed in
3532 different locations. Because the front of sound wave has fixed speed in
3533 normal conditions, the phasing of microphones can vary and depends on
3534 their location and interposition. The best sound mix can be achieved
3535 when these microphones are in phase (synchronized). Note that a
3536 distance of ~30 cm between microphones makes one microphone capture the
3537 signal in antiphase to the other microphone. That makes the final mix
3538 sound moody. This filter helps to solve phasing problems by adding
3539 different delays to each microphone track and make them synchronized.
3540 The best result can be reached when you take one track as base and
3542 synchronize other tracks one by one with it. Remember that
3543 synchronization/delay tolerance depends on sample rate, too. Higher
3544 sample rates will give more tolerance.
3545 The filter accepts the following parameters:
3547 mm Set millimeters distance. This is compensation distance for fine
3549 tuning. Default is 0.
3550 cm Set cm distance. This is compensation distance for tightening
3552 distance setup. Default is 0.
3553 m Set meters distance. This is compensation distance for hard
3555 distance setup. Default is 0.
3556 dry Set dry amount. Amount of unprocessed (dry) signal. Default is 0.
3558 wet Set wet amount. Amount of processed (wet) signal. Default is 1.
3560 temp
3562 Set temperature in degrees Celsius. This is the temperature of the
3563 environment. Default is 20.
3564 crossfeed
3566 Apply headphone crossfeed filter.
3567 Crossfeed is the process of blending the left and right channels of
3569 stereo audio recording. It is mainly used to reduce extreme stereo
3570 separation of low frequencies.
3571 The intent is to produce more speaker like sound to the listener.
3573 The filter accepts the following options:
3575 strength
3577 Set strength of crossfeed. Default is 0.2. Allowed range is from 0
3578 to 1. This sets gain of low shelf filter for side part of stereo
3579 image. Default is -6dB. Max allowed is -30db when strength is set
3580 to 1.
3581 range
3583 Set soundstage wideness. Default is 0.5. Allowed range is from 0 to
3584 1. This sets cut off frequency of low shelf filter. Default is cut
3585 off near 1550 Hz. With range set to 1 cut off frequency is set to
3586 2100 Hz.
3587 slope
3589 Set curve slope of low shelf filter. Default is 0.5. Allowed range
3590 is from 0.01 to 1.
3591 level_in
3593 Set input gain. Default is 0.9.
3594 level_out
3596 Set output gain. Default is 1.
3597 Commands
3599 This filter supports the all above options as commands.
3601 crystalizer
3603 Simple algorithm for audio noise sharpening.
3604 This filter linearly increases differences betweeen each audio sample.
3606 The filter accepts the following options:
3608 i Sets the intensity of effect (default: 2.0). Must be in range
3610 between -10.0 to 0 (unchanged sound) to 10.0 (maximum effect). To
3611 inverse filtering use negative value.
3612 c Enable clipping. By default is enabled.
3614 Commands
3616 This filter supports the all above options as commands.
3618 dcshift
3620 Apply a DC shift to the audio.
3621 This can be useful to remove a DC offset (caused perhaps by a hardware
3623 problem in the recording chain) from the audio. The effect of a DC
3624 offset is reduced headroom and hence volume. The astats filter can be
3625 used to determine if a signal has a DC offset.
3626 shift
3628 Set the DC shift, allowed range is [-1, 1]. It indicates the amount
3629 to shift the audio.
3630 limitergain
3632 Optional. It should have a value much less than 1 (e.g. 0.05 or
3633 0.02) and is used to prevent clipping.
3634 deesser
3636 Apply de-essing to the audio samples.
3637 i Set intensity for triggering de-essing. Allowed range is from 0 to
3639 1. Default is 0.
3640 m Set amount of ducking on treble part of sound. Allowed range is
3642 from 0 to 1. Default is 0.5.
3643 f How much of original frequency content to keep when de-essing.
3645 Allowed range is from 0 to 1. Default is 0.5.
3646 s Set the output mode.
3648 It accepts the following values:
3650 i Pass input unchanged.
3652 o Pass ess filtered out.
3654 e Pass only ess.
3656 Default value is o.
3658 drmeter
3660 Measure audio dynamic range.
3661 DR values of 14 and higher is found in very dynamic material. DR of 8
3663 to 13 is found in transition material. And anything less that 8 have
3664 very poor dynamics and is very compressed.
3665 The filter accepts the following options:
3667 length
3669 Set window length in seconds used to split audio into segments of
3670 equal length. Default is 3 seconds.
3671 dynaudnorm
3673 Dynamic Audio Normalizer.
3674 This filter applies a certain amount of gain to the input audio in
3676 order to bring its peak magnitude to a target level (e.g. 0 dBFS).
3677 However, in contrast to more "simple" normalization algorithms, the
3678 Dynamic Audio Normalizer *dynamically* re-adjusts the gain factor to
3679 the input audio. This allows for applying extra gain to the "quiet"
3680 sections of the audio while avoiding distortions or clipping the "loud"
3681 sections. In other words: The Dynamic Audio Normalizer will "even out"
3682 the volume of quiet and loud sections, in the sense that the volume of
3683 each section is brought to the same target level. Note, however, that
3684 the Dynamic Audio Normalizer achieves this goal *without* applying
3685 "dynamic range compressing". It will retain 100% of the dynamic range
3686 *within* each section of the audio file.
3687 framelen, f
3689 Set the frame length in milliseconds. In range from 10 to 8000
3690 milliseconds. Default is 500 milliseconds. The Dynamic Audio
3691 Normalizer processes the input audio in small chunks, referred to
3692 as frames. This is required, because a peak magnitude has no
3693 meaning for just a single sample value. Instead, we need to
3694 determine the peak magnitude for a contiguous sequence of sample
3695 values. While a "standard" normalizer would simply use the peak
3696 magnitude of the complete file, the Dynamic Audio Normalizer
3697 determines the peak magnitude individually for each frame. The
3698 length of a frame is specified in milliseconds. By default, the
3699 Dynamic Audio Normalizer uses a frame length of 500 milliseconds,
3700 which has been found to give good results with most files. Note
3701 that the exact frame length, in number of samples, will be
3702 determined automatically, based on the sampling rate of the
3703 individual input audio file.
3704 gausssize, g
3706 Set the Gaussian filter window size. In range from 3 to 301, must
3707 be odd number. Default is 31. Probably the most important
3708 parameter of the Dynamic Audio Normalizer is the "window size" of
3709 the Gaussian smoothing filter. The filter's window size is
3710 specified in frames, centered around the current frame. For the
3711 sake of simplicity, this must be an odd number. Consequently, the
3712 default value of 31 takes into account the current frame, as well
3713 as the 15 preceding frames and the 15 subsequent frames. Using a
3714 larger window results in a stronger smoothing effect and thus in
3715 less gain variation, i.e. slower gain adaptation. Conversely, using
3716 a smaller window results in a weaker smoothing effect and thus in
3717 more gain variation, i.e. faster gain adaptation. In other words,
3718 the more you increase this value, the more the Dynamic Audio
3719 Normalizer will behave like a "traditional" normalization filter.
3720 On the contrary, the more you decrease this value, the more the
3721 Dynamic Audio Normalizer will behave like a dynamic range
3722 compressor.
3723 peak, p
3725 Set the target peak value. This specifies the highest permissible
3726 magnitude level for the normalized audio input. This filter will
3727 try to approach the target peak magnitude as closely as possible,
3728 but at the same time it also makes sure that the normalized signal
3729 will never exceed the peak magnitude. A frame's maximum local gain
3730 factor is imposed directly by the target peak magnitude. The
3731 default value is 0.95 and thus leaves a headroom of 5%*. It is not
3732 recommended to go above this value.
3733 maxgain, m
3735 Set the maximum gain factor. In range from 1.0 to 100.0. Default is
3736 10.0. The Dynamic Audio Normalizer determines the maximum possible
3737 (local) gain factor for each input frame, i.e. the maximum gain
3738 factor that does not result in clipping or distortion. The maximum
3739 gain factor is determined by the frame's highest magnitude sample.
3740 However, the Dynamic Audio Normalizer additionally bounds the
3741 frame's maximum gain factor by a predetermined (global) maximum
3742 gain factor. This is done in order to avoid excessive gain factors
3743 in "silent" or almost silent frames. By default, the maximum gain
3744 factor is 10.0, For most inputs the default value should be
3745 sufficient and it usually is not recommended to increase this
3746 value. Though, for input with an extremely low overall volume
3747 level, it may be necessary to allow even higher gain factors. Note,
3748 however, that the Dynamic Audio Normalizer does not simply apply a
3749 "hard" threshold (i.e. cut off values above the threshold).
3750 Instead, a "sigmoid" threshold function will be applied. This way,
3751 the gain factors will smoothly approach the threshold value, but
3752 never exceed that value.
3753 targetrms, r
3755 Set the target RMS. In range from 0.0 to 1.0. Default is 0.0 -
3756 disabled. By default, the Dynamic Audio Normalizer performs "peak"
3757 normalization. This means that the maximum local gain factor for
3758 each frame is defined (only) by the frame's highest magnitude
3759 sample. This way, the samples can be amplified as much as possible
3760 without exceeding the maximum signal level, i.e. without clipping.
3761 Optionally, however, the Dynamic Audio Normalizer can also take
3762 into account the frame's root mean square, abbreviated RMS. In
3763 electrical engineering, the RMS is commonly used to determine the
3764 power of a time-varying signal. It is therefore considered that the
3765 RMS is a better approximation of the "perceived loudness" than just
3766 looking at the signal's peak magnitude. Consequently, by adjusting
3767 all frames to a constant RMS value, a uniform "perceived loudness"
3768 can be established. If a target RMS value has been specified, a
3769 frame's local gain factor is defined as the factor that would
3770 result in exactly that RMS value. Note, however, that the maximum
3771 local gain factor is still restricted by the frame's highest
3772 magnitude sample, in order to prevent clipping.
3773 coupling, n
3775 Enable channels coupling. By default is enabled. By default, the
3776 Dynamic Audio Normalizer will amplify all channels by the same
3777 amount. This means the same gain factor will be applied to all
3778 channels, i.e. the maximum possible gain factor is determined by
3779 the "loudest" channel. However, in some recordings, it may happen
3780 that the volume of the different channels is uneven, e.g. one
3781 channel may be "quieter" than the other one(s). In this case, this
3782 option can be used to disable the channel coupling. This way, the
3783 gain factor will be determined independently for each channel,
3784 depending only on the individual channel's highest magnitude
3785 sample. This allows for harmonizing the volume of the different
3786 channels.
3787 correctdc, c
3789 Enable DC bias correction. By default is disabled. An audio signal
3790 (in the time domain) is a sequence of sample values. In the
3791 Dynamic Audio Normalizer these sample values are represented in the
3792 -1.0 to 1.0 range, regardless of the original input format.
3793 Normally, the audio signal, or "waveform", should be centered
3794 around the zero point. That means if we calculate the mean value
3795 of all samples in a file, or in a single frame, then the result
3796 should be 0.0 or at least very close to that value. If, however,
3797 there is a significant deviation of the mean value from 0.0, in
3798 either positive or negative direction, this is referred to as a DC
3799 bias or DC offset. Since a DC bias is clearly undesirable, the
3800 Dynamic Audio Normalizer provides optional DC bias correction.
3801 With DC bias correction enabled, the Dynamic Audio Normalizer will
3802 determine the mean value, or "DC correction" offset, of each input
3803 frame and subtract that value from all of the frame's sample values
3804 which ensures those samples are centered around 0.0 again. Also, in
3805 order to avoid "gaps" at the frame boundaries, the DC correction
3806 offset values will be interpolated smoothly between neighbouring
3807 frames.
3808 altboundary, b
3810 Enable alternative boundary mode. By default is disabled. The
3811 Dynamic Audio Normalizer takes into account a certain neighbourhood
3812 around each frame. This includes the preceding frames as well as
3813 the subsequent frames. However, for the "boundary" frames, located
3814 at the very beginning and at the very end of the audio file, not
3815 all neighbouring frames are available. In particular, for the first
3816 few frames in the audio file, the preceding frames are not known.
3817 And, similarly, for the last few frames in the audio file, the
3818 subsequent frames are not known. Thus, the question arises which
3819 gain factors should be assumed for the missing frames in the
3820 "boundary" region. The Dynamic Audio Normalizer implements two
3821 modes to deal with this situation. The default boundary mode
3822 assumes a gain factor of exactly 1.0 for the missing frames,
3823 resulting in a smooth "fade in" and "fade out" at the beginning and
3824 at the end of the input, respectively.
3825 compress, s
3827 Set the compress factor. In range from 0.0 to 30.0. Default is 0.0.
3828 By default, the Dynamic Audio Normalizer does not apply
3829 "traditional" compression. This means that signal peaks will not be
3830 pruned and thus the full dynamic range will be retained within each
3831 local neighbourhood. However, in some cases it may be desirable to
3832 combine the Dynamic Audio Normalizer's normalization algorithm with
3833 a more "traditional" compression. For this purpose, the Dynamic
3834 Audio Normalizer provides an optional compression (thresholding)
3835 function. If (and only if) the compression feature is enabled, all
3836 input frames will be processed by a soft knee thresholding function
3837 prior to the actual normalization process. Put simply, the
3838 thresholding function is going to prune all samples whose magnitude
3839 exceeds a certain threshold value. However, the Dynamic Audio
3840 Normalizer does not simply apply a fixed threshold value. Instead,
3841 the threshold value will be adjusted for each individual frame. In
3842 general, smaller parameters result in stronger compression, and
3843 vice versa. Values below 3.0 are not recommended, because audible
3844 distortion may appear.
3845 threshold, t
3847 Set the target threshold value. This specifies the lowest
3848 permissible magnitude level for the audio input which will be
3849 normalized. If input frame volume is above this value frame will
3850 be normalized. Otherwise frame may not be normalized at all. The
3851 default value is set to 0, which means all input frames will be
3852 normalized. This option is mostly useful if digital noise is not
3853 wanted to be amplified.
3854 Commands
3856 This filter supports the all above options as commands.
3858 earwax
3860 Make audio easier to listen to on headphones.
3861 This filter adds `cues' to 44.1kHz stereo (i.e. audio CD format) audio
3863 so that when listened to on headphones the stereo image is moved from
3864 inside your head (standard for headphones) to outside and in front of
3865 the listener (standard for speakers).
3866 Ported from SoX.
3868 equalizer
3870 Apply a two-pole peaking equalisation (EQ) filter. With this filter,
3871 the signal-level at and around a selected frequency can be increased or
3872 decreased, whilst (unlike bandpass and bandreject filters) that at all
3873 other frequencies is unchanged.
3874 In order to produce complex equalisation curves, this filter can be
3876 given several times, each with a different central frequency.
3877 The filter accepts the following options:
3879 frequency, f
3881 Set the filter's central frequency in Hz.
3882 width_type, t
3884 Set method to specify band-width of filter.
3885 h Hz
3887 q Q-Factor
3889 o octave
3891 s slope
3893 k kHz
3895 width, w
3897 Specify the band-width of a filter in width_type units.
3898 gain, g
3900 Set the required gain or attenuation in dB. Beware of clipping
3901 when using a positive gain.
3902 mix, m
3904 How much to use filtered signal in output. Default is 1. Range is
3905 between 0 and 1.
3906 channels, c
3908 Specify which channels to filter, by default all available are
3909 filtered.
3910 normalize, n
3912 Normalize biquad coefficients, by default is disabled. Enabling it
3913 will normalize magnitude response at DC to 0dB.
3914 transform, a
3916 Set transform type of IIR filter.
3917 di
3919 dii
3920 tdii
3921 latt
3922 svf
3923 precision, r
3924 Set precison of filtering.
3925 auto
3927 Pick automatic sample format depending on surround filters.
3928 s16 Always use signed 16-bit.
3930 s32 Always use signed 32-bit.
3932 f32 Always use float 32-bit.
3934 f64 Always use float 64-bit.
3936 Examples
3938 • Attenuate 10 dB at 1000 Hz, with a bandwidth of 200 Hz:
3940 equalizer=f=1000:t=h:width=200:g=-10
3942 • Apply 2 dB gain at 1000 Hz with Q 1 and attenuate 5 dB at 100 Hz
3944 with Q 2:
3945 equalizer=f=1000:t=q:w=1:g=2,equalizer=f=100:t=q:w=2:g=-5
3947 Commands
3949 This filter supports the following commands:
3951 frequency, f
3953 Change equalizer frequency. Syntax for the command is :
3954 "frequency"
3955 width_type, t
3957 Change equalizer width_type. Syntax for the command is :
3958 "width_type"
3959 width, w
3961 Change equalizer width. Syntax for the command is : "width"
3962 gain, g
3964 Change equalizer gain. Syntax for the command is : "gain"
3965 mix, m
3967 Change equalizer mix. Syntax for the command is : "mix"
3968 extrastereo
3970 Linearly increases the difference between left and right channels which
3971 adds some sort of "live" effect to playback.
3972 The filter accepts the following options:
3974 m Sets the difference coefficient (default: 2.5). 0.0 means mono
3976 sound (average of both channels), with 1.0 sound will be unchanged,
3977 with -1.0 left and right channels will be swapped.
3978 c Enable clipping. By default is enabled.
3980 Commands
3982 This filter supports the all above options as commands.
3984 firequalizer
3986 Apply FIR Equalization using arbitrary frequency response.
3987 The filter accepts the following option:
3989 gain
3991 Set gain curve equation (in dB). The expression can contain
3992 variables:
3993 f the evaluated frequency
3995 sr sample rate
3997 ch channel number, set to 0 when multichannels evaluation is
3999 disabled
4000 chid
4002 channel id, see libavutil/channel_layout.h, set to the first
4003 channel id when multichannels evaluation is disabled
4004 chs number of channels
4006 chlayout
4008 channel_layout, see libavutil/channel_layout.h
4009 and functions:
4011 gain_interpolate(f)
4013 interpolate gain on frequency f based on gain_entry
4014 cubic_interpolate(f)
4016 same as gain_interpolate, but smoother
4017 This option is also available as command. Default is
4019 gain_interpolate(f).
4020 gain_entry
4022 Set gain entry for gain_interpolate function. The expression can
4023 contain functions:
4024 entry(f, g)
4026 store gain entry at frequency f with value g
4027 This option is also available as command.
4029 delay
4031 Set filter delay in seconds. Higher value means more accurate.
4032 Default is 0.01.
4033 accuracy
4035 Set filter accuracy in Hz. Lower value means more accurate.
4036 Default is 5.
4037 wfunc
4039 Set window function. Acceptable values are:
4040 rectangular
4042 rectangular window, useful when gain curve is already smooth
4043 hann
4045 hann window (default)
4046 hamming
4048 hamming window
4049 blackman
4051 blackman window
4052 nuttall3
4054 3-terms continuous 1st derivative nuttall window
4055 mnuttall3
4057 minimum 3-terms discontinuous nuttall window
4058 nuttall
4060 4-terms continuous 1st derivative nuttall window
4061 bnuttall
4063 minimum 4-terms discontinuous nuttall (blackman-nuttall) window
4064 bharris
4066 blackman-harris window
4067 tukey
4069 tukey window
4070 fixed
4072 If enabled, use fixed number of audio samples. This improves speed
4073 when filtering with large delay. Default is disabled.
4074 multi
4076 Enable multichannels evaluation on gain. Default is disabled.
4077 zero_phase
4079 Enable zero phase mode by subtracting timestamp to compensate
4080 delay. Default is disabled.
4081 scale
4083 Set scale used by gain. Acceptable values are:
4084 linlin
4086 linear frequency, linear gain
4087 linlog
4089 linear frequency, logarithmic (in dB) gain (default)
4090 loglin
4092 logarithmic (in octave scale where 20 Hz is 0) frequency,
4093 linear gain
4094 loglog
4096 logarithmic frequency, logarithmic gain
4097 dumpfile
4099 Set file for dumping, suitable for gnuplot.
4100 dumpscale
4102 Set scale for dumpfile. Acceptable values are same with scale
4103 option. Default is linlog.
4104 fft2
4106 Enable 2-channel convolution using complex FFT. This improves speed
4107 significantly. Default is disabled.
4108 min_phase
4110 Enable minimum phase impulse response. Default is disabled.
4111 Examples
4113 • lowpass at 1000 Hz:
4115 firequalizer=gain='if(lt(f,1000), 0, -INF)'
4117 • lowpass at 1000 Hz with gain_entry:
4119 firequalizer=gain_entry='entry(1000,0); entry(1001, -INF)'
4121 • custom equalization:
4123 firequalizer=gain_entry='entry(100,0); entry(400, -4); entry(1000, -6); entry(2000, 0)'
4125 • higher delay with zero phase to compensate delay:
4127 firequalizer=delay=0.1:fixed=on:zero_phase=on
4129 • lowpass on left channel, highpass on right channel:
4131 firequalizer=gain='if(eq(chid,1), gain_interpolate(f), if(eq(chid,2), gain_interpolate(1e6+f), 0))'
4133 :gain_entry='entry(1000, 0); entry(1001,-INF); entry(1e6+1000,0)':multi=on
4134 flanger
4136 Apply a flanging effect to the audio.
4137 The filter accepts the following options:
4139 delay
4141 Set base delay in milliseconds. Range from 0 to 30. Default value
4142 is 0.
4143 depth
4145 Set added sweep delay in milliseconds. Range from 0 to 10. Default
4146 value is 2.
4147 regen
4149 Set percentage regeneration (delayed signal feedback). Range from
4150 -95 to 95. Default value is 0.
4151 width
4153 Set percentage of delayed signal mixed with original. Range from 0
4154 to 100. Default value is 71.
4155 speed
4157 Set sweeps per second (Hz). Range from 0.1 to 10. Default value is
4158 0.5.
4159 shape
4161 Set swept wave shape, can be triangular or sinusoidal. Default
4162 value is sinusoidal.
4163 phase
4165 Set swept wave percentage-shift for multi channel. Range from 0 to
4166 100. Default value is 25.
4167 interp
4169 Set delay-line interpolation, linear or quadratic. Default is
4170 linear.
4171 haas
4173 Apply Haas effect to audio.
4174 Note that this makes most sense to apply on mono signals. With this
4176 filter applied to mono signals it give some directionality and
4177 stretches its stereo image.
4178 The filter accepts the following options:
4180 level_in
4182 Set input level. By default is 1, or 0dB
4183 level_out
4185 Set output level. By default is 1, or 0dB.
4186 side_gain
4188 Set gain applied to side part of signal. By default is 1.
4189 middle_source
4191 Set kind of middle source. Can be one of the following:
4192 left
4194 Pick left channel.
4195 right
4197 Pick right channel.
4198 mid Pick middle part signal of stereo image.
4200 side
4202 Pick side part signal of stereo image.
4203 middle_phase
4205 Change middle phase. By default is disabled.
4206 left_delay
4208 Set left channel delay. By default is 2.05 milliseconds.
4209 left_balance
4211 Set left channel balance. By default is -1.
4212 left_gain
4214 Set left channel gain. By default is 1.
4215 left_phase
4217 Change left phase. By default is disabled.
4218 right_delay
4220 Set right channel delay. By defaults is 2.12 milliseconds.
4221 right_balance
4223 Set right channel balance. By default is 1.
4224 right_gain
4226 Set right channel gain. By default is 1.
4227 right_phase
4229 Change right phase. By default is enabled.
4230 hdcd
4232 Decodes High Definition Compatible Digital (HDCD) data. A 16-bit PCM
4233 stream with embedded HDCD codes is expanded into a 20-bit PCM stream.
4234 The filter supports the Peak Extend and Low-level Gain Adjustment
4236 features of HDCD, and detects the Transient Filter flag.
4237 ffmpeg -i HDCD16.flac -af hdcd OUT24.flac
4239 When using the filter with wav, note the default encoding for wav is
4241 16-bit, so the resulting 20-bit stream will be truncated back to
4242 16-bit. Use something like -acodec pcm_s24le after the filter to get
4243 24-bit PCM output.
4244 ffmpeg -i HDCD16.wav -af hdcd OUT16.wav
4246 ffmpeg -i HDCD16.wav -af hdcd -c:a pcm_s24le OUT24.wav
4247 The filter accepts the following options:
4249 disable_autoconvert
4251 Disable any automatic format conversion or resampling in the filter
4252 graph.
4253 process_stereo
4255 Process the stereo channels together. If target_gain does not match
4256 between channels, consider it invalid and use the last valid
4257 target_gain.
4258 cdt_ms
4260 Set the code detect timer period in ms.
4261 force_pe
4263 Always extend peaks above -3dBFS even if PE isn't signaled.
4264 analyze_mode
4266 Replace audio with a solid tone and adjust the amplitude to signal
4267 some specific aspect of the decoding process. The output file can
4268 be loaded in an audio editor alongside the original to aid
4269 analysis.
4270 "analyze_mode=pe:force_pe=true" can be used to see all samples
4272 above the PE level.
4273 Modes are:
4275 0, off
4277 Disabled
4278 1, lle
4280 Gain adjustment level at each sample
4281 2, pe
4283 Samples where peak extend occurs
4284 3, cdt
4286 Samples where the code detect timer is active
4287 4, tgm
4289 Samples where the target gain does not match between channels
4290 headphone
4292 Apply head-related transfer functions (HRTFs) to create virtual
4293 loudspeakers around the user for binaural listening via headphones.
4294 The HRIRs are provided via additional streams, for each channel one
4295 stereo input stream is needed.
4296 The filter accepts the following options:
4298 map Set mapping of input streams for convolution. The argument is a
4300 '|'-separated list of channel names in order as they are given as
4301 additional stream inputs for filter. This also specify number of
4302 input streams. Number of input streams must be not less than number
4303 of channels in first stream plus one.
4304 gain
4306 Set gain applied to audio. Value is in dB. Default is 0.
4307 type
4309 Set processing type. Can be time or freq. time is processing audio
4310 in time domain which is slow. freq is processing audio in
4311 frequency domain which is fast. Default is freq.
4312 lfe Set custom gain for LFE channels. Value is in dB. Default is 0.
4314 size
4316 Set size of frame in number of samples which will be processed at
4317 once. Default value is 1024. Allowed range is from 1024 to 96000.
4318 hrir
4320 Set format of hrir stream. Default value is stereo. Alternative
4321 value is multich. If value is set to stereo, number of additional
4322 streams should be greater or equal to number of input channels in
4323 first input stream. Also each additional stream should have stereo
4324 number of channels. If value is set to multich, number of
4325 additional streams should be exactly one. Also number of input
4326 channels of additional stream should be equal or greater than twice
4327 number of channels of first input stream.
4328 Examples
4330 • Full example using wav files as coefficients with amovie filters
4332 for 7.1 downmix, each amovie filter use stereo file with IR
4333 coefficients as input. The files give coefficients for each
4334 position of virtual loudspeaker:
4335 ffmpeg -i input.wav
4337 -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"
4338 output.wav
4339 • Full example using wav files as coefficients with amovie filters
4341 for 7.1 downmix, but now in multich hrir format.
4342 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"
4344 output.wav
4345 highpass
4347 Apply a high-pass filter with 3dB point frequency. The filter can be
4348 either single-pole, or double-pole (the default). The filter roll off
4349 at 6dB per pole per octave (20dB per pole per decade).
4350 The filter accepts the following options:
4352 frequency, f
4354 Set frequency in Hz. Default is 3000.
4355 poles, p
4357 Set number of poles. Default is 2.
4358 width_type, t
4360 Set method to specify band-width of filter.
4361 h Hz
4363 q Q-Factor
4365 o octave
4367 s slope
4369 k kHz
4371 width, w
4373 Specify the band-width of a filter in width_type units. Applies
4374 only to double-pole filter. The default is 0.707q and gives a
4375 Butterworth response.
4376 mix, m
4378 How much to use filtered signal in output. Default is 1. Range is
4379 between 0 and 1.
4380 channels, c
4382 Specify which channels to filter, by default all available are
4383 filtered.
4384 normalize, n
4386 Normalize biquad coefficients, by default is disabled. Enabling it
4387 will normalize magnitude response at DC to 0dB.
4388 transform, a
4390 Set transform type of IIR filter.
4391 di
4393 dii
4394 tdii
4395 latt
4396 svf
4397 precision, r
4398 Set precison of filtering.
4399 auto
4401 Pick automatic sample format depending on surround filters.
4402 s16 Always use signed 16-bit.
4404 s32 Always use signed 32-bit.
4406 f32 Always use float 32-bit.
4408 f64 Always use float 64-bit.
4410 Commands
4412 This filter supports the following commands:
4414 frequency, f
4416 Change highpass frequency. Syntax for the command is : "frequency"
4417 width_type, t
4419 Change highpass width_type. Syntax for the command is :
4420 "width_type"
4421 width, w
4423 Change highpass width. Syntax for the command is : "width"
4424 mix, m
4426 Change highpass mix. Syntax for the command is : "mix"
4427 join
4429 Join multiple input streams into one multi-channel stream.
4430 It accepts the following parameters:
4432 inputs
4434 The number of input streams. It defaults to 2.
4435 channel_layout
4437 The desired output channel layout. It defaults to stereo.
4438 map Map channels from inputs to output. The argument is a '|'-separated
4440 list of mappings, each in the "input_idx.in_channel-out_channel"
4441 form. input_idx is the 0-based index of the input stream.
4442 in_channel can be either the name of the input channel (e.g. FL for
4443 front left) or its index in the specified input stream. out_channel
4444 is the name of the output channel.
4445 The filter will attempt to guess the mappings when they are not
4447 specified explicitly. It does so by first trying to find an unused
4448 matching input channel and if that fails it picks the first unused
4449 input channel.
4450 Join 3 inputs (with properly set channel layouts):
4452 ffmpeg -i INPUT1 -i INPUT2 -i INPUT3 -filter_complex join=inputs=3 OUTPUT
4454 Build a 5.1 output from 6 single-channel streams:
4456 ffmpeg -i fl -i fr -i fc -i sl -i sr -i lfe -filter_complex
4458 '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'
4459 out
4460 ladspa
4462 Load a LADSPA (Linux Audio Developer's Simple Plugin API) plugin.
4463 To enable compilation of this filter you need to configure FFmpeg with
4465 "--enable-ladspa".
4466 file, f
4468 Specifies the name of LADSPA plugin library to load. If the
4469 environment variable LADSPA_PATH is defined, the LADSPA plugin is
4470 searched in each one of the directories specified by the colon
4471 separated list in LADSPA_PATH, otherwise in the standard LADSPA
4472 paths, which are in this order: HOME/.ladspa/lib/,
4473 /usr/local/lib/ladspa/, /usr/lib/ladspa/.
4474 plugin, p
4476 Specifies the plugin within the library. Some libraries contain
4477 only one plugin, but others contain many of them. If this is not
4478 set filter will list all available plugins within the specified
4479 library.
4480 controls, c
4482 Set the '|' separated list of controls which are zero or more
4483 floating point values that determine the behavior of the loaded
4484 plugin (for example delay, threshold or gain). Controls need to be
4485 defined using the following syntax:
4486 c0=value0|c1=value1|c2=value2|..., where valuei is the value set on
4487 the i-th control. Alternatively they can be also defined using the
4488 following syntax: value0|value1|value2|..., where valuei is the
4489 value set on the i-th control. If controls is set to "help", all
4490 available controls and their valid ranges are printed.
4491 sample_rate, s
4493 Specify the sample rate, default to 44100. Only used if plugin have
4494 zero inputs.
4495 nb_samples, n
4497 Set the number of samples per channel per each output frame,
4498 default is 1024. Only used if plugin have zero inputs.
4499 duration, d
4501 Set the minimum duration of the sourced audio. See the Time
4502 duration section in the ffmpeg-utils(1) manual for the accepted
4503 syntax. Note that the resulting duration may be greater than the
4504 specified duration, as the generated audio is always cut at the end
4505 of a complete frame. If not specified, or the expressed duration
4506 is negative, the audio is supposed to be generated forever. Only
4507 used if plugin have zero inputs.
4508 latency, l
4510 Enable latency compensation, by default is disabled. Only used if
4511 plugin have inputs.
4512 Examples
4514 • List all available plugins within amp (LADSPA example plugin)
4516 library:
4517 ladspa=file=amp
4519 • List all available controls and their valid ranges for "vcf_notch"
4521 plugin from "VCF" library:
4522 ladspa=f=vcf:p=vcf_notch:c=help
4524 • Simulate low quality audio equipment using "Computer Music Toolkit"
4526 (CMT) plugin library:
4527 ladspa=file=cmt:plugin=lofi:controls=c0=22|c1=12|c2=12
4529 • Add reverberation to the audio using TAP-plugins (Tom's Audio
4531 Processing plugins):
4532 ladspa=file=tap_reverb:tap_reverb
4534 • Generate white noise, with 0.2 amplitude:
4536 ladspa=file=cmt:noise_source_white:c=c0=.2
4538 • Generate 20 bpm clicks using plugin "C* Click - Metronome" from the
4540 "C* Audio Plugin Suite" (CAPS) library:
4541 ladspa=file=caps:Click:c=c1=20'
4543 • Apply "C* Eq10X2 - Stereo 10-band equaliser" effect:
4545 ladspa=caps:Eq10X2:c=c0=-48|c9=-24|c3=12|c4=2
4547 • Increase volume by 20dB using fast lookahead limiter from Steve
4549 Harris "SWH Plugins" collection:
4550 ladspa=fast_lookahead_limiter_1913:fastLookaheadLimiter:20|0|2
4552 • Attenuate low frequencies using Multiband EQ from Steve Harris "SWH
4554 Plugins" collection:
4555 ladspa=mbeq_1197:mbeq:-24|-24|-24|0|0|0|0|0|0|0|0|0|0|0|0
4557 • Reduce stereo image using "Narrower" from the "C* Audio Plugin
4559 Suite" (CAPS) library:
4560 ladspa=caps:Narrower
4562 • Another white noise, now using "C* Audio Plugin Suite" (CAPS)
4564 library:
4565 ladspa=caps:White:.2
4567 • Some fractal noise, using "C* Audio Plugin Suite" (CAPS) library:
4569 ladspa=caps:Fractal:c=c1=1
4571 • Dynamic volume normalization using "VLevel" plugin:
4573 ladspa=vlevel-ladspa:vlevel_mono
4575 Commands
4577 This filter supports the following commands:
4579 cN Modify the N-th control value.
4581 If the specified value is not valid, it is ignored and prior one is
4583 kept.
4584 loudnorm
4586 EBU R128 loudness normalization. Includes both dynamic and linear
4587 normalization modes. Support for both single pass (livestreams, files)
4588 and double pass (files) modes. This algorithm can target IL, LRA, and
4589 maximum true peak. In dynamic mode, to accurately detect true peaks,
4590 the audio stream will be upsampled to 192 kHz. Use the "-ar" option or
4591 "aresample" filter to explicitly set an output sample rate.
4592 The filter accepts the following options:
4594 I, i
4596 Set integrated loudness target. Range is -70.0 - -5.0. Default
4597 value is -24.0.
4598 LRA, lra
4600 Set loudness range target. Range is 1.0 - 20.0. Default value is
4601 7.0.
4602 TP, tp
4604 Set maximum true peak. Range is -9.0 - +0.0. Default value is
4605 -2.0.
4606 measured_I, measured_i
4608 Measured IL of input file. Range is -99.0 - +0.0.
4609 measured_LRA, measured_lra
4611 Measured LRA of input file. Range is 0.0 - 99.0.
4612 measured_TP, measured_tp
4614 Measured true peak of input file. Range is -99.0 - +99.0.
4615 measured_thresh
4617 Measured threshold of input file. Range is -99.0 - +0.0.
4618 offset
4620 Set offset gain. Gain is applied before the true-peak limiter.
4621 Range is -99.0 - +99.0. Default is +0.0.
4622 linear
4624 Normalize by linearly scaling the source audio. "measured_I",
4625 "measured_LRA", "measured_TP", and "measured_thresh" must all be
4626 specified. Target LRA shouldn't be lower than source LRA and the
4627 change in integrated loudness shouldn't result in a true peak which
4628 exceeds the target TP. If any of these conditions aren't met,
4629 normalization mode will revert to dynamic. Options are "true" or
4630 "false". Default is "true".
4631 dual_mono
4633 Treat mono input files as "dual-mono". If a mono file is intended
4634 for playback on a stereo system, its EBU R128 measurement will be
4635 perceptually incorrect. If set to "true", this option will
4636 compensate for this effect. Multi-channel input files are not
4637 affected by this option. Options are true or false. Default is
4638 false.
4639 print_format
4641 Set print format for stats. Options are summary, json, or none.
4642 Default value is none.
4643 lowpass
4645 Apply a low-pass filter with 3dB point frequency. The filter can be
4646 either single-pole or double-pole (the default). The filter roll off
4647 at 6dB per pole per octave (20dB per pole per decade).
4648 The filter accepts the following options:
4650 frequency, f
4652 Set frequency in Hz. Default is 500.
4653 poles, p
4655 Set number of poles. Default is 2.
4656 width_type, t
4658 Set method to specify band-width of filter.
4659 h Hz
4661 q Q-Factor
4663 o octave
4665 s slope
4667 k kHz
4669 width, w
4671 Specify the band-width of a filter in width_type units. Applies
4672 only to double-pole filter. The default is 0.707q and gives a
4673 Butterworth response.
4674 mix, m
4676 How much to use filtered signal in output. Default is 1. Range is
4677 between 0 and 1.
4678 channels, c
4680 Specify which channels to filter, by default all available are
4681 filtered.
4682 normalize, n
4684 Normalize biquad coefficients, by default is disabled. Enabling it
4685 will normalize magnitude response at DC to 0dB.
4686 transform, a
4688 Set transform type of IIR filter.
4689 di
4691 dii
4692 tdii
4693 latt
4694 svf
4695 precision, r
4696 Set precison of filtering.
4697 auto
4699 Pick automatic sample format depending on surround filters.
4700 s16 Always use signed 16-bit.
4702 s32 Always use signed 32-bit.
4704 f32 Always use float 32-bit.
4706 f64 Always use float 64-bit.
4708 Examples
4710 • Lowpass only LFE channel, it LFE is not present it does nothing:
4712 lowpass=c=LFE
4714 Commands
4716 This filter supports the following commands:
4718 frequency, f
4720 Change lowpass frequency. Syntax for the command is : "frequency"
4721 width_type, t
4723 Change lowpass width_type. Syntax for the command is :
4724 "width_type"
4725 width, w
4727 Change lowpass width. Syntax for the command is : "width"
4728 mix, m
4730 Change lowpass mix. Syntax for the command is : "mix"
4731 lv2
4733 Load a LV2 (LADSPA Version 2) plugin.
4734 To enable compilation of this filter you need to configure FFmpeg with
4736 "--enable-lv2".
4737 plugin, p
4739 Specifies the plugin URI. You may need to escape ':'.
4740 controls, c
4742 Set the '|' separated list of controls which are zero or more
4743 floating point values that determine the behavior of the loaded
4744 plugin (for example delay, threshold or gain). If controls is set
4745 to "help", all available controls and their valid ranges are
4746 printed.
4747 sample_rate, s
4749 Specify the sample rate, default to 44100. Only used if plugin have
4750 zero inputs.
4751 nb_samples, n
4753 Set the number of samples per channel per each output frame,
4754 default is 1024. Only used if plugin have zero inputs.
4755 duration, d
4757 Set the minimum duration of the sourced audio. See the Time
4758 duration section in the ffmpeg-utils(1) manual for the accepted
4759 syntax. Note that the resulting duration may be greater than the
4760 specified duration, as the generated audio is always cut at the end
4761 of a complete frame. If not specified, or the expressed duration
4762 is negative, the audio is supposed to be generated forever. Only
4763 used if plugin have zero inputs.
4764 Examples
4766 • Apply bass enhancer plugin from Calf:
4768 lv2=p=http\\\\://calf.sourceforge.net/plugins/BassEnhancer:c=amount=2
4770 • Apply vinyl plugin from Calf:
4772 lv2=p=http\\\\://calf.sourceforge.net/plugins/Vinyl:c=drone=0.2|aging=0.5
4774 • Apply bit crusher plugin from ArtyFX:
4776 lv2=p=http\\\\://www.openavproductions.com/artyfx#bitta:c=crush=0.3
4778 mcompand
4780 Multiband Compress or expand the audio's dynamic range.
4781 The input audio is divided into bands using 4th order Linkwitz-Riley
4783 IIRs. This is akin to the crossover of a loudspeaker, and results in
4784 flat frequency response when absent compander action.
4785 It accepts the following parameters:
4787 args
4789 This option syntax is: attack,decay,[attack,decay..] soft-knee
4790 points crossover_frequency [delay [initial_volume [gain]]] |
4791 attack,decay ... For explanation of each item refer to compand
4792 filter documentation.
4793 pan
4795 Mix channels with specific gain levels. The filter accepts the output
4796 channel layout followed by a set of channels definitions.
4797 This filter is also designed to efficiently remap the channels of an
4799 audio stream.
4800 The filter accepts parameters of the form: "l|outdef|outdef|..."
4802 l output channel layout or number of channels
4804 outdef
4806 output channel specification, of the form:
4807 "out_name=[gain*]in_name[(+-)[gain*]in_name...]"
4808 out_name
4810 output channel to define, either a channel name (FL, FR, etc.) or a
4811 channel number (c0, c1, etc.)
4812 gain
4814 multiplicative coefficient for the channel, 1 leaving the volume
4815 unchanged
4816 in_name
4818 input channel to use, see out_name for details; it is not possible
4819 to mix named and numbered input channels
4820 If the `=' in a channel specification is replaced by `<', then the
4822 gains for that specification will be renormalized so that the total is
4823 1, thus avoiding clipping noise.
4824 Mixing examples
4826 For example, if you want to down-mix from stereo to mono, but with a
4828 bigger factor for the left channel:
4829 pan=1c|c0=0.9*c0+0.1*c1
4831 A customized down-mix to stereo that works automatically for 3-, 4-, 5-
4833 and 7-channels surround:
4834 pan=stereo| FL < FL + 0.5*FC + 0.6*BL + 0.6*SL | FR < FR + 0.5*FC + 0.6*BR + 0.6*SR
4836 Note that ffmpeg integrates a default down-mix (and up-mix) system that
4838 should be preferred (see "-ac" option) unless you have very specific
4839 needs.
4840 Remapping examples
4842 The channel remapping will be effective if, and only if:
4844 *<gain coefficients are zeroes or ones,>
4846 *<only one input per channel output,>
4847 If all these conditions are satisfied, the filter will notify the user
4849 ("Pure channel mapping detected"), and use an optimized and lossless
4850 method to do the remapping.
4851 For example, if you have a 5.1 source and want a stereo audio stream by
4853 dropping the extra channels:
4854 pan="stereo| c0=FL | c1=FR"
4856 Given the same source, you can also switch front left and front right
4858 channels and keep the input channel layout:
4859 pan="5.1| c0=c1 | c1=c0 | c2=c2 | c3=c3 | c4=c4 | c5=c5"
4861 If the input is a stereo audio stream, you can mute the front left
4863 channel (and still keep the stereo channel layout) with:
4864 pan="stereo|c1=c1"
4866 Still with a stereo audio stream input, you can copy the right channel
4868 in both front left and right:
4869 pan="stereo| c0=FR | c1=FR"
4871 replaygain
4873 ReplayGain scanner filter. This filter takes an audio stream as an
4874 input and outputs it unchanged. At end of filtering it displays
4875 "track_gain" and "track_peak".
4876 resample
4878 Convert the audio sample format, sample rate and channel layout. It is
4879 not meant to be used directly.
4880 rubberband
4882 Apply time-stretching and pitch-shifting with librubberband.
4883 To enable compilation of this filter, you need to configure FFmpeg with
4885 "--enable-librubberband".
4886 The filter accepts the following options:
4888 tempo
4890 Set tempo scale factor.
4891 pitch
4893 Set pitch scale factor.
4894 transients
4896 Set transients detector. Possible values are:
4897 crisp
4899 mixed
4900 smooth
4901 detector
4902 Set detector. Possible values are:
4903 compound
4905 percussive
4906 soft
4907 phase
4908 Set phase. Possible values are:
4909 laminar
4911 independent
4912 window
4913 Set processing window size. Possible values are:
4914 standard
4916 short
4917 long
4918 smoothing
4919 Set smoothing. Possible values are:
4920 off
4922 on
4923 formant
4924 Enable formant preservation when shift pitching. Possible values
4925 are:
4926 shifted
4928 preserved
4929 pitchq
4930 Set pitch quality. Possible values are:
4931 quality
4933 speed
4934 consistency
4935 channels
4936 Set channels. Possible values are:
4937 apart
4939 together
4940 Commands
4942 This filter supports the following commands:
4944 tempo
4946 Change filter tempo scale factor. Syntax for the command is :
4947 "tempo"
4948 pitch
4950 Change filter pitch scale factor. Syntax for the command is :
4951 "pitch"
4952 sidechaincompress
4954 This filter acts like normal compressor but has the ability to compress
4955 detected signal using second input signal. It needs two input streams
4956 and returns one output stream. First input stream will be processed
4957 depending on second stream signal. The filtered signal then can be
4958 filtered with other filters in later stages of processing. See pan and
4959 amerge filter.
4960 The filter accepts the following options:
4962 level_in
4964 Set input gain. Default is 1. Range is between 0.015625 and 64.
4965 mode
4967 Set mode of compressor operation. Can be "upward" or "downward".
4968 Default is "downward".
4969 threshold
4971 If a signal of second stream raises above this level it will affect
4972 the gain reduction of first stream. By default is 0.125. Range is
4973 between 0.00097563 and 1.
4974 ratio
4976 Set a ratio about which the signal is reduced. 1:2 means that if
4977 the level raised 4dB above the threshold, it will be only 2dB above
4978 after the reduction. Default is 2. Range is between 1 and 20.
4979 attack
4981 Amount of milliseconds the signal has to rise above the threshold
4982 before gain reduction starts. Default is 20. Range is between 0.01
4983 and 2000.
4984 release
4986 Amount of milliseconds the signal has to fall below the threshold
4987 before reduction is decreased again. Default is 250. Range is
4988 between 0.01 and 9000.
4989 makeup
4991 Set the amount by how much signal will be amplified after
4992 processing. Default is 1. Range is from 1 to 64.
4993 knee
4995 Curve the sharp knee around the threshold to enter gain reduction
4996 more softly. Default is 2.82843. Range is between 1 and 8.
4997 link
4999 Choose if the "average" level between all channels of side-chain
5000 stream or the louder("maximum") channel of side-chain stream
5001 affects the reduction. Default is "average".
5002 detection
5004 Should the exact signal be taken in case of "peak" or an RMS one in
5005 case of "rms". Default is "rms" which is mainly smoother.
5006 level_sc
5008 Set sidechain gain. Default is 1. Range is between 0.015625 and 64.
5009 mix How much to use compressed signal in output. Default is 1. Range
5011 is between 0 and 1.
5012 Commands
5014 This filter supports the all above options as commands.
5016 Examples
5018 • Full ffmpeg example taking 2 audio inputs, 1st input to be
5020 compressed depending on the signal of 2nd input and later
5021 compressed signal to be merged with 2nd input:
5022 ffmpeg -i main.flac -i sidechain.flac -filter_complex "[1:a]asplit=2[sc][mix];[0:a][sc]sidechaincompress[compr];[compr][mix]amerge"
5024 sidechaingate
5026 A sidechain gate acts like a normal (wideband) gate but has the ability
5027 to filter the detected signal before sending it to the gain reduction
5028 stage. Normally a gate uses the full range signal to detect a level
5029 above the threshold. For example: If you cut all lower frequencies
5030 from your sidechain signal the gate will decrease the volume of your
5031 track only if not enough highs appear. With this technique you are able
5032 to reduce the resonation of a natural drum or remove "rumbling" of
5033 muted strokes from a heavily distorted guitar. It needs two input
5034 streams and returns one output stream. First input stream will be
5035 processed depending on second stream signal.
5036 The filter accepts the following options:
5038 level_in
5040 Set input level before filtering. Default is 1. Allowed range is
5041 from 0.015625 to 64.
5042 mode
5044 Set the mode of operation. Can be "upward" or "downward". Default
5045 is "downward". If set to "upward" mode, higher parts of signal will
5046 be amplified, expanding dynamic range in upward direction.
5047 Otherwise, in case of "downward" lower parts of signal will be
5048 reduced.
5049 range
5051 Set the level of gain reduction when the signal is below the
5052 threshold. Default is 0.06125. Allowed range is from 0 to 1.
5053 Setting this to 0 disables reduction and then filter behaves like
5054 expander.
5055 threshold
5057 If a signal rises above this level the gain reduction is released.
5058 Default is 0.125. Allowed range is from 0 to 1.
5059 ratio
5061 Set a ratio about which the signal is reduced. Default is 2.
5062 Allowed range is from 1 to 9000.
5063 attack
5065 Amount of milliseconds the signal has to rise above the threshold
5066 before gain reduction stops. Default is 20 milliseconds. Allowed
5067 range is from 0.01 to 9000.
5068 release
5070 Amount of milliseconds the signal has to fall below the threshold
5071 before the reduction is increased again. Default is 250
5072 milliseconds. Allowed range is from 0.01 to 9000.
5073 makeup
5075 Set amount of amplification of signal after processing. Default is
5076 1. Allowed range is from 1 to 64.
5077 knee
5079 Curve the sharp knee around the threshold to enter gain reduction
5080 more softly. Default is 2.828427125. Allowed range is from 1 to 8.
5081 detection
5083 Choose if exact signal should be taken for detection or an RMS like
5084 one. Default is rms. Can be peak or rms.
5085 link
5087 Choose if the average level between all channels or the louder
5088 channel affects the reduction. Default is average. Can be average
5089 or maximum.
5090 level_sc
5092 Set sidechain gain. Default is 1. Range is from 0.015625 to 64.
5093 Commands
5095 This filter supports the all above options as commands.
5097 silencedetect
5099 Detect silence in an audio stream.
5100 This filter logs a message when it detects that the input audio volume
5102 is less or equal to a noise tolerance value for a duration greater or
5103 equal to the minimum detected noise duration.
5104 The printed times and duration are expressed in seconds. The
5106 "lavfi.silence_start" or "lavfi.silence_start.X" metadata key is set on
5107 the first frame whose timestamp equals or exceeds the detection
5108 duration and it contains the timestamp of the first frame of the
5109 silence.
5110 The "lavfi.silence_duration" or "lavfi.silence_duration.X" and
5112 "lavfi.silence_end" or "lavfi.silence_end.X" metadata keys are set on
5113 the first frame after the silence. If mono is enabled, and each channel
5114 is evaluated separately, the ".X" suffixed keys are used, and "X"
5115 corresponds to the channel number.
5116 The filter accepts the following options:
5118 noise, n
5120 Set noise tolerance. Can be specified in dB (in case "dB" is
5121 appended to the specified value) or amplitude ratio. Default is
5122 -60dB, or 0.001.
5123 duration, d
5125 Set silence duration until notification (default is 2 seconds). See
5126 the Time duration section in the ffmpeg-utils(1) manual for the
5127 accepted syntax.
5128 mono, m
5130 Process each channel separately, instead of combined. By default is
5131 disabled.
5132 Examples
5134 • Detect 5 seconds of silence with -50dB noise tolerance:
5136 silencedetect=n=-50dB:d=5
5138 • Complete example with ffmpeg to detect silence with 0.0001 noise
5140 tolerance in silence.mp3:
5141 ffmpeg -i silence.mp3 -af silencedetect=noise=0.0001 -f null -
5143 silenceremove
5145 Remove silence from the beginning, middle or end of the audio.
5146 The filter accepts the following options:
5148 start_periods
5150 This value is used to indicate if audio should be trimmed at
5151 beginning of the audio. A value of zero indicates no silence should
5152 be trimmed from the beginning. When specifying a non-zero value, it
5153 trims audio up until it finds non-silence. Normally, when trimming
5154 silence from beginning of audio the start_periods will be 1 but it
5155 can be increased to higher values to trim all audio up to specific
5156 count of non-silence periods. Default value is 0.
5157 start_duration
5159 Specify the amount of time that non-silence must be detected before
5160 it stops trimming audio. By increasing the duration, bursts of
5161 noises can be treated as silence and trimmed off. Default value is
5162 0.
5163 start_threshold
5165 This indicates what sample value should be treated as silence. For
5166 digital audio, a value of 0 may be fine but for audio recorded from
5167 analog, you may wish to increase the value to account for
5168 background noise. Can be specified in dB (in case "dB" is appended
5169 to the specified value) or amplitude ratio. Default value is 0.
5170 start_silence
5172 Specify max duration of silence at beginning that will be kept
5173 after trimming. Default is 0, which is equal to trimming all
5174 samples detected as silence.
5175 start_mode
5177 Specify mode of detection of silence end in start of multi-channel
5178 audio. Can be any or all. Default is any. With any, any sample
5179 that is detected as non-silence will cause stopped trimming of
5180 silence. With all, only if all channels are detected as non-
5181 silence will cause stopped trimming of silence.
5182 stop_periods
5184 Set the count for trimming silence from the end of audio. To
5185 remove silence from the middle of a file, specify a stop_periods
5186 that is negative. This value is then treated as a positive value
5187 and is used to indicate the effect should restart processing as
5188 specified by start_periods, making it suitable for removing periods
5189 of silence in the middle of the audio. Default value is 0.
5190 stop_duration
5192 Specify a duration of silence that must exist before audio is not
5193 copied any more. By specifying a higher duration, silence that is
5194 wanted can be left in the audio. Default value is 0.
5195 stop_threshold
5197 This is the same as start_threshold but for trimming silence from
5198 the end of audio. Can be specified in dB (in case "dB" is appended
5199 to the specified value) or amplitude ratio. Default value is 0.
5200 stop_silence
5202 Specify max duration of silence at end that will be kept after
5203 trimming. Default is 0, which is equal to trimming all samples
5204 detected as silence.
5205 stop_mode
5207 Specify mode of detection of silence start in end of multi-channel
5208 audio. Can be any or all. Default is any. With any, any sample
5209 that is detected as non-silence will cause stopped trimming of
5210 silence. With all, only if all channels are detected as non-
5211 silence will cause stopped trimming of silence.
5212 detection
5214 Set how is silence detected. Can be "rms" or "peak". Second is
5215 faster and works better with digital silence which is exactly 0.
5216 Default value is "rms".
5217 window
5219 Set duration in number of seconds used to calculate size of window
5220 in number of samples for detecting silence. Default value is 0.02.
5221 Allowed range is from 0 to 10.
5222 Examples
5224 • The following example shows how this filter can be used to start a
5226 recording that does not contain the delay at the start which
5227 usually occurs between pressing the record button and the start of
5228 the performance:
5229 silenceremove=start_periods=1:start_duration=5:start_threshold=0.02
5231 • Trim all silence encountered from beginning to end where there is
5233 more than 1 second of silence in audio:
5234 silenceremove=stop_periods=-1:stop_duration=1:stop_threshold=-90dB
5236 • Trim all digital silence samples, using peak detection, from
5238 beginning to end where there is more than 0 samples of digital
5239 silence in audio and digital silence is detected in all channels at
5240 same positions in stream:
5241 silenceremove=window=0:detection=peak:stop_mode=all:start_mode=all:stop_periods=-1:stop_threshold=0
5243 sofalizer
5245 SOFAlizer uses head-related transfer functions (HRTFs) to create
5246 virtual loudspeakers around the user for binaural listening via
5247 headphones (audio formats up to 9 channels supported). The HRTFs are
5248 stored in SOFA files (see <http://www.sofacoustics.org/> for a
5249 database). SOFAlizer is developed at the Acoustics Research Institute
5250 (ARI) of the Austrian Academy of Sciences.
5251 To enable compilation of this filter you need to configure FFmpeg with
5253 "--enable-libmysofa".
5254 The filter accepts the following options:
5256 sofa
5258 Set the SOFA file used for rendering.
5259 gain
5261 Set gain applied to audio. Value is in dB. Default is 0.
5262 rotation
5264 Set rotation of virtual loudspeakers in deg. Default is 0.
5265 elevation
5267 Set elevation of virtual speakers in deg. Default is 0.
5268 radius
5270 Set distance in meters between loudspeakers and the listener with
5271 near-field HRTFs. Default is 1.
5272 type
5274 Set processing type. Can be time or freq. time is processing audio
5275 in time domain which is slow. freq is processing audio in
5276 frequency domain which is fast. Default is freq.
5277 speakers
5279 Set custom positions of virtual loudspeakers. Syntax for this
5280 option is: <CH> <AZIM> <ELEV>[|<CH> <AZIM> <ELEV>|...]. Each
5281 virtual loudspeaker is described with short channel name following
5282 with azimuth and elevation in degrees. Each virtual loudspeaker
5283 description is separated by '|'. For example to override front
5284 left and front right channel positions use: 'speakers=FL 45 15|FR
5285 345 15'. Descriptions with unrecognised channel names are ignored.
5286 lfegain
5288 Set custom gain for LFE channels. Value is in dB. Default is 0.
5289 framesize
5291 Set custom frame size in number of samples. Default is 1024.
5292 Allowed range is from 1024 to 96000. Only used if option type is
5293 set to freq.
5294 normalize
5296 Should all IRs be normalized upon importing SOFA file. By default
5297 is enabled.
5298 interpolate
5300 Should nearest IRs be interpolated with neighbor IRs if exact
5301 position does not match. By default is disabled.
5302 minphase
5304 Minphase all IRs upon loading of SOFA file. By default is disabled.
5305 anglestep
5307 Set neighbor search angle step. Only used if option interpolate is
5308 enabled.
5309 radstep
5311 Set neighbor search radius step. Only used if option interpolate is
5312 enabled.
5313 Examples
5315 • Using ClubFritz6 sofa file:
5317 sofalizer=sofa=/path/to/ClubFritz6.sofa:type=freq:radius=1
5319 • Using ClubFritz12 sofa file and bigger radius with small rotation:
5321 sofalizer=sofa=/path/to/ClubFritz12.sofa:type=freq:radius=2:rotation=5
5323 • Similar as above but with custom speaker positions for front left,
5325 front right, back left and back right and also with custom gain:
5326 "sofalizer=sofa=/path/to/ClubFritz6.sofa:type=freq:radius=2:speakers=FL 45|FR 315|BL 135|BR 225:gain=28"
5328 speechnorm
5330 Speech Normalizer.
5331 This filter expands or compresses each half-cycle of audio samples
5333 (local set of samples all above or all below zero and between two
5334 nearest zero crossings) depending on threshold value, so audio reaches
5335 target peak value under conditions controlled by below options.
5336 The filter accepts the following options:
5338 peak, p
5340 Set the expansion target peak value. This specifies the highest
5341 allowed absolute amplitude level for the normalized audio input.
5342 Default value is 0.95. Allowed range is from 0.0 to 1.0.
5343 expansion, e
5345 Set the maximum expansion factor. Allowed range is from 1.0 to
5346 50.0. Default value is 2.0. This option controls maximum local
5347 half-cycle of samples expansion. The maximum expansion would be
5348 such that local peak value reaches target peak value but never to
5349 surpass it and that ratio between new and previous peak value does
5350 not surpass this option value.
5351 compression, c
5353 Set the maximum compression factor. Allowed range is from 1.0 to
5354 50.0. Default value is 2.0. This option controls maximum local
5355 half-cycle of samples compression. This option is used only if
5356 threshold option is set to value greater than 0.0, then in such
5357 cases when local peak is lower or same as value set by threshold
5358 all samples belonging to that peak's half-cycle will be compressed
5359 by current compression factor.
5360 threshold, t
5362 Set the threshold value. Default value is 0.0. Allowed range is
5363 from 0.0 to 1.0. This option specifies which half-cycles of
5364 samples will be compressed and which will be expanded. Any half-
5365 cycle samples with their local peak value below or same as this
5366 option value will be compressed by current compression factor,
5367 otherwise, if greater than threshold value they will be expanded
5368 with expansion factor so that it could reach peak target value but
5369 never surpass it.
5370 raise, r
5372 Set the expansion raising amount per each half-cycle of samples.
5373 Default value is 0.001. Allowed range is from 0.0 to 1.0. This
5374 controls how fast expansion factor is raised per each new half-
5375 cycle until it reaches expansion value. Setting this options too
5376 high may lead to distortions.
5377 fall, f
5379 Set the compression raising amount per each half-cycle of samples.
5380 Default value is 0.001. Allowed range is from 0.0 to 1.0. This
5381 controls how fast compression factor is raised per each new half-
5382 cycle until it reaches compression value.
5383 channels, h
5385 Specify which channels to filter, by default all available channels
5386 are filtered.
5387 invert, i
5389 Enable inverted filtering, by default is disabled. This inverts
5390 interpretation of threshold option. When enabled any half-cycle of
5391 samples with their local peak value below or same as threshold
5392 option will be expanded otherwise it will be compressed.
5393 link, l
5395 Link channels when calculating gain applied to each filtered
5396 channel sample, by default is disabled. When disabled each
5397 filtered channel gain calculation is independent, otherwise when
5398 this option is enabled the minimum of all possible gains for each
5399 filtered channel is used.
5400 Commands
5402 This filter supports the all above options as commands.
5404 stereotools
5406 This filter has some handy utilities to manage stereo signals, for
5407 converting M/S stereo recordings to L/R signal while having control
5408 over the parameters or spreading the stereo image of master track.
5409 The filter accepts the following options:
5411 level_in
5413 Set input level before filtering for both channels. Defaults is 1.
5414 Allowed range is from 0.015625 to 64.
5415 level_out
5417 Set output level after filtering for both channels. Defaults is 1.
5418 Allowed range is from 0.015625 to 64.
5419 balance_in
5421 Set input balance between both channels. Default is 0. Allowed
5422 range is from -1 to 1.
5423 balance_out
5425 Set output balance between both channels. Default is 0. Allowed
5426 range is from -1 to 1.
5427 softclip
5429 Enable softclipping. Results in analog distortion instead of harsh
5430 digital 0dB clipping. Disabled by default.
5431 mutel
5433 Mute the left channel. Disabled by default.
5434 muter
5436 Mute the right channel. Disabled by default.
5437 phasel
5439 Change the phase of the left channel. Disabled by default.
5440 phaser
5442 Change the phase of the right channel. Disabled by default.
5443 mode
5445 Set stereo mode. Available values are:
5446 lr>lr
5448 Left/Right to Left/Right, this is default.
5449 lr>ms
5451 Left/Right to Mid/Side.
5452 ms>lr
5454 Mid/Side to Left/Right.
5455 lr>ll
5457 Left/Right to Left/Left.
5458 lr>rr
5460 Left/Right to Right/Right.
5461 lr>l+r
5463 Left/Right to Left + Right.
5464 lr>rl
5466 Left/Right to Right/Left.
5467 ms>ll
5469 Mid/Side to Left/Left.
5470 ms>rr
5472 Mid/Side to Right/Right.
5473 ms>rl
5475 Mid/Side to Right/Left.
5476 lr>l-r
5478 Left/Right to Left - Right.
5479 slev
5481 Set level of side signal. Default is 1. Allowed range is from
5482 0.015625 to 64.
5483 sbal
5485 Set balance of side signal. Default is 0. Allowed range is from -1
5486 to 1.
5487 mlev
5489 Set level of the middle signal. Default is 1. Allowed range is
5490 from 0.015625 to 64.
5491 mpan
5493 Set middle signal pan. Default is 0. Allowed range is from -1 to 1.
5494 base
5496 Set stereo base between mono and inversed channels. Default is 0.
5497 Allowed range is from -1 to 1.
5498 delay
5500 Set delay in milliseconds how much to delay left from right channel
5501 and vice versa. Default is 0. Allowed range is from -20 to 20.
5502 sclevel
5504 Set S/C level. Default is 1. Allowed range is from 1 to 100.
5505 phase
5507 Set the stereo phase in degrees. Default is 0. Allowed range is
5508 from 0 to 360.
5509 bmode_in, bmode_out
5511 Set balance mode for balance_in/balance_out option.
5512 Can be one of the following:
5514 balance
5516 Classic balance mode. Attenuate one channel at time. Gain is
5517 raised up to 1.
5518 amplitude
5520 Similar as classic mode above but gain is raised up to 2.
5521 power
5523 Equal power distribution, from -6dB to +6dB range.
5524 Commands
5526 This filter supports the all above options as commands.
5528 Examples
5530 • Apply karaoke like effect:
5532 stereotools=mlev=0.015625
5534 • Convert M/S signal to L/R:
5536 "stereotools=mode=ms>lr"
5538 stereowiden
5540 This filter enhance the stereo effect by suppressing signal common to
5541 both channels and by delaying the signal of left into right and vice
5542 versa, thereby widening the stereo effect.
5543 The filter accepts the following options:
5545 delay
5547 Time in milliseconds of the delay of left signal into right and
5548 vice versa. Default is 20 milliseconds.
5549 feedback
5551 Amount of gain in delayed signal into right and vice versa. Gives a
5552 delay effect of left signal in right output and vice versa which
5553 gives widening effect. Default is 0.3.
5554 crossfeed
5556 Cross feed of left into right with inverted phase. This helps in
5557 suppressing the mono. If the value is 1 it will cancel all the
5558 signal common to both channels. Default is 0.3.
5559 drymix
5561 Set level of input signal of original channel. Default is 0.8.
5562 Commands
5564 This filter supports the all above options except "delay" as commands.
5566 superequalizer
5568 Apply 18 band equalizer.
5569 The filter accepts the following options:
5571 1b Set 65Hz band gain.
5573 2b Set 92Hz band gain.
5575 3b Set 131Hz band gain.
5577 4b Set 185Hz band gain.
5579 5b Set 262Hz band gain.
5581 6b Set 370Hz band gain.
5583 7b Set 523Hz band gain.
5585 8b Set 740Hz band gain.
5587 9b Set 1047Hz band gain.
5589 10b Set 1480Hz band gain.
5591 11b Set 2093Hz band gain.
5593 12b Set 2960Hz band gain.
5595 13b Set 4186Hz band gain.
5597 14b Set 5920Hz band gain.
5599 15b Set 8372Hz band gain.
5601 16b Set 11840Hz band gain.
5603 17b Set 16744Hz band gain.
5605 18b Set 20000Hz band gain.
5607 surround
5609 Apply audio surround upmix filter.
5610 This filter allows to produce multichannel output from audio stream.
5612 The filter accepts the following options:
5614 chl_out
5616 Set output channel layout. By default, this is 5.1.
5617 See the Channel Layout section in the ffmpeg-utils(1) manual for
5619 the required syntax.
5620 chl_in
5622 Set input channel layout. By default, this is stereo.
5623 See the Channel Layout section in the ffmpeg-utils(1) manual for
5625 the required syntax.
5626 level_in
5628 Set input volume level. By default, this is 1.
5629 level_out
5631 Set output volume level. By default, this is 1.
5632 lfe Enable LFE channel output if output channel layout has it. By
5634 default, this is enabled.
5635 lfe_low
5637 Set LFE low cut off frequency. By default, this is 128 Hz.
5638 lfe_high
5640 Set LFE high cut off frequency. By default, this is 256 Hz.
5641 lfe_mode
5643 Set LFE mode, can be add or sub. Default is add. In add mode, LFE
5644 channel is created from input audio and added to output. In sub
5645 mode, LFE channel is created from input audio and added to output
5646 but also all non-LFE output channels are subtracted with output LFE
5647 channel.
5648 angle
5650 Set angle of stereo surround transform, Allowed range is from 0 to
5651 360. Default is 90.
5652 fc_in
5654 Set front center input volume. By default, this is 1.
5655 fc_out
5657 Set front center output volume. By default, this is 1.
5658 fl_in
5660 Set front left input volume. By default, this is 1.
5661 fl_out
5663 Set front left output volume. By default, this is 1.
5664 fr_in
5666 Set front right input volume. By default, this is 1.
5667 fr_out
5669 Set front right output volume. By default, this is 1.
5670 sl_in
5672 Set side left input volume. By default, this is 1.
5673 sl_out
5675 Set side left output volume. By default, this is 1.
5676 sr_in
5678 Set side right input volume. By default, this is 1.
5679 sr_out
5681 Set side right output volume. By default, this is 1.
5682 bl_in
5684 Set back left input volume. By default, this is 1.
5685 bl_out
5687 Set back left output volume. By default, this is 1.
5688 br_in
5690 Set back right input volume. By default, this is 1.
5691 br_out
5693 Set back right output volume. By default, this is 1.
5694 bc_in
5696 Set back center input volume. By default, this is 1.
5697 bc_out
5699 Set back center output volume. By default, this is 1.
5700 lfe_in
5702 Set LFE input volume. By default, this is 1.
5703 lfe_out
5705 Set LFE output volume. By default, this is 1.
5706 allx
5708 Set spread usage of stereo image across X axis for all channels.
5709 ally
5711 Set spread usage of stereo image across Y axis for all channels.
5712 fcx, flx, frx, blx, brx, slx, srx, bcx
5714 Set spread usage of stereo image across X axis for each channel.
5715 fcy, fly, fry, bly, bry, sly, sry, bcy
5717 Set spread usage of stereo image across Y axis for each channel.
5718 win_size
5720 Set window size. Allowed range is from 1024 to 65536. Default size
5721 is 4096.
5722 win_func
5724 Set window function.
5725 It accepts the following values:
5727 rect
5729 bartlett
5730 hann, hanning
5731 hamming
5732 blackman
5733 welch
5734 flattop
5735 bharris
5736 bnuttall
5737 bhann
5738 sine
5739 nuttall
5740 lanczos
5741 gauss
5742 tukey
5743 dolph
5744 cauchy
5745 parzen
5746 poisson
5747 bohman
5748 Default is "hann".
5750 overlap
5752 Set window overlap. If set to 1, the recommended overlap for
5753 selected window function will be picked. Default is 0.5.
5754 treble, highshelf
5756 Boost or cut treble (upper) frequencies of the audio using a two-pole
5757 shelving filter with a response similar to that of a standard hi-fi's
5758 tone-controls. This is also known as shelving equalisation (EQ).
5759 The filter accepts the following options:
5761 gain, g
5763 Give the gain at whichever is the lower of ~22 kHz and the Nyquist
5764 frequency. Its useful range is about -20 (for a large cut) to +20
5765 (for a large boost). Beware of clipping when using a positive gain.
5766 frequency, f
5768 Set the filter's central frequency and so can be used to extend or
5769 reduce the frequency range to be boosted or cut. The default value
5770 is 3000 Hz.
5771 width_type, t
5773 Set method to specify band-width of filter.
5774 h Hz
5776 q Q-Factor
5778 o octave
5780 s slope
5782 k kHz
5784 width, w
5786 Determine how steep is the filter's shelf transition.
5787 poles, p
5789 Set number of poles. Default is 2.
5790 mix, m
5792 How much to use filtered signal in output. Default is 1. Range is
5793 between 0 and 1.
5794 channels, c
5796 Specify which channels to filter, by default all available are
5797 filtered.
5798 normalize, n
5800 Normalize biquad coefficients, by default is disabled. Enabling it
5801 will normalize magnitude response at DC to 0dB.
5802 transform, a
5804 Set transform type of IIR filter.
5805 di
5807 dii
5808 tdii
5809 latt
5810 svf
5811 precision, r
5812 Set precison of filtering.
5813 auto
5815 Pick automatic sample format depending on surround filters.
5816 s16 Always use signed 16-bit.
5818 s32 Always use signed 32-bit.
5820 f32 Always use float 32-bit.
5822 f64 Always use float 64-bit.
5824 Commands
5826 This filter supports the following commands:
5828 frequency, f
5830 Change treble frequency. Syntax for the command is : "frequency"
5831 width_type, t
5833 Change treble width_type. Syntax for the command is : "width_type"
5834 width, w
5836 Change treble width. Syntax for the command is : "width"
5837 gain, g
5839 Change treble gain. Syntax for the command is : "gain"
5840 mix, m
5842 Change treble mix. Syntax for the command is : "mix"
5843 tremolo
5845 Sinusoidal amplitude modulation.
5846 The filter accepts the following options:
5848 f Modulation frequency in Hertz. Modulation frequencies in the
5850 subharmonic range (20 Hz or lower) will result in a tremolo effect.
5851 This filter may also be used as a ring modulator by specifying a
5852 modulation frequency higher than 20 Hz. Range is 0.1 - 20000.0.
5853 Default value is 5.0 Hz.
5854 d Depth of modulation as a percentage. Range is 0.0 - 1.0. Default
5856 value is 0.5.
5857 vibrato
5859 Sinusoidal phase modulation.
5860 The filter accepts the following options:
5862 f Modulation frequency in Hertz. Range is 0.1 - 20000.0. Default
5864 value is 5.0 Hz.
5865 d Depth of modulation as a percentage. Range is 0.0 - 1.0. Default
5867 value is 0.5.
5868 volume
5870 Adjust the input audio volume.
5871 It accepts the following parameters:
5873 volume
5875 Set audio volume expression.
5876 Output values are clipped to the maximum value.
5878 The output audio volume is given by the relation:
5880 <output_volume> = <volume> * <input_volume>
5882 The default value for volume is "1.0".
5884 precision
5886 This parameter represents the mathematical precision.
5887 It determines which input sample formats will be allowed, which
5889 affects the precision of the volume scaling.
5890 fixed
5892 8-bit fixed-point; this limits input sample format to U8, S16,
5893 and S32.
5894 float
5896 32-bit floating-point; this limits input sample format to FLT.
5897 (default)
5898 double
5900 64-bit floating-point; this limits input sample format to DBL.
5901 replaygain
5903 Choose the behaviour on encountering ReplayGain side data in input
5904 frames.
5905 drop
5907 Remove ReplayGain side data, ignoring its contents (the
5908 default).
5909 ignore
5911 Ignore ReplayGain side data, but leave it in the frame.
5912 track
5914 Prefer the track gain, if present.
5915 album
5917 Prefer the album gain, if present.
5918 replaygain_preamp
5920 Pre-amplification gain in dB to apply to the selected replaygain
5921 gain.
5922 Default value for replaygain_preamp is 0.0.
5924 replaygain_noclip
5926 Prevent clipping by limiting the gain applied.
5927 Default value for replaygain_noclip is 1.
5929 eval
5931 Set when the volume expression is evaluated.
5932 It accepts the following values:
5934 once
5936 only evaluate expression once during the filter initialization,
5937 or when the volume command is sent
5938 frame
5940 evaluate expression for each incoming frame
5941 Default value is once.
5943 The volume expression can contain the following parameters.
5945 n frame number (starting at zero)
5947 nb_channels
5949 number of channels
5950 nb_consumed_samples
5952 number of samples consumed by the filter
5953 nb_samples
5955 number of samples in the current frame
5956 pos original frame position in the file
5958 pts frame PTS
5960 sample_rate
5962 sample rate
5963 startpts
5965 PTS at start of stream
5966 startt
5968 time at start of stream
5969 t frame time
5971 tb timestamp timebase
5973 volume
5975 last set volume value
5976 Note that when eval is set to once only the sample_rate and tb
5978 variables are available, all other variables will evaluate to NAN.
5979 Commands
5981 This filter supports the following commands:
5983 volume
5985 Modify the volume expression. The command accepts the same syntax
5986 of the corresponding option.
5987 If the specified expression is not valid, it is kept at its current
5989 value.
5990 Examples
5992 • Halve the input audio volume:
5994 volume=volume=0.5
5996 volume=volume=1/2
5997 volume=volume=-6.0206dB
5998 In all the above example the named key for volume can be omitted,
6000 for example like in:
6001 volume=0.5
6003 • Increase input audio power by 6 decibels using fixed-point
6005 precision:
6006 volume=volume=6dB:precision=fixed
6008 • Fade volume after time 10 with an annihilation period of 5 seconds:
6010 volume='if(lt(t,10),1,max(1-(t-10)/5,0))':eval=frame
6012 volumedetect
6014 Detect the volume of the input video.
6015 The filter has no parameters. The input is not modified. Statistics
6017 about the volume will be printed in the log when the input stream end
6018 is reached.
6019 In particular it will show the mean volume (root mean square), maximum
6021 volume (on a per-sample basis), and the beginning of a histogram of the
6022 registered volume values (from the maximum value to a cumulated 1/1000
6023 of the samples).
6024 All volumes are in decibels relative to the maximum PCM value.
6026 Examples
6028 Here is an excerpt of the output:
6030 [Parsed_volumedetect_0 0xa23120] mean_volume: -27 dB
6032 [Parsed_volumedetect_0 0xa23120] max_volume: -4 dB
6033 [Parsed_volumedetect_0 0xa23120] histogram_4db: 6
6034 [Parsed_volumedetect_0 0xa23120] histogram_5db: 62
6035 [Parsed_volumedetect_0 0xa23120] histogram_6db: 286
6036 [Parsed_volumedetect_0 0xa23120] histogram_7db: 1042
6037 [Parsed_volumedetect_0 0xa23120] histogram_8db: 2551
6038 [Parsed_volumedetect_0 0xa23120] histogram_9db: 4609
6039 [Parsed_volumedetect_0 0xa23120] histogram_10db: 8409
6040 It means that:
6042 • The mean square energy is approximately -27 dB, or 10^-2.7.
6044 • The largest sample is at -4 dB, or more precisely between -4 dB and
6046 -5 dB.
6047 • There are 6 samples at -4 dB, 62 at -5 dB, 286 at -6 dB, etc.
6049 In other words, raising the volume by +4 dB does not cause any
6051 clipping, raising it by +5 dB causes clipping for 6 samples, etc.
6052
6054 Below is a description of the currently available audio sources.
6055 abuffer
6057 Buffer audio frames, and make them available to the filter chain.
6058 This source is mainly intended for a programmatic use, in particular
6060 through the interface defined in libavfilter/buffersrc.h.
6061 It accepts the following parameters:
6063 time_base
6065 The timebase which will be used for timestamps of submitted frames.
6066 It must be either a floating-point number or in
6067 numerator/denominator form.
6068 sample_rate
6070 The sample rate of the incoming audio buffers.
6071 sample_fmt
6073 The sample format of the incoming audio buffers. Either a sample
6074 format name or its corresponding integer representation from the
6075 enum AVSampleFormat in libavutil/samplefmt.h
6076 channel_layout
6078 The channel layout of the incoming audio buffers. Either a channel
6079 layout name from channel_layout_map in libavutil/channel_layout.c
6080 or its corresponding integer representation from the AV_CH_LAYOUT_*
6081 macros in libavutil/channel_layout.h
6082 channels
6084 The number of channels of the incoming audio buffers. If both
6085 channels and channel_layout are specified, then they must be
6086 consistent.
6087 Examples
6089 abuffer=sample_rate=44100:sample_fmt=s16p:channel_layout=stereo
6091 will instruct the source to accept planar 16bit signed stereo at
6093 44100Hz. Since the sample format with name "s16p" corresponds to the
6094 number 6 and the "stereo" channel layout corresponds to the value 0x3,
6095 this is equivalent to:
6096 abuffer=sample_rate=44100:sample_fmt=6:channel_layout=0x3
6098 aevalsrc
6100 Generate an audio signal specified by an expression.
6101 This source accepts in input one or more expressions (one for each
6103 channel), which are evaluated and used to generate a corresponding
6104 audio signal.
6105 This source accepts the following options:
6107 exprs
6109 Set the '|'-separated expressions list for each separate channel.
6110 In case the channel_layout option is not specified, the selected
6111 channel layout depends on the number of provided expressions.
6112 Otherwise the last specified expression is applied to the remaining
6113 output channels.
6114 channel_layout, c
6116 Set the channel layout. The number of channels in the specified
6117 layout must be equal to the number of specified expressions.
6118 duration, d
6120 Set the minimum duration of the sourced audio. See the Time
6121 duration section in the ffmpeg-utils(1) manual for the accepted
6122 syntax. Note that the resulting duration may be greater than the
6123 specified duration, as the generated audio is always cut at the end
6124 of a complete frame.
6125 If not specified, or the expressed duration is negative, the audio
6127 is supposed to be generated forever.
6128 nb_samples, n
6130 Set the number of samples per channel per each output frame,
6131 default to 1024.
6132 sample_rate, s
6134 Specify the sample rate, default to 44100.
6135 Each expression in exprs can contain the following constants:
6137 n number of the evaluated sample, starting from 0
6139 t time of the evaluated sample expressed in seconds, starting from 0
6141 s sample rate
6143 Examples
6145 • Generate silence:
6147 aevalsrc=0
6149 • Generate a sin signal with frequency of 440 Hz, set sample rate to
6151 8000 Hz:
6152 aevalsrc="sin(440*2*PI*t):s=8000"
6154 • Generate a two channels signal, specify the channel layout (Front
6156 Center + Back Center) explicitly:
6157 aevalsrc="sin(420*2*PI*t)|cos(430*2*PI*t):c=FC|BC"
6159 • Generate white noise:
6161 aevalsrc="-2+random(0)"
6163 • Generate an amplitude modulated signal:
6165 aevalsrc="sin(10*2*PI*t)*sin(880*2*PI*t)"
6167 • Generate 2.5 Hz binaural beats on a 360 Hz carrier:
6169 aevalsrc="0.1*sin(2*PI*(360-2.5/2)*t) | 0.1*sin(2*PI*(360+2.5/2)*t)"
6171 afirsrc
6173 Generate a FIR coefficients using frequency sampling method.
6174 The resulting stream can be used with afir filter for filtering the
6176 audio signal.
6177 The filter accepts the following options:
6179 taps, t
6181 Set number of filter coefficents in output audio stream. Default
6182 value is 1025.
6183 frequency, f
6185 Set frequency points from where magnitude and phase are set. This
6186 must be in non decreasing order, and first element must be 0, while
6187 last element must be 1. Elements are separated by white spaces.
6188 magnitude, m
6190 Set magnitude value for every frequency point set by frequency.
6191 Number of values must be same as number of frequency points.
6192 Values are separated by white spaces.
6193 phase, p
6195 Set phase value for every frequency point set by frequency. Number
6196 of values must be same as number of frequency points. Values are
6197 separated by white spaces.
6198 sample_rate, r
6200 Set sample rate, default is 44100.
6201 nb_samples, n
6203 Set number of samples per each frame. Default is 1024.
6204 win_func, w
6206 Set window function. Default is blackman.
6207 anullsrc
6209 The null audio source, return unprocessed audio frames. It is mainly
6210 useful as a template and to be employed in analysis / debugging tools,
6211 or as the source for filters which ignore the input data (for example
6212 the sox synth filter).
6213 This source accepts the following options:
6215 channel_layout, cl
6217 Specifies the channel layout, and can be either an integer or a
6218 string representing a channel layout. The default value of
6219 channel_layout is "stereo".
6220 Check the channel_layout_map definition in
6222 libavutil/channel_layout.c for the mapping between strings and
6223 channel layout values.
6224 sample_rate, r
6226 Specifies the sample rate, and defaults to 44100.
6227 nb_samples, n
6229 Set the number of samples per requested frames.
6230 duration, d
6232 Set the duration of the sourced audio. See the Time duration
6233 section in the ffmpeg-utils(1) manual for the accepted syntax.
6234 If not specified, or the expressed duration is negative, the audio
6236 is supposed to be generated forever.
6237 Examples
6239 • Set the sample rate to 48000 Hz and the channel layout to
6241 AV_CH_LAYOUT_MONO.
6242 anullsrc=r=48000:cl=4
6244 • Do the same operation with a more obvious syntax:
6246 anullsrc=r=48000:cl=mono
6248 All the parameters need to be explicitly defined.
6250 flite
6252 Synthesize a voice utterance using the libflite library.
6253 To enable compilation of this filter you need to configure FFmpeg with
6255 "--enable-libflite".
6256 Note that versions of the flite library prior to 2.0 are not thread-
6258 safe.
6259 The filter accepts the following options:
6261 list_voices
6263 If set to 1, list the names of the available voices and exit
6264 immediately. Default value is 0.
6265 nb_samples, n
6267 Set the maximum number of samples per frame. Default value is 512.
6268 textfile
6270 Set the filename containing the text to speak.
6271 text
6273 Set the text to speak.
6274 voice, v
6276 Set the voice to use for the speech synthesis. Default value is
6277 "kal". See also the list_voices option.
6278 Examples
6280 • Read from file speech.txt, and synthesize the text using the
6282 standard flite voice:
6283 flite=textfile=speech.txt
6285 • Read the specified text selecting the "slt" voice:
6287 flite=text='So fare thee well, poor devil of a Sub-Sub, whose commentator I am':voice=slt
6289 • Input text to ffmpeg:
6291 ffmpeg -f lavfi -i flite=text='So fare thee well, poor devil of a Sub-Sub, whose commentator I am':voice=slt
6293 • Make ffplay speak the specified text, using "flite" and the "lavfi"
6295 device:
6296 ffplay -f lavfi flite=text='No more be grieved for which that thou hast done.'
6298 For more information about libflite, check:
6300 <http://www.festvox.org/flite/>
6301 anoisesrc
6303 Generate a noise audio signal.
6304 The filter accepts the following options:
6306 sample_rate, r
6308 Specify the sample rate. Default value is 48000 Hz.
6309 amplitude, a
6311 Specify the amplitude (0.0 - 1.0) of the generated audio stream.
6312 Default value is 1.0.
6313 duration, d
6315 Specify the duration of the generated audio stream. Not specifying
6316 this option results in noise with an infinite length.
6317 color, colour, c
6319 Specify the color of noise. Available noise colors are white, pink,
6320 brown, blue, violet and velvet. Default color is white.
6321 seed, s
6323 Specify a value used to seed the PRNG.
6324 nb_samples, n
6326 Set the number of samples per each output frame, default is 1024.
6327 Examples
6329 • Generate 60 seconds of pink noise, with a 44.1 kHz sampling rate
6331 and an amplitude of 0.5:
6332 anoisesrc=d=60:c=pink:r=44100:a=0.5
6334 hilbert
6336 Generate odd-tap Hilbert transform FIR coefficients.
6337 The resulting stream can be used with afir filter for phase-shifting
6339 the signal by 90 degrees.
6340 This is used in many matrix coding schemes and for analytic signal
6342 generation. The process is often written as a multiplication by i (or
6343 j), the imaginary unit.
6344 The filter accepts the following options:
6346 sample_rate, s
6348 Set sample rate, default is 44100.
6349 taps, t
6351 Set length of FIR filter, default is 22051.
6352 nb_samples, n
6354 Set number of samples per each frame.
6355 win_func, w
6357 Set window function to be used when generating FIR coefficients.
6358 sinc
6360 Generate a sinc kaiser-windowed low-pass, high-pass, band-pass, or
6361 band-reject FIR coefficients.
6362 The resulting stream can be used with afir filter for filtering the
6364 audio signal.
6365 The filter accepts the following options:
6367 sample_rate, r
6369 Set sample rate, default is 44100.
6370 nb_samples, n
6372 Set number of samples per each frame. Default is 1024.
6373 hp Set high-pass frequency. Default is 0.
6375 lp Set low-pass frequency. Default is 0. If high-pass frequency is
6377 lower than low-pass frequency and low-pass frequency is higher than
6378 0 then filter will create band-pass filter coefficients, otherwise
6379 band-reject filter coefficients.
6380 phase
6382 Set filter phase response. Default is 50. Allowed range is from 0
6383 to 100.
6384 beta
6386 Set Kaiser window beta.
6387 att Set stop-band attenuation. Default is 120dB, allowed range is from
6389 40 to 180 dB.
6390 round
6392 Enable rounding, by default is disabled.
6393 hptaps
6395 Set number of taps for high-pass filter.
6396 lptaps
6398 Set number of taps for low-pass filter.
6399 sine
6401 Generate an audio signal made of a sine wave with amplitude 1/8.
6402 The audio signal is bit-exact.
6404 The filter accepts the following options:
6406 frequency, f
6408 Set the carrier frequency. Default is 440 Hz.
6409 beep_factor, b
6411 Enable a periodic beep every second with frequency beep_factor
6412 times the carrier frequency. Default is 0, meaning the beep is
6413 disabled.
6414 sample_rate, r
6416 Specify the sample rate, default is 44100.
6417 duration, d
6419 Specify the duration of the generated audio stream.
6420 samples_per_frame
6422 Set the number of samples per output frame.
6423 The expression can contain the following constants:
6425 n The (sequential) number of the output audio frame, starting
6427 from 0.
6428 pts The PTS (Presentation TimeStamp) of the output audio frame,
6430 expressed in TB units.
6431 t The PTS of the output audio frame, expressed in seconds.
6433 TB The timebase of the output audio frames.
6435 Default is 1024.
6437 Examples
6439 • Generate a simple 440 Hz sine wave:
6441 sine
6443 • Generate a 220 Hz sine wave with a 880 Hz beep each second, for 5
6445 seconds:
6446 sine=220:4:d=5
6448 sine=f=220:b=4:d=5
6449 sine=frequency=220:beep_factor=4:duration=5
6450 • Generate a 1 kHz sine wave following "1602,1601,1602,1601,1602"
6452 NTSC pattern:
6453 sine=1000:samples_per_frame='st(0,mod(n,5)); 1602-not(not(eq(ld(0),1)+eq(ld(0),3)))'
6455
6457 Below is a description of the currently available audio sinks.
6458 abuffersink
6460 Buffer audio frames, and make them available to the end of filter
6461 chain.
6462 This sink is mainly intended for programmatic use, in particular
6464 through the interface defined in libavfilter/buffersink.h or the
6465 options system.
6466 It accepts a pointer to an AVABufferSinkContext structure, which
6468 defines the incoming buffers' formats, to be passed as the opaque
6469 parameter to "avfilter_init_filter" for initialization.
6470 anullsink
6472 Null audio sink; do absolutely nothing with the input audio. It is
6473 mainly useful as a template and for use in analysis / debugging tools.
6474
6476 When you configure your FFmpeg build, you can disable any of the
6477 existing filters using "--disable-filters". The configure output will
6478 show the video filters included in your build.
6479 Below is a description of the currently available video filters.
6481 addroi
6483 Mark a region of interest in a video frame.
6484 The frame data is passed through unchanged, but metadata is attached to
6486 the frame indicating regions of interest which can affect the behaviour
6487 of later encoding. Multiple regions can be marked by applying the
6488 filter multiple times.
6489 x Region distance in pixels from the left edge of the frame.
6491 y Region distance in pixels from the top edge of the frame.
6493 w Region width in pixels.
6495 h Region height in pixels.
6497 The parameters x, y, w and h are expressions, and may contain the
6499 following variables:
6500 iw Width of the input frame.
6502 ih Height of the input frame.
6504 qoffset
6506 Quantisation offset to apply within the region.
6507 This must be a real value in the range -1 to +1. A value of zero
6509 indicates no quality change. A negative value asks for better
6510 quality (less quantisation), while a positive value asks for worse
6511 quality (greater quantisation).
6512 The range is calibrated so that the extreme values indicate the
6514 largest possible offset - if the rest of the frame is encoded with
6515 the worst possible quality, an offset of -1 indicates that this
6516 region should be encoded with the best possible quality anyway.
6517 Intermediate values are then interpolated in some codec-dependent
6518 way.
6519 For example, in 10-bit H.264 the quantisation parameter varies
6521 between -12 and 51. A typical qoffset value of -1/10 therefore
6522 indicates that this region should be encoded with a QP around one-
6523 tenth of the full range better than the rest of the frame. So, if
6524 most of the frame were to be encoded with a QP of around 30, this
6525 region would get a QP of around 24 (an offset of approximately
6526 -1/10 * (51 - -12) = -6.3). An extreme value of -1 would indicate
6527 that this region should be encoded with the best possible quality
6528 regardless of the treatment of the rest of the frame - that is,
6529 should be encoded at a QP of -12.
6530 clear
6532 If set to true, remove any existing regions of interest marked on
6533 the frame before adding the new one.
6534 Examples
6536 • Mark the centre quarter of the frame as interesting.
6538 addroi=iw/4:ih/4:iw/2:ih/2:-1/10
6540 • Mark the 100-pixel-wide region on the left edge of the frame as
6542 very uninteresting (to be encoded at much lower quality than the
6543 rest of the frame).
6544 addroi=0:0:100:ih:+1/5
6546 alphaextract
6548 Extract the alpha component from the input as a grayscale video. This
6549 is especially useful with the alphamerge filter.
6550 alphamerge
6552 Add or replace the alpha component of the primary input with the
6553 grayscale value of a second input. This is intended for use with
6554 alphaextract to allow the transmission or storage of frame sequences
6555 that have alpha in a format that doesn't support an alpha channel.
6556 For example, to reconstruct full frames from a normal YUV-encoded video
6558 and a separate video created with alphaextract, you might use:
6559 movie=in_alpha.mkv [alpha]; [in][alpha] alphamerge [out]
6561 amplify
6563 Amplify differences between current pixel and pixels of adjacent frames
6564 in same pixel location.
6565 This filter accepts the following options:
6567 radius
6569 Set frame radius. Default is 2. Allowed range is from 1 to 63. For
6570 example radius of 3 will instruct filter to calculate average of 7
6571 frames.
6572 factor
6574 Set factor to amplify difference. Default is 2. Allowed range is
6575 from 0 to 65535.
6576 threshold
6578 Set threshold for difference amplification. Any difference greater
6579 or equal to this value will not alter source pixel. Default is 10.
6580 Allowed range is from 0 to 65535.
6581 tolerance
6583 Set tolerance for difference amplification. Any difference lower to
6584 this value will not alter source pixel. Default is 0. Allowed
6585 range is from 0 to 65535.
6586 low Set lower limit for changing source pixel. Default is 65535.
6588 Allowed range is from 0 to 65535. This option controls maximum
6589 possible value that will decrease source pixel value.
6590 high
6592 Set high limit for changing source pixel. Default is 65535. Allowed
6593 range is from 0 to 65535. This option controls maximum possible
6594 value that will increase source pixel value.
6595 planes
6597 Set which planes to filter. Default is all. Allowed range is from 0
6598 to 15.
6599 Commands
6601 This filter supports the following commands that corresponds to option
6603 of same name:
6604 factor
6606 threshold
6607 tolerance
6608 low
6609 high
6610 planes
6611 ass
6613 Same as the subtitles filter, except that it doesn't require libavcodec
6614 and libavformat to work. On the other hand, it is limited to ASS
6615 (Advanced Substation Alpha) subtitles files.
6616 This filter accepts the following option in addition to the common
6618 options from the subtitles filter:
6619 shaping
6621 Set the shaping engine
6622 Available values are:
6624 auto
6626 The default libass shaping engine, which is the best available.
6627 simple
6629 Fast, font-agnostic shaper that can do only substitutions
6630 complex
6632 Slower shaper using OpenType for substitutions and positioning
6633 The default is "auto".
6635 atadenoise
6637 Apply an Adaptive Temporal Averaging Denoiser to the video input.
6638 The filter accepts the following options:
6640 0a Set threshold A for 1st plane. Default is 0.02. Valid range is 0
6642 to 0.3.
6643 0b Set threshold B for 1st plane. Default is 0.04. Valid range is 0
6645 to 5.
6646 1a Set threshold A for 2nd plane. Default is 0.02. Valid range is 0
6648 to 0.3.
6649 1b Set threshold B for 2nd plane. Default is 0.04. Valid range is 0
6651 to 5.
6652 2a Set threshold A for 3rd plane. Default is 0.02. Valid range is 0
6654 to 0.3.
6655 2b Set threshold B for 3rd plane. Default is 0.04. Valid range is 0
6657 to 5.
6658 Threshold A is designed to react on abrupt changes in the input
6660 signal and threshold B is designed to react on continuous changes
6661 in the input signal.
6662 s Set number of frames filter will use for averaging. Default is 9.
6664 Must be odd number in range [5, 129].
6665 p Set what planes of frame filter will use for averaging. Default is
6667 all.
6668 a Set what variant of algorithm filter will use for averaging.
6670 Default is "p" parallel. Alternatively can be set to "s" serial.
6671 Parallel can be faster then serial, while other way around is never
6673 true. Parallel will abort early on first change being greater then
6674 thresholds, while serial will continue processing other side of
6675 frames if they are equal or below thresholds.
6676 0s
6678 1s
6679 2s Set sigma for 1st plane, 2nd plane or 3rd plane. Default is 32767.
6680 Valid range is from 0 to 32767. This options controls weight for
6681 each pixel in radius defined by size. Default value means every
6682 pixel have same weight. Setting this option to 0 effectively
6683 disables filtering.
6684 Commands
6686 This filter supports same commands as options except option "s". The
6688 command accepts the same syntax of the corresponding option.
6689 avgblur
6691 Apply average blur filter.
6692 The filter accepts the following options:
6694 sizeX
6696 Set horizontal radius size.
6697 planes
6699 Set which planes to filter. By default all planes are filtered.
6700 sizeY
6702 Set vertical radius size, if zero it will be same as "sizeX".
6703 Default is 0.
6704 Commands
6706 This filter supports same commands as options. The command accepts the
6708 same syntax of the corresponding option.
6709 If the specified expression is not valid, it is kept at its current
6711 value.
6712 bbox
6714 Compute the bounding box for the non-black pixels in the input frame
6715 luminance plane.
6716 This filter computes the bounding box containing all the pixels with a
6718 luminance value greater than the minimum allowed value. The parameters
6719 describing the bounding box are printed on the filter log.
6720 The filter accepts the following option:
6722 min_val
6724 Set the minimal luminance value. Default is 16.
6725 Commands
6727 This filter supports the all above options as commands.
6729 bilateral
6731 Apply bilateral filter, spatial smoothing while preserving edges.
6732 The filter accepts the following options:
6734 sigmaS
6736 Set sigma of gaussian function to calculate spatial weight.
6737 Allowed range is 0 to 512. Default is 0.1.
6738 sigmaR
6740 Set sigma of gaussian function to calculate range weight. Allowed
6741 range is 0 to 1. Default is 0.1.
6742 planes
6744 Set planes to filter. Default is first only.
6745 Commands
6747 This filter supports the all above options as commands.
6749 bitplanenoise
6751 Show and measure bit plane noise.
6752 The filter accepts the following options:
6754 bitplane
6756 Set which plane to analyze. Default is 1.
6757 filter
6759 Filter out noisy pixels from "bitplane" set above. Default is
6760 disabled.
6761 blackdetect
6763 Detect video intervals that are (almost) completely black. Can be
6764 useful to detect chapter transitions, commercials, or invalid
6765 recordings.
6766 The filter outputs its detection analysis to both the log as well as
6768 frame metadata. If a black segment of at least the specified minimum
6769 duration is found, a line with the start and end timestamps as well as
6770 duration is printed to the log with level "info". In addition, a log
6771 line with level "debug" is printed per frame showing the black amount
6772 detected for that frame.
6773 The filter also attaches metadata to the first frame of a black segment
6775 with key "lavfi.black_start" and to the first frame after the black
6776 segment ends with key "lavfi.black_end". The value is the frame's
6777 timestamp. This metadata is added regardless of the minimum duration
6778 specified.
6779 The filter accepts the following options:
6781 black_min_duration, d
6783 Set the minimum detected black duration expressed in seconds. It
6784 must be a non-negative floating point number.
6785 Default value is 2.0.
6787 picture_black_ratio_th, pic_th
6789 Set the threshold for considering a picture "black". Express the
6790 minimum value for the ratio:
6791 <nb_black_pixels> / <nb_pixels>
6793 for which a picture is considered black. Default value is 0.98.
6795 pixel_black_th, pix_th
6797 Set the threshold for considering a pixel "black".
6798 The threshold expresses the maximum pixel luminance value for which
6800 a pixel is considered "black". The provided value is scaled
6801 according to the following equation:
6802 <absolute_threshold> = <luminance_minimum_value> + <pixel_black_th> * <luminance_range_size>
6804 luminance_range_size and luminance_minimum_value depend on the
6806 input video format, the range is [0-255] for YUV full-range formats
6807 and [16-235] for YUV non full-range formats.
6808 Default value is 0.10.
6810 The following example sets the maximum pixel threshold to the minimum
6812 value, and detects only black intervals of 2 or more seconds:
6813 blackdetect=d=2:pix_th=0.00
6815 blackframe
6817 Detect frames that are (almost) completely black. Can be useful to
6818 detect chapter transitions or commercials. Output lines consist of the
6819 frame number of the detected frame, the percentage of blackness, the
6820 position in the file if known or -1 and the timestamp in seconds.
6821 In order to display the output lines, you need to set the loglevel at
6823 least to the AV_LOG_INFO value.
6824 This filter exports frame metadata "lavfi.blackframe.pblack". The
6826 value represents the percentage of pixels in the picture that are below
6827 the threshold value.
6828 It accepts the following parameters:
6830 amount
6832 The percentage of the pixels that have to be below the threshold;
6833 it defaults to 98.
6834 threshold, thresh
6836 The threshold below which a pixel value is considered black; it
6837 defaults to 32.
6838 blend
6840 Blend two video frames into each other.
6841 The "blend" filter takes two input streams and outputs one stream, the
6843 first input is the "top" layer and second input is "bottom" layer. By
6844 default, the output terminates when the longest input terminates.
6845 The "tblend" (time blend) filter takes two consecutive frames from one
6847 single stream, and outputs the result obtained by blending the new
6848 frame on top of the old frame.
6849 A description of the accepted options follows.
6851 c0_mode
6853 c1_mode
6854 c2_mode
6855 c3_mode
6856 all_mode
6857 Set blend mode for specific pixel component or all pixel components
6858 in case of all_mode. Default value is "normal".
6859 Available values for component modes are:
6861 addition
6863 and
6864 average
6865 bleach
6866 burn
6867 darken
6868 difference
6869 divide
6870 dodge
6871 exclusion
6872 extremity
6873 freeze
6874 geometric
6875 glow
6876 grainextract
6877 grainmerge
6878 hardlight
6879 hardmix
6880 hardoverlay
6881 harmonic
6882 heat
6883 interpolate
6884 lighten
6885 linearlight
6886 multiply
6887 multiply128
6888 negation
6889 normal
6890 or
6891 overlay
6892 phoenix
6893 pinlight
6894 reflect
6895 screen
6896 softdifference
6897 softlight
6898 stain
6899 subtract
6900 vividlight
6901 xor
6902 c0_opacity
6903 c1_opacity
6904 c2_opacity
6905 c3_opacity
6906 all_opacity
6907 Set blend opacity for specific pixel component or all pixel
6908 components in case of all_opacity. Only used in combination with
6909 pixel component blend modes.
6910 c0_expr
6912 c1_expr
6913 c2_expr
6914 c3_expr
6915 all_expr
6916 Set blend expression for specific pixel component or all pixel
6917 components in case of all_expr. Note that related mode options will
6918 be ignored if those are set.
6919 The expressions can use the following variables:
6921 N The sequential number of the filtered frame, starting from 0.
6923 X
6925 Y the coordinates of the current sample
6926 W
6928 H the width and height of currently filtered plane
6929 SW
6931 SH Width and height scale for the plane being filtered. It is the
6932 ratio between the dimensions of the current plane to the luma
6933 plane, e.g. for a "yuv420p" frame, the values are "1,1" for the
6934 luma plane and "0.5,0.5" for the chroma planes.
6935 T Time of the current frame, expressed in seconds.
6937 TOP, A
6939 Value of pixel component at current location for first video
6940 frame (top layer).
6941 BOTTOM, B
6943 Value of pixel component at current location for second video
6944 frame (bottom layer).
6945 The "blend" filter also supports the framesync options.
6947 Examples
6949 • Apply transition from bottom layer to top layer in first 10
6951 seconds:
6952 blend=all_expr='A*(if(gte(T,10),1,T/10))+B*(1-(if(gte(T,10),1,T/10)))'
6954 • Apply linear horizontal transition from top layer to bottom layer:
6956 blend=all_expr='A*(X/W)+B*(1-X/W)'
6958 • Apply 1x1 checkerboard effect:
6960 blend=all_expr='if(eq(mod(X,2),mod(Y,2)),A,B)'
6962 • Apply uncover left effect:
6964 blend=all_expr='if(gte(N*SW+X,W),A,B)'
6966 • Apply uncover down effect:
6968 blend=all_expr='if(gte(Y-N*SH,0),A,B)'
6970 • Apply uncover up-left effect:
6972 blend=all_expr='if(gte(T*SH*40+Y,H)*gte((T*40*SW+X)*W/H,W),A,B)'
6974 • Split diagonally video and shows top and bottom layer on each side:
6976 blend=all_expr='if(gt(X,Y*(W/H)),A,B)'
6978 • Display differences between the current and the previous frame:
6980 tblend=all_mode=grainextract
6982 Commands
6984 This filter supports same commands as options.
6986 bm3d
6988 Denoise frames using Block-Matching 3D algorithm.
6989 The filter accepts the following options.
6991 sigma
6993 Set denoising strength. Default value is 1. Allowed range is from
6994 0 to 999.9. The denoising algorithm is very sensitive to sigma, so
6995 adjust it according to the source.
6996 block
6998 Set local patch size. This sets dimensions in 2D.
6999 bstep
7001 Set sliding step for processing blocks. Default value is 4.
7002 Allowed range is from 1 to 64. Smaller values allows processing
7003 more reference blocks and is slower.
7004 group
7006 Set maximal number of similar blocks for 3rd dimension. Default
7007 value is 1. When set to 1, no block matching is done. Larger
7008 values allows more blocks in single group. Allowed range is from 1
7009 to 256.
7010 range
7012 Set radius for search block matching. Default is 9. Allowed range
7013 is from 1 to INT32_MAX.
7014 mstep
7016 Set step between two search locations for block matching. Default
7017 is 1. Allowed range is from 1 to 64. Smaller is slower.
7018 thmse
7020 Set threshold of mean square error for block matching. Valid range
7021 is 0 to INT32_MAX.
7022 hdthr
7024 Set thresholding parameter for hard thresholding in 3D transformed
7025 domain. Larger values results in stronger hard-thresholding
7026 filtering in frequency domain.
7027 estim
7029 Set filtering estimation mode. Can be "basic" or "final". Default
7030 is "basic".
7031 ref If enabled, filter will use 2nd stream for block matching. Default
7033 is disabled for "basic" value of estim option, and always enabled
7034 if value of estim is "final".
7035 planes
7037 Set planes to filter. Default is all available except alpha.
7038 Examples
7040 • Basic filtering with bm3d:
7042 bm3d=sigma=3:block=4:bstep=2:group=1:estim=basic
7044 • Same as above, but filtering only luma:
7046 bm3d=sigma=3:block=4:bstep=2:group=1:estim=basic:planes=1
7048 • Same as above, but with both estimation modes:
7050 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
7052 • Same as above, but prefilter with nlmeans filter instead:
7054 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
7056 boxblur
7058 Apply a boxblur algorithm to the input video.
7059 It accepts the following parameters:
7061 luma_radius, lr
7063 luma_power, lp
7064 chroma_radius, cr
7065 chroma_power, cp
7066 alpha_radius, ar
7067 alpha_power, ap
7068 A description of the accepted options follows.
7070 luma_radius, lr
7072 chroma_radius, cr
7073 alpha_radius, ar
7074 Set an expression for the box radius in pixels used for blurring
7075 the corresponding input plane.
7076 The radius value must be a non-negative number, and must not be
7078 greater than the value of the expression "min(w,h)/2" for the luma
7079 and alpha planes, and of "min(cw,ch)/2" for the chroma planes.
7080 Default value for luma_radius is "2". If not specified,
7082 chroma_radius and alpha_radius default to the corresponding value
7083 set for luma_radius.
7084 The expressions can contain the following constants:
7086 w
7088 h The input width and height in pixels.
7089 cw
7091 ch The input chroma image width and height in pixels.
7092 hsub
7094 vsub
7095 The horizontal and vertical chroma subsample values. For
7096 example, for the pixel format "yuv422p", hsub is 2 and vsub is
7097 1.
7098 luma_power, lp
7100 chroma_power, cp
7101 alpha_power, ap
7102 Specify how many times the boxblur filter is applied to the
7103 corresponding plane.
7104 Default value for luma_power is 2. If not specified, chroma_power
7106 and alpha_power default to the corresponding value set for
7107 luma_power.
7108 A value of 0 will disable the effect.
7110 Examples
7112 • Apply a boxblur filter with the luma, chroma, and alpha radii set
7114 to 2:
7115 boxblur=luma_radius=2:luma_power=1
7117 boxblur=2:1
7118 • Set the luma radius to 2, and alpha and chroma radius to 0:
7120 boxblur=2:1:cr=0:ar=0
7122 • Set the luma and chroma radii to a fraction of the video dimension:
7124 boxblur=luma_radius=min(h\,w)/10:luma_power=1:chroma_radius=min(cw\,ch)/10:chroma_power=1
7126 bwdif
7128 Deinterlace the input video ("bwdif" stands for "Bob Weaver
7129 Deinterlacing Filter").
7130 Motion adaptive deinterlacing based on yadif with the use of w3fdif and
7132 cubic interpolation algorithms. It accepts the following parameters:
7133 mode
7135 The interlacing mode to adopt. It accepts one of the following
7136 values:
7137 0, send_frame
7139 Output one frame for each frame.
7140 1, send_field
7142 Output one frame for each field.
7143 The default value is "send_field".
7145 parity
7147 The picture field parity assumed for the input interlaced video. It
7148 accepts one of the following values:
7149 0, tff
7151 Assume the top field is first.
7152 1, bff
7154 Assume the bottom field is first.
7155 -1, auto
7157 Enable automatic detection of field parity.
7158 The default value is "auto". If the interlacing is unknown or the
7160 decoder does not export this information, top field first will be
7161 assumed.
7162 deint
7164 Specify which frames to deinterlace. Accepts one of the following
7165 values:
7166 0, all
7168 Deinterlace all frames.
7169 1, interlaced
7171 Only deinterlace frames marked as interlaced.
7172 The default value is "all".
7174 cas
7176 Apply Contrast Adaptive Sharpen filter to video stream.
7177 The filter accepts the following options:
7179 strength
7181 Set the sharpening strength. Default value is 0.
7182 planes
7184 Set planes to filter. Default value is to filter all planes except
7185 alpha plane.
7186 Commands
7188 This filter supports same commands as options.
7190 chromahold
7192 Remove all color information for all colors except for certain one.
7193 The filter accepts the following options:
7195 color
7197 The color which will not be replaced with neutral chroma.
7198 similarity
7200 Similarity percentage with the above color. 0.01 matches only the
7201 exact key color, while 1.0 matches everything.
7202 blend
7204 Blend percentage. 0.0 makes pixels either fully gray, or not gray
7205 at all. Higher values result in more preserved color.
7206 yuv Signals that the color passed is already in YUV instead of RGB.
7208 Literal colors like "green" or "red" don't make sense with this
7210 enabled anymore. This can be used to pass exact YUV values as
7211 hexadecimal numbers.
7212 Commands
7214 This filter supports same commands as options. The command accepts the
7216 same syntax of the corresponding option.
7217 If the specified expression is not valid, it is kept at its current
7219 value.
7220 chromakey
7222 YUV colorspace color/chroma keying.
7223 The filter accepts the following options:
7225 color
7227 The color which will be replaced with transparency.
7228 similarity
7230 Similarity percentage with the key color.
7231 0.01 matches only the exact key color, while 1.0 matches
7233 everything.
7234 blend
7236 Blend percentage.
7237 0.0 makes pixels either fully transparent, or not transparent at
7239 all.
7240 Higher values result in semi-transparent pixels, with a higher
7242 transparency the more similar the pixels color is to the key color.
7243 yuv Signals that the color passed is already in YUV instead of RGB.
7245 Literal colors like "green" or "red" don't make sense with this
7247 enabled anymore. This can be used to pass exact YUV values as
7248 hexadecimal numbers.
7249 Commands
7251 This filter supports same commands as options. The command accepts the
7253 same syntax of the corresponding option.
7254 If the specified expression is not valid, it is kept at its current
7256 value.
7257 Examples
7259 • Make every green pixel in the input image transparent:
7261 ffmpeg -i input.png -vf chromakey=green out.png
7263 • Overlay a greenscreen-video on top of a static black background.
7265 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
7267 chromanr
7269 Reduce chrominance noise.
7270 The filter accepts the following options:
7272 thres
7274 Set threshold for averaging chrominance values. Sum of absolute
7275 difference of Y, U and V pixel components of current pixel and
7276 neighbour pixels lower than this threshold will be used in
7277 averaging. Luma component is left unchanged and is copied to
7278 output. Default value is 30. Allowed range is from 1 to 200.
7279 sizew
7281 Set horizontal radius of rectangle used for averaging. Allowed
7282 range is from 1 to 100. Default value is 5.
7283 sizeh
7285 Set vertical radius of rectangle used for averaging. Allowed range
7286 is from 1 to 100. Default value is 5.
7287 stepw
7289 Set horizontal step when averaging. Default value is 1. Allowed
7290 range is from 1 to 50. Mostly useful to speed-up filtering.
7291 steph
7293 Set vertical step when averaging. Default value is 1. Allowed
7294 range is from 1 to 50. Mostly useful to speed-up filtering.
7295 threy
7297 Set Y threshold for averaging chrominance values. Set finer
7298 control for max allowed difference between Y components of current
7299 pixel and neigbour pixels. Default value is 200. Allowed range is
7300 from 1 to 200.
7301 threu
7303 Set U threshold for averaging chrominance values. Set finer
7304 control for max allowed difference between U components of current
7305 pixel and neigbour pixels. Default value is 200. Allowed range is
7306 from 1 to 200.
7307 threv
7309 Set V threshold for averaging chrominance values. Set finer
7310 control for max allowed difference between V components of current
7311 pixel and neigbour pixels. Default value is 200. Allowed range is
7312 from 1 to 200.
7313 distance
7315 Set distance type used in calculations.
7316 manhattan
7318 Absolute difference.
7319 euclidean
7321 Difference squared.
7322 Default distance type is manhattan.
7324 Commands
7326 This filter supports same commands as options. The command accepts the
7328 same syntax of the corresponding option.
7329 chromashift
7331 Shift chroma pixels horizontally and/or vertically.
7332 The filter accepts the following options:
7334 cbh Set amount to shift chroma-blue horizontally.
7336 cbv Set amount to shift chroma-blue vertically.
7338 crh Set amount to shift chroma-red horizontally.
7340 crv Set amount to shift chroma-red vertically.
7342 edge
7344 Set edge mode, can be smear, default, or warp.
7345 Commands
7347 This filter supports the all above options as commands.
7349 ciescope
7351 Display CIE color diagram with pixels overlaid onto it.
7352 The filter accepts the following options:
7354 system
7356 Set color system.
7357 ntsc, 470m
7359 ebu, 470bg
7360 smpte
7361 240m
7362 apple
7363 widergb
7364 cie1931
7365 rec709, hdtv
7366 uhdtv, rec2020
7367 dcip3
7368 cie Set CIE system.
7369 xyy
7371 ucs
7372 luv
7373 gamuts
7374 Set what gamuts to draw.
7375 See "system" option for available values.
7377 size, s
7379 Set ciescope size, by default set to 512.
7380 intensity, i
7382 Set intensity used to map input pixel values to CIE diagram.
7383 contrast
7385 Set contrast used to draw tongue colors that are out of active
7386 color system gamut.
7387 corrgamma
7389 Correct gamma displayed on scope, by default enabled.
7390 showwhite
7392 Show white point on CIE diagram, by default disabled.
7393 gamma
7395 Set input gamma. Used only with XYZ input color space.
7396 codecview
7398 Visualize information exported by some codecs.
7399 Some codecs can export information through frames using side-data or
7401 other means. For example, some MPEG based codecs export motion vectors
7402 through the export_mvs flag in the codec flags2 option.
7403 The filter accepts the following option:
7405 block
7407 Display block partition structure using the luma plane.
7408 mv Set motion vectors to visualize.
7410 Available flags for mv are:
7412 pf forward predicted MVs of P-frames
7414 bf forward predicted MVs of B-frames
7416 bb backward predicted MVs of B-frames
7418 qp Display quantization parameters using the chroma planes.
7420 mv_type, mvt
7422 Set motion vectors type to visualize. Includes MVs from all frames
7423 unless specified by frame_type option.
7424 Available flags for mv_type are:
7426 fp forward predicted MVs
7428 bp backward predicted MVs
7430 frame_type, ft
7432 Set frame type to visualize motion vectors of.
7433 Available flags for frame_type are:
7435 if intra-coded frames (I-frames)
7437 pf predicted frames (P-frames)
7439 bf bi-directionally predicted frames (B-frames)
7441 Examples
7443 • Visualize forward predicted MVs of all frames using ffplay:
7445 ffplay -flags2 +export_mvs input.mp4 -vf codecview=mv_type=fp
7447 • Visualize multi-directionals MVs of P and B-Frames using ffplay:
7449 ffplay -flags2 +export_mvs input.mp4 -vf codecview=mv=pf+bf+bb
7451 colorbalance
7453 Modify intensity of primary colors (red, green and blue) of input
7454 frames.
7455 The filter allows an input frame to be adjusted in the shadows,
7457 midtones or highlights regions for the red-cyan, green-magenta or blue-
7458 yellow balance.
7459 A positive adjustment value shifts the balance towards the primary
7461 color, a negative value towards the complementary color.
7462 The filter accepts the following options:
7464 rs
7466 gs
7467 bs Adjust red, green and blue shadows (darkest pixels).
7468 rm
7470 gm
7471 bm Adjust red, green and blue midtones (medium pixels).
7472 rh
7474 gh
7475 bh Adjust red, green and blue highlights (brightest pixels).
7476 Allowed ranges for options are "[-1.0, 1.0]". Defaults are 0.
7478 pl Preserve lightness when changing color balance. Default is
7480 disabled.
7481 Examples
7483 • Add red color cast to shadows:
7485 colorbalance=rs=.3
7487 Commands
7489 This filter supports the all above options as commands.
7491 colorcontrast
7493 Adjust color contrast between RGB components.
7494 The filter accepts the following options:
7496 rc Set the red-cyan contrast. Defaults is 0.0. Allowed range is from
7498 -1.0 to 1.0.
7499 gm Set the green-magenta contrast. Defaults is 0.0. Allowed range is
7501 from -1.0 to 1.0.
7502 by Set the blue-yellow contrast. Defaults is 0.0. Allowed range is
7504 from -1.0 to 1.0.
7505 rcw
7507 gmw
7508 byw Set the weight of each "rc", "gm", "by" option value. Default value
7509 is 0.0. Allowed range is from 0.0 to 1.0. If all weights are 0.0
7510 filtering is disabled.
7511 pl Set the amount of preserving lightness. Default value is 0.0.
7513 Allowed range is from 0.0 to 1.0.
7514 Commands
7516 This filter supports the all above options as commands.
7518 colorcorrect
7520 Adjust color white balance selectively for blacks and whites. This
7521 filter operates in YUV colorspace.
7522 The filter accepts the following options:
7524 rl Set the red shadow spot. Allowed range is from -1.0 to 1.0.
7526 Default value is 0.
7527 bl Set the blue shadow spot. Allowed range is from -1.0 to 1.0.
7529 Default value is 0.
7530 rh Set the red highlight spot. Allowed range is from -1.0 to 1.0.
7532 Default value is 0.
7533 bh Set the red highlight spot. Allowed range is from -1.0 to 1.0.
7535 Default value is 0.
7536 saturation
7538 Set the amount of saturation. Allowed range is from -3.0 to 3.0.
7539 Default value is 1.
7540 analyze
7542 If set to anything other than "manual" it will analyze every frame
7543 and use derived parameters for filtering output frame.
7544 Possible values are:
7546 manual
7548 average
7549 minmax
7550 median
7551 Default value is "manual".
7553 Commands
7555 This filter supports the all above options as commands.
7557 colorchannelmixer
7559 Adjust video input frames by re-mixing color channels.
7560 This filter modifies a color channel by adding the values associated to
7562 the other channels of the same pixels. For example if the value to
7563 modify is red, the output value will be:
7564 <red>=<red>*<rr> + <blue>*<rb> + <green>*<rg> + <alpha>*<ra>
7566 The filter accepts the following options:
7568 rr
7570 rg
7571 rb
7572 ra Adjust contribution of input red, green, blue and alpha channels
7573 for output red channel. Default is 1 for rr, and 0 for rg, rb and
7574 ra.
7575 gr
7577 gg
7578 gb
7579 ga Adjust contribution of input red, green, blue and alpha channels
7580 for output green channel. Default is 1 for gg, and 0 for gr, gb
7581 and ga.
7582 br
7584 bg
7585 bb
7586 ba Adjust contribution of input red, green, blue and alpha channels
7587 for output blue channel. Default is 1 for bb, and 0 for br, bg and
7588 ba.
7589 ar
7591 ag
7592 ab
7593 aa Adjust contribution of input red, green, blue and alpha channels
7594 for output alpha channel. Default is 1 for aa, and 0 for ar, ag
7595 and ab.
7596 Allowed ranges for options are "[-2.0, 2.0]".
7598 pc Set preserve color mode. The accepted values are:
7600 none
7602 Disable color preserving, this is default.
7603 lum Preserve luminance.
7605 max Preserve max value of RGB triplet.
7607 avg Preserve average value of RGB triplet.
7609 sum Preserve sum value of RGB triplet.
7611 nrm Preserve normalized value of RGB triplet.
7613 pwr Preserve power value of RGB triplet.
7615 pa Set the preserve color amount when changing colors. Allowed range
7617 is from "[0.0, 1.0]". Default is 0.0, thus disabled.
7618 Examples
7620 • Convert source to grayscale:
7622 colorchannelmixer=.3:.4:.3:0:.3:.4:.3:0:.3:.4:.3
7624 • Simulate sepia tones:
7626 colorchannelmixer=.393:.769:.189:0:.349:.686:.168:0:.272:.534:.131
7628 Commands
7630 This filter supports the all above options as commands.
7632 colorize
7634 Overlay a solid color on the video stream.
7635 The filter accepts the following options:
7637 hue Set the color hue. Allowed range is from 0 to 360. Default value
7639 is 0.
7640 saturation
7642 Set the color saturation. Allowed range is from 0 to 1. Default
7643 value is 0.5.
7644 lightness
7646 Set the color lightness. Allowed range is from 0 to 1. Default
7647 value is 0.5.
7648 mix Set the mix of source lightness. By default is set to 1.0. Allowed
7650 range is from 0.0 to 1.0.
7651 Commands
7653 This filter supports the all above options as commands.
7655 colorkey
7657 RGB colorspace color keying. This filter operates on 8-bit RGB format
7658 frames by setting the alpha component of each pixel which falls within
7659 the similarity radius of the key color to 0. The alpha value for pixels
7660 outside the similarity radius depends on the value of the blend option.
7661 The filter accepts the following options:
7663 color
7665 Set the color for which alpha will be set to 0 (full transparency).
7666 See "Color" section in the ffmpeg-utils manual. Default is
7667 "black".
7668 similarity
7670 Set the radius from the key color within which other colors also
7671 have full transparency. The computed distance is related to the
7672 unit fractional distance in 3D space between the RGB values of the
7673 key color and the pixel's color. Range is 0.01 to 1.0. 0.01 matches
7674 within a very small radius around the exact key color, while 1.0
7675 matches everything. Default is 0.01.
7676 blend
7678 Set how the alpha value for pixels that fall outside the similarity
7679 radius is computed. 0.0 makes pixels either fully transparent or
7680 fully opaque. Higher values result in semi-transparent pixels,
7681 with greater transparency the more similar the pixel color is to
7682 the key color. Range is 0.0 to 1.0. Default is 0.0.
7683 Examples
7685 • Make every green pixel in the input image transparent:
7687 ffmpeg -i input.png -vf colorkey=green out.png
7689 • Overlay a greenscreen-video on top of a static background image.
7691 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
7693 Commands
7695 This filter supports same commands as options. The command accepts the
7697 same syntax of the corresponding option.
7698 If the specified expression is not valid, it is kept at its current
7700 value.
7701 colorhold
7703 Remove all color information for all RGB colors except for certain one.
7704 The filter accepts the following options:
7706 color
7708 The color which will not be replaced with neutral gray.
7709 similarity
7711 Similarity percentage with the above color. 0.01 matches only the
7712 exact key color, while 1.0 matches everything.
7713 blend
7715 Blend percentage. 0.0 makes pixels fully gray. Higher values
7716 result in more preserved color.
7717 Commands
7719 This filter supports same commands as options. The command accepts the
7721 same syntax of the corresponding option.
7722 If the specified expression is not valid, it is kept at its current
7724 value.
7725 colorlevels
7727 Adjust video input frames using levels.
7728 The filter accepts the following options:
7730 rimin
7732 gimin
7733 bimin
7734 aimin
7735 Adjust red, green, blue and alpha input black point. Allowed
7736 ranges for options are "[-1.0, 1.0]". Defaults are 0.
7737 rimax
7739 gimax
7740 bimax
7741 aimax
7742 Adjust red, green, blue and alpha input white point. Allowed
7743 ranges for options are "[-1.0, 1.0]". Defaults are 1.
7744 Input levels are used to lighten highlights (bright tones), darken
7746 shadows (dark tones), change the balance of bright and dark tones.
7747 romin
7749 gomin
7750 bomin
7751 aomin
7752 Adjust red, green, blue and alpha output black point. Allowed
7753 ranges for options are "[0, 1.0]". Defaults are 0.
7754 romax
7756 gomax
7757 bomax
7758 aomax
7759 Adjust red, green, blue and alpha output white point. Allowed
7760 ranges for options are "[0, 1.0]". Defaults are 1.
7761 Output levels allows manual selection of a constrained output level
7763 range.
7764 preserve
7766 Set preserve color mode. The accepted values are:
7767 none
7769 Disable color preserving, this is default.
7770 lum Preserve luminance.
7772 max Preserve max value of RGB triplet.
7774 avg Preserve average value of RGB triplet.
7776 sum Preserve sum value of RGB triplet.
7778 nrm Preserve normalized value of RGB triplet.
7780 pwr Preserve power value of RGB triplet.
7782 Examples
7784 • Make video output darker:
7786 colorlevels=rimin=0.058:gimin=0.058:bimin=0.058
7788 • Increase contrast:
7790 colorlevels=rimin=0.039:gimin=0.039:bimin=0.039:rimax=0.96:gimax=0.96:bimax=0.96
7792 • Make video output lighter:
7794 colorlevels=rimax=0.902:gimax=0.902:bimax=0.902
7796 • Increase brightness:
7798 colorlevels=romin=0.5:gomin=0.5:bomin=0.5
7800 Commands
7802 This filter supports the all above options as commands.
7804 colormatrix
7806 Convert color matrix.
7807 The filter accepts the following options:
7809 src
7811 dst Specify the source and destination color matrix. Both values must
7812 be specified.
7813 The accepted values are:
7815 bt709
7817 BT.709
7818 fcc FCC
7820 bt601
7822 BT.601
7823 bt470
7825 BT.470
7826 bt470bg
7828 BT.470BG
7829 smpte170m
7831 SMPTE-170M
7832 smpte240m
7834 SMPTE-240M
7835 bt2020
7837 BT.2020
7838 For example to convert from BT.601 to SMPTE-240M, use the command:
7840 colormatrix=bt601:smpte240m
7842 colorspace
7844 Convert colorspace, transfer characteristics or color primaries. Input
7845 video needs to have an even size.
7846 The filter accepts the following options:
7848 all Specify all color properties at once.
7850 The accepted values are:
7852 bt470m
7854 BT.470M
7855 bt470bg
7857 BT.470BG
7858 bt601-6-525
7860 BT.601-6 525
7861 bt601-6-625
7863 BT.601-6 625
7864 bt709
7866 BT.709
7867 smpte170m
7869 SMPTE-170M
7870 smpte240m
7872 SMPTE-240M
7873 bt2020
7875 BT.2020
7876 space
7878 Specify output colorspace.
7879 The accepted values are:
7881 bt709
7883 BT.709
7884 fcc FCC
7886 bt470bg
7888 BT.470BG or BT.601-6 625
7889 smpte170m
7891 SMPTE-170M or BT.601-6 525
7892 smpte240m
7894 SMPTE-240M
7895 ycgco
7897 YCgCo
7898 bt2020ncl
7900 BT.2020 with non-constant luminance
7901 trc Specify output transfer characteristics.
7903 The accepted values are:
7905 bt709
7907 BT.709
7908 bt470m
7910 BT.470M
7911 bt470bg
7913 BT.470BG
7914 gamma22
7916 Constant gamma of 2.2
7917 gamma28
7919 Constant gamma of 2.8
7920 smpte170m
7922 SMPTE-170M, BT.601-6 625 or BT.601-6 525
7923 smpte240m
7925 SMPTE-240M
7926 srgb
7928 SRGB
7929 iec61966-2-1
7931 iec61966-2-1
7932 iec61966-2-4
7934 iec61966-2-4
7935 xvycc
7937 xvycc
7938 bt2020-10
7940 BT.2020 for 10-bits content
7941 bt2020-12
7943 BT.2020 for 12-bits content
7944 primaries
7946 Specify output color primaries.
7947 The accepted values are:
7949 bt709
7951 BT.709
7952 bt470m
7954 BT.470M
7955 bt470bg
7957 BT.470BG or BT.601-6 625
7958 smpte170m
7960 SMPTE-170M or BT.601-6 525
7961 smpte240m
7963 SMPTE-240M
7964 film
7966 film
7967 smpte431
7969 SMPTE-431
7970 smpte432
7972 SMPTE-432
7973 bt2020
7975 BT.2020
7976 jedec-p22
7978 JEDEC P22 phosphors
7979 range
7981 Specify output color range.
7982 The accepted values are:
7984 tv TV (restricted) range
7986 mpeg
7988 MPEG (restricted) range
7989 pc PC (full) range
7991 jpeg
7993 JPEG (full) range
7994 format
7996 Specify output color format.
7997 The accepted values are:
7999 yuv420p
8001 YUV 4:2:0 planar 8-bits
8002 yuv420p10
8004 YUV 4:2:0 planar 10-bits
8005 yuv420p12
8007 YUV 4:2:0 planar 12-bits
8008 yuv422p
8010 YUV 4:2:2 planar 8-bits
8011 yuv422p10
8013 YUV 4:2:2 planar 10-bits
8014 yuv422p12
8016 YUV 4:2:2 planar 12-bits
8017 yuv444p
8019 YUV 4:4:4 planar 8-bits
8020 yuv444p10
8022 YUV 4:4:4 planar 10-bits
8023 yuv444p12
8025 YUV 4:4:4 planar 12-bits
8026 fast
8028 Do a fast conversion, which skips gamma/primary correction. This
8029 will take significantly less CPU, but will be mathematically
8030 incorrect. To get output compatible with that produced by the
8031 colormatrix filter, use fast=1.
8032 dither
8034 Specify dithering mode.
8035 The accepted values are:
8037 none
8039 No dithering
8040 fsb Floyd-Steinberg dithering
8042 wpadapt
8044 Whitepoint adaptation mode.
8045 The accepted values are:
8047 bradford
8049 Bradford whitepoint adaptation
8050 vonkries
8052 von Kries whitepoint adaptation
8053 identity
8055 identity whitepoint adaptation (i.e. no whitepoint adaptation)
8056 iall
8058 Override all input properties at once. Same accepted values as all.
8059 ispace
8061 Override input colorspace. Same accepted values as space.
8062 iprimaries
8064 Override input color primaries. Same accepted values as primaries.
8065 itrc
8067 Override input transfer characteristics. Same accepted values as
8068 trc.
8069 irange
8071 Override input color range. Same accepted values as range.
8072 The filter converts the transfer characteristics, color space and color
8074 primaries to the specified user values. The output value, if not
8075 specified, is set to a default value based on the "all" property. If
8076 that property is also not specified, the filter will log an error. The
8077 output color range and format default to the same value as the input
8078 color range and format. The input transfer characteristics, color
8079 space, color primaries and color range should be set on the input data.
8080 If any of these are missing, the filter will log an error and no
8081 conversion will take place.
8082 For example to convert the input to SMPTE-240M, use the command:
8084 colorspace=smpte240m
8086 colortemperature
8088 Adjust color temperature in video to simulate variations in ambient
8089 color temperature.
8090 The filter accepts the following options:
8092 temperature
8094 Set the temperature in Kelvin. Allowed range is from 1000 to 40000.
8095 Default value is 6500 K.
8096 mix Set mixing with filtered output. Allowed range is from 0 to 1.
8098 Default value is 1.
8099 pl Set the amount of preserving lightness. Allowed range is from 0 to
8101 1. Default value is 0.
8102 Commands
8104 This filter supports same commands as options.
8106 convolution
8108 Apply convolution of 3x3, 5x5, 7x7 or horizontal/vertical up to 49
8109 elements.
8110 The filter accepts the following options:
8112 0m
8114 1m
8115 2m
8116 3m Set matrix for each plane. Matrix is sequence of 9, 25 or 49
8117 signed integers in square mode, and from 1 to 49 odd number of
8118 signed integers in row mode.
8119 0rdiv
8121 1rdiv
8122 2rdiv
8123 3rdiv
8124 Set multiplier for calculated value for each plane. If unset or 0,
8125 it will be sum of all matrix elements.
8126 0bias
8128 1bias
8129 2bias
8130 3bias
8131 Set bias for each plane. This value is added to the result of the
8132 multiplication. Useful for making the overall image brighter or
8133 darker. Default is 0.0.
8134 0mode
8136 1mode
8137 2mode
8138 3mode
8139 Set matrix mode for each plane. Can be square, row or column.
8140 Default is square.
8141 Commands
8143 This filter supports the all above options as commands.
8145 Examples
8147 • Apply sharpen:
8149 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"
8151 • Apply blur:
8153 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"
8155 • Apply edge enhance:
8157 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"
8159 • Apply edge detect:
8161 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"
8163 • Apply laplacian edge detector which includes diagonals:
8165 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"
8167 • Apply emboss:
8169 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"
8171 convolve
8173 Apply 2D convolution of video stream in frequency domain using second
8174 stream as impulse.
8175 The filter accepts the following options:
8177 planes
8179 Set which planes to process.
8180 impulse
8182 Set which impulse video frames will be processed, can be first or
8183 all. Default is all.
8184 The "convolve" filter also supports the framesync options.
8186 copy
8188 Copy the input video source unchanged to the output. This is mainly
8189 useful for testing purposes.
8190 coreimage
8192 Video filtering on GPU using Apple's CoreImage API on OSX.
8193 Hardware acceleration is based on an OpenGL context. Usually, this
8195 means it is processed by video hardware. However, software-based OpenGL
8196 implementations exist which means there is no guarantee for hardware
8197 processing. It depends on the respective OSX.
8198 There are many filters and image generators provided by Apple that come
8200 with a large variety of options. The filter has to be referenced by its
8201 name along with its options.
8202 The coreimage filter accepts the following options:
8204 list_filters
8206 List all available filters and generators along with all their
8207 respective options as well as possible minimum and maximum values
8208 along with the default values.
8209 list_filters=true
8211 filter
8213 Specify all filters by their respective name and options. Use
8214 list_filters to determine all valid filter names and options.
8215 Numerical options are specified by a float value and are
8216 automatically clamped to their respective value range. Vector and
8217 color options have to be specified by a list of space separated
8218 float values. Character escaping has to be done. A special option
8219 name "default" is available to use default options for a filter.
8220 It is required to specify either "default" or at least one of the
8222 filter options. All omitted options are used with their default
8223 values. The syntax of the filter string is as follows:
8224 filter=<NAME>@<OPTION>=<VALUE>[@<OPTION>=<VALUE>][@...][#<NAME>@<OPTION>=<VALUE>[@<OPTION>=<VALUE>][@...]][#...]
8226 output_rect
8228 Specify a rectangle where the output of the filter chain is copied
8229 into the input image. It is given by a list of space separated
8230 float values:
8231 output_rect=x\ y\ width\ height
8233 If not given, the output rectangle equals the dimensions of the
8235 input image. The output rectangle is automatically cropped at the
8236 borders of the input image. Negative values are valid for each
8237 component.
8238 output_rect=25\ 25\ 100\ 100
8240 Several filters can be chained for successive processing without GPU-
8242 HOST transfers allowing for fast processing of complex filter chains.
8243 Currently, only filters with zero (generators) or exactly one (filters)
8244 input image and one output image are supported. Also, transition
8245 filters are not yet usable as intended.
8246 Some filters generate output images with additional padding depending
8248 on the respective filter kernel. The padding is automatically removed
8249 to ensure the filter output has the same size as the input image.
8250 For image generators, the size of the output image is determined by the
8252 previous output image of the filter chain or the input image of the
8253 whole filterchain, respectively. The generators do not use the pixel
8254 information of this image to generate their output. However, the
8255 generated output is blended onto this image, resulting in partial or
8256 complete coverage of the output image.
8257 The coreimagesrc video source can be used for generating input images
8259 which are directly fed into the filter chain. By using it, providing
8260 input images by another video source or an input video is not required.
8261 Examples
8263 • List all filters available:
8265 coreimage=list_filters=true
8267 • Use the CIBoxBlur filter with default options to blur an image:
8269 coreimage=filter=CIBoxBlur@default
8271 • Use a filter chain with CISepiaTone at default values and
8273 CIVignetteEffect with its center at 100x100 and a radius of 50
8274 pixels:
8275 coreimage=filter=CIBoxBlur@default#CIVignetteEffect@inputCenter=100\ 100@inputRadius=50
8277 • Use nullsrc and CIQRCodeGenerator to create a QR code for the
8279 FFmpeg homepage, given as complete and escaped command-line for
8280 Apple's standard bash shell:
8281 ffmpeg -f lavfi -i nullsrc=s=100x100,coreimage=filter=CIQRCodeGenerator@inputMessage=https\\\\\://FFmpeg.org/@inputCorrectionLevel=H -frames:v 1 QRCode.png
8283 cover_rect
8285 Cover a rectangular object
8286 It accepts the following options:
8288 cover
8290 Filepath of the optional cover image, needs to be in yuv420.
8291 mode
8293 Set covering mode.
8294 It accepts the following values:
8296 cover
8298 cover it by the supplied image
8299 blur
8301 cover it by interpolating the surrounding pixels
8302 Default value is blur.
8304 Examples
8306 • Cover a rectangular object by the supplied image of a given video
8308 using ffmpeg:
8309 ffmpeg -i file.ts -vf find_rect=newref.pgm,cover_rect=cover.jpg:mode=cover new.mkv
8311 crop
8313 Crop the input video to given dimensions.
8314 It accepts the following parameters:
8316 w, out_w
8318 The width of the output video. It defaults to "iw". This
8319 expression is evaluated only once during the filter configuration,
8320 or when the w or out_w command is sent.
8321 h, out_h
8323 The height of the output video. It defaults to "ih". This
8324 expression is evaluated only once during the filter configuration,
8325 or when the h or out_h command is sent.
8326 x The horizontal position, in the input video, of the left edge of
8328 the output video. It defaults to "(in_w-out_w)/2". This expression
8329 is evaluated per-frame.
8330 y The vertical position, in the input video, of the top edge of the
8332 output video. It defaults to "(in_h-out_h)/2". This expression is
8333 evaluated per-frame.
8334 keep_aspect
8336 If set to 1 will force the output display aspect ratio to be the
8337 same of the input, by changing the output sample aspect ratio. It
8338 defaults to 0.
8339 exact
8341 Enable exact cropping. If enabled, subsampled videos will be
8342 cropped at exact width/height/x/y as specified and will not be
8343 rounded to nearest smaller value. It defaults to 0.
8344 The out_w, out_h, x, y parameters are expressions containing the
8346 following constants:
8347 x
8349 y The computed values for x and y. They are evaluated for each new
8350 frame.
8351 in_w
8353 in_h
8354 The input width and height.
8355 iw
8357 ih These are the same as in_w and in_h.
8358 out_w
8360 out_h
8361 The output (cropped) width and height.
8362 ow
8364 oh These are the same as out_w and out_h.
8365 a same as iw / ih
8367 sar input sample aspect ratio
8369 dar input display aspect ratio, it is the same as (iw / ih) * sar
8371 hsub
8373 vsub
8374 horizontal and vertical chroma subsample values. For example for
8375 the pixel format "yuv422p" hsub is 2 and vsub is 1.
8376 n The number of the input frame, starting from 0.
8378 pos the position in the file of the input frame, NAN if unknown
8380 t The timestamp expressed in seconds. It's NAN if the input timestamp
8382 is unknown.
8383 The expression for out_w may depend on the value of out_h, and the
8385 expression for out_h may depend on out_w, but they cannot depend on x
8386 and y, as x and y are evaluated after out_w and out_h.
8387 The x and y parameters specify the expressions for the position of the
8389 top-left corner of the output (non-cropped) area. They are evaluated
8390 for each frame. If the evaluated value is not valid, it is approximated
8391 to the nearest valid value.
8392 The expression for x may depend on y, and the expression for y may
8394 depend on x.
8395 Examples
8397 • Crop area with size 100x100 at position (12,34).
8399 crop=100:100:12:34
8401 Using named options, the example above becomes:
8403 crop=w=100:h=100:x=12:y=34
8405 • Crop the central input area with size 100x100:
8407 crop=100:100
8409 • Crop the central input area with size 2/3 of the input video:
8411 crop=2/3*in_w:2/3*in_h
8413 • Crop the input video central square:
8415 crop=out_w=in_h
8417 crop=in_h
8418 • Delimit the rectangle with the top-left corner placed at position
8420 100:100 and the right-bottom corner corresponding to the right-
8421 bottom corner of the input image.
8422 crop=in_w-100:in_h-100:100:100
8424 • Crop 10 pixels from the left and right borders, and 20 pixels from
8426 the top and bottom borders
8427 crop=in_w-2*10:in_h-2*20
8429 • Keep only the bottom right quarter of the input image:
8431 crop=in_w/2:in_h/2:in_w/2:in_h/2
8433 • Crop height for getting Greek harmony:
8435 crop=in_w:1/PHI*in_w
8437 • Apply trembling effect:
8439 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)
8441 • Apply erratic camera effect depending on timestamp:
8443 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)"
8445 • Set x depending on the value of y:
8447 crop=in_w/2:in_h/2:y:10+10*sin(n/10)
8449 Commands
8451 This filter supports the following commands:
8453 w, out_w
8455 h, out_h
8456 x
8457 y Set width/height of the output video and the horizontal/vertical
8458 position in the input video. The command accepts the same syntax
8459 of the corresponding option.
8460 If the specified expression is not valid, it is kept at its current
8462 value.
8463 cropdetect
8465 Auto-detect the crop size.
8466 It calculates the necessary cropping parameters and prints the
8468 recommended parameters via the logging system. The detected dimensions
8469 correspond to the non-black area of the input video.
8470 It accepts the following parameters:
8472 limit
8474 Set higher black value threshold, which can be optionally specified
8475 from nothing (0) to everything (255 for 8-bit based formats). An
8476 intensity value greater to the set value is considered non-black.
8477 It defaults to 24. You can also specify a value between 0.0 and
8478 1.0 which will be scaled depending on the bitdepth of the pixel
8479 format.
8480 round
8482 The value which the width/height should be divisible by. It
8483 defaults to 16. The offset is automatically adjusted to center the
8484 video. Use 2 to get only even dimensions (needed for 4:2:2 video).
8485 16 is best when encoding to most video codecs.
8486 skip
8488 Set the number of initial frames for which evaluation is skipped.
8489 Default is 2. Range is 0 to INT_MAX.
8490 reset_count, reset
8492 Set the counter that determines after how many frames cropdetect
8493 will reset the previously detected largest video area and start
8494 over to detect the current optimal crop area. Default value is 0.
8495 This can be useful when channel logos distort the video area. 0
8497 indicates 'never reset', and returns the largest area encountered
8498 during playback.
8499 cue
8501 Delay video filtering until a given wallclock timestamp. The filter
8502 first passes on preroll amount of frames, then it buffers at most
8503 buffer amount of frames and waits for the cue. After reaching the cue
8504 it forwards the buffered frames and also any subsequent frames coming
8505 in its input.
8506 The filter can be used synchronize the output of multiple ffmpeg
8508 processes for realtime output devices like decklink. By putting the
8509 delay in the filtering chain and pre-buffering frames the process can
8510 pass on data to output almost immediately after the target wallclock
8511 timestamp is reached.
8512 Perfect frame accuracy cannot be guaranteed, but the result is good
8514 enough for some use cases.
8515 cue The cue timestamp expressed in a UNIX timestamp in microseconds.
8517 Default is 0.
8518 preroll
8520 The duration of content to pass on as preroll expressed in seconds.
8521 Default is 0.
8522 buffer
8524 The maximum duration of content to buffer before waiting for the
8525 cue expressed in seconds. Default is 0.
8526 curves
8528 Apply color adjustments using curves.
8529 This filter is similar to the Adobe Photoshop and GIMP curves tools.
8531 Each component (red, green and blue) has its values defined by N key
8532 points tied from each other using a smooth curve. The x-axis represents
8533 the pixel values from the input frame, and the y-axis the new pixel
8534 values to be set for the output frame.
8535 By default, a component curve is defined by the two points (0;0) and
8537 (1;1). This creates a straight line where each original pixel value is
8538 "adjusted" to its own value, which means no change to the image.
8539 The filter allows you to redefine these two points and add some more. A
8541 new curve (using a natural cubic spline interpolation) will be define
8542 to pass smoothly through all these new coordinates. The new defined
8543 points needs to be strictly increasing over the x-axis, and their x and
8544 y values must be in the [0;1] interval. If the computed curves
8545 happened to go outside the vector spaces, the values will be clipped
8546 accordingly.
8547 The filter accepts the following options:
8549 preset
8551 Select one of the available color presets. This option can be used
8552 in addition to the r, g, b parameters; in this case, the later
8553 options takes priority on the preset values. Available presets
8554 are:
8555 none
8557 color_negative
8558 cross_process
8559 darker
8560 increase_contrast
8561 lighter
8562 linear_contrast
8563 medium_contrast
8564 negative
8565 strong_contrast
8566 vintage
8567 Default is "none".
8569 master, m
8571 Set the master key points. These points will define a second pass
8572 mapping. It is sometimes called a "luminance" or "value" mapping.
8573 It can be used with r, g, b or all since it acts like a post-
8574 processing LUT.
8575 red, r
8577 Set the key points for the red component.
8578 green, g
8580 Set the key points for the green component.
8581 blue, b
8583 Set the key points for the blue component.
8584 all Set the key points for all components (not including master). Can
8586 be used in addition to the other key points component options. In
8587 this case, the unset component(s) will fallback on this all
8588 setting.
8589 psfile
8591 Specify a Photoshop curves file (".acv") to import the settings
8592 from.
8593 plot
8595 Save Gnuplot script of the curves in specified file.
8596 To avoid some filtergraph syntax conflicts, each key points list need
8598 to be defined using the following syntax: "x0/y0 x1/y1 x2/y2 ...".
8599 Commands
8601 This filter supports same commands as options.
8603 Examples
8605 • Increase slightly the middle level of blue:
8607 curves=blue='0/0 0.5/0.58 1/1'
8609 • Vintage effect:
8611 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'
8613 Here we obtain the following coordinates for each components:
8615 red "(0;0.11) (0.42;0.51) (1;0.95)"
8617 green
8619 "(0;0) (0.50;0.48) (1;1)"
8620 blue
8622 "(0;0.22) (0.49;0.44) (1;0.80)"
8623 • The previous example can also be achieved with the associated
8625 built-in preset:
8626 curves=preset=vintage
8628 • Or simply:
8630 curves=vintage
8632 • Use a Photoshop preset and redefine the points of the green
8634 component:
8635 curves=psfile='MyCurvesPresets/purple.acv':green='0/0 0.45/0.53 1/1'
8637 • Check out the curves of the "cross_process" profile using ffmpeg
8639 and gnuplot:
8640 ffmpeg -f lavfi -i color -vf curves=cross_process:plot=/tmp/curves.plt -frames:v 1 -f null -
8642 gnuplot -p /tmp/curves.plt
8643 datascope
8645 Video data analysis filter.
8646 This filter shows hexadecimal pixel values of part of video.
8648 The filter accepts the following options:
8650 size, s
8652 Set output video size.
8653 x Set x offset from where to pick pixels.
8655 y Set y offset from where to pick pixels.
8657 mode
8659 Set scope mode, can be one of the following:
8660 mono
8662 Draw hexadecimal pixel values with white color on black
8663 background.
8664 color
8666 Draw hexadecimal pixel values with input video pixel color on
8667 black background.
8668 color2
8670 Draw hexadecimal pixel values on color background picked from
8671 input video, the text color is picked in such way so its always
8672 visible.
8673 axis
8675 Draw rows and columns numbers on left and top of video.
8676 opacity
8678 Set background opacity.
8679 format
8681 Set display number format. Can be "hex", or "dec". Default is
8682 "hex".
8683 components
8685 Set pixel components to display. By default all pixel components
8686 are displayed.
8687 Commands
8689 This filter supports same commands as options excluding "size" option.
8691 dblur
8693 Apply Directional blur filter.
8694 The filter accepts the following options:
8696 angle
8698 Set angle of directional blur. Default is 45.
8699 radius
8701 Set radius of directional blur. Default is 5.
8702 planes
8704 Set which planes to filter. By default all planes are filtered.
8705 Commands
8707 This filter supports same commands as options. The command accepts the
8709 same syntax of the corresponding option.
8710 If the specified expression is not valid, it is kept at its current
8712 value.
8713 dctdnoiz
8715 Denoise frames using 2D DCT (frequency domain filtering).
8716 This filter is not designed for real time.
8718 The filter accepts the following options:
8720 sigma, s
8722 Set the noise sigma constant.
8723 This sigma defines a hard threshold of "3 * sigma"; every DCT
8725 coefficient (absolute value) below this threshold with be dropped.
8726 If you need a more advanced filtering, see expr.
8728 Default is 0.
8730 overlap
8732 Set number overlapping pixels for each block. Since the filter can
8733 be slow, you may want to reduce this value, at the cost of a less
8734 effective filter and the risk of various artefacts.
8735 If the overlapping value doesn't permit processing the whole input
8737 width or height, a warning will be displayed and according borders
8738 won't be denoised.
8739 Default value is blocksize-1, which is the best possible setting.
8741 expr, e
8743 Set the coefficient factor expression.
8744 For each coefficient of a DCT block, this expression will be
8746 evaluated as a multiplier value for the coefficient.
8747 If this is option is set, the sigma option will be ignored.
8749 The absolute value of the coefficient can be accessed through the c
8751 variable.
8752 n Set the blocksize using the number of bits. "1<<n" defines the
8754 blocksize, which is the width and height of the processed blocks.
8755 The default value is 3 (8x8) and can be raised to 4 for a blocksize
8757 of 16x16. Note that changing this setting has huge consequences on
8758 the speed processing. Also, a larger block size does not
8759 necessarily means a better de-noising.
8760 Examples
8762 Apply a denoise with a sigma of 4.5:
8764 dctdnoiz=4.5
8766 The same operation can be achieved using the expression system:
8768 dctdnoiz=e='gte(c, 4.5*3)'
8770 Violent denoise using a block size of "16x16":
8772 dctdnoiz=15:n=4
8774 deband
8776 Remove banding artifacts from input video. It works by replacing
8777 banded pixels with average value of referenced pixels.
8778 The filter accepts the following options:
8780 1thr
8782 2thr
8783 3thr
8784 4thr
8785 Set banding detection threshold for each plane. Default is 0.02.
8786 Valid range is 0.00003 to 0.5. If difference between current pixel
8787 and reference pixel is less than threshold, it will be considered
8788 as banded.
8789 range, r
8791 Banding detection range in pixels. Default is 16. If positive,
8792 random number in range 0 to set value will be used. If negative,
8793 exact absolute value will be used. The range defines square of
8794 four pixels around current pixel.
8795 direction, d
8797 Set direction in radians from which four pixel will be compared. If
8798 positive, random direction from 0 to set direction will be picked.
8799 If negative, exact of absolute value will be picked. For example
8800 direction 0, -PI or -2*PI radians will pick only pixels on same row
8801 and -PI/2 will pick only pixels on same column.
8802 blur, b
8804 If enabled, current pixel is compared with average value of all
8805 four surrounding pixels. The default is enabled. If disabled
8806 current pixel is compared with all four surrounding pixels. The
8807 pixel is considered banded if only all four differences with
8808 surrounding pixels are less than threshold.
8809 coupling, c
8811 If enabled, current pixel is changed if and only if all pixel
8812 components are banded, e.g. banding detection threshold is
8813 triggered for all color components. The default is disabled.
8814 Commands
8816 This filter supports the all above options as commands.
8818 deblock
8820 Remove blocking artifacts from input video.
8821 The filter accepts the following options:
8823 filter
8825 Set filter type, can be weak or strong. Default is strong. This
8826 controls what kind of deblocking is applied.
8827 block
8829 Set size of block, allowed range is from 4 to 512. Default is 8.
8830 alpha
8832 beta
8833 gamma
8834 delta
8835 Set blocking detection thresholds. Allowed range is 0 to 1.
8836 Defaults are: 0.098 for alpha and 0.05 for the rest. Using higher
8837 threshold gives more deblocking strength. Setting alpha controls
8838 threshold detection at exact edge of block. Remaining options
8839 controls threshold detection near the edge. Each one for
8840 below/above or left/right. Setting any of those to 0 disables
8841 deblocking.
8842 planes
8844 Set planes to filter. Default is to filter all available planes.
8845 Examples
8847 • Deblock using weak filter and block size of 4 pixels.
8849 deblock=filter=weak:block=4
8851 • Deblock using strong filter, block size of 4 pixels and custom
8853 thresholds for deblocking more edges.
8854 deblock=filter=strong:block=4:alpha=0.12:beta=0.07:gamma=0.06:delta=0.05
8856 • Similar as above, but filter only first plane.
8858 deblock=filter=strong:block=4:alpha=0.12:beta=0.07:gamma=0.06:delta=0.05:planes=1
8860 • Similar as above, but filter only second and third plane.
8862 deblock=filter=strong:block=4:alpha=0.12:beta=0.07:gamma=0.06:delta=0.05:planes=6
8864 Commands
8866 This filter supports the all above options as commands.
8868 decimate
8870 Drop duplicated frames at regular intervals.
8871 The filter accepts the following options:
8873 cycle
8875 Set the number of frames from which one will be dropped. Setting
8876 this to N means one frame in every batch of N frames will be
8877 dropped. Default is 5.
8878 dupthresh
8880 Set the threshold for duplicate detection. If the difference metric
8881 for a frame is less than or equal to this value, then it is
8882 declared as duplicate. Default is 1.1
8883 scthresh
8885 Set scene change threshold. Default is 15.
8886 blockx
8888 blocky
8889 Set the size of the x and y-axis blocks used during metric
8890 calculations. Larger blocks give better noise suppression, but
8891 also give worse detection of small movements. Must be a power of
8892 two. Default is 32.
8893 ppsrc
8895 Mark main input as a pre-processed input and activate clean source
8896 input stream. This allows the input to be pre-processed with
8897 various filters to help the metrics calculation while keeping the
8898 frame selection lossless. When set to 1, the first stream is for
8899 the pre-processed input, and the second stream is the clean source
8900 from where the kept frames are chosen. Default is 0.
8901 chroma
8903 Set whether or not chroma is considered in the metric calculations.
8904 Default is 1.
8905 deconvolve
8907 Apply 2D deconvolution of video stream in frequency domain using second
8908 stream as impulse.
8909 The filter accepts the following options:
8911 planes
8913 Set which planes to process.
8914 impulse
8916 Set which impulse video frames will be processed, can be first or
8917 all. Default is all.
8918 noise
8920 Set noise when doing divisions. Default is 0.0000001. Useful when
8921 width and height are not same and not power of 2 or if stream prior
8922 to convolving had noise.
8923 The "deconvolve" filter also supports the framesync options.
8925 dedot
8927 Reduce cross-luminance (dot-crawl) and cross-color (rainbows) from
8928 video.
8929 It accepts the following options:
8931 m Set mode of operation. Can be combination of dotcrawl for cross-
8933 luminance reduction and/or rainbows for cross-color reduction.
8934 lt Set spatial luma threshold. Lower values increases reduction of
8936 cross-luminance.
8937 tl Set tolerance for temporal luma. Higher values increases reduction
8939 of cross-luminance.
8940 tc Set tolerance for chroma temporal variation. Higher values
8942 increases reduction of cross-color.
8943 ct Set temporal chroma threshold. Lower values increases reduction of
8945 cross-color.
8946 deflate
8948 Apply deflate effect to the video.
8949 This filter replaces the pixel by the local(3x3) average by taking into
8951 account only values lower than the pixel.
8952 It accepts the following options:
8954 threshold0
8956 threshold1
8957 threshold2
8958 threshold3
8959 Limit the maximum change for each plane, default is 65535. If 0,
8960 plane will remain unchanged.
8961 Commands
8963 This filter supports the all above options as commands.
8965 deflicker
8967 Remove temporal frame luminance variations.
8968 It accepts the following options:
8970 size, s
8972 Set moving-average filter size in frames. Default is 5. Allowed
8973 range is 2 - 129.
8974 mode, m
8976 Set averaging mode to smooth temporal luminance variations.
8977 Available values are:
8979 am Arithmetic mean
8981 gm Geometric mean
8983 hm Harmonic mean
8985 qm Quadratic mean
8987 cm Cubic mean
8989 pm Power mean
8991 median
8993 Median
8994 bypass
8996 Do not actually modify frame. Useful when one only wants metadata.
8997 dejudder
8999 Remove judder produced by partially interlaced telecined content.
9000 Judder can be introduced, for instance, by pullup filter. If the
9002 original source was partially telecined content then the output of
9003 "pullup,dejudder" will have a variable frame rate. May change the
9004 recorded frame rate of the container. Aside from that change, this
9005 filter will not affect constant frame rate video.
9006 The option available in this filter is:
9008 cycle
9010 Specify the length of the window over which the judder repeats.
9011 Accepts any integer greater than 1. Useful values are:
9013 4 If the original was telecined from 24 to 30 fps (Film to NTSC).
9015 5 If the original was telecined from 25 to 30 fps (PAL to NTSC).
9017 20 If a mixture of the two.
9019 The default is 4.
9021 delogo
9023 Suppress a TV station logo by a simple interpolation of the surrounding
9024 pixels. Just set a rectangle covering the logo and watch it disappear
9025 (and sometimes something even uglier appear - your mileage may vary).
9026 It accepts the following parameters:
9028 x
9030 y Specify the top left corner coordinates of the logo. They must be
9031 specified.
9032 w
9034 h Specify the width and height of the logo to clear. They must be
9035 specified.
9036 show
9038 When set to 1, a green rectangle is drawn on the screen to simplify
9039 finding the right x, y, w, and h parameters. The default value is
9040 0.
9041 The rectangle is drawn on the outermost pixels which will be
9043 (partly) replaced with interpolated values. The values of the next
9044 pixels immediately outside this rectangle in each direction will be
9045 used to compute the interpolated pixel values inside the rectangle.
9046 Examples
9048 • Set a rectangle covering the area with top left corner coordinates
9050 0,0 and size 100x77:
9051 delogo=x=0:y=0:w=100:h=77
9053 derain
9055 Remove the rain in the input image/video by applying the derain methods
9056 based on convolutional neural networks. Supported models:
9057 • Recurrent Squeeze-and-Excitation Context Aggregation Net (RESCAN).
9059 See
9060 <http://openaccess.thecvf.com/content_ECCV_2018/papers/Xia_Li_Recurrent_Squeeze-and-Excitation_Context_ECCV_2018_paper.pdf>.
9061 Training as well as model generation scripts are provided in the
9063 repository at <https://github.com/XueweiMeng/derain_filter.git>.
9064 Native model files (.model) can be generated from TensorFlow model
9066 files (.pb) by using tools/python/convert.py
9067 The filter accepts the following options:
9069 filter_type
9071 Specify which filter to use. This option accepts the following
9072 values:
9073 derain
9075 Derain filter. To conduct derain filter, you need to use a
9076 derain model.
9077 dehaze
9079 Dehaze filter. To conduct dehaze filter, you need to use a
9080 dehaze model.
9081 Default value is derain.
9083 dnn_backend
9085 Specify which DNN backend to use for model loading and execution.
9086 This option accepts the following values:
9087 native
9089 Native implementation of DNN loading and execution.
9090 tensorflow
9092 TensorFlow backend. To enable this backend you need to install
9093 the TensorFlow for C library (see
9094 <https://www.tensorflow.org/install/lang_c>) and configure
9095 FFmpeg with "--enable-libtensorflow"
9096 Default value is native.
9098 model
9100 Set path to model file specifying network architecture and its
9101 parameters. Note that different backends use different file
9102 formats. TensorFlow and native backend can load files for only its
9103 format.
9104 To get full functionality (such as async execution), please use the
9106 dnn_processing filter.
9107 deshake
9109 Attempt to fix small changes in horizontal and/or vertical shift. This
9110 filter helps remove camera shake from hand-holding a camera, bumping a
9111 tripod, moving on a vehicle, etc.
9112 The filter accepts the following options:
9114 x
9116 y
9117 w
9118 h Specify a rectangular area where to limit the search for motion
9119 vectors. If desired the search for motion vectors can be limited
9120 to a rectangular area of the frame defined by its top left corner,
9121 width and height. These parameters have the same meaning as the
9122 drawbox filter which can be used to visualise the position of the
9123 bounding box.
9124 This is useful when simultaneous movement of subjects within the
9126 frame might be confused for camera motion by the motion vector
9127 search.
9128 If any or all of x, y, w and h are set to -1 then the full frame is
9130 used. This allows later options to be set without specifying the
9131 bounding box for the motion vector search.
9132 Default - search the whole frame.
9134 rx
9136 ry Specify the maximum extent of movement in x and y directions in the
9137 range 0-64 pixels. Default 16.
9138 edge
9140 Specify how to generate pixels to fill blanks at the edge of the
9141 frame. Available values are:
9142 blank, 0
9144 Fill zeroes at blank locations
9145 original, 1
9147 Original image at blank locations
9148 clamp, 2
9150 Extruded edge value at blank locations
9151 mirror, 3
9153 Mirrored edge at blank locations
9154 Default value is mirror.
9156 blocksize
9158 Specify the blocksize to use for motion search. Range 4-128 pixels,
9159 default 8.
9160 contrast
9162 Specify the contrast threshold for blocks. Only blocks with more
9163 than the specified contrast (difference between darkest and
9164 lightest pixels) will be considered. Range 1-255, default 125.
9165 search
9167 Specify the search strategy. Available values are:
9168 exhaustive, 0
9170 Set exhaustive search
9171 less, 1
9173 Set less exhaustive search.
9174 Default value is exhaustive.
9176 filename
9178 If set then a detailed log of the motion search is written to the
9179 specified file.
9180 despill
9182 Remove unwanted contamination of foreground colors, caused by reflected
9183 color of greenscreen or bluescreen.
9184 This filter accepts the following options:
9186 type
9188 Set what type of despill to use.
9189 mix Set how spillmap will be generated.
9191 expand
9193 Set how much to get rid of still remaining spill.
9194 red Controls amount of red in spill area.
9196 green
9198 Controls amount of green in spill area. Should be -1 for
9199 greenscreen.
9200 blue
9202 Controls amount of blue in spill area. Should be -1 for
9203 bluescreen.
9204 brightness
9206 Controls brightness of spill area, preserving colors.
9207 alpha
9209 Modify alpha from generated spillmap.
9210 Commands
9212 This filter supports the all above options as commands.
9214 detelecine
9216 Apply an exact inverse of the telecine operation. It requires a
9217 predefined pattern specified using the pattern option which must be the
9218 same as that passed to the telecine filter.
9219 This filter accepts the following options:
9221 first_field
9223 top, t
9224 top field first
9225 bottom, b
9227 bottom field first The default value is "top".
9228 pattern
9230 A string of numbers representing the pulldown pattern you wish to
9231 apply. The default value is 23.
9232 start_frame
9234 A number representing position of the first frame with respect to
9235 the telecine pattern. This is to be used if the stream is cut. The
9236 default value is 0.
9237 dilation
9239 Apply dilation effect to the video.
9240 This filter replaces the pixel by the local(3x3) maximum.
9242 It accepts the following options:
9244 threshold0
9246 threshold1
9247 threshold2
9248 threshold3
9249 Limit the maximum change for each plane, default is 65535. If 0,
9250 plane will remain unchanged.
9251 coordinates
9253 Flag which specifies the pixel to refer to. Default is 255 i.e. all
9254 eight pixels are used.
9255 Flags to local 3x3 coordinates maps like this:
9257 1 2 3
9259 4 5
9260 6 7 8
9261 Commands
9263 This filter supports the all above options as commands.
9265 displace
9267 Displace pixels as indicated by second and third input stream.
9268 It takes three input streams and outputs one stream, the first input is
9270 the source, and second and third input are displacement maps.
9271 The second input specifies how much to displace pixels along the
9273 x-axis, while the third input specifies how much to displace pixels
9274 along the y-axis. If one of displacement map streams terminates, last
9275 frame from that displacement map will be used.
9276 Note that once generated, displacements maps can be reused over and
9278 over again.
9279 A description of the accepted options follows.
9281 edge
9283 Set displace behavior for pixels that are out of range.
9284 Available values are:
9286 blank
9288 Missing pixels are replaced by black pixels.
9289 smear
9291 Adjacent pixels will spread out to replace missing pixels.
9292 wrap
9294 Out of range pixels are wrapped so they point to pixels of
9295 other side.
9296 mirror
9298 Out of range pixels will be replaced with mirrored pixels.
9299 Default is smear.
9301 Examples
9303 • Add ripple effect to rgb input of video size hd720:
9305 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
9307 • Add wave effect to rgb input of video size hd720:
9309 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
9311 dnn_classify
9313 Do classification with deep neural networks based on bounding boxes.
9314 The filter accepts the following options:
9316 dnn_backend
9318 Specify which DNN backend to use for model loading and execution.
9319 This option accepts only openvino now, tensorflow backends will be
9320 added.
9321 model
9323 Set path to model file specifying network architecture and its
9324 parameters. Note that different backends use different file
9325 formats.
9326 input
9328 Set the input name of the dnn network.
9329 output
9331 Set the output name of the dnn network.
9332 confidence
9334 Set the confidence threshold (default: 0.5).
9335 labels
9337 Set path to label file specifying the mapping between label id and
9338 name. Each label name is written in one line, tailing spaces and
9339 empty lines are skipped. The first line is the name of label id 0,
9340 and the second line is the name of label id 1, etc. The label id
9341 is considered as name if the label file is not provided.
9342 backend_configs
9344 Set the configs to be passed into backend
9345 For tensorflow backend, you can set its configs with sess_config
9347 options, please use tools/python/tf_sess_config.py to get the
9348 configs for your system.
9349 dnn_detect
9351 Do object detection with deep neural networks.
9352 The filter accepts the following options:
9354 dnn_backend
9356 Specify which DNN backend to use for model loading and execution.
9357 This option accepts only openvino now, tensorflow backends will be
9358 added.
9359 model
9361 Set path to model file specifying network architecture and its
9362 parameters. Note that different backends use different file
9363 formats.
9364 input
9366 Set the input name of the dnn network.
9367 output
9369 Set the output name of the dnn network.
9370 confidence
9372 Set the confidence threshold (default: 0.5).
9373 labels
9375 Set path to label file specifying the mapping between label id and
9376 name. Each label name is written in one line, tailing spaces and
9377 empty lines are skipped. The first line is the name of label id 0
9378 (usually it is 'background'), and the second line is the name of
9379 label id 1, etc. The label id is considered as name if the label
9380 file is not provided.
9381 backend_configs
9383 Set the configs to be passed into backend. To use async execution,
9384 set async (default: set). Roll back to sync execution if the
9385 backend does not support async.
9386 dnn_processing
9388 Do image processing with deep neural networks. It works together with
9389 another filter which converts the pixel format of the Frame to what the
9390 dnn network requires.
9391 The filter accepts the following options:
9393 dnn_backend
9395 Specify which DNN backend to use for model loading and execution.
9396 This option accepts the following values:
9397 native
9399 Native implementation of DNN loading and execution.
9400 tensorflow
9402 TensorFlow backend. To enable this backend you need to install
9403 the TensorFlow for C library (see
9404 <https://www.tensorflow.org/install/lang_c>) and configure
9405 FFmpeg with "--enable-libtensorflow"
9406 openvino
9408 OpenVINO backend. To enable this backend you need to build and
9409 install the OpenVINO for C library (see
9410 <https://github.com/openvinotoolkit/openvino/blob/master/build-instruction.md>)
9411 and configure FFmpeg with "--enable-libopenvino"
9412 (--extra-cflags=-I... --extra-ldflags=-L... might be needed if
9413 the header files and libraries are not installed into system
9414 path)
9415 Default value is native.
9417 model
9419 Set path to model file specifying network architecture and its
9420 parameters. Note that different backends use different file
9421 formats. TensorFlow, OpenVINO and native backend can load files for
9422 only its format.
9423 Native model file (.model) can be generated from TensorFlow model
9425 file (.pb) by using tools/python/convert.py
9426 input
9428 Set the input name of the dnn network.
9429 output
9431 Set the output name of the dnn network.
9432 backend_configs
9434 Set the configs to be passed into backend. To use async execution,
9435 set async (default: set). Roll back to sync execution if the
9436 backend does not support async.
9437 For tensorflow backend, you can set its configs with sess_config
9439 options, please use tools/python/tf_sess_config.py to get the
9440 configs of TensorFlow backend for your system.
9441 Examples
9443 • Remove rain in rgb24 frame with can.pb (see derain filter):
9445 ./ffmpeg -i rain.jpg -vf format=rgb24,dnn_processing=dnn_backend=tensorflow:model=can.pb:input=x:output=y derain.jpg
9447 • Halve the pixel value of the frame with format gray32f:
9449 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
9451 • Handle the Y channel with srcnn.pb (see sr filter) for frame with
9453 yuv420p (planar YUV formats supported):
9454 ./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
9456 • Handle the Y channel with espcn.pb (see sr filter), which changes
9458 frame size, for format yuv420p (planar YUV formats supported),
9459 please use tools/python/tf_sess_config.py to get the configs of
9460 TensorFlow backend for your system.
9461 ./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
9463 drawbox
9465 Draw a colored box on the input image.
9466 It accepts the following parameters:
9468 x
9470 y The expressions which specify the top left corner coordinates of
9471 the box. It defaults to 0.
9472 width, w
9474 height, h
9475 The expressions which specify the width and height of the box; if 0
9476 they are interpreted as the input width and height. It defaults to
9477 0.
9478 color, c
9480 Specify the color of the box to write. For the general syntax of
9481 this option, check the "Color" section in the ffmpeg-utils manual.
9482 If the special value "invert" is used, the box edge color is the
9483 same as the video with inverted luma.
9484 thickness, t
9486 The expression which sets the thickness of the box edge. A value
9487 of "fill" will create a filled box. Default value is 3.
9488 See below for the list of accepted constants.
9490 replace
9492 Applicable if the input has alpha. With value 1, the pixels of the
9493 painted box will overwrite the video's color and alpha pixels.
9494 Default is 0, which composites the box onto the input, leaving the
9495 video's alpha intact.
9496 The parameters for x, y, w and h and t are expressions containing the
9498 following constants:
9499 dar The input display aspect ratio, it is the same as (w / h) * sar.
9501 hsub
9503 vsub
9504 horizontal and vertical chroma subsample values. For example for
9505 the pixel format "yuv422p" hsub is 2 and vsub is 1.
9506 in_h, ih
9508 in_w, iw
9509 The input width and height.
9510 sar The input sample aspect ratio.
9512 x
9514 y The x and y offset coordinates where the box is drawn.
9515 w
9517 h The width and height of the drawn box.
9518 box_source
9520 Box source can be set as side_data_detection_bboxes if you want to
9521 use box data in detection bboxes of side data.
9522 If box_source is set, the x, y, width and height will be ignored
9524 and still use box data in detection bboxes of side data. So please
9525 do not use this parameter if you were not sure about the box
9526 source.
9527 t The thickness of the drawn box.
9529 These constants allow the x, y, w, h and t expressions to refer to
9531 each other, so you may for example specify "y=x/dar" or "h=w/dar".
9532 Examples
9534 • Draw a black box around the edge of the input image:
9536 drawbox
9538 • Draw a box with color red and an opacity of 50%:
9540 drawbox=10:20:200:60:red@0.5
9542 The previous example can be specified as:
9544 drawbox=x=10:y=20:w=200:h=60:color=red@0.5
9546 • Fill the box with pink color:
9548 drawbox=x=10:y=10:w=100:h=100:color=pink@0.5:t=fill
9550 • Draw a 2-pixel red 2.40:1 mask:
9552 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
9554 Commands
9556 This filter supports same commands as options. The command accepts the
9558 same syntax of the corresponding option.
9559 If the specified expression is not valid, it is kept at its current
9561 value.
9562 drawgraph
9564 Draw a graph using input video metadata.
9565 It accepts the following parameters:
9567 m1 Set 1st frame metadata key from which metadata values will be used
9569 to draw a graph.
9570 fg1 Set 1st foreground color expression.
9572 m2 Set 2nd frame metadata key from which metadata values will be used
9574 to draw a graph.
9575 fg2 Set 2nd foreground color expression.
9577 m3 Set 3rd frame metadata key from which metadata values will be used
9579 to draw a graph.
9580 fg3 Set 3rd foreground color expression.
9582 m4 Set 4th frame metadata key from which metadata values will be used
9584 to draw a graph.
9585 fg4 Set 4th foreground color expression.
9587 min Set minimal value of metadata value.
9589 max Set maximal value of metadata value.
9591 bg Set graph background color. Default is white.
9593 mode
9595 Set graph mode.
9596 Available values for mode is:
9598 bar
9600 dot
9601 line
9602 Default is "line".
9604 slide
9606 Set slide mode.
9607 Available values for slide is:
9609 frame
9611 Draw new frame when right border is reached.
9612 replace
9614 Replace old columns with new ones.
9615 scroll
9617 Scroll from right to left.
9618 rscroll
9620 Scroll from left to right.
9621 picture
9623 Draw single picture.
9624 Default is "frame".
9626 size
9628 Set size of graph video. For the syntax of this option, check the
9629 "Video size" section in the ffmpeg-utils manual. The default value
9630 is "900x256".
9631 rate, r
9633 Set the output frame rate. Default value is 25.
9634 The foreground color expressions can use the following variables:
9636 MIN Minimal value of metadata value.
9638 MAX Maximal value of metadata value.
9640 VAL Current metadata key value.
9642 The color is defined as 0xAABBGGRR.
9644 Example using metadata from signalstats filter:
9646 signalstats,drawgraph=lavfi.signalstats.YAVG:min=0:max=255
9648 Example using metadata from ebur128 filter:
9650 ebur128=metadata=1,adrawgraph=lavfi.r128.M:min=-120:max=5
9652 drawgrid
9654 Draw a grid on the input image.
9655 It accepts the following parameters:
9657 x
9659 y The expressions which specify the coordinates of some point of grid
9660 intersection (meant to configure offset). Both default to 0.
9661 width, w
9663 height, h
9664 The expressions which specify the width and height of the grid
9665 cell, if 0 they are interpreted as the input width and height,
9666 respectively, minus "thickness", so image gets framed. Default to
9667 0.
9668 color, c
9670 Specify the color of the grid. For the general syntax of this
9671 option, check the "Color" section in the ffmpeg-utils manual. If
9672 the special value "invert" is used, the grid color is the same as
9673 the video with inverted luma.
9674 thickness, t
9676 The expression which sets the thickness of the grid line. Default
9677 value is 1.
9678 See below for the list of accepted constants.
9680 replace
9682 Applicable if the input has alpha. With 1 the pixels of the painted
9683 grid will overwrite the video's color and alpha pixels. Default is
9684 0, which composites the grid onto the input, leaving the video's
9685 alpha intact.
9686 The parameters for x, y, w and h and t are expressions containing the
9688 following constants:
9689 dar The input display aspect ratio, it is the same as (w / h) * sar.
9691 hsub
9693 vsub
9694 horizontal and vertical chroma subsample values. For example for
9695 the pixel format "yuv422p" hsub is 2 and vsub is 1.
9696 in_h, ih
9698 in_w, iw
9699 The input grid cell width and height.
9700 sar The input sample aspect ratio.
9702 x
9704 y The x and y coordinates of some point of grid intersection (meant
9705 to configure offset).
9706 w
9708 h The width and height of the drawn cell.
9709 t The thickness of the drawn cell.
9711 These constants allow the x, y, w, h and t expressions to refer to
9713 each other, so you may for example specify "y=x/dar" or "h=w/dar".
9714 Examples
9716 • Draw a grid with cell 100x100 pixels, thickness 2 pixels, with
9718 color red and an opacity of 50%:
9719 drawgrid=width=100:height=100:thickness=2:color=red@0.5
9721 • Draw a white 3x3 grid with an opacity of 50%:
9723 drawgrid=w=iw/3:h=ih/3:t=2:c=white@0.5
9725 Commands
9727 This filter supports same commands as options. The command accepts the
9729 same syntax of the corresponding option.
9730 If the specified expression is not valid, it is kept at its current
9732 value.
9733 drawtext
9735 Draw a text string or text from a specified file on top of a video,
9736 using the libfreetype library.
9737 To enable compilation of this filter, you need to configure FFmpeg with
9739 "--enable-libfreetype". To enable default font fallback and the font
9740 option you need to configure FFmpeg with "--enable-libfontconfig". To
9741 enable the text_shaping option, you need to configure FFmpeg with
9742 "--enable-libfribidi".
9743 Syntax
9745 It accepts the following parameters:
9747 box Used to draw a box around text using the background color. The
9749 value must be either 1 (enable) or 0 (disable). The default value
9750 of box is 0.
9751 boxborderw
9753 Set the width of the border to be drawn around the box using
9754 boxcolor. The default value of boxborderw is 0.
9755 boxcolor
9757 The color to be used for drawing box around text. For the syntax of
9758 this option, check the "Color" section in the ffmpeg-utils manual.
9759 The default value of boxcolor is "white".
9761 line_spacing
9763 Set the line spacing in pixels of the border to be drawn around the
9764 box using box. The default value of line_spacing is 0.
9765 borderw
9767 Set the width of the border to be drawn around the text using
9768 bordercolor. The default value of borderw is 0.
9769 bordercolor
9771 Set the color to be used for drawing border around text. For the
9772 syntax of this option, check the "Color" section in the ffmpeg-
9773 utils manual.
9774 The default value of bordercolor is "black".
9776 expansion
9778 Select how the text is expanded. Can be either "none", "strftime"
9779 (deprecated) or "normal" (default). See the drawtext_expansion,
9780 Text expansion section below for details.
9781 basetime
9783 Set a start time for the count. Value is in microseconds. Only
9784 applied in the deprecated strftime expansion mode. To emulate in
9785 normal expansion mode use the "pts" function, supplying the start
9786 time (in seconds) as the second argument.
9787 fix_bounds
9789 If true, check and fix text coords to avoid clipping.
9790 fontcolor
9792 The color to be used for drawing fonts. For the syntax of this
9793 option, check the "Color" section in the ffmpeg-utils manual.
9794 The default value of fontcolor is "black".
9796 fontcolor_expr
9798 String which is expanded the same way as text to obtain dynamic
9799 fontcolor value. By default this option has empty value and is not
9800 processed. When this option is set, it overrides fontcolor option.
9801 font
9803 The font family to be used for drawing text. By default Sans.
9804 fontfile
9806 The font file to be used for drawing text. The path must be
9807 included. This parameter is mandatory if the fontconfig support is
9808 disabled.
9809 alpha
9811 Draw the text applying alpha blending. The value can be a number
9812 between 0.0 and 1.0. The expression accepts the same variables x,
9813 y as well. The default value is 1. Please see fontcolor_expr.
9814 fontsize
9816 The font size to be used for drawing text. The default value of
9817 fontsize is 16.
9818 text_shaping
9820 If set to 1, attempt to shape the text (for example, reverse the
9821 order of right-to-left text and join Arabic characters) before
9822 drawing it. Otherwise, just draw the text exactly as given. By
9823 default 1 (if supported).
9824 ft_load_flags
9826 The flags to be used for loading the fonts.
9827 The flags map the corresponding flags supported by libfreetype, and
9829 are a combination of the following values:
9830 default
9832 no_scale
9833 no_hinting
9834 render
9835 no_bitmap
9836 vertical_layout
9837 force_autohint
9838 crop_bitmap
9839 pedantic
9840 ignore_global_advance_width
9841 no_recurse
9842 ignore_transform
9843 monochrome
9844 linear_design
9845 no_autohint
9846 Default value is "default".
9848 For more information consult the documentation for the FT_LOAD_*
9850 libfreetype flags.
9851 shadowcolor
9853 The color to be used for drawing a shadow behind the drawn text.
9854 For the syntax of this option, check the "Color" section in the
9855 ffmpeg-utils manual.
9856 The default value of shadowcolor is "black".
9858 shadowx
9860 shadowy
9861 The x and y offsets for the text shadow position with respect to
9862 the position of the text. They can be either positive or negative
9863 values. The default value for both is "0".
9864 start_number
9866 The starting frame number for the n/frame_num variable. The default
9867 value is "0".
9868 tabsize
9870 The size in number of spaces to use for rendering the tab. Default
9871 value is 4.
9872 timecode
9874 Set the initial timecode representation in "hh:mm:ss[:;.]ff"
9875 format. It can be used with or without text parameter.
9876 timecode_rate option must be specified.
9877 timecode_rate, rate, r
9879 Set the timecode frame rate (timecode only). Value will be rounded
9880 to nearest integer. Minimum value is "1". Drop-frame timecode is
9881 supported for frame rates 30 & 60.
9882 tc24hmax
9884 If set to 1, the output of the timecode option will wrap around at
9885 24 hours. Default is 0 (disabled).
9886 text
9888 The text string to be drawn. The text must be a sequence of UTF-8
9889 encoded characters. This parameter is mandatory if no file is
9890 specified with the parameter textfile.
9891 textfile
9893 A text file containing text to be drawn. The text must be a
9894 sequence of UTF-8 encoded characters.
9895 This parameter is mandatory if no text string is specified with the
9897 parameter text.
9898 If both text and textfile are specified, an error is thrown.
9900 text_source
9902 Text source should be set as side_data_detection_bboxes if you want
9903 to use text data in detection bboxes of side data.
9904 If text source is set, text and textfile will be ignored and still
9906 use text data in detection bboxes of side data. So please do not
9907 use this parameter if you are not sure about the text source.
9908 reload
9910 If set to 1, the textfile will be reloaded before each frame. Be
9911 sure to update it atomically, or it may be read partially, or even
9912 fail.
9913 x
9915 y The expressions which specify the offsets where text will be drawn
9916 within the video frame. They are relative to the top/left border of
9917 the output image.
9918 The default value of x and y is "0".
9920 See below for the list of accepted constants and functions.
9922 The parameters for x and y are expressions containing the following
9924 constants and functions:
9925 dar input display aspect ratio, it is the same as (w / h) * sar
9927 hsub
9929 vsub
9930 horizontal and vertical chroma subsample values. For example for
9931 the pixel format "yuv422p" hsub is 2 and vsub is 1.
9932 line_h, lh
9934 the height of each text line
9935 main_h, h, H
9937 the input height
9938 main_w, w, W
9940 the input width
9941 max_glyph_a, ascent
9943 the maximum distance from the baseline to the highest/upper grid
9944 coordinate used to place a glyph outline point, for all the
9945 rendered glyphs. It is a positive value, due to the grid's
9946 orientation with the Y axis upwards.
9947 max_glyph_d, descent
9949 the maximum distance from the baseline to the lowest grid
9950 coordinate used to place a glyph outline point, for all the
9951 rendered glyphs. This is a negative value, due to the grid's
9952 orientation, with the Y axis upwards.
9953 max_glyph_h
9955 maximum glyph height, that is the maximum height for all the glyphs
9956 contained in the rendered text, it is equivalent to ascent -
9957 descent.
9958 max_glyph_w
9960 maximum glyph width, that is the maximum width for all the glyphs
9961 contained in the rendered text
9962 n the number of input frame, starting from 0
9964 rand(min, max)
9966 return a random number included between min and max
9967 sar The input sample aspect ratio.
9969 t timestamp expressed in seconds, NAN if the input timestamp is
9971 unknown
9972 text_h, th
9974 the height of the rendered text
9975 text_w, tw
9977 the width of the rendered text
9978 x
9980 y the x and y offset coordinates where the text is drawn.
9981 These parameters allow the x and y expressions to refer to each
9983 other, so you can for example specify "y=x/dar".
9984 pict_type
9986 A one character description of the current frame's picture type.
9987 pkt_pos
9989 The current packet's position in the input file or stream (in
9990 bytes, from the start of the input). A value of -1 indicates this
9991 info is not available.
9992 pkt_duration
9994 The current packet's duration, in seconds.
9995 pkt_size
9997 The current packet's size (in bytes).
9998 Text expansion
10000 If expansion is set to "strftime", the filter recognizes strftime()
10002 sequences in the provided text and expands them accordingly. Check the
10003 documentation of strftime(). This feature is deprecated.
10004 If expansion is set to "none", the text is printed verbatim.
10006 If expansion is set to "normal" (which is the default), the following
10008 expansion mechanism is used.
10009 The backslash character \, followed by any character, always expands to
10011 the second character.
10012 Sequences of the form "%{...}" are expanded. The text between the
10014 braces is a function name, possibly followed by arguments separated by
10015 ':'. If the arguments contain special characters or delimiters (':' or
10016 '}'), they should be escaped.
10017 Note that they probably must also be escaped as the value for the text
10019 option in the filter argument string and as the filter argument in the
10020 filtergraph description, and possibly also for the shell, that makes up
10021 to four levels of escaping; using a text file avoids these problems.
10022 The following functions are available:
10024 expr, e
10026 The expression evaluation result.
10027 It must take one argument specifying the expression to be
10029 evaluated, which accepts the same constants and functions as the x
10030 and y values. Note that not all constants should be used, for
10031 example the text size is not known when evaluating the expression,
10032 so the constants text_w and text_h will have an undefined value.
10033 expr_int_format, eif
10035 Evaluate the expression's value and output as formatted integer.
10036 The first argument is the expression to be evaluated, just as for
10038 the expr function. The second argument specifies the output
10039 format. Allowed values are x, X, d and u. They are treated exactly
10040 as in the "printf" function. The third parameter is optional and
10041 sets the number of positions taken by the output. It can be used
10042 to add padding with zeros from the left.
10043 gmtime
10045 The time at which the filter is running, expressed in UTC. It can
10046 accept an argument: a strftime() format string.
10047 localtime
10049 The time at which the filter is running, expressed in the local
10050 time zone. It can accept an argument: a strftime() format string.
10051 metadata
10053 Frame metadata. Takes one or two arguments.
10054 The first argument is mandatory and specifies the metadata key.
10056 The second argument is optional and specifies a default value, used
10058 when the metadata key is not found or empty.
10059 Available metadata can be identified by inspecting entries starting
10061 with TAG included within each frame section printed by running
10062 "ffprobe -show_frames".
10063 String metadata generated in filters leading to the drawtext filter
10065 are also available.
10066 n, frame_num
10068 The frame number, starting from 0.
10069 pict_type
10071 A one character description of the current picture type.
10072 pts The timestamp of the current frame. It can take up to three
10074 arguments.
10075 The first argument is the format of the timestamp; it defaults to
10077 "flt" for seconds as a decimal number with microsecond accuracy;
10078 "hms" stands for a formatted [-]HH:MM:SS.mmm timestamp with
10079 millisecond accuracy. "gmtime" stands for the timestamp of the
10080 frame formatted as UTC time; "localtime" stands for the timestamp
10081 of the frame formatted as local time zone time.
10082 The second argument is an offset added to the timestamp.
10084 If the format is set to "hms", a third argument "24HH" may be
10086 supplied to present the hour part of the formatted timestamp in 24h
10087 format (00-23).
10088 If the format is set to "localtime" or "gmtime", a third argument
10090 may be supplied: a strftime() format string. By default, YYYY-MM-
10091 DD HH:MM:SS format will be used.
10092 Commands
10094 This filter supports altering parameters via commands:
10096 reinit
10098 Alter existing filter parameters.
10099 Syntax for the argument is the same as for filter invocation, e.g.
10101 fontsize=56:fontcolor=green:text='Hello World'
10103 Full filter invocation with sendcmd would look like this:
10105 sendcmd=c='56.0 drawtext reinit fontsize=56\:fontcolor=green\:text=Hello\\ World'
10107 If the entire argument can't be parsed or applied as valid values then
10109 the filter will continue with its existing parameters.
10110 Examples
10112 • Draw "Test Text" with font FreeSerif, using the default values for
10114 the optional parameters.
10115 drawtext="fontfile=/usr/share/fonts/truetype/freefont/FreeSerif.ttf: text='Test Text'"
10117 • Draw 'Test Text' with font FreeSerif of size 24 at position x=100
10119 and y=50 (counting from the top-left corner of the screen), text is
10120 yellow with a red box around it. Both the text and the box have an
10121 opacity of 20%.
10122 drawtext="fontfile=/usr/share/fonts/truetype/freefont/FreeSerif.ttf: text='Test Text':\
10124 x=100: y=50: fontsize=24: fontcolor=yellow@0.2: box=1: boxcolor=red@0.2"
10125 Note that the double quotes are not necessary if spaces are not
10127 used within the parameter list.
10128 • Show the text at the center of the video frame:
10130 drawtext="fontsize=30:fontfile=FreeSerif.ttf:text='hello world':x=(w-text_w)/2:y=(h-text_h)/2"
10132 • Show the text at a random position, switching to a new position
10134 every 30 seconds:
10135 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)"
10137 • Show a text line sliding from right to left in the last row of the
10139 video frame. The file LONG_LINE is assumed to contain a single line
10140 with no newlines.
10141 drawtext="fontsize=15:fontfile=FreeSerif.ttf:text=LONG_LINE:y=h-line_h:x=-50*t"
10143 • Show the content of file CREDITS off the bottom of the frame and
10145 scroll up.
10146 drawtext="fontsize=20:fontfile=FreeSerif.ttf:textfile=CREDITS:y=h-20*t"
10148 • Draw a single green letter "g", at the center of the input video.
10150 The glyph baseline is placed at half screen height.
10151 drawtext="fontsize=60:fontfile=FreeSerif.ttf:fontcolor=green:text=g:x=(w-max_glyph_w)/2:y=h/2-ascent"
10153 • Show text for 1 second every 3 seconds:
10155 drawtext="fontfile=FreeSerif.ttf:fontcolor=white:x=100:y=x/dar:enable=lt(mod(t\,3)\,1):text='blink'"
10157 • Use fontconfig to set the font. Note that the colons need to be
10159 escaped.
10160 drawtext='fontfile=Linux Libertine O-40\:style=Semibold:text=FFmpeg'
10162 • Draw "Test Text" with font size dependent on height of the video.
10164 drawtext="text='Test Text': fontsize=h/30: x=(w-text_w)/2: y=(h-text_h*2)"
10166 • Print the date of a real-time encoding (see strftime(3)):
10168 drawtext='fontfile=FreeSans.ttf:text=%{localtime\:%a %b %d %Y}'
10170 • Show text fading in and out (appearing/disappearing):
10172 #!/bin/sh
10174 DS=1.0 # display start
10175 DE=10.0 # display end
10176 FID=1.5 # fade in duration
10177 FOD=5 # fade out duration
10178 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 }"
10179 • Horizontally align multiple separate texts. Note that max_glyph_a
10181 and the fontsize value are included in the y offset.
10182 drawtext=fontfile=FreeSans.ttf:text=DOG:fontsize=24:x=10:y=20+24-max_glyph_a,
10184 drawtext=fontfile=FreeSans.ttf:text=cow:fontsize=24:x=80:y=20+24-max_glyph_a
10185 • Plot special lavf.image2dec.source_basename metadata onto each
10187 frame if such metadata exists. Otherwise, plot the string "NA".
10188 Note that image2 demuxer must have option -export_path_metadata 1
10189 for the special metadata fields to be available for filters.
10190 drawtext="fontsize=20:fontcolor=white:fontfile=FreeSans.ttf:text='%{metadata\:lavf.image2dec.source_basename\:NA}':x=10:y=10"
10192 For more information about libfreetype, check:
10194 <http://www.freetype.org/>.
10195 For more information about fontconfig, check:
10197 <http://freedesktop.org/software/fontconfig/fontconfig-user.html>.
10198 For more information about libfribidi, check: <http://fribidi.org/>.
10200 edgedetect
10202 Detect and draw edges. The filter uses the Canny Edge Detection
10203 algorithm.
10204 The filter accepts the following options:
10206 low
10208 high
10209 Set low and high threshold values used by the Canny thresholding
10210 algorithm.
10211 The high threshold selects the "strong" edge pixels, which are then
10213 connected through 8-connectivity with the "weak" edge pixels
10214 selected by the low threshold.
10215 low and high threshold values must be chosen in the range [0,1],
10217 and low should be lesser or equal to high.
10218 Default value for low is "20/255", and default value for high is
10220 "50/255".
10221 mode
10223 Define the drawing mode.
10224 wires
10226 Draw white/gray wires on black background.
10227 colormix
10229 Mix the colors to create a paint/cartoon effect.
10230 canny
10232 Apply Canny edge detector on all selected planes.
10233 Default value is wires.
10235 planes
10237 Select planes for filtering. By default all available planes are
10238 filtered.
10239 Examples
10241 • Standard edge detection with custom values for the hysteresis
10243 thresholding:
10244 edgedetect=low=0.1:high=0.4
10246 • Painting effect without thresholding:
10248 edgedetect=mode=colormix:high=0
10250 elbg
10252 Apply a posterize effect using the ELBG (Enhanced LBG) algorithm.
10253 For each input image, the filter will compute the optimal mapping from
10255 the input to the output given the codebook length, that is the number
10256 of distinct output colors.
10257 This filter accepts the following options.
10259 codebook_length, l
10261 Set codebook length. The value must be a positive integer, and
10262 represents the number of distinct output colors. Default value is
10263 256.
10264 nb_steps, n
10266 Set the maximum number of iterations to apply for computing the
10267 optimal mapping. The higher the value the better the result and the
10268 higher the computation time. Default value is 1.
10269 seed, s
10271 Set a random seed, must be an integer included between 0 and
10272 UINT32_MAX. If not specified, or if explicitly set to -1, the
10273 filter will try to use a good random seed on a best effort basis.
10274 pal8
10276 Set pal8 output pixel format. This option does not work with
10277 codebook length greater than 256. Default is disabled.
10278 use_alpha
10280 Include alpha values in the quantization calculation. Allows
10281 creating palettized output images (e.g. PNG8) with multiple alpha
10282 smooth blending.
10283 entropy
10285 Measure graylevel entropy in histogram of color channels of video
10286 frames.
10287 It accepts the following parameters:
10289 mode
10291 Can be either normal or diff. Default is normal.
10292 diff mode measures entropy of histogram delta values, absolute
10294 differences between neighbour histogram values.
10295 epx
10297 Apply the EPX magnification filter which is designed for pixel art.
10298 It accepts the following option:
10300 n Set the scaling dimension: 2 for "2xEPX", 3 for "3xEPX". Default
10302 is 3.
10303 eq
10305 Set brightness, contrast, saturation and approximate gamma adjustment.
10306 The filter accepts the following options:
10308 contrast
10310 Set the contrast expression. The value must be a float value in
10311 range "-1000.0" to 1000.0. The default value is "1".
10312 brightness
10314 Set the brightness expression. The value must be a float value in
10315 range "-1.0" to 1.0. The default value is "0".
10316 saturation
10318 Set the saturation expression. The value must be a float in range
10319 0.0 to 3.0. The default value is "1".
10320 gamma
10322 Set the gamma expression. The value must be a float in range 0.1 to
10323 10.0. The default value is "1".
10324 gamma_r
10326 Set the gamma expression for red. The value must be a float in
10327 range 0.1 to 10.0. The default value is "1".
10328 gamma_g
10330 Set the gamma expression for green. The value must be a float in
10331 range 0.1 to 10.0. The default value is "1".
10332 gamma_b
10334 Set the gamma expression for blue. The value must be a float in
10335 range 0.1 to 10.0. The default value is "1".
10336 gamma_weight
10338 Set the gamma weight expression. It can be used to reduce the
10339 effect of a high gamma value on bright image areas, e.g. keep them
10340 from getting overamplified and just plain white. The value must be
10341 a float in range 0.0 to 1.0. A value of 0.0 turns the gamma
10342 correction all the way down while 1.0 leaves it at its full
10343 strength. Default is "1".
10344 eval
10346 Set when the expressions for brightness, contrast, saturation and
10347 gamma expressions are evaluated.
10348 It accepts the following values:
10350 init
10352 only evaluate expressions once during the filter initialization
10353 or when a command is processed
10354 frame
10356 evaluate expressions for each incoming frame
10357 Default value is init.
10359 The expressions accept the following parameters:
10361 n frame count of the input frame starting from 0
10363 pos byte position of the corresponding packet in the input file, NAN if
10365 unspecified
10366 r frame rate of the input video, NAN if the input frame rate is
10368 unknown
10369 t timestamp expressed in seconds, NAN if the input timestamp is
10371 unknown
10372 Commands
10374 The filter supports the following commands:
10376 contrast
10378 Set the contrast expression.
10379 brightness
10381 Set the brightness expression.
10382 saturation
10384 Set the saturation expression.
10385 gamma
10387 Set the gamma expression.
10388 gamma_r
10390 Set the gamma_r expression.
10391 gamma_g
10393 Set gamma_g expression.
10394 gamma_b
10396 Set gamma_b expression.
10397 gamma_weight
10399 Set gamma_weight expression.
10400 The command accepts the same syntax of the corresponding option.
10402 If the specified expression is not valid, it is kept at its current
10404 value.
10405 erosion
10407 Apply erosion effect to the video.
10408 This filter replaces the pixel by the local(3x3) minimum.
10410 It accepts the following options:
10412 threshold0
10414 threshold1
10415 threshold2
10416 threshold3
10417 Limit the maximum change for each plane, default is 65535. If 0,
10418 plane will remain unchanged.
10419 coordinates
10421 Flag which specifies the pixel to refer to. Default is 255 i.e. all
10422 eight pixels are used.
10423 Flags to local 3x3 coordinates maps like this:
10425 1 2 3
10427 4 5
10428 6 7 8
10429 Commands
10431 This filter supports the all above options as commands.
10433 estdif
10435 Deinterlace the input video ("estdif" stands for "Edge Slope Tracing
10436 Deinterlacing Filter").
10437 Spatial only filter that uses edge slope tracing algorithm to
10439 interpolate missing lines. It accepts the following parameters:
10440 mode
10442 The interlacing mode to adopt. It accepts one of the following
10443 values:
10444 frame
10446 Output one frame for each frame.
10447 field
10449 Output one frame for each field.
10450 The default value is "field".
10452 parity
10454 The picture field parity assumed for the input interlaced video. It
10455 accepts one of the following values:
10456 tff Assume the top field is first.
10458 bff Assume the bottom field is first.
10460 auto
10462 Enable automatic detection of field parity.
10463 The default value is "auto". If the interlacing is unknown or the
10465 decoder does not export this information, top field first will be
10466 assumed.
10467 deint
10469 Specify which frames to deinterlace. Accepts one of the following
10470 values:
10471 all Deinterlace all frames.
10473 interlaced
10475 Only deinterlace frames marked as interlaced.
10476 The default value is "all".
10478 rslope
10480 Specify the search radius for edge slope tracing. Default value is
10481 1. Allowed range is from 1 to 15.
10482 redge
10484 Specify the search radius for best edge matching. Default value is
10485 2. Allowed range is from 0 to 15.
10486 ecost
10488 Specify the edge cost for edge matching. Default value is 0.03125.
10489 Allowed range is from 0 to 1.
10490 mcost
10492 Specify the middle cost for edge matching. Default value is 0.5.
10493 Allowed range is from 0 to 1.
10494 dcost
10496 Specify the distance cost for edge matching. Default value is 0.5.
10497 Allowed range is from 0 to 1.
10498 interp
10500 Specify the interpolation used. Default is 4-point interpolation.
10501 It accepts one of the following values:
10502 2p Two-point interpolation.
10504 4p Four-point interpolation.
10506 6p Six-point interpolation.
10508 Commands
10510 This filter supports same commands as options.
10512 exposure
10514 Adjust exposure of the video stream.
10515 The filter accepts the following options:
10517 exposure
10519 Set the exposure correction in EV. Allowed range is from -3.0 to
10520 3.0 EV Default value is 0 EV.
10521 black
10523 Set the black level correction. Allowed range is from -1.0 to 1.0.
10524 Default value is 0.
10525 Commands
10527 This filter supports same commands as options.
10529 extractplanes
10531 Extract color channel components from input video stream into separate
10532 grayscale video streams.
10533 The filter accepts the following option:
10535 planes
10537 Set plane(s) to extract.
10538 Available values for planes are:
10540 y
10542 u
10543 v
10544 a
10545 r
10546 g
10547 b
10548 Choosing planes not available in the input will result in an error.
10550 That means you cannot select "r", "g", "b" planes with "y", "u",
10551 "v" planes at same time.
10552 Examples
10554 • Extract luma, u and v color channel component from input video
10556 frame into 3 grayscale outputs:
10557 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
10559 fade
10561 Apply a fade-in/out effect to the input video.
10562 It accepts the following parameters:
10564 type, t
10566 The effect type can be either "in" for a fade-in, or "out" for a
10567 fade-out effect. Default is "in".
10568 start_frame, s
10570 Specify the number of the frame to start applying the fade effect
10571 at. Default is 0.
10572 nb_frames, n
10574 The number of frames that the fade effect lasts. At the end of the
10575 fade-in effect, the output video will have the same intensity as
10576 the input video. At the end of the fade-out transition, the output
10577 video will be filled with the selected color. Default is 25.
10578 alpha
10580 If set to 1, fade only alpha channel, if one exists on the input.
10581 Default value is 0.
10582 start_time, st
10584 Specify the timestamp (in seconds) of the frame to start to apply
10585 the fade effect. If both start_frame and start_time are specified,
10586 the fade will start at whichever comes last. Default is 0.
10587 duration, d
10589 The number of seconds for which the fade effect has to last. At the
10590 end of the fade-in effect the output video will have the same
10591 intensity as the input video, at the end of the fade-out transition
10592 the output video will be filled with the selected color. If both
10593 duration and nb_frames are specified, duration is used. Default is
10594 0 (nb_frames is used by default).
10595 color, c
10597 Specify the color of the fade. Default is "black".
10598 Examples
10600 • Fade in the first 30 frames of video:
10602 fade=in:0:30
10604 The command above is equivalent to:
10606 fade=t=in:s=0:n=30
10608 • Fade out the last 45 frames of a 200-frame video:
10610 fade=out:155:45
10612 fade=type=out:start_frame=155:nb_frames=45
10613 • Fade in the first 25 frames and fade out the last 25 frames of a
10615 1000-frame video:
10616 fade=in:0:25, fade=out:975:25
10618 • Make the first 5 frames yellow, then fade in from frame 5-24:
10620 fade=in:5:20:color=yellow
10622 • Fade in alpha over first 25 frames of video:
10624 fade=in:0:25:alpha=1
10626 • Make the first 5.5 seconds black, then fade in for 0.5 seconds:
10628 fade=t=in:st=5.5:d=0.5
10630 fftdnoiz
10632 Denoise frames using 3D FFT (frequency domain filtering).
10633 The filter accepts the following options:
10635 sigma
10637 Set the noise sigma constant. This sets denoising strength.
10638 Default value is 1. Allowed range is from 0 to 30. Using very high
10639 sigma with low overlap may give blocking artifacts.
10640 amount
10642 Set amount of denoising. By default all detected noise is reduced.
10643 Default value is 1. Allowed range is from 0 to 1.
10644 block
10646 Set size of block, Default is 4, can be 3, 4, 5 or 6. Actual size
10647 of block in pixels is 2 to power of block, so by default block size
10648 in pixels is 2^4 which is 16.
10649 overlap
10651 Set block overlap. Default is 0.5. Allowed range is from 0.2 to
10652 0.8.
10653 prev
10655 Set number of previous frames to use for denoising. By default is
10656 set to 0.
10657 next
10659 Set number of next frames to to use for denoising. By default is
10660 set to 0.
10661 planes
10663 Set planes which will be filtered, by default are all available
10664 filtered except alpha.
10665 fftfilt
10667 Apply arbitrary expressions to samples in frequency domain
10668 dc_Y
10670 Adjust the dc value (gain) of the luma plane of the image. The
10671 filter accepts an integer value in range 0 to 1000. The default
10672 value is set to 0.
10673 dc_U
10675 Adjust the dc value (gain) of the 1st chroma plane of the image.
10676 The filter accepts an integer value in range 0 to 1000. The default
10677 value is set to 0.
10678 dc_V
10680 Adjust the dc value (gain) of the 2nd chroma plane of the image.
10681 The filter accepts an integer value in range 0 to 1000. The default
10682 value is set to 0.
10683 weight_Y
10685 Set the frequency domain weight expression for the luma plane.
10686 weight_U
10688 Set the frequency domain weight expression for the 1st chroma
10689 plane.
10690 weight_V
10692 Set the frequency domain weight expression for the 2nd chroma
10693 plane.
10694 eval
10696 Set when the expressions are evaluated.
10697 It accepts the following values:
10699 init
10701 Only evaluate expressions once during the filter
10702 initialization.
10703 frame
10705 Evaluate expressions for each incoming frame.
10706 Default value is init.
10708 The filter accepts the following variables:
10710 X
10712 Y The coordinates of the current sample.
10713 W
10715 H The width and height of the image.
10716 N The number of input frame, starting from 0.
10718 WS
10720 HS The size of FFT array for horizontal and vertical processing.
10721 Examples
10723 • High-pass:
10725 fftfilt=dc_Y=128:weight_Y='squish(1-(Y+X)/100)'
10727 • Low-pass:
10729 fftfilt=dc_Y=0:weight_Y='squish((Y+X)/100-1)'
10731 • Sharpen:
10733 fftfilt=dc_Y=0:weight_Y='1+squish(1-(Y+X)/100)'
10735 • Blur:
10737 fftfilt=dc_Y=0:weight_Y='exp(-4 * ((Y+X)/(W+H)))'
10739 field
10741 Extract a single field from an interlaced image using stride arithmetic
10742 to avoid wasting CPU time. The output frames are marked as non-
10743 interlaced.
10744 The filter accepts the following options:
10746 type
10748 Specify whether to extract the top (if the value is 0 or "top") or
10749 the bottom field (if the value is 1 or "bottom").
10750 fieldhint
10752 Create new frames by copying the top and bottom fields from surrounding
10753 frames supplied as numbers by the hint file.
10754 hint
10756 Set file containing hints: absolute/relative frame numbers.
10757 There must be one line for each frame in a clip. Each line must
10759 contain two numbers separated by the comma, optionally followed by
10760 "-" or "+". Numbers supplied on each line of file can not be out
10761 of [N-1,N+1] where N is current frame number for "absolute" mode or
10762 out of [-1, 1] range for "relative" mode. First number tells from
10763 which frame to pick up top field and second number tells from which
10764 frame to pick up bottom field.
10765 If optionally followed by "+" output frame will be marked as
10767 interlaced, else if followed by "-" output frame will be marked as
10768 progressive, else it will be marked same as input frame. If
10769 optionally followed by "t" output frame will use only top field, or
10770 in case of "b" it will use only bottom field. If line starts with
10771 "#" or ";" that line is skipped.
10772 mode
10774 Can be item "absolute" or "relative". Default is "absolute".
10775 Example of first several lines of "hint" file for "relative" mode:
10777 0,0 - # first frame
10779 1,0 - # second frame, use third's frame top field and second's frame bottom field
10780 1,0 - # third frame, use fourth's frame top field and third's frame bottom field
10781 1,0 -
10782 0,0 -
10783 0,0 -
10784 1,0 -
10785 1,0 -
10786 1,0 -
10787 0,0 -
10788 0,0 -
10789 1,0 -
10790 1,0 -
10791 1,0 -
10792 0,0 -
10793 fieldmatch
10795 Field matching filter for inverse telecine. It is meant to reconstruct
10796 the progressive frames from a telecined stream. The filter does not
10797 drop duplicated frames, so to achieve a complete inverse telecine
10798 "fieldmatch" needs to be followed by a decimation filter such as
10799 decimate in the filtergraph.
10800 The separation of the field matching and the decimation is notably
10802 motivated by the possibility of inserting a de-interlacing filter
10803 fallback between the two. If the source has mixed telecined and real
10804 interlaced content, "fieldmatch" will not be able to match fields for
10805 the interlaced parts. But these remaining combed frames will be marked
10806 as interlaced, and thus can be de-interlaced by a later filter such as
10807 yadif before decimation.
10808 In addition to the various configuration options, "fieldmatch" can take
10810 an optional second stream, activated through the ppsrc option. If
10811 enabled, the frames reconstruction will be based on the fields and
10812 frames from this second stream. This allows the first input to be pre-
10813 processed in order to help the various algorithms of the filter, while
10814 keeping the output lossless (assuming the fields are matched properly).
10815 Typically, a field-aware denoiser, or brightness/contrast adjustments
10816 can help.
10817 Note that this filter uses the same algorithms as TIVTC/TFM (AviSynth
10819 project) and VIVTC/VFM (VapourSynth project). The later is a light
10820 clone of TFM from which "fieldmatch" is based on. While the semantic
10821 and usage are very close, some behaviour and options names can differ.
10822 The decimate filter currently only works for constant frame rate input.
10824 If your input has mixed telecined (30fps) and progressive content with
10825 a lower framerate like 24fps use the following filterchain to produce
10826 the necessary cfr stream:
10827 "dejudder,fps=30000/1001,fieldmatch,decimate".
10828 The filter accepts the following options:
10830 order
10832 Specify the assumed field order of the input stream. Available
10833 values are:
10834 auto
10836 Auto detect parity (use FFmpeg's internal parity value).
10837 bff Assume bottom field first.
10839 tff Assume top field first.
10841 Note that it is sometimes recommended not to trust the parity
10843 announced by the stream.
10844 Default value is auto.
10846 mode
10848 Set the matching mode or strategy to use. pc mode is the safest in
10849 the sense that it won't risk creating jerkiness due to duplicate
10850 frames when possible, but if there are bad edits or blended fields
10851 it will end up outputting combed frames when a good match might
10852 actually exist. On the other hand, pcn_ub mode is the most risky in
10853 terms of creating jerkiness, but will almost always find a good
10854 frame if there is one. The other values are all somewhere in
10855 between pc and pcn_ub in terms of risking jerkiness and creating
10856 duplicate frames versus finding good matches in sections with bad
10857 edits, orphaned fields, blended fields, etc.
10858 More details about p/c/n/u/b are available in p/c/n/u/b meaning
10860 section.
10861 Available values are:
10863 pc 2-way matching (p/c)
10865 pc_n
10867 2-way matching, and trying 3rd match if still combed (p/c + n)
10868 pc_u
10870 2-way matching, and trying 3rd match (same order) if still
10871 combed (p/c + u)
10872 pc_n_ub
10874 2-way matching, trying 3rd match if still combed, and trying
10875 4th/5th matches if still combed (p/c + n + u/b)
10876 pcn 3-way matching (p/c/n)
10878 pcn_ub
10880 3-way matching, and trying 4th/5th matches if all 3 of the
10881 original matches are detected as combed (p/c/n + u/b)
10882 The parenthesis at the end indicate the matches that would be used
10884 for that mode assuming order=tff (and field on auto or top).
10885 In terms of speed pc mode is by far the fastest and pcn_ub is the
10887 slowest.
10888 Default value is pc_n.
10890 ppsrc
10892 Mark the main input stream as a pre-processed input, and enable the
10893 secondary input stream as the clean source to pick the fields from.
10894 See the filter introduction for more details. It is similar to the
10895 clip2 feature from VFM/TFM.
10896 Default value is 0 (disabled).
10898 field
10900 Set the field to match from. It is recommended to set this to the
10901 same value as order unless you experience matching failures with
10902 that setting. In certain circumstances changing the field that is
10903 used to match from can have a large impact on matching performance.
10904 Available values are:
10905 auto
10907 Automatic (same value as order).
10908 bottom
10910 Match from the bottom field.
10911 top Match from the top field.
10913 Default value is auto.
10915 mchroma
10917 Set whether or not chroma is included during the match comparisons.
10918 In most cases it is recommended to leave this enabled. You should
10919 set this to 0 only if your clip has bad chroma problems such as
10920 heavy rainbowing or other artifacts. Setting this to 0 could also
10921 be used to speed things up at the cost of some accuracy.
10922 Default value is 1.
10924 y0
10926 y1 These define an exclusion band which excludes the lines between y0
10927 and y1 from being included in the field matching decision. An
10928 exclusion band can be used to ignore subtitles, a logo, or other
10929 things that may interfere with the matching. y0 sets the starting
10930 scan line and y1 sets the ending line; all lines in between y0 and
10931 y1 (including y0 and y1) will be ignored. Setting y0 and y1 to the
10932 same value will disable the feature. y0 and y1 defaults to 0.
10933 scthresh
10935 Set the scene change detection threshold as a percentage of maximum
10936 change on the luma plane. Good values are in the "[8.0, 14.0]"
10937 range. Scene change detection is only relevant in case
10938 combmatch=sc. The range for scthresh is "[0.0, 100.0]".
10939 Default value is 12.0.
10941 combmatch
10943 When combatch is not none, "fieldmatch" will take into account the
10944 combed scores of matches when deciding what match to use as the
10945 final match. Available values are:
10946 none
10948 No final matching based on combed scores.
10949 sc Combed scores are only used when a scene change is detected.
10951 full
10953 Use combed scores all the time.
10954 Default is sc.
10956 combdbg
10958 Force "fieldmatch" to calculate the combed metrics for certain
10959 matches and print them. This setting is known as micout in TFM/VFM
10960 vocabulary. Available values are:
10961 none
10963 No forced calculation.
10964 pcn Force p/c/n calculations.
10966 pcnub
10968 Force p/c/n/u/b calculations.
10969 Default value is none.
10971 cthresh
10973 This is the area combing threshold used for combed frame detection.
10974 This essentially controls how "strong" or "visible" combing must be
10975 to be detected. Larger values mean combing must be more visible
10976 and smaller values mean combing can be less visible or strong and
10977 still be detected. Valid settings are from "-1" (every pixel will
10978 be detected as combed) to 255 (no pixel will be detected as
10979 combed). This is basically a pixel difference value. A good range
10980 is "[8, 12]".
10981 Default value is 9.
10983 chroma
10985 Sets whether or not chroma is considered in the combed frame
10986 decision. Only disable this if your source has chroma problems
10987 (rainbowing, etc.) that are causing problems for the combed frame
10988 detection with chroma enabled. Actually, using chroma=0 is usually
10989 more reliable, except for the case where there is chroma only
10990 combing in the source.
10991 Default value is 0.
10993 blockx
10995 blocky
10996 Respectively set the x-axis and y-axis size of the window used
10997 during combed frame detection. This has to do with the size of the
10998 area in which combpel pixels are required to be detected as combed
10999 for a frame to be declared combed. See the combpel parameter
11000 description for more info. Possible values are any number that is
11001 a power of 2 starting at 4 and going up to 512.
11002 Default value is 16.
11004 combpel
11006 The number of combed pixels inside any of the blocky by blockx size
11007 blocks on the frame for the frame to be detected as combed. While
11008 cthresh controls how "visible" the combing must be, this setting
11009 controls "how much" combing there must be in any localized area (a
11010 window defined by the blockx and blocky settings) on the frame.
11011 Minimum value is 0 and maximum is "blocky x blockx" (at which point
11012 no frames will ever be detected as combed). This setting is known
11013 as MI in TFM/VFM vocabulary.
11014 Default value is 80.
11016 p/c/n/u/b meaning
11018 p/c/n
11020 We assume the following telecined stream:
11022 Top fields: 1 2 2 3 4
11024 Bottom fields: 1 2 3 4 4
11025 The numbers correspond to the progressive frame the fields relate to.
11027 Here, the first two frames are progressive, the 3rd and 4th are combed,
11028 and so on.
11029 When "fieldmatch" is configured to run a matching from bottom
11031 (field=bottom) this is how this input stream get transformed:
11032 Input stream:
11034 T 1 2 2 3 4
11035 B 1 2 3 4 4 <-- matching reference
11036 Matches: c c n n c
11038 Output stream:
11040 T 1 2 3 4 4
11041 B 1 2 3 4 4
11042 As a result of the field matching, we can see that some frames get
11044 duplicated. To perform a complete inverse telecine, you need to rely
11045 on a decimation filter after this operation. See for instance the
11046 decimate filter.
11047 The same operation now matching from top fields (field=top) looks like
11049 this:
11050 Input stream:
11052 T 1 2 2 3 4 <-- matching reference
11053 B 1 2 3 4 4
11054 Matches: c c p p c
11056 Output stream:
11058 T 1 2 2 3 4
11059 B 1 2 2 3 4
11060 In these examples, we can see what p, c and n mean; basically, they
11062 refer to the frame and field of the opposite parity:
11063 *<p matches the field of the opposite parity in the previous frame>
11065 *<c matches the field of the opposite parity in the current frame>
11066 *<n matches the field of the opposite parity in the next frame>
11067 u/b
11069 The u and b matching are a bit special in the sense that they match
11071 from the opposite parity flag. In the following examples, we assume
11072 that we are currently matching the 2nd frame (Top:2, bottom:2).
11073 According to the match, a 'x' is placed above and below each matched
11074 fields.
11075 With bottom matching (field=bottom):
11077 Match: c p n b u
11079 x x x x x
11081 Top 1 2 2 1 2 2 1 2 2 1 2 2 1 2 2
11082 Bottom 1 2 3 1 2 3 1 2 3 1 2 3 1 2 3
11083 x x x x x
11084 Output frames:
11086 2 1 2 2 2
11087 2 2 2 1 3
11088 With top matching (field=top):
11090 Match: c p n b u
11092 x x x x x
11094 Top 1 2 2 1 2 2 1 2 2 1 2 2 1 2 2
11095 Bottom 1 2 3 1 2 3 1 2 3 1 2 3 1 2 3
11096 x x x x x
11097 Output frames:
11099 2 2 2 1 2
11100 2 1 3 2 2
11101 Examples
11103 Simple IVTC of a top field first telecined stream:
11105 fieldmatch=order=tff:combmatch=none, decimate
11107 Advanced IVTC, with fallback on yadif for still combed frames:
11109 fieldmatch=order=tff:combmatch=full, yadif=deint=interlaced, decimate
11111 fieldorder
11113 Transform the field order of the input video.
11114 It accepts the following parameters:
11116 order
11118 The output field order. Valid values are tff for top field first or
11119 bff for bottom field first.
11120 The default value is tff.
11122 The transformation is done by shifting the picture content up or down
11124 by one line, and filling the remaining line with appropriate picture
11125 content. This method is consistent with most broadcast field order
11126 converters.
11127 If the input video is not flagged as being interlaced, or it is already
11129 flagged as being of the required output field order, then this filter
11130 does not alter the incoming video.
11131 It is very useful when converting to or from PAL DV material, which is
11133 bottom field first.
11134 For example:
11136 ffmpeg -i in.vob -vf "fieldorder=bff" out.dv
11138 fifo, afifo
11140 Buffer input images and send them when they are requested.
11141 It is mainly useful when auto-inserted by the libavfilter framework.
11143 It does not take parameters.
11145 fillborders
11147 Fill borders of the input video, without changing video stream
11148 dimensions. Sometimes video can have garbage at the four edges and you
11149 may not want to crop video input to keep size multiple of some number.
11150 This filter accepts the following options:
11152 left
11154 Number of pixels to fill from left border.
11155 right
11157 Number of pixels to fill from right border.
11158 top Number of pixels to fill from top border.
11160 bottom
11162 Number of pixels to fill from bottom border.
11163 mode
11165 Set fill mode.
11166 It accepts the following values:
11168 smear
11170 fill pixels using outermost pixels
11171 mirror
11173 fill pixels using mirroring (half sample symmetric)
11174 fixed
11176 fill pixels with constant value
11177 reflect
11179 fill pixels using reflecting (whole sample symmetric)
11180 wrap
11182 fill pixels using wrapping
11183 fade
11185 fade pixels to constant value
11186 margins
11188 fill pixels at top and bottom with weighted averages pixels
11189 near borders
11190 Default is smear.
11192 color
11194 Set color for pixels in fixed or fade mode. Default is black.
11195 Commands
11197 This filter supports same commands as options. The command accepts the
11199 same syntax of the corresponding option.
11200 If the specified expression is not valid, it is kept at its current
11202 value.
11203 find_rect
11205 Find a rectangular object
11206 It accepts the following options:
11208 object
11210 Filepath of the object image, needs to be in gray8.
11211 threshold
11213 Detection threshold, default is 0.5.
11214 mipmaps
11216 Number of mipmaps, default is 3.
11217 xmin, ymin, xmax, ymax
11219 Specifies the rectangle in which to search.
11220 discard
11222 Discard frames where object is not detected. Default is disabled.
11223 Examples
11225 • Cover a rectangular object by the supplied image of a given video
11227 using ffmpeg:
11228 ffmpeg -i file.ts -vf find_rect=newref.pgm,cover_rect=cover.jpg:mode=cover new.mkv
11230 floodfill
11232 Flood area with values of same pixel components with another values.
11233 It accepts the following options:
11235 x Set pixel x coordinate.
11237 y Set pixel y coordinate.
11239 s0 Set source #0 component value.
11241 s1 Set source #1 component value.
11243 s2 Set source #2 component value.
11245 s3 Set source #3 component value.
11247 d0 Set destination #0 component value.
11249 d1 Set destination #1 component value.
11251 d2 Set destination #2 component value.
11253 d3 Set destination #3 component value.
11255 format
11257 Convert the input video to one of the specified pixel formats.
11258 Libavfilter will try to pick one that is suitable as input to the next
11259 filter.
11260 It accepts the following parameters:
11262 pix_fmts
11264 A '|'-separated list of pixel format names, such as
11265 "pix_fmts=yuv420p|monow|rgb24".
11266 Examples
11268 • Convert the input video to the yuv420p format
11270 format=pix_fmts=yuv420p
11272 Convert the input video to any of the formats in the list
11274 format=pix_fmts=yuv420p|yuv444p|yuv410p
11276 fps
11278 Convert the video to specified constant frame rate by duplicating or
11279 dropping frames as necessary.
11280 It accepts the following parameters:
11282 fps The desired output frame rate. It accepts expressions containing
11284 the following constants:
11285 source_fps
11287 The input's frame rate
11288 ntsc
11290 NTSC frame rate of "30000/1001"
11291 pal PAL frame rate of 25.0
11293 film
11295 Film frame rate of 24.0
11296 ntsc_film
11298 NTSC-film frame rate of "24000/1001"
11299 The default is 25.
11301 start_time
11303 Assume the first PTS should be the given value, in seconds. This
11304 allows for padding/trimming at the start of stream. By default, no
11305 assumption is made about the first frame's expected PTS, so no
11306 padding or trimming is done. For example, this could be set to 0
11307 to pad the beginning with duplicates of the first frame if a video
11308 stream starts after the audio stream or to trim any frames with a
11309 negative PTS.
11310 round
11312 Timestamp (PTS) rounding method.
11313 Possible values are:
11315 zero
11317 round towards 0
11318 inf round away from 0
11320 down
11322 round towards -infinity
11323 up round towards +infinity
11325 near
11327 round to nearest
11328 The default is "near".
11330 eof_action
11332 Action performed when reading the last frame.
11333 Possible values are:
11335 round
11337 Use same timestamp rounding method as used for other frames.
11338 pass
11340 Pass through last frame if input duration has not been reached
11341 yet.
11342 The default is "round".
11344 Alternatively, the options can be specified as a flat string:
11346 fps[:start_time[:round]].
11347 See also the setpts filter.
11349 Examples
11351 • A typical usage in order to set the fps to 25:
11353 fps=fps=25
11355 • Sets the fps to 24, using abbreviation and rounding method to round
11357 to nearest:
11358 fps=fps=film:round=near
11360 framepack
11362 Pack two different video streams into a stereoscopic video, setting
11363 proper metadata on supported codecs. The two views should have the same
11364 size and framerate and processing will stop when the shorter video
11365 ends. Please note that you may conveniently adjust view properties with
11366 the scale and fps filters.
11367 It accepts the following parameters:
11369 format
11371 The desired packing format. Supported values are:
11372 sbs The views are next to each other (default).
11374 tab The views are on top of each other.
11376 lines
11378 The views are packed by line.
11379 columns
11381 The views are packed by column.
11382 frameseq
11384 The views are temporally interleaved.
11385 Some examples:
11387 # Convert left and right views into a frame-sequential video
11389 ffmpeg -i LEFT -i RIGHT -filter_complex framepack=frameseq OUTPUT
11390 # Convert views into a side-by-side video with the same output resolution as the input
11392 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
11393 framerate
11395 Change the frame rate by interpolating new video output frames from the
11396 source frames.
11397 This filter is not designed to function correctly with interlaced
11399 media. If you wish to change the frame rate of interlaced media then
11400 you are required to deinterlace before this filter and re-interlace
11401 after this filter.
11402 A description of the accepted options follows.
11404 fps Specify the output frames per second. This option can also be
11406 specified as a value alone. The default is 50.
11407 interp_start
11409 Specify the start of a range where the output frame will be created
11410 as a linear interpolation of two frames. The range is [0-255], the
11411 default is 15.
11412 interp_end
11414 Specify the end of a range where the output frame will be created
11415 as a linear interpolation of two frames. The range is [0-255], the
11416 default is 240.
11417 scene
11419 Specify the level at which a scene change is detected as a value
11420 between 0 and 100 to indicate a new scene; a low value reflects a
11421 low probability for the current frame to introduce a new scene,
11422 while a higher value means the current frame is more likely to be
11423 one. The default is 8.2.
11424 flags
11426 Specify flags influencing the filter process.
11427 Available value for flags is:
11429 scene_change_detect, scd
11431 Enable scene change detection using the value of the option
11432 scene. This flag is enabled by default.
11433 framestep
11435 Select one frame every N-th frame.
11436 This filter accepts the following option:
11438 step
11440 Select frame after every "step" frames. Allowed values are
11441 positive integers higher than 0. Default value is 1.
11442 freezedetect
11444 Detect frozen video.
11445 This filter logs a message and sets frame metadata when it detects that
11447 the input video has no significant change in content during a specified
11448 duration. Video freeze detection calculates the mean average absolute
11449 difference of all the components of video frames and compares it to a
11450 noise floor.
11451 The printed times and duration are expressed in seconds. The
11453 "lavfi.freezedetect.freeze_start" metadata key is set on the first
11454 frame whose timestamp equals or exceeds the detection duration and it
11455 contains the timestamp of the first frame of the freeze. The
11456 "lavfi.freezedetect.freeze_duration" and
11457 "lavfi.freezedetect.freeze_end" metadata keys are set on the first
11458 frame after the freeze.
11459 The filter accepts the following options:
11461 noise, n
11463 Set noise tolerance. Can be specified in dB (in case "dB" is
11464 appended to the specified value) or as a difference ratio between 0
11465 and 1. Default is -60dB, or 0.001.
11466 duration, d
11468 Set freeze duration until notification (default is 2 seconds).
11469 freezeframes
11471 Freeze video frames.
11472 This filter freezes video frames using frame from 2nd input.
11474 The filter accepts the following options:
11476 first
11478 Set number of first frame from which to start freeze.
11479 last
11481 Set number of last frame from which to end freeze.
11482 replace
11484 Set number of frame from 2nd input which will be used instead of
11485 replaced frames.
11486 frei0r
11488 Apply a frei0r effect to the input video.
11489 To enable the compilation of this filter, you need to install the
11491 frei0r header and configure FFmpeg with "--enable-frei0r".
11492 It accepts the following parameters:
11494 filter_name
11496 The name of the frei0r effect to load. If the environment variable
11497 FREI0R_PATH is defined, the frei0r effect is searched for in each
11498 of the directories specified by the colon-separated list in
11499 FREI0R_PATH. Otherwise, the standard frei0r paths are searched, in
11500 this order: HOME/.frei0r-1/lib/, /usr/local/lib/frei0r-1/,
11501 /usr/lib/frei0r-1/.
11502 filter_params
11504 A '|'-separated list of parameters to pass to the frei0r effect.
11505 A frei0r effect parameter can be a boolean (its value is either "y" or
11507 "n"), a double, a color (specified as R/G/B, where R, G, and B are
11508 floating point numbers between 0.0 and 1.0, inclusive) or a color
11509 description as specified in the "Color" section in the ffmpeg-utils
11510 manual, a position (specified as X/Y, where X and Y are floating point
11511 numbers) and/or a string.
11512 The number and types of parameters depend on the loaded effect. If an
11514 effect parameter is not specified, the default value is set.
11515 Examples
11517 • Apply the distort0r effect, setting the first two double
11519 parameters:
11520 frei0r=filter_name=distort0r:filter_params=0.5|0.01
11522 • Apply the colordistance effect, taking a color as the first
11524 parameter:
11525 frei0r=colordistance:0.2/0.3/0.4
11527 frei0r=colordistance:violet
11528 frei0r=colordistance:0x112233
11529 • Apply the perspective effect, specifying the top left and top right
11531 image positions:
11532 frei0r=perspective:0.2/0.2|0.8/0.2
11534 For more information, see <http://frei0r.dyne.org>
11536 Commands
11538 This filter supports the filter_params option as commands.
11540 fspp
11542 Apply fast and simple postprocessing. It is a faster version of spp.
11543 It splits (I)DCT into horizontal/vertical passes. Unlike the simple
11545 post- processing filter, one of them is performed once per block, not
11546 per pixel. This allows for much higher speed.
11547 The filter accepts the following options:
11549 quality
11551 Set quality. This option defines the number of levels for
11552 averaging. It accepts an integer in the range 4-5. Default value is
11553 4.
11554 qp Force a constant quantization parameter. It accepts an integer in
11556 range 0-63. If not set, the filter will use the QP from the video
11557 stream (if available).
11558 strength
11560 Set filter strength. It accepts an integer in range -15 to 32.
11561 Lower values mean more details but also more artifacts, while
11562 higher values make the image smoother but also blurrier. Default
11563 value is 0 X PSNR optimal.
11564 use_bframe_qp
11566 Enable the use of the QP from the B-Frames if set to 1. Using this
11567 option may cause flicker since the B-Frames have often larger QP.
11568 Default is 0 (not enabled).
11569 gblur
11571 Apply Gaussian blur filter.
11572 The filter accepts the following options:
11574 sigma
11576 Set horizontal sigma, standard deviation of Gaussian blur. Default
11577 is 0.5.
11578 steps
11580 Set number of steps for Gaussian approximation. Default is 1.
11581 planes
11583 Set which planes to filter. By default all planes are filtered.
11584 sigmaV
11586 Set vertical sigma, if negative it will be same as "sigma".
11587 Default is "-1".
11588 Commands
11590 This filter supports same commands as options. The command accepts the
11592 same syntax of the corresponding option.
11593 If the specified expression is not valid, it is kept at its current
11595 value.
11596 geq
11598 Apply generic equation to each pixel.
11599 The filter accepts the following options:
11601 lum_expr, lum
11603 Set the luminance expression.
11604 cb_expr, cb
11606 Set the chrominance blue expression.
11607 cr_expr, cr
11609 Set the chrominance red expression.
11610 alpha_expr, a
11612 Set the alpha expression.
11613 red_expr, r
11615 Set the red expression.
11616 green_expr, g
11618 Set the green expression.
11619 blue_expr, b
11621 Set the blue expression.
11622 The colorspace is selected according to the specified options. If one
11624 of the lum_expr, cb_expr, or cr_expr options is specified, the filter
11625 will automatically select a YCbCr colorspace. If one of the red_expr,
11626 green_expr, or blue_expr options is specified, it will select an RGB
11627 colorspace.
11628 If one of the chrominance expression is not defined, it falls back on
11630 the other one. If no alpha expression is specified it will evaluate to
11631 opaque value. If none of chrominance expressions are specified, they
11632 will evaluate to the luminance expression.
11633 The expressions can use the following variables and functions:
11635 N The sequential number of the filtered frame, starting from 0.
11637 X
11639 Y The coordinates of the current sample.
11640 W
11642 H The width and height of the image.
11643 SW
11645 SH Width and height scale depending on the currently filtered plane.
11646 It is the ratio between the corresponding luma plane number of
11647 pixels and the current plane ones. E.g. for YUV4:2:0 the values are
11648 "1,1" for the luma plane, and "0.5,0.5" for chroma planes.
11649 T Time of the current frame, expressed in seconds.
11651 p(x, y)
11653 Return the value of the pixel at location (x,y) of the current
11654 plane.
11655 lum(x, y)
11657 Return the value of the pixel at location (x,y) of the luminance
11658 plane.
11659 cb(x, y)
11661 Return the value of the pixel at location (x,y) of the blue-
11662 difference chroma plane. Return 0 if there is no such plane.
11663 cr(x, y)
11665 Return the value of the pixel at location (x,y) of the red-
11666 difference chroma plane. Return 0 if there is no such plane.
11667 r(x, y)
11669 g(x, y)
11670 b(x, y)
11671 Return the value of the pixel at location (x,y) of the
11672 red/green/blue component. Return 0 if there is no such component.
11673 alpha(x, y)
11675 Return the value of the pixel at location (x,y) of the alpha plane.
11676 Return 0 if there is no such plane.
11677 psum(x,y), lumsum(x, y), cbsum(x,y), crsum(x,y), rsum(x,y), gsum(x,y),
11679 bsum(x,y), alphasum(x,y)
11680 Sum of sample values in the rectangle from (0,0) to (x,y), this
11681 allows obtaining sums of samples within a rectangle. See the
11682 functions without the sum postfix.
11683 interpolation
11685 Set one of interpolation methods:
11686 nearest, n
11688 bilinear, b
11689 Default is bilinear.
11691 For functions, if x and y are outside the area, the value will be
11693 automatically clipped to the closer edge.
11694 Please note that this filter can use multiple threads in which case
11696 each slice will have its own expression state. If you want to use only
11697 a single expression state because your expressions depend on previous
11698 state then you should limit the number of filter threads to 1.
11699 Examples
11701 • Flip the image horizontally:
11703 geq=p(W-X\,Y)
11705 • Generate a bidimensional sine wave, with angle "PI/3" and a
11707 wavelength of 100 pixels:
11708 geq=128 + 100*sin(2*(PI/100)*(cos(PI/3)*(X-50*T) + sin(PI/3)*Y)):128:128
11710 • Generate a fancy enigmatic moving light:
11712 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
11714 • Generate a quick emboss effect:
11716 format=gray,geq=lum_expr='(p(X,Y)+(256-p(X-4,Y-4)))/2'
11718 • Modify RGB components depending on pixel position:
11720 geq=r='X/W*r(X,Y)':g='(1-X/W)*g(X,Y)':b='(H-Y)/H*b(X,Y)'
11722 • Create a radial gradient that is the same size as the input (also
11724 see the vignette filter):
11725 geq=lum=255*gauss((X/W-0.5)*3)*gauss((Y/H-0.5)*3)/gauss(0)/gauss(0),format=gray
11727 gradfun
11729 Fix the banding artifacts that are sometimes introduced into nearly
11730 flat regions by truncation to 8-bit color depth. Interpolate the
11731 gradients that should go where the bands are, and dither them.
11732 It is designed for playback only. Do not use it prior to lossy
11734 compression, because compression tends to lose the dither and bring
11735 back the bands.
11736 It accepts the following parameters:
11738 strength
11740 The maximum amount by which the filter will change any one pixel.
11741 This is also the threshold for detecting nearly flat regions.
11742 Acceptable values range from .51 to 64; the default value is 1.2.
11743 Out-of-range values will be clipped to the valid range.
11744 radius
11746 The neighborhood to fit the gradient to. A larger radius makes for
11747 smoother gradients, but also prevents the filter from modifying the
11748 pixels near detailed regions. Acceptable values are 8-32; the
11749 default value is 16. Out-of-range values will be clipped to the
11750 valid range.
11751 Alternatively, the options can be specified as a flat string:
11753 strength[:radius]
11754 Examples
11756 • Apply the filter with a 3.5 strength and radius of 8:
11758 gradfun=3.5:8
11760 • Specify radius, omitting the strength (which will fall-back to the
11762 default value):
11763 gradfun=radius=8
11765 graphmonitor
11767 Show various filtergraph stats.
11768 With this filter one can debug complete filtergraph. Especially issues
11770 with links filling with queued frames.
11771 The filter accepts the following options:
11773 size, s
11775 Set video output size. Default is hd720.
11776 opacity, o
11778 Set video opacity. Default is 0.9. Allowed range is from 0 to 1.
11779 mode, m
11781 Set output mode, can be fulll or compact. In compact mode only
11782 filters with some queued frames have displayed stats.
11783 flags, f
11785 Set flags which enable which stats are shown in video.
11786 Available values for flags are:
11788 queue
11790 Display number of queued frames in each link.
11791 frame_count_in
11793 Display number of frames taken from filter.
11794 frame_count_out
11796 Display number of frames given out from filter.
11797 pts Display current filtered frame pts.
11799 time
11801 Display current filtered frame time.
11802 timebase
11804 Display time base for filter link.
11805 format
11807 Display used format for filter link.
11808 size
11810 Display video size or number of audio channels in case of audio
11811 used by filter link.
11812 rate
11814 Display video frame rate or sample rate in case of audio used
11815 by filter link.
11816 eof Display link output status.
11818 sample_count_in
11820 Display number of samples taken from filter.
11821 sample_count_out
11823 Display number of samples given out from filter.
11824 rate, r
11826 Set upper limit for video rate of output stream, Default value is
11827 25. This guarantee that output video frame rate will not be higher
11828 than this value.
11829 grayworld
11831 A color constancy filter that applies color correction based on the
11832 grayworld assumption
11833 See:
11835 <https://www.researchgate.net/publication/275213614_A_New_Color_Correction_Method_for_Underwater_Imaging>
11836 The algorithm uses linear light, so input data should be linearized
11838 beforehand (and possibly correctly tagged).
11839 ffmpeg -i INPUT -vf zscale=transfer=linear,grayworld,zscale=transfer=bt709,format=yuv420p OUTPUT
11841 greyedge
11843 A color constancy variation filter which estimates scene illumination
11844 via grey edge algorithm and corrects the scene colors accordingly.
11845 See: <https://staff.science.uva.nl/th.gevers/pub/GeversTIP07.pdf>
11847 The filter accepts the following options:
11849 difford
11851 The order of differentiation to be applied on the scene. Must be
11852 chosen in the range [0,2] and default value is 1.
11853 minknorm
11855 The Minkowski parameter to be used for calculating the Minkowski
11856 distance. Must be chosen in the range [0,20] and default value is
11857 1. Set to 0 for getting max value instead of calculating Minkowski
11858 distance.
11859 sigma
11861 The standard deviation of Gaussian blur to be applied on the scene.
11862 Must be chosen in the range [0,1024.0] and default value = 1.
11863 floor( sigma * break_off_sigma(3) ) can't be equal to 0 if difford
11864 is greater than 0.
11865 Examples
11867 • Grey Edge:
11869 greyedge=difford=1:minknorm=5:sigma=2
11871 • Max Edge:
11873 greyedge=difford=1:minknorm=0:sigma=2
11875 guided
11877 Apply guided filter for edge-preserving smoothing, dehazing and so on.
11878 The filter accepts the following options:
11880 radius
11882 Set the box radius in pixels. Allowed range is 1 to 20. Default is
11883 3.
11884 eps Set regularization parameter (with square). Allowed range is 0 to
11886 1. Default is 0.01.
11887 mode
11889 Set filter mode. Can be "basic" or "fast". Default is "basic".
11890 sub Set subsampling ratio for "fast" mode. Range is 2 to 64. Default
11892 is 4. No subsampling occurs in "basic" mode.
11893 guidance
11895 Set guidance mode. Can be "off" or "on". Default is "off". If
11896 "off", single input is required. If "on", two inputs of the same
11897 resolution and pixel format are required. The second input serves
11898 as the guidance.
11899 planes
11901 Set planes to filter. Default is first only.
11902 Commands
11904 This filter supports the all above options as commands.
11906 Examples
11908 • Edge-preserving smoothing with guided filter:
11910 ffmpeg -i in.png -vf guided out.png
11912 • Dehazing, structure-transferring filtering, detail enhancement with
11914 guided filter. For the generation of guidance image, refer to
11915 paper "Guided Image Filtering". See:
11916 <http://kaiminghe.com/publications/pami12guidedfilter.pdf>.
11917 ffmpeg -i in.png -i guidance.png -filter_complex guided=guidance=on out.png
11919 haldclut
11921 Apply a Hald CLUT to a video stream.
11922 First input is the video stream to process, and second one is the Hald
11924 CLUT. The Hald CLUT input can be a simple picture or a complete video
11925 stream.
11926 The filter accepts the following options:
11928 shortest
11930 Force termination when the shortest input terminates. Default is 0.
11931 repeatlast
11933 Continue applying the last CLUT after the end of the stream. A
11934 value of 0 disable the filter after the last frame of the CLUT is
11935 reached. Default is 1.
11936 "haldclut" also has the same interpolation options as lut3d (both
11938 filters share the same internals).
11939 This filter also supports the framesync options.
11941 More information about the Hald CLUT can be found on Eskil Steenberg's
11943 website (Hald CLUT author) at
11944 <http://www.quelsolaar.com/technology/clut.html>.
11945 Commands
11947 This filter supports the "interp" option as commands.
11949 Workflow examples
11951 Hald CLUT video stream
11953 Generate an identity Hald CLUT stream altered with various effects:
11955 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
11957 Note: make sure you use a lossless codec.
11959 Then use it with "haldclut" to apply it on some random stream:
11961 ffmpeg -f lavfi -i mandelbrot -i clut.nut -filter_complex '[0][1] haldclut' -t 20 mandelclut.mkv
11963 The Hald CLUT will be applied to the 10 first seconds (duration of
11965 clut.nut), then the latest picture of that CLUT stream will be applied
11966 to the remaining frames of the "mandelbrot" stream.
11967 Hald CLUT with preview
11969 A Hald CLUT is supposed to be a squared image of "Level*Level*Level" by
11971 "Level*Level*Level" pixels. For a given Hald CLUT, FFmpeg will select
11972 the biggest possible square starting at the top left of the picture.
11973 The remaining padding pixels (bottom or right) will be ignored. This
11974 area can be used to add a preview of the Hald CLUT.
11975 Typically, the following generated Hald CLUT will be supported by the
11977 "haldclut" filter:
11978 ffmpeg -f lavfi -i B<haldclutsrc>=8 -vf "
11980 pad=iw+320 [padded_clut];
11981 smptebars=s=320x256, split [a][b];
11982 [padded_clut][a] overlay=W-320:h, curves=color_negative [main];
11983 [main][b] overlay=W-320" -frames:v 1 clut.png
11984 It contains the original and a preview of the effect of the CLUT: SMPTE
11986 color bars are displayed on the right-top, and below the same color
11987 bars processed by the color changes.
11988 Then, the effect of this Hald CLUT can be visualized with:
11990 ffplay input.mkv -vf "movie=clut.png, [in] haldclut"
11992 hflip
11994 Flip the input video horizontally.
11995 For example, to horizontally flip the input video with ffmpeg:
11997 ffmpeg -i in.avi -vf "hflip" out.avi
11999 histeq
12001 This filter applies a global color histogram equalization on a per-
12002 frame basis.
12003 It can be used to correct video that has a compressed range of pixel
12005 intensities. The filter redistributes the pixel intensities to
12006 equalize their distribution across the intensity range. It may be
12007 viewed as an "automatically adjusting contrast filter". This filter is
12008 useful only for correcting degraded or poorly captured source video.
12009 The filter accepts the following options:
12011 strength
12013 Determine the amount of equalization to be applied. As the
12014 strength is reduced, the distribution of pixel intensities more-
12015 and-more approaches that of the input frame. The value must be a
12016 float number in the range [0,1] and defaults to 0.200.
12017 intensity
12019 Set the maximum intensity that can generated and scale the output
12020 values appropriately. The strength should be set as desired and
12021 then the intensity can be limited if needed to avoid washing-out.
12022 The value must be a float number in the range [0,1] and defaults to
12023 0.210.
12024 antibanding
12026 Set the antibanding level. If enabled the filter will randomly vary
12027 the luminance of output pixels by a small amount to avoid banding
12028 of the histogram. Possible values are "none", "weak" or "strong".
12029 It defaults to "none".
12030 histogram
12032 Compute and draw a color distribution histogram for the input video.
12033 The computed histogram is a representation of the color component
12035 distribution in an image.
12036 Standard histogram displays the color components distribution in an
12038 image. Displays color graph for each color component. Shows
12039 distribution of the Y, U, V, A or R, G, B components, depending on
12040 input format, in the current frame. Below each graph a color component
12041 scale meter is shown.
12042 The filter accepts the following options:
12044 level_height
12046 Set height of level. Default value is 200. Allowed range is [50,
12047 2048].
12048 scale_height
12050 Set height of color scale. Default value is 12. Allowed range is
12051 [0, 40].
12052 display_mode
12054 Set display mode. It accepts the following values:
12055 stack
12057 Per color component graphs are placed below each other.
12058 parade
12060 Per color component graphs are placed side by side.
12061 overlay
12063 Presents information identical to that in the "parade", except
12064 that the graphs representing color components are superimposed
12065 directly over one another.
12066 Default is "stack".
12068 levels_mode
12070 Set mode. Can be either "linear", or "logarithmic". Default is
12071 "linear".
12072 components
12074 Set what color components to display. Default is 7.
12075 fgopacity
12077 Set foreground opacity. Default is 0.7.
12078 bgopacity
12080 Set background opacity. Default is 0.5.
12081 colors_mode
12083 Set colors mode. It accepts the following values:
12084 whiteonblack
12086 blackonwhite
12087 whiteongray
12088 blackongray
12089 coloronblack
12090 coloronwhite
12091 colorongray
12092 blackoncolor
12093 whiteoncolor
12094 grayoncolor
12095 Default is "whiteonblack".
12097 Examples
12099 • Calculate and draw histogram:
12101 ffplay -i input -vf histogram
12103 hqdn3d
12105 This is a high precision/quality 3d denoise filter. It aims to reduce
12106 image noise, producing smooth images and making still images really
12107 still. It should enhance compressibility.
12108 It accepts the following optional parameters:
12110 luma_spatial
12112 A non-negative floating point number which specifies spatial luma
12113 strength. It defaults to 4.0.
12114 chroma_spatial
12116 A non-negative floating point number which specifies spatial chroma
12117 strength. It defaults to 3.0*luma_spatial/4.0.
12118 luma_tmp
12120 A floating point number which specifies luma temporal strength. It
12121 defaults to 6.0*luma_spatial/4.0.
12122 chroma_tmp
12124 A floating point number which specifies chroma temporal strength.
12125 It defaults to luma_tmp*chroma_spatial/luma_spatial.
12126 Commands
12128 This filter supports same commands as options. The command accepts the
12130 same syntax of the corresponding option.
12131 If the specified expression is not valid, it is kept at its current
12133 value.
12134 hwdownload
12136 Download hardware frames to system memory.
12137 The input must be in hardware frames, and the output a non-hardware
12139 format. Not all formats will be supported on the output - it may be
12140 necessary to insert an additional format filter immediately following
12141 in the graph to get the output in a supported format.
12142 hwmap
12144 Map hardware frames to system memory or to another device.
12145 This filter has several different modes of operation; which one is used
12147 depends on the input and output formats:
12148 • Hardware frame input, normal frame output
12150 Map the input frames to system memory and pass them to the output.
12152 If the original hardware frame is later required (for example,
12153 after overlaying something else on part of it), the hwmap filter
12154 can be used again in the next mode to retrieve it.
12155 • Normal frame input, hardware frame output
12157 If the input is actually a software-mapped hardware frame, then
12159 unmap it - that is, return the original hardware frame.
12160 Otherwise, a device must be provided. Create new hardware surfaces
12162 on that device for the output, then map them back to the software
12163 format at the input and give those frames to the preceding filter.
12164 This will then act like the hwupload filter, but may be able to
12165 avoid an additional copy when the input is already in a compatible
12166 format.
12167 • Hardware frame input and output
12169 A device must be supplied for the output, either directly or with
12171 the derive_device option. The input and output devices must be of
12172 different types and compatible - the exact meaning of this is
12173 system-dependent, but typically it means that they must refer to
12174 the same underlying hardware context (for example, refer to the
12175 same graphics card).
12176 If the input frames were originally created on the output device,
12178 then unmap to retrieve the original frames.
12179 Otherwise, map the frames to the output device - create new
12181 hardware frames on the output corresponding to the frames on the
12182 input.
12183 The following additional parameters are accepted:
12185 mode
12187 Set the frame mapping mode. Some combination of:
12188 read
12190 The mapped frame should be readable.
12191 write
12193 The mapped frame should be writeable.
12194 overwrite
12196 The mapping will always overwrite the entire frame.
12197 This may improve performance in some cases, as the original
12199 contents of the frame need not be loaded.
12200 direct
12202 The mapping must not involve any copying.
12203 Indirect mappings to copies of frames are created in some cases
12205 where either direct mapping is not possible or it would have
12206 unexpected properties. Setting this flag ensures that the
12207 mapping is direct and will fail if that is not possible.
12208 Defaults to read+write if not specified.
12210 derive_device type
12212 Rather than using the device supplied at initialisation, instead
12213 derive a new device of type type from the device the input frames
12214 exist on.
12215 reverse
12217 In a hardware to hardware mapping, map in reverse - create frames
12218 in the sink and map them back to the source. This may be necessary
12219 in some cases where a mapping in one direction is required but only
12220 the opposite direction is supported by the devices being used.
12221 This option is dangerous - it may break the preceding filter in
12223 undefined ways if there are any additional constraints on that
12224 filter's output. Do not use it without fully understanding the
12225 implications of its use.
12226 hwupload
12228 Upload system memory frames to hardware surfaces.
12229 The device to upload to must be supplied when the filter is
12231 initialised. If using ffmpeg, select the appropriate device with the
12232 -filter_hw_device option or with the derive_device option. The input
12233 and output devices must be of different types and compatible - the
12234 exact meaning of this is system-dependent, but typically it means that
12235 they must refer to the same underlying hardware context (for example,
12236 refer to the same graphics card).
12237 The following additional parameters are accepted:
12239 derive_device type
12241 Rather than using the device supplied at initialisation, instead
12242 derive a new device of type type from the device the input frames
12243 exist on.
12244 hwupload_cuda
12246 Upload system memory frames to a CUDA device.
12247 It accepts the following optional parameters:
12249 device
12251 The number of the CUDA device to use
12252 hqx
12254 Apply a high-quality magnification filter designed for pixel art. This
12255 filter was originally created by Maxim Stepin.
12256 It accepts the following option:
12258 n Set the scaling dimension: 2 for "hq2x", 3 for "hq3x" and 4 for
12260 "hq4x". Default is 3.
12261 hstack
12263 Stack input videos horizontally.
12264 All streams must be of same pixel format and of same height.
12266 Note that this filter is faster than using overlay and pad filter to
12268 create same output.
12269 The filter accepts the following option:
12271 inputs
12273 Set number of input streams. Default is 2.
12274 shortest
12276 If set to 1, force the output to terminate when the shortest input
12277 terminates. Default value is 0.
12278 hsvhold
12280 Turns a certain HSV range into gray values.
12281 This filter measures color difference between set HSV color in options
12283 and ones measured in video stream. Depending on options, output colors
12284 can be changed to be gray or not.
12285 The filter accepts the following options:
12287 hue Set the hue value which will be used in color difference
12289 calculation. Allowed range is from -360 to 360. Default value is
12290 0.
12291 sat Set the saturation value which will be used in color difference
12293 calculation. Allowed range is from -1 to 1. Default value is 0.
12294 val Set the value which will be used in color difference calculation.
12296 Allowed range is from -1 to 1. Default value is 0.
12297 similarity
12299 Set similarity percentage with the key color. Allowed range is
12300 from 0 to 1. Default value is 0.01.
12301 0.00001 matches only the exact key color, while 1.0 matches
12303 everything.
12304 blend
12306 Blend percentage. Allowed range is from 0 to 1. Default value is
12307 0.
12308 0.0 makes pixels either fully gray, or not gray at all.
12310 Higher values result in more gray pixels, with a higher gray pixel
12312 the more similar the pixels color is to the key color.
12313 hsvkey
12315 Turns a certain HSV range into transparency.
12316 This filter measures color difference between set HSV color in options
12318 and ones measured in video stream. Depending on options, output colors
12319 can be changed to transparent by adding alpha channel.
12320 The filter accepts the following options:
12322 hue Set the hue value which will be used in color difference
12324 calculation. Allowed range is from -360 to 360. Default value is
12325 0.
12326 sat Set the saturation value which will be used in color difference
12328 calculation. Allowed range is from -1 to 1. Default value is 0.
12329 val Set the value which will be used in color difference calculation.
12331 Allowed range is from -1 to 1. Default value is 0.
12332 similarity
12334 Set similarity percentage with the key color. Allowed range is
12335 from 0 to 1. Default value is 0.01.
12336 0.00001 matches only the exact key color, while 1.0 matches
12338 everything.
12339 blend
12341 Blend percentage. Allowed range is from 0 to 1. Default value is
12342 0.
12343 0.0 makes pixels either fully transparent, or not transparent at
12345 all.
12346 Higher values result in semi-transparent pixels, with a higher
12348 transparency the more similar the pixels color is to the key color.
12349 hue
12351 Modify the hue and/or the saturation of the input.
12352 It accepts the following parameters:
12354 h Specify the hue angle as a number of degrees. It accepts an
12356 expression, and defaults to "0".
12357 s Specify the saturation in the [-10,10] range. It accepts an
12359 expression and defaults to "1".
12360 H Specify the hue angle as a number of radians. It accepts an
12362 expression, and defaults to "0".
12363 b Specify the brightness in the [-10,10] range. It accepts an
12365 expression and defaults to "0".
12366 h and H are mutually exclusive, and can't be specified at the same
12368 time.
12369 The b, h, H and s option values are expressions containing the
12371 following constants:
12372 n frame count of the input frame starting from 0
12374 pts presentation timestamp of the input frame expressed in time base
12376 units
12377 r frame rate of the input video, NAN if the input frame rate is
12379 unknown
12380 t timestamp expressed in seconds, NAN if the input timestamp is
12382 unknown
12383 tb time base of the input video
12385 Examples
12387 • Set the hue to 90 degrees and the saturation to 1.0:
12389 hue=h=90:s=1
12391 • Same command but expressing the hue in radians:
12393 hue=H=PI/2:s=1
12395 • Rotate hue and make the saturation swing between 0 and 2 over a
12397 period of 1 second:
12398 hue="H=2*PI*t: s=sin(2*PI*t)+1"
12400 • Apply a 3 seconds saturation fade-in effect starting at 0:
12402 hue="s=min(t/3\,1)"
12404 The general fade-in expression can be written as:
12406 hue="s=min(0\, max((t-START)/DURATION\, 1))"
12408 • Apply a 3 seconds saturation fade-out effect starting at 5 seconds:
12410 hue="s=max(0\, min(1\, (8-t)/3))"
12412 The general fade-out expression can be written as:
12414 hue="s=max(0\, min(1\, (START+DURATION-t)/DURATION))"
12416 Commands
12418 This filter supports the following commands:
12420 b
12422 s
12423 h
12424 H Modify the hue and/or the saturation and/or brightness of the input
12425 video. The command accepts the same syntax of the corresponding
12426 option.
12427 If the specified expression is not valid, it is kept at its current
12429 value.
12430 huesaturation
12432 Apply hue-saturation-intensity adjustments to input video stream.
12433 This filter operates in RGB colorspace.
12435 This filter accepts the following options:
12437 hue Set the hue shift in degrees to apply. Default is 0. Allowed range
12439 is from -180 to 180.
12440 saturation
12442 Set the saturation shift. Default is 0. Allowed range is from -1
12443 to 1.
12444 intensity
12446 Set the intensity shift. Default is 0. Allowed range is from -1 to
12447 1.
12448 colors
12450 Set which primary and complementary colors are going to be
12451 adjusted. This options is set by providing one or multiple values.
12452 This can select multiple colors at once. By default all colors are
12453 selected.
12454 r Adjust reds.
12456 y Adjust yellows.
12458 g Adjust greens.
12460 c Adjust cyans.
12462 b Adjust blues.
12464 m Adjust magentas.
12466 a Adjust all colors.
12468 strength
12470 Set strength of filtering. Allowed range is from 0 to 100. Default
12471 value is 1.
12472 rw, gw, bw
12474 Set weight for each RGB component. Allowed range is from 0 to 1.
12475 By default is set to 0.333, 0.334, 0.333. Those options are used
12476 in saturation and lightess processing.
12477 lightness
12479 Set preserving lightness, by default is disabled. Adjusting hues
12480 can change lightness from original RGB triplet, with this option
12481 enabled lightness is kept at same value.
12482 hysteresis
12484 Grow first stream into second stream by connecting components. This
12485 makes it possible to build more robust edge masks.
12486 This filter accepts the following options:
12488 planes
12490 Set which planes will be processed as bitmap, unprocessed planes
12491 will be copied from first stream. By default value 0xf, all planes
12492 will be processed.
12493 threshold
12495 Set threshold which is used in filtering. If pixel component value
12496 is higher than this value filter algorithm for connecting
12497 components is activated. By default value is 0.
12498 The "hysteresis" filter also supports the framesync options.
12500 identity
12502 Obtain the identity score between two input videos.
12503 This filter takes two input videos.
12505 Both input videos must have the same resolution and pixel format for
12507 this filter to work correctly. Also it assumes that both inputs have
12508 the same number of frames, which are compared one by one.
12509 The obtained per component, average, min and max identity score is
12511 printed through the logging system.
12512 The filter stores the calculated identity scores of each frame in frame
12514 metadata.
12515 In the below example the input file main.mpg being processed is
12517 compared with the reference file ref.mpg.
12518 ffmpeg -i main.mpg -i ref.mpg -lavfi identity -f null -
12520 idet
12522 Detect video interlacing type.
12523 This filter tries to detect if the input frames are interlaced,
12525 progressive, top or bottom field first. It will also try to detect
12526 fields that are repeated between adjacent frames (a sign of telecine).
12527 Single frame detection considers only immediately adjacent frames when
12529 classifying each frame. Multiple frame detection incorporates the
12530 classification history of previous frames.
12531 The filter will log these metadata values:
12533 single.current_frame
12535 Detected type of current frame using single-frame detection. One
12536 of: ``tff'' (top field first), ``bff'' (bottom field first),
12537 ``progressive'', or ``undetermined''
12538 single.tff
12540 Cumulative number of frames detected as top field first using
12541 single-frame detection.
12542 multiple.tff
12544 Cumulative number of frames detected as top field first using
12545 multiple-frame detection.
12546 single.bff
12548 Cumulative number of frames detected as bottom field first using
12549 single-frame detection.
12550 multiple.current_frame
12552 Detected type of current frame using multiple-frame detection. One
12553 of: ``tff'' (top field first), ``bff'' (bottom field first),
12554 ``progressive'', or ``undetermined''
12555 multiple.bff
12557 Cumulative number of frames detected as bottom field first using
12558 multiple-frame detection.
12559 single.progressive
12561 Cumulative number of frames detected as progressive using single-
12562 frame detection.
12563 multiple.progressive
12565 Cumulative number of frames detected as progressive using multiple-
12566 frame detection.
12567 single.undetermined
12569 Cumulative number of frames that could not be classified using
12570 single-frame detection.
12571 multiple.undetermined
12573 Cumulative number of frames that could not be classified using
12574 multiple-frame detection.
12575 repeated.current_frame
12577 Which field in the current frame is repeated from the last. One of
12578 ``neither'', ``top'', or ``bottom''.
12579 repeated.neither
12581 Cumulative number of frames with no repeated field.
12582 repeated.top
12584 Cumulative number of frames with the top field repeated from the
12585 previous frame's top field.
12586 repeated.bottom
12588 Cumulative number of frames with the bottom field repeated from the
12589 previous frame's bottom field.
12590 The filter accepts the following options:
12592 intl_thres
12594 Set interlacing threshold.
12595 prog_thres
12597 Set progressive threshold.
12598 rep_thres
12600 Threshold for repeated field detection.
12601 half_life
12603 Number of frames after which a given frame's contribution to the
12604 statistics is halved (i.e., it contributes only 0.5 to its
12605 classification). The default of 0 means that all frames seen are
12606 given full weight of 1.0 forever.
12607 analyze_interlaced_flag
12609 When this is not 0 then idet will use the specified number of
12610 frames to determine if the interlaced flag is accurate, it will not
12611 count undetermined frames. If the flag is found to be accurate it
12612 will be used without any further computations, if it is found to be
12613 inaccurate it will be cleared without any further computations.
12614 This allows inserting the idet filter as a low computational method
12615 to clean up the interlaced flag
12616 il
12618 Deinterleave or interleave fields.
12619 This filter allows one to process interlaced images fields without
12621 deinterlacing them. Deinterleaving splits the input frame into 2 fields
12622 (so called half pictures). Odd lines are moved to the top half of the
12623 output image, even lines to the bottom half. You can process (filter)
12624 them independently and then re-interleave them.
12625 The filter accepts the following options:
12627 luma_mode, l
12629 chroma_mode, c
12630 alpha_mode, a
12631 Available values for luma_mode, chroma_mode and alpha_mode are:
12632 none
12634 Do nothing.
12635 deinterleave, d
12637 Deinterleave fields, placing one above the other.
12638 interleave, i
12640 Interleave fields. Reverse the effect of deinterleaving.
12641 Default value is "none".
12643 luma_swap, ls
12645 chroma_swap, cs
12646 alpha_swap, as
12647 Swap luma/chroma/alpha fields. Exchange even & odd lines. Default
12648 value is 0.
12649 Commands
12651 This filter supports the all above options as commands.
12653 inflate
12655 Apply inflate effect to the video.
12656 This filter replaces the pixel by the local(3x3) average by taking into
12658 account only values higher than the pixel.
12659 It accepts the following options:
12661 threshold0
12663 threshold1
12664 threshold2
12665 threshold3
12666 Limit the maximum change for each plane, default is 65535. If 0,
12667 plane will remain unchanged.
12668 Commands
12670 This filter supports the all above options as commands.
12672 interlace
12674 Simple interlacing filter from progressive contents. This interleaves
12675 upper (or lower) lines from odd frames with lower (or upper) lines from
12676 even frames, halving the frame rate and preserving image height.
12677 Original Original New Frame
12679 Frame 'j' Frame 'j+1' (tff)
12680 ========== =========== ==================
12681 Line 0 --------------------> Frame 'j' Line 0
12682 Line 1 Line 1 ----> Frame 'j+1' Line 1
12683 Line 2 ---------------------> Frame 'j' Line 2
12684 Line 3 Line 3 ----> Frame 'j+1' Line 3
12685 ... ... ...
12686 New Frame + 1 will be generated by Frame 'j+2' and Frame 'j+3' and so on
12687 It accepts the following optional parameters:
12689 scan
12691 This determines whether the interlaced frame is taken from the even
12692 (tff - default) or odd (bff) lines of the progressive frame.
12693 lowpass
12695 Vertical lowpass filter to avoid twitter interlacing and reduce
12696 moire patterns.
12697 0, off
12699 Disable vertical lowpass filter
12700 1, linear
12702 Enable linear filter (default)
12703 2, complex
12705 Enable complex filter. This will slightly less reduce twitter
12706 and moire but better retain detail and subjective sharpness
12707 impression.
12708 kerndeint
12710 Deinterlace input video by applying Donald Graft's adaptive kernel
12711 deinterling. Work on interlaced parts of a video to produce progressive
12712 frames.
12713 The description of the accepted parameters follows.
12715 thresh
12717 Set the threshold which affects the filter's tolerance when
12718 determining if a pixel line must be processed. It must be an
12719 integer in the range [0,255] and defaults to 10. A value of 0 will
12720 result in applying the process on every pixels.
12721 map Paint pixels exceeding the threshold value to white if set to 1.
12723 Default is 0.
12724 order
12726 Set the fields order. Swap fields if set to 1, leave fields alone
12727 if 0. Default is 0.
12728 sharp
12730 Enable additional sharpening if set to 1. Default is 0.
12731 twoway
12733 Enable twoway sharpening if set to 1. Default is 0.
12734 Examples
12736 • Apply default values:
12738 kerndeint=thresh=10:map=0:order=0:sharp=0:twoway=0
12740 • Enable additional sharpening:
12742 kerndeint=sharp=1
12744 • Paint processed pixels in white:
12746 kerndeint=map=1
12748 kirsch
12750 Apply kirsch operator to input video stream.
12751 The filter accepts the following option:
12753 planes
12755 Set which planes will be processed, unprocessed planes will be
12756 copied. By default value 0xf, all planes will be processed.
12757 scale
12759 Set value which will be multiplied with filtered result.
12760 delta
12762 Set value which will be added to filtered result.
12763 Commands
12765 This filter supports the all above options as commands.
12767 lagfun
12769 Slowly update darker pixels.
12770 This filter makes short flashes of light appear longer. This filter
12772 accepts the following options:
12773 decay
12775 Set factor for decaying. Default is .95. Allowed range is from 0 to
12776 1.
12777 planes
12779 Set which planes to filter. Default is all. Allowed range is from 0
12780 to 15.
12781 Commands
12783 This filter supports the all above options as commands.
12785 lenscorrection
12787 Correct radial lens distortion
12788 This filter can be used to correct for radial distortion as can result
12790 from the use of wide angle lenses, and thereby re-rectify the image. To
12791 find the right parameters one can use tools available for example as
12792 part of opencv or simply trial-and-error. To use opencv use the
12793 calibration sample (under samples/cpp) from the opencv sources and
12794 extract the k1 and k2 coefficients from the resulting matrix.
12795 Note that effectively the same filter is available in the open-source
12797 tools Krita and Digikam from the KDE project.
12798 In contrast to the vignette filter, which can also be used to
12800 compensate lens errors, this filter corrects the distortion of the
12801 image, whereas vignette corrects the brightness distribution, so you
12802 may want to use both filters together in certain cases, though you will
12803 have to take care of ordering, i.e. whether vignetting should be
12804 applied before or after lens correction.
12805 Options
12807 The filter accepts the following options:
12809 cx Relative x-coordinate of the focal point of the image, and thereby
12811 the center of the distortion. This value has a range [0,1] and is
12812 expressed as fractions of the image width. Default is 0.5.
12813 cy Relative y-coordinate of the focal point of the image, and thereby
12815 the center of the distortion. This value has a range [0,1] and is
12816 expressed as fractions of the image height. Default is 0.5.
12817 k1 Coefficient of the quadratic correction term. This value has a
12819 range [-1,1]. 0 means no correction. Default is 0.
12820 k2 Coefficient of the double quadratic correction term. This value has
12822 a range [-1,1]. 0 means no correction. Default is 0.
12823 i Set interpolation type. Can be "nearest" or "bilinear". Default is
12825 "nearest".
12826 fc Specify the color of the unmapped pixels. For the syntax of this
12828 option, check the "Color" section in the ffmpeg-utils manual.
12829 Default color is "black@0".
12830 The formula that generates the correction is:
12832 r_src = r_tgt * (1 + k1 * (r_tgt / r_0)^2 + k2 * (r_tgt / r_0)^4)
12834 where r_0 is halve of the image diagonal and r_src and r_tgt are the
12836 distances from the focal point in the source and target images,
12837 respectively.
12838 Commands
12840 This filter supports the all above options as commands.
12842 lensfun
12844 Apply lens correction via the lensfun library
12845 (<http://lensfun.sourceforge.net/>).
12846 The "lensfun" filter requires the camera make, camera model, and lens
12848 model to apply the lens correction. The filter will load the lensfun
12849 database and query it to find the corresponding camera and lens entries
12850 in the database. As long as these entries can be found with the given
12851 options, the filter can perform corrections on frames. Note that
12852 incomplete strings will result in the filter choosing the best match
12853 with the given options, and the filter will output the chosen camera
12854 and lens models (logged with level "info"). You must provide the make,
12855 camera model, and lens model as they are required.
12856 The filter accepts the following options:
12858 make
12860 The make of the camera (for example, "Canon"). This option is
12861 required.
12862 model
12864 The model of the camera (for example, "Canon EOS 100D"). This
12865 option is required.
12866 lens_model
12868 The model of the lens (for example, "Canon EF-S 18-55mm f/3.5-5.6
12869 IS STM"). This option is required.
12870 mode
12872 The type of correction to apply. The following values are valid
12873 options:
12874 vignetting
12876 Enables fixing lens vignetting.
12877 geometry
12879 Enables fixing lens geometry. This is the default.
12880 subpixel
12882 Enables fixing chromatic aberrations.
12883 vig_geo
12885 Enables fixing lens vignetting and lens geometry.
12886 vig_subpixel
12888 Enables fixing lens vignetting and chromatic aberrations.
12889 distortion
12891 Enables fixing both lens geometry and chromatic aberrations.
12892 all Enables all possible corrections.
12894 focal_length
12896 The focal length of the image/video (zoom; expected constant for
12897 video). For example, a 18--55mm lens has focal length range of
12898 [18--55], so a value in that range should be chosen when using that
12899 lens. Default 18.
12900 aperture
12902 The aperture of the image/video (expected constant for video). Note
12903 that aperture is only used for vignetting correction. Default 3.5.
12904 focus_distance
12906 The focus distance of the image/video (expected constant for
12907 video). Note that focus distance is only used for vignetting and
12908 only slightly affects the vignetting correction process. If
12909 unknown, leave it at the default value (which is 1000).
12910 scale
12912 The scale factor which is applied after transformation. After
12913 correction the video is no longer necessarily rectangular. This
12914 parameter controls how much of the resulting image is visible. The
12915 value 0 means that a value will be chosen automatically such that
12916 there is little or no unmapped area in the output image. 1.0 means
12917 that no additional scaling is done. Lower values may result in more
12918 of the corrected image being visible, while higher values may avoid
12919 unmapped areas in the output.
12920 target_geometry
12922 The target geometry of the output image/video. The following values
12923 are valid options:
12924 rectilinear (default)
12926 fisheye
12927 panoramic
12928 equirectangular
12929 fisheye_orthographic
12930 fisheye_stereographic
12931 fisheye_equisolid
12932 fisheye_thoby
12933 reverse
12934 Apply the reverse of image correction (instead of correcting
12935 distortion, apply it).
12936 interpolation
12938 The type of interpolation used when correcting distortion. The
12939 following values are valid options:
12940 nearest
12942 linear (default)
12943 lanczos
12944 Examples
12946 • Apply lens correction with make "Canon", camera model "Canon EOS
12948 100D", and lens model "Canon EF-S 18-55mm f/3.5-5.6 IS STM" with
12949 focal length of "18" and aperture of "8.0".
12950 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
12952 • Apply the same as before, but only for the first 5 seconds of
12954 video.
12955 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
12957 libvmaf
12959 Obtain the VMAF (Video Multi-Method Assessment Fusion) score between
12960 two input videos.
12961 The first input is the encoded video, and the second input is the
12963 reference video.
12964 The obtained VMAF score is printed through the logging system.
12966 It requires Netflix's vmaf library (libvmaf) as a pre-requisite. After
12968 installing the library it can be enabled using: "./configure
12969 --enable-libvmaf". If no model path is specified it uses the default
12970 model: "vmaf_v0.6.1.pkl".
12971 The filter has following options:
12973 model_path
12975 Set the model path which is to be used for SVM. Default value:
12976 "/usr/local/share/model/vmaf_v0.6.1.pkl"
12977 log_path
12979 Set the file path to be used to store logs.
12980 log_fmt
12982 Set the format of the log file (csv, json or xml).
12983 enable_transform
12985 This option can enable/disable the "score_transform" applied to the
12986 final predicted VMAF score, if you have specified score_transform
12987 option in the input parameter file passed to "run_vmaf_training.py"
12988 Default value: "false"
12989 phone_model
12991 Invokes the phone model which will generate VMAF scores higher than
12992 in the regular model, which is more suitable for laptop, TV, etc.
12993 viewing conditions. Default value: "false"
12994 psnr
12996 Enables computing psnr along with vmaf. Default value: "false"
12997 ssim
12999 Enables computing ssim along with vmaf. Default value: "false"
13000 ms_ssim
13002 Enables computing ms_ssim along with vmaf. Default value: "false"
13003 pool
13005 Set the pool method to be used for computing vmaf. Options are
13006 "min", "harmonic_mean" or "mean" (default).
13007 n_threads
13009 Set number of threads to be used when computing vmaf. Default
13010 value: 0, which makes use of all available logical processors.
13011 n_subsample
13013 Set interval for frame subsampling used when computing vmaf.
13014 Default value: 1
13015 enable_conf_interval
13017 Enables confidence interval. Default value: "false"
13018 This filter also supports the framesync options.
13020 Examples
13022 • On the below examples the input file main.mpg being processed is
13024 compared with the reference file ref.mpg.
13025 ffmpeg -i main.mpg -i ref.mpg -lavfi libvmaf -f null -
13027 • Example with options:
13029 ffmpeg -i main.mpg -i ref.mpg -lavfi libvmaf="psnr=1:log_fmt=json" -f null -
13031 • Example with options and different containers:
13033 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]libvmaf=psnr=1:log_fmt=json" -f null -
13035 limitdiff
13037 Apply limited difference filter using second and optionally third video
13038 stream.
13039 The filter accepts the following options:
13041 threshold
13043 Set the threshold to use when allowing certain differences between
13044 video streams. Any absolute difference value lower or exact than
13045 this threshold will pick pixel components from first video stream.
13046 elasticity
13048 Set the elasticity of soft thresholding when processing video
13049 streams. This value multiplied with first one sets second
13050 threshold. Any absolute difference value greater or exact than
13051 second threshold will pick pixel components from second video
13052 stream. For values between those two threshold linear interpolation
13053 between first and second video stream will be used.
13054 reference
13056 Enable the reference (third) video stream processing. By default is
13057 disabled. If set, this video stream will be used for calculating
13058 absolute difference with first video stream.
13059 planes
13061 Specify which planes will be processed. Defaults to all available.
13062 Commands
13064 This filter supports the all above options as commands except option
13066 reference.
13067 limiter
13069 Limits the pixel components values to the specified range [min, max].
13070 The filter accepts the following options:
13072 min Lower bound. Defaults to the lowest allowed value for the input.
13074 max Upper bound. Defaults to the highest allowed value for the input.
13076 planes
13078 Specify which planes will be processed. Defaults to all available.
13079 Commands
13081 This filter supports the all above options as commands.
13083 loop
13085 Loop video frames.
13086 The filter accepts the following options:
13088 loop
13090 Set the number of loops. Setting this value to -1 will result in
13091 infinite loops. Default is 0.
13092 size
13094 Set maximal size in number of frames. Default is 0.
13095 start
13097 Set first frame of loop. Default is 0.
13098 Examples
13100 • Loop single first frame infinitely:
13102 loop=loop=-1:size=1:start=0
13104 • Loop single first frame 10 times:
13106 loop=loop=10:size=1:start=0
13108 • Loop 10 first frames 5 times:
13110 loop=loop=5:size=10:start=0
13112 lut1d
13114 Apply a 1D LUT to an input video.
13115 The filter accepts the following options:
13117 file
13119 Set the 1D LUT file name.
13120 Currently supported formats:
13122 cube
13124 Iridas
13125 csp cineSpace
13127 interp
13129 Select interpolation mode.
13130 Available values are:
13132 nearest
13134 Use values from the nearest defined point.
13135 linear
13137 Interpolate values using the linear interpolation.
13138 cosine
13140 Interpolate values using the cosine interpolation.
13141 cubic
13143 Interpolate values using the cubic interpolation.
13144 spline
13146 Interpolate values using the spline interpolation.
13147 Commands
13149 This filter supports the all above options as commands.
13151 lut3d
13153 Apply a 3D LUT to an input video.
13154 The filter accepts the following options:
13156 file
13158 Set the 3D LUT file name.
13159 Currently supported formats:
13161 3dl AfterEffects
13163 cube
13165 Iridas
13166 dat DaVinci
13168 m3d Pandora
13170 csp cineSpace
13172 interp
13174 Select interpolation mode.
13175 Available values are:
13177 nearest
13179 Use values from the nearest defined point.
13180 trilinear
13182 Interpolate values using the 8 points defining a cube.
13183 tetrahedral
13185 Interpolate values using a tetrahedron.
13186 pyramid
13188 Interpolate values using a pyramid.
13189 prism
13191 Interpolate values using a prism.
13192 Commands
13194 This filter supports the "interp" option as commands.
13196 lumakey
13198 Turn certain luma values into transparency.
13199 The filter accepts the following options:
13201 threshold
13203 Set the luma which will be used as base for transparency. Default
13204 value is 0.
13205 tolerance
13207 Set the range of luma values to be keyed out. Default value is
13208 0.01.
13209 softness
13211 Set the range of softness. Default value is 0. Use this to control
13212 gradual transition from zero to full transparency.
13213 Commands
13215 This filter supports same commands as options. The command accepts the
13217 same syntax of the corresponding option.
13218 If the specified expression is not valid, it is kept at its current
13220 value.
13221 lut, lutrgb, lutyuv
13223 Compute a look-up table for binding each pixel component input value to
13224 an output value, and apply it to the input video.
13225 lutyuv applies a lookup table to a YUV input video, lutrgb to an RGB
13227 input video.
13228 These filters accept the following parameters:
13230 c0 set first pixel component expression
13232 c1 set second pixel component expression
13234 c2 set third pixel component expression
13236 c3 set fourth pixel component expression, corresponds to the alpha
13238 component
13239 r set red component expression
13241 g set green component expression
13243 b set blue component expression
13245 a alpha component expression
13247 y set Y/luminance component expression
13249 u set U/Cb component expression
13251 v set V/Cr component expression
13253 Each of them specifies the expression to use for computing the lookup
13255 table for the corresponding pixel component values.
13256 The exact component associated to each of the c* options depends on the
13258 format in input.
13259 The lut filter requires either YUV or RGB pixel formats in input,
13261 lutrgb requires RGB pixel formats in input, and lutyuv requires YUV.
13262 The expressions can contain the following constants and functions:
13264 w
13266 h The input width and height.
13267 val The input value for the pixel component.
13269 clipval
13271 The input value, clipped to the minval-maxval range.
13272 maxval
13274 The maximum value for the pixel component.
13275 minval
13277 The minimum value for the pixel component.
13278 negval
13280 The negated value for the pixel component value, clipped to the
13281 minval-maxval range; it corresponds to the expression
13282 "maxval-clipval+minval".
13283 clip(val)
13285 The computed value in val, clipped to the minval-maxval range.
13286 gammaval(gamma)
13288 The computed gamma correction value of the pixel component value,
13289 clipped to the minval-maxval range. It corresponds to the
13290 expression
13291 "pow((clipval-minval)/(maxval-minval)\,gamma)*(maxval-minval)+minval"
13292 All expressions default to "val".
13294 Commands
13296 This filter supports same commands as options.
13298 Examples
13300 • Negate input video:
13302 lutrgb="r=maxval+minval-val:g=maxval+minval-val:b=maxval+minval-val"
13304 lutyuv="y=maxval+minval-val:u=maxval+minval-val:v=maxval+minval-val"
13305 The above is the same as:
13307 lutrgb="r=negval:g=negval:b=negval"
13309 lutyuv="y=negval:u=negval:v=negval"
13310 • Negate luminance:
13312 lutyuv=y=negval
13314 • Remove chroma components, turning the video into a graytone image:
13316 lutyuv="u=128:v=128"
13318 • Apply a luma burning effect:
13320 lutyuv="y=2*val"
13322 • Remove green and blue components:
13324 lutrgb="g=0:b=0"
13326 • Set a constant alpha channel value on input:
13328 format=rgba,lutrgb=a="maxval-minval/2"
13330 • Correct luminance gamma by a factor of 0.5:
13332 lutyuv=y=gammaval(0.5)
13334 • Discard least significant bits of luma:
13336 lutyuv=y='bitand(val, 128+64+32)'
13338 • Technicolor like effect:
13340 lutyuv=u='(val-maxval/2)*2+maxval/2':v='(val-maxval/2)*2+maxval/2'
13342 lut2, tlut2
13344 The "lut2" filter takes two input streams and outputs one stream.
13345 The "tlut2" (time lut2) filter takes two consecutive frames from one
13347 single stream.
13348 This filter accepts the following parameters:
13350 c0 set first pixel component expression
13352 c1 set second pixel component expression
13354 c2 set third pixel component expression
13356 c3 set fourth pixel component expression, corresponds to the alpha
13358 component
13359 d set output bit depth, only available for "lut2" filter. By default
13361 is 0, which means bit depth is automatically picked from first
13362 input format.
13363 The "lut2" filter also supports the framesync options.
13365 Each of them specifies the expression to use for computing the lookup
13367 table for the corresponding pixel component values.
13368 The exact component associated to each of the c* options depends on the
13370 format in inputs.
13371 The expressions can contain the following constants:
13373 w
13375 h The input width and height.
13376 x The first input value for the pixel component.
13378 y The second input value for the pixel component.
13380 bdx The first input video bit depth.
13382 bdy The second input video bit depth.
13384 All expressions default to "x".
13386 Commands
13388 This filter supports the all above options as commands except option
13390 "d".
13391 Examples
13393 • Highlight differences between two RGB video streams:
13395 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)'
13397 • Highlight differences between two YUV video streams:
13399 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)'
13401 • Show max difference between two video streams:
13403 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)))'
13405 maskedclamp
13407 Clamp the first input stream with the second input and third input
13408 stream.
13409 Returns the value of first stream to be between second input stream -
13411 "undershoot" and third input stream + "overshoot".
13412 This filter accepts the following options:
13414 undershoot
13416 Default value is 0.
13417 overshoot
13419 Default value is 0.
13420 planes
13422 Set which planes will be processed as bitmap, unprocessed planes
13423 will be copied from first stream. By default value 0xf, all planes
13424 will be processed.
13425 Commands
13427 This filter supports the all above options as commands.
13429 maskedmax
13431 Merge the second and third input stream into output stream using
13432 absolute differences between second input stream and first input stream
13433 and absolute difference between third input stream and first input
13434 stream. The picked value will be from second input stream if second
13435 absolute difference is greater than first one or from third input
13436 stream otherwise.
13437 This filter accepts the following options:
13439 planes
13441 Set which planes will be processed as bitmap, unprocessed planes
13442 will be copied from first stream. By default value 0xf, all planes
13443 will be processed.
13444 Commands
13446 This filter supports the all above options as commands.
13448 maskedmerge
13450 Merge the first input stream with the second input stream using per
13451 pixel weights in the third input stream.
13452 A value of 0 in the third stream pixel component means that pixel
13454 component from first stream is returned unchanged, while maximum value
13455 (eg. 255 for 8-bit videos) means that pixel component from second
13456 stream is returned unchanged. Intermediate values define the amount of
13457 merging between both input stream's pixel components.
13458 This filter accepts the following options:
13460 planes
13462 Set which planes will be processed as bitmap, unprocessed planes
13463 will be copied from first stream. By default value 0xf, all planes
13464 will be processed.
13465 Commands
13467 This filter supports the all above options as commands.
13469 maskedmin
13471 Merge the second and third input stream into output stream using
13472 absolute differences between second input stream and first input stream
13473 and absolute difference between third input stream and first input
13474 stream. The picked value will be from second input stream if second
13475 absolute difference is less than first one or from third input stream
13476 otherwise.
13477 This filter accepts the following options:
13479 planes
13481 Set which planes will be processed as bitmap, unprocessed planes
13482 will be copied from first stream. By default value 0xf, all planes
13483 will be processed.
13484 Commands
13486 This filter supports the all above options as commands.
13488 maskedthreshold
13490 Pick pixels comparing absolute difference of two video streams with
13491 fixed threshold.
13492 If absolute difference between pixel component of first and second
13494 video stream is equal or lower than user supplied threshold than pixel
13495 component from first video stream is picked, otherwise pixel component
13496 from second video stream is picked.
13497 This filter accepts the following options:
13499 threshold
13501 Set threshold used when picking pixels from absolute difference
13502 from two input video streams.
13503 planes
13505 Set which planes will be processed as bitmap, unprocessed planes
13506 will be copied from second stream. By default value 0xf, all
13507 planes will be processed.
13508 Commands
13510 This filter supports the all above options as commands.
13512 maskfun
13514 Create mask from input video.
13515 For example it is useful to create motion masks after "tblend" filter.
13517 This filter accepts the following options:
13519 low Set low threshold. Any pixel component lower or exact than this
13521 value will be set to 0.
13522 high
13524 Set high threshold. Any pixel component higher than this value will
13525 be set to max value allowed for current pixel format.
13526 planes
13528 Set planes to filter, by default all available planes are filtered.
13529 fill
13531 Fill all frame pixels with this value.
13532 sum Set max average pixel value for frame. If sum of all pixel
13534 components is higher that this average, output frame will be
13535 completely filled with value set by fill option. Typically useful
13536 for scene changes when used in combination with "tblend" filter.
13537 Commands
13539 This filter supports the all above options as commands.
13541 mcdeint
13543 Apply motion-compensation deinterlacing.
13544 It needs one field per frame as input and must thus be used together
13546 with yadif=1/3 or equivalent.
13547 This filter is only available in ffmpeg version 4.4 or earlier.
13549 This filter accepts the following options:
13551 mode
13553 Set the deinterlacing mode.
13554 It accepts one of the following values:
13556 fast
13558 medium
13559 slow
13560 use iterative motion estimation
13561 extra_slow
13563 like slow, but use multiple reference frames.
13564 Default value is fast.
13566 parity
13568 Set the picture field parity assumed for the input video. It must
13569 be one of the following values:
13570 0, tff
13572 assume top field first
13573 1, bff
13575 assume bottom field first
13576 Default value is bff.
13578 qp Set per-block quantization parameter (QP) used by the internal
13580 encoder.
13581 Higher values should result in a smoother motion vector field but
13583 less optimal individual vectors. Default value is 1.
13584 median
13586 Pick median pixel from certain rectangle defined by radius.
13587 This filter accepts the following options:
13589 radius
13591 Set horizontal radius size. Default value is 1. Allowed range is
13592 integer from 1 to 127.
13593 planes
13595 Set which planes to process. Default is 15, which is all available
13596 planes.
13597 radiusV
13599 Set vertical radius size. Default value is 0. Allowed range is
13600 integer from 0 to 127. If it is 0, value will be picked from
13601 horizontal "radius" option.
13602 percentile
13604 Set median percentile. Default value is 0.5. Default value of 0.5
13605 will pick always median values, while 0 will pick minimum values,
13606 and 1 maximum values.
13607 Commands
13609 This filter supports same commands as options. The command accepts the
13611 same syntax of the corresponding option.
13612 If the specified expression is not valid, it is kept at its current
13614 value.
13615 mergeplanes
13617 Merge color channel components from several video streams.
13618 The filter accepts up to 4 input streams, and merge selected input
13620 planes to the output video.
13621 This filter accepts the following options:
13623 mapping
13625 Set input to output plane mapping. Default is 0.
13626 The mappings is specified as a bitmap. It should be specified as a
13628 hexadecimal number in the form 0xAa[Bb[Cc[Dd]]]. 'Aa' describes the
13629 mapping for the first plane of the output stream. 'A' sets the
13630 number of the input stream to use (from 0 to 3), and 'a' the plane
13631 number of the corresponding input to use (from 0 to 3). The rest of
13632 the mappings is similar, 'Bb' describes the mapping for the output
13633 stream second plane, 'Cc' describes the mapping for the output
13634 stream third plane and 'Dd' describes the mapping for the output
13635 stream fourth plane.
13636 format
13638 Set output pixel format. Default is "yuva444p".
13639 Examples
13641 • Merge three gray video streams of same width and height into single
13643 video stream:
13644 [a0][a1][a2]mergeplanes=0x001020:yuv444p
13646 • Merge 1st yuv444p stream and 2nd gray video stream into yuva444p
13648 video stream:
13649 [a0][a1]mergeplanes=0x00010210:yuva444p
13651 • Swap Y and A plane in yuva444p stream:
13653 format=yuva444p,mergeplanes=0x03010200:yuva444p
13655 • Swap U and V plane in yuv420p stream:
13657 format=yuv420p,mergeplanes=0x000201:yuv420p
13659 • Cast a rgb24 clip to yuv444p:
13661 format=rgb24,mergeplanes=0x000102:yuv444p
13663 mestimate
13665 Estimate and export motion vectors using block matching algorithms.
13666 Motion vectors are stored in frame side data to be used by other
13667 filters.
13668 This filter accepts the following options:
13670 method
13672 Specify the motion estimation method. Accepts one of the following
13673 values:
13674 esa Exhaustive search algorithm.
13676 tss Three step search algorithm.
13678 tdls
13680 Two dimensional logarithmic search algorithm.
13681 ntss
13683 New three step search algorithm.
13684 fss Four step search algorithm.
13686 ds Diamond search algorithm.
13688 hexbs
13690 Hexagon-based search algorithm.
13691 epzs
13693 Enhanced predictive zonal search algorithm.
13694 umh Uneven multi-hexagon search algorithm.
13696 Default value is esa.
13698 mb_size
13700 Macroblock size. Default 16.
13701 search_param
13703 Search parameter. Default 7.
13704 midequalizer
13706 Apply Midway Image Equalization effect using two video streams.
13707 Midway Image Equalization adjusts a pair of images to have the same
13709 histogram, while maintaining their dynamics as much as possible. It's
13710 useful for e.g. matching exposures from a pair of stereo cameras.
13711 This filter has two inputs and one output, which must be of same pixel
13713 format, but may be of different sizes. The output of filter is first
13714 input adjusted with midway histogram of both inputs.
13715 This filter accepts the following option:
13717 planes
13719 Set which planes to process. Default is 15, which is all available
13720 planes.
13721 minterpolate
13723 Convert the video to specified frame rate using motion interpolation.
13724 This filter accepts the following options:
13726 fps Specify the output frame rate. This can be rational e.g.
13728 "60000/1001". Frames are dropped if fps is lower than source fps.
13729 Default 60.
13730 mi_mode
13732 Motion interpolation mode. Following values are accepted:
13733 dup Duplicate previous or next frame for interpolating new ones.
13735 blend
13737 Blend source frames. Interpolated frame is mean of previous and
13738 next frames.
13739 mci Motion compensated interpolation. Following options are
13741 effective when this mode is selected:
13742 mc_mode
13744 Motion compensation mode. Following values are accepted:
13745 obmc
13747 Overlapped block motion compensation.
13748 aobmc
13750 Adaptive overlapped block motion compensation. Window
13751 weighting coefficients are controlled adaptively
13752 according to the reliabilities of the neighboring
13753 motion vectors to reduce oversmoothing.
13754 Default mode is obmc.
13756 me_mode
13758 Motion estimation mode. Following values are accepted:
13759 bidir
13761 Bidirectional motion estimation. Motion vectors are
13762 estimated for each source frame in both forward and
13763 backward directions.
13764 bilat
13766 Bilateral motion estimation. Motion vectors are
13767 estimated directly for interpolated frame.
13768 Default mode is bilat.
13770 me The algorithm to be used for motion estimation. Following
13772 values are accepted:
13773 esa Exhaustive search algorithm.
13775 tss Three step search algorithm.
13777 tdls
13779 Two dimensional logarithmic search algorithm.
13780 ntss
13782 New three step search algorithm.
13783 fss Four step search algorithm.
13785 ds Diamond search algorithm.
13787 hexbs
13789 Hexagon-based search algorithm.
13790 epzs
13792 Enhanced predictive zonal search algorithm.
13793 umh Uneven multi-hexagon search algorithm.
13795 Default algorithm is epzs.
13797 mb_size
13799 Macroblock size. Default 16.
13800 search_param
13802 Motion estimation search parameter. Default 32.
13803 vsbmc
13805 Enable variable-size block motion compensation. Motion
13806 estimation is applied with smaller block sizes at object
13807 boundaries in order to make the them less blur. Default is
13808 0 (disabled).
13809 scd Scene change detection method. Scene change leads motion vectors to
13811 be in random direction. Scene change detection replace interpolated
13812 frames by duplicate ones. May not be needed for other modes.
13813 Following values are accepted:
13814 none
13816 Disable scene change detection.
13817 fdiff
13819 Frame difference. Corresponding pixel values are compared and
13820 if it satisfies scd_threshold scene change is detected.
13821 Default method is fdiff.
13823 scd_threshold
13825 Scene change detection threshold. Default is 10..
13826 mix
13828 Mix several video input streams into one video stream.
13829 A description of the accepted options follows.
13831 inputs
13833 The number of inputs. If unspecified, it defaults to 2.
13834 weights
13836 Specify weight of each input video stream as sequence. Each weight
13837 is separated by space. If number of weights is smaller than number
13838 of frames last specified weight will be used for all remaining
13839 unset weights.
13840 scale
13842 Specify scale, if it is set it will be multiplied with sum of each
13843 weight multiplied with pixel values to give final destination pixel
13844 value. By default scale is auto scaled to sum of weights.
13845 duration
13847 Specify how end of stream is determined.
13848 longest
13850 The duration of the longest input. (default)
13851 shortest
13853 The duration of the shortest input.
13854 first
13856 The duration of the first input.
13857 Commands
13859 This filter supports the following commands:
13861 weights
13863 scale
13864 Syntax is same as option with same name.
13865 monochrome
13867 Convert video to gray using custom color filter.
13868 A description of the accepted options follows.
13870 cb Set the chroma blue spot. Allowed range is from -1 to 1. Default
13872 value is 0.
13873 cr Set the chroma red spot. Allowed range is from -1 to 1. Default
13875 value is 0.
13876 size
13878 Set the color filter size. Allowed range is from .1 to 10. Default
13879 value is 1.
13880 high
13882 Set the highlights strength. Allowed range is from 0 to 1. Default
13883 value is 0.
13884 Commands
13886 This filter supports the all above options as commands.
13888 morpho
13890 This filter allows to apply main morphological grayscale transforms,
13891 erode and dilate with arbitrary structures set in second input stream.
13892 Unlike naive implementation and much slower performance in erosion and
13894 dilation filters, when speed is critical "morpho" filter should be used
13895 instead.
13896 A description of accepted options follows,
13898 mode
13900 Set morphological transform to apply, can be:
13901 erode
13903 dilate
13904 open
13905 close
13906 gradient
13907 tophat
13908 blackhat
13909 Default is "erode".
13911 planes
13913 Set planes to filter, by default all planes except alpha are
13914 filtered.
13915 structure
13917 Set which structure video frames will be processed from second
13918 input stream, can be first or all. Default is all.
13919 The "morpho" filter also supports the framesync options.
13921 Commands
13923 This filter supports same commands as options.
13925 mpdecimate
13927 Drop frames that do not differ greatly from the previous frame in order
13928 to reduce frame rate.
13929 The main use of this filter is for very-low-bitrate encoding (e.g.
13931 streaming over dialup modem), but it could in theory be used for fixing
13932 movies that were inverse-telecined incorrectly.
13933 A description of the accepted options follows.
13935 max Set the maximum number of consecutive frames which can be dropped
13937 (if positive), or the minimum interval between dropped frames (if
13938 negative). If the value is 0, the frame is dropped disregarding the
13939 number of previous sequentially dropped frames.
13940 Default value is 0.
13942 hi
13944 lo
13945 frac
13946 Set the dropping threshold values.
13947 Values for hi and lo are for 8x8 pixel blocks and represent actual
13949 pixel value differences, so a threshold of 64 corresponds to 1 unit
13950 of difference for each pixel, or the same spread out differently
13951 over the block.
13952 A frame is a candidate for dropping if no 8x8 blocks differ by more
13954 than a threshold of hi, and if no more than frac blocks (1 meaning
13955 the whole image) differ by more than a threshold of lo.
13956 Default value for hi is 64*12, default value for lo is 64*5, and
13958 default value for frac is 0.33.
13959 msad
13961 Obtain the MSAD (Mean Sum of Absolute Differences) between two input
13962 videos.
13963 This filter takes two input videos.
13965 Both input videos must have the same resolution and pixel format for
13967 this filter to work correctly. Also it assumes that both inputs have
13968 the same number of frames, which are compared one by one.
13969 The obtained per component, average, min and max MSAD is printed
13971 through the logging system.
13972 The filter stores the calculated MSAD of each frame in frame metadata.
13974 In the below example the input file main.mpg being processed is
13976 compared with the reference file ref.mpg.
13977 ffmpeg -i main.mpg -i ref.mpg -lavfi msad -f null -
13979 negate
13981 Negate (invert) the input video.
13982 It accepts the following option:
13984 components
13986 Set components to negate.
13987 Available values for components are:
13989 y
13991 u
13992 v
13993 a
13994 r
13995 g
13996 b
13997 negate_alpha
13998 With value 1, it negates the alpha component, if present. Default
13999 value is 0.
14000 Commands
14002 This filter supports same commands as options.
14004 nlmeans
14006 Denoise frames using Non-Local Means algorithm.
14007 Each pixel is adjusted by looking for other pixels with similar
14009 contexts. This context similarity is defined by comparing their
14010 surrounding patches of size pxp. Patches are searched in an area of rxr
14011 around the pixel.
14012 Note that the research area defines centers for patches, which means
14014 some patches will be made of pixels outside that research area.
14015 The filter accepts the following options.
14017 s Set denoising strength. Default is 1.0. Must be in range [1.0,
14019 30.0].
14020 p Set patch size. Default is 7. Must be odd number in range [0, 99].
14022 pc Same as p but for chroma planes.
14024 The default value is 0 and means automatic.
14026 r Set research size. Default is 15. Must be odd number in range [0,
14028 99].
14029 rc Same as r but for chroma planes.
14031 The default value is 0 and means automatic.
14033 nnedi
14035 Deinterlace video using neural network edge directed interpolation.
14036 This filter accepts the following options:
14038 weights
14040 Mandatory option, without binary file filter can not work.
14041 Currently file can be found here:
14042 https://github.com/dubhater/vapoursynth-nnedi3/blob/master/src/nnedi3_weights.bin
14043 deint
14045 Set which frames to deinterlace, by default it is "all". Can be
14046 "all" or "interlaced".
14047 field
14049 Set mode of operation.
14050 Can be one of the following:
14052 af Use frame flags, both fields.
14054 a Use frame flags, single field.
14056 t Use top field only.
14058 b Use bottom field only.
14060 tf Use both fields, top first.
14062 bf Use both fields, bottom first.
14064 planes
14066 Set which planes to process, by default filter process all frames.
14067 nsize
14069 Set size of local neighborhood around each pixel, used by the
14070 predictor neural network.
14071 Can be one of the following:
14073 s8x6
14075 s16x6
14076 s32x6
14077 s48x6
14078 s8x4
14079 s16x4
14080 s32x4
14081 nns Set the number of neurons in predictor neural network. Can be one
14082 of the following:
14083 n16
14085 n32
14086 n64
14087 n128
14088 n256
14089 qual
14090 Controls the number of different neural network predictions that
14091 are blended together to compute the final output value. Can be
14092 "fast", default or "slow".
14093 etype
14095 Set which set of weights to use in the predictor. Can be one of
14096 the following:
14097 a, abs
14099 weights trained to minimize absolute error
14100 s, mse
14102 weights trained to minimize squared error
14103 pscrn
14105 Controls whether or not the prescreener neural network is used to
14106 decide which pixels should be processed by the predictor neural
14107 network and which can be handled by simple cubic interpolation.
14108 The prescreener is trained to know whether cubic interpolation will
14109 be sufficient for a pixel or whether it should be predicted by the
14110 predictor nn. The computational complexity of the prescreener nn
14111 is much less than that of the predictor nn. Since most pixels can
14112 be handled by cubic interpolation, using the prescreener generally
14113 results in much faster processing. The prescreener is pretty
14114 accurate, so the difference between using it and not using it is
14115 almost always unnoticeable.
14116 Can be one of the following:
14118 none
14120 original
14121 new
14122 new2
14123 new3
14124 Default is "new".
14126 Commands
14128 This filter supports same commands as options, excluding weights
14130 option.
14131 noformat
14133 Force libavfilter not to use any of the specified pixel formats for the
14134 input to the next filter.
14135 It accepts the following parameters:
14137 pix_fmts
14139 A '|'-separated list of pixel format names, such as
14140 pix_fmts=yuv420p|monow|rgb24".
14141 Examples
14143 • Force libavfilter to use a format different from yuv420p for the
14145 input to the vflip filter:
14146 noformat=pix_fmts=yuv420p,vflip
14148 • Convert the input video to any of the formats not contained in the
14150 list:
14151 noformat=yuv420p|yuv444p|yuv410p
14153 noise
14155 Add noise on video input frame.
14156 The filter accepts the following options:
14158 all_seed
14160 c0_seed
14161 c1_seed
14162 c2_seed
14163 c3_seed
14164 Set noise seed for specific pixel component or all pixel components
14165 in case of all_seed. Default value is 123457.
14166 all_strength, alls
14168 c0_strength, c0s
14169 c1_strength, c1s
14170 c2_strength, c2s
14171 c3_strength, c3s
14172 Set noise strength for specific pixel component or all pixel
14173 components in case all_strength. Default value is 0. Allowed range
14174 is [0, 100].
14175 all_flags, allf
14177 c0_flags, c0f
14178 c1_flags, c1f
14179 c2_flags, c2f
14180 c3_flags, c3f
14181 Set pixel component flags or set flags for all components if
14182 all_flags. Available values for component flags are:
14183 a averaged temporal noise (smoother)
14185 p mix random noise with a (semi)regular pattern
14187 t temporal noise (noise pattern changes between frames)
14189 u uniform noise (gaussian otherwise)
14191 Examples
14193 Add temporal and uniform noise to input video:
14195 noise=alls=20:allf=t+u
14197 normalize
14199 Normalize RGB video (aka histogram stretching, contrast stretching).
14200 See: https://en.wikipedia.org/wiki/Normalization_(image_processing)
14201 For each channel of each frame, the filter computes the input range and
14203 maps it linearly to the user-specified output range. The output range
14204 defaults to the full dynamic range from pure black to pure white.
14205 Temporal smoothing can be used on the input range to reduce flickering
14207 (rapid changes in brightness) caused when small dark or bright objects
14208 enter or leave the scene. This is similar to the auto-exposure
14209 (automatic gain control) on a video camera, and, like a video camera,
14210 it may cause a period of over- or under-exposure of the video.
14211 The R,G,B channels can be normalized independently, which may cause
14213 some color shifting, or linked together as a single channel, which
14214 prevents color shifting. Linked normalization preserves hue.
14215 Independent normalization does not, so it can be used to remove some
14216 color casts. Independent and linked normalization can be combined in
14217 any ratio.
14218 The normalize filter accepts the following options:
14220 blackpt
14222 whitept
14223 Colors which define the output range. The minimum input value is
14224 mapped to the blackpt. The maximum input value is mapped to the
14225 whitept. The defaults are black and white respectively. Specifying
14226 white for blackpt and black for whitept will give color-inverted,
14227 normalized video. Shades of grey can be used to reduce the dynamic
14228 range (contrast). Specifying saturated colors here can create some
14229 interesting effects.
14230 smoothing
14232 The number of previous frames to use for temporal smoothing. The
14233 input range of each channel is smoothed using a rolling average
14234 over the current frame and the smoothing previous frames. The
14235 default is 0 (no temporal smoothing).
14236 independence
14238 Controls the ratio of independent (color shifting) channel
14239 normalization to linked (color preserving) normalization. 0.0 is
14240 fully linked, 1.0 is fully independent. Defaults to 1.0 (fully
14241 independent).
14242 strength
14244 Overall strength of the filter. 1.0 is full strength. 0.0 is a
14245 rather expensive no-op. Defaults to 1.0 (full strength).
14246 Commands
14248 This filter supports same commands as options, excluding smoothing
14250 option. The command accepts the same syntax of the corresponding
14251 option.
14252 If the specified expression is not valid, it is kept at its current
14254 value.
14255 Examples
14257 Stretch video contrast to use the full dynamic range, with no temporal
14259 smoothing; may flicker depending on the source content:
14260 normalize=blackpt=black:whitept=white:smoothing=0
14262 As above, but with 50 frames of temporal smoothing; flicker should be
14264 reduced, depending on the source content:
14265 normalize=blackpt=black:whitept=white:smoothing=50
14267 As above, but with hue-preserving linked channel normalization:
14269 normalize=blackpt=black:whitept=white:smoothing=50:independence=0
14271 As above, but with half strength:
14273 normalize=blackpt=black:whitept=white:smoothing=50:independence=0:strength=0.5
14275 Map the darkest input color to red, the brightest input color to cyan:
14277 normalize=blackpt=red:whitept=cyan
14279 null
14281 Pass the video source unchanged to the output.
14282 ocr
14284 Optical Character Recognition
14285 This filter uses Tesseract for optical character recognition. To enable
14287 compilation of this filter, you need to configure FFmpeg with
14288 "--enable-libtesseract".
14289 It accepts the following options:
14291 datapath
14293 Set datapath to tesseract data. Default is to use whatever was set
14294 at installation.
14295 language
14297 Set language, default is "eng".
14298 whitelist
14300 Set character whitelist.
14301 blacklist
14303 Set character blacklist.
14304 The filter exports recognized text as the frame metadata
14306 "lavfi.ocr.text". The filter exports confidence of recognized words as
14307 the frame metadata "lavfi.ocr.confidence".
14308 ocv
14310 Apply a video transform using libopencv.
14311 To enable this filter, install the libopencv library and headers and
14313 configure FFmpeg with "--enable-libopencv".
14314 It accepts the following parameters:
14316 filter_name
14318 The name of the libopencv filter to apply.
14319 filter_params
14321 The parameters to pass to the libopencv filter. If not specified,
14322 the default values are assumed.
14323 Refer to the official libopencv documentation for more precise
14325 information:
14326 <http://docs.opencv.org/master/modules/imgproc/doc/filtering.html>
14327 Several libopencv filters are supported; see the following subsections.
14329 dilate
14331 Dilate an image by using a specific structuring element. It
14333 corresponds to the libopencv function "cvDilate".
14334 It accepts the parameters: struct_el|nb_iterations.
14336 struct_el represents a structuring element, and has the syntax:
14338 colsxrows+anchor_xxanchor_y/shape
14339 cols and rows represent the number of columns and rows of the
14341 structuring element, anchor_x and anchor_y the anchor point, and shape
14342 the shape for the structuring element. shape must be "rect", "cross",
14343 "ellipse", or "custom".
14344 If the value for shape is "custom", it must be followed by a string of
14346 the form "=filename". The file with name filename is assumed to
14347 represent a binary image, with each printable character corresponding
14348 to a bright pixel. When a custom shape is used, cols and rows are
14349 ignored, the number or columns and rows of the read file are assumed
14350 instead.
14351 The default value for struct_el is "3x3+0x0/rect".
14353 nb_iterations specifies the number of times the transform is applied to
14355 the image, and defaults to 1.
14356 Some examples:
14358 # Use the default values
14360 ocv=dilate
14361 # Dilate using a structuring element with a 5x5 cross, iterating two times
14363 ocv=filter_name=dilate:filter_params=5x5+2x2/cross|2
14364 # Read the shape from the file diamond.shape, iterating two times.
14366 # The file diamond.shape may contain a pattern of characters like this
14367 # *
14368 # ***
14369 # *****
14370 # ***
14371 # *
14372 # The specified columns and rows are ignored
14373 # but the anchor point coordinates are not
14374 ocv=dilate:0x0+2x2/custom=diamond.shape|2
14375 erode
14377 Erode an image by using a specific structuring element. It corresponds
14379 to the libopencv function "cvErode".
14380 It accepts the parameters: struct_el:nb_iterations, with the same
14382 syntax and semantics as the dilate filter.
14383 smooth
14385 Smooth the input video.
14387 The filter takes the following parameters:
14389 type|param1|param2|param3|param4.
14390 type is the type of smooth filter to apply, and must be one of the
14392 following values: "blur", "blur_no_scale", "median", "gaussian", or
14393 "bilateral". The default value is "gaussian".
14394 The meaning of param1, param2, param3, and param4 depends on the smooth
14396 type. param1 and param2 accept integer positive values or 0. param3 and
14397 param4 accept floating point values.
14398 The default value for param1 is 3. The default value for the other
14400 parameters is 0.
14401 These parameters correspond to the parameters assigned to the libopencv
14403 function "cvSmooth".
14404 oscilloscope
14406 2D Video Oscilloscope.
14407 Useful to measure spatial impulse, step responses, chroma delays, etc.
14409 It accepts the following parameters:
14411 x Set scope center x position.
14413 y Set scope center y position.
14415 s Set scope size, relative to frame diagonal.
14417 t Set scope tilt/rotation.
14419 o Set trace opacity.
14421 tx Set trace center x position.
14423 ty Set trace center y position.
14425 tw Set trace width, relative to width of frame.
14427 th Set trace height, relative to height of frame.
14429 c Set which components to trace. By default it traces first three
14431 components.
14432 g Draw trace grid. By default is enabled.
14434 st Draw some statistics. By default is enabled.
14436 sc Draw scope. By default is enabled.
14438 Commands
14440 This filter supports same commands as options. The command accepts the
14442 same syntax of the corresponding option.
14443 If the specified expression is not valid, it is kept at its current
14445 value.
14446 Examples
14448 • Inspect full first row of video frame.
14450 oscilloscope=x=0.5:y=0:s=1
14452 • Inspect full last row of video frame.
14454 oscilloscope=x=0.5:y=1:s=1
14456 • Inspect full 5th line of video frame of height 1080.
14458 oscilloscope=x=0.5:y=5/1080:s=1
14460 • Inspect full last column of video frame.
14462 oscilloscope=x=1:y=0.5:s=1:t=1
14464 overlay
14466 Overlay one video on top of another.
14467 It takes two inputs and has one output. The first input is the "main"
14469 video on which the second input is overlaid.
14470 It accepts the following parameters:
14472 A description of the accepted options follows.
14474 x
14476 y Set the expression for the x and y coordinates of the overlaid
14477 video on the main video. Default value is "0" for both expressions.
14478 In case the expression is invalid, it is set to a huge value
14479 (meaning that the overlay will not be displayed within the output
14480 visible area).
14481 eof_action
14483 See framesync.
14484 eval
14486 Set when the expressions for x, and y are evaluated.
14487 It accepts the following values:
14489 init
14491 only evaluate expressions once during the filter initialization
14492 or when a command is processed
14493 frame
14495 evaluate expressions for each incoming frame
14496 Default value is frame.
14498 shortest
14500 See framesync.
14501 format
14503 Set the format for the output video.
14504 It accepts the following values:
14506 yuv420
14508 force YUV420 output
14509 yuv420p10
14511 force YUV420p10 output
14512 yuv422
14514 force YUV422 output
14515 yuv422p10
14517 force YUV422p10 output
14518 yuv444
14520 force YUV444 output
14521 rgb force packed RGB output
14523 gbrp
14525 force planar RGB output
14526 auto
14528 automatically pick format
14529 Default value is yuv420.
14531 repeatlast
14533 See framesync.
14534 alpha
14536 Set format of alpha of the overlaid video, it can be straight or
14537 premultiplied. Default is straight.
14538 The x, and y expressions can contain the following parameters.
14540 main_w, W
14542 main_h, H
14543 The main input width and height.
14544 overlay_w, w
14546 overlay_h, h
14547 The overlay input width and height.
14548 x
14550 y The computed values for x and y. They are evaluated for each new
14551 frame.
14552 hsub
14554 vsub
14555 horizontal and vertical chroma subsample values of the output
14556 format. For example for the pixel format "yuv422p" hsub is 2 and
14557 vsub is 1.
14558 n the number of input frame, starting from 0
14560 pos the position in the file of the input frame, NAN if unknown
14562 t The timestamp, expressed in seconds. It's NAN if the input
14564 timestamp is unknown.
14565 This filter also supports the framesync options.
14567 Note that the n, pos, t variables are available only when evaluation is
14569 done per frame, and will evaluate to NAN when eval is set to init.
14570 Be aware that frames are taken from each input video in timestamp
14572 order, hence, if their initial timestamps differ, it is a good idea to
14573 pass the two inputs through a setpts=PTS-STARTPTS filter to have them
14574 begin in the same zero timestamp, as the example for the movie filter
14575 does.
14576 You can chain together more overlays but you should test the efficiency
14578 of such approach.
14579 Commands
14581 This filter supports the following commands:
14583 x
14585 y Modify the x and y of the overlay input. The command accepts the
14586 same syntax of the corresponding option.
14587 If the specified expression is not valid, it is kept at its current
14589 value.
14590 Examples
14592 • Draw the overlay at 10 pixels from the bottom right corner of the
14594 main video:
14595 overlay=main_w-overlay_w-10:main_h-overlay_h-10
14597 Using named options the example above becomes:
14599 overlay=x=main_w-overlay_w-10:y=main_h-overlay_h-10
14601 • Insert a transparent PNG logo in the bottom left corner of the
14603 input, using the ffmpeg tool with the "-filter_complex" option:
14604 ffmpeg -i input -i logo -filter_complex 'overlay=10:main_h-overlay_h-10' output
14606 • Insert 2 different transparent PNG logos (second logo on bottom
14608 right corner) using the ffmpeg tool:
14609 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
14611 • Add a transparent color layer on top of the main video; "WxH" must
14613 specify the size of the main input to the overlay filter:
14614 color=color=red@.3:size=WxH [over]; [in][over] overlay [out]
14616 • Play an original video and a filtered version (here with the
14618 deshake filter) side by side using the ffplay tool:
14619 ffplay input.avi -vf 'split[a][b]; [a]pad=iw*2:ih[src]; [b]deshake[filt]; [src][filt]overlay=w'
14621 The above command is the same as:
14623 ffplay input.avi -vf 'split[b], pad=iw*2[src], [b]deshake, [src]overlay=w'
14625 • Make a sliding overlay appearing from the left to the right top
14627 part of the screen starting since time 2:
14628 overlay=x='if(gte(t,2), -w+(t-2)*20, NAN)':y=0
14630 • Compose output by putting two input videos side to side:
14632 ffmpeg -i left.avi -i right.avi -filter_complex "
14634 nullsrc=size=200x100 [background];
14635 [0:v] setpts=PTS-STARTPTS, scale=100x100 [left];
14636 [1:v] setpts=PTS-STARTPTS, scale=100x100 [right];
14637 [background][left] overlay=shortest=1 [background+left];
14638 [background+left][right] overlay=shortest=1:x=100 [left+right]
14639 "
14640 • Mask 10-20 seconds of a video by applying the delogo filter to a
14642 section
14643 ffmpeg -i test.avi -codec:v:0 wmv2 -ar 11025 -b:v 9000k
14645 -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]'
14646 masked.avi
14647 • Chain several overlays in cascade:
14649 nullsrc=s=200x200 [bg];
14651 testsrc=s=100x100, split=4 [in0][in1][in2][in3];
14652 [in0] lutrgb=r=0, [bg] overlay=0:0 [mid0];
14653 [in1] lutrgb=g=0, [mid0] overlay=100:0 [mid1];
14654 [in2] lutrgb=b=0, [mid1] overlay=0:100 [mid2];
14655 [in3] null, [mid2] overlay=100:100 [out0]
14656 overlay_cuda
14658 Overlay one video on top of another.
14659 This is the CUDA variant of the overlay filter. It only accepts CUDA
14661 frames. The underlying input pixel formats have to match.
14662 It takes two inputs and has one output. The first input is the "main"
14664 video on which the second input is overlaid.
14665 It accepts the following parameters:
14667 x
14669 y Set expressions for the x and y coordinates of the overlaid video
14670 on the main video.
14671 They can contain the following parameters:
14673 main_w, W
14675 main_h, H
14676 The main input width and height.
14677 overlay_w, w
14679 overlay_h, h
14680 The overlay input width and height.
14681 x
14683 y The computed values for x and y. They are evaluated for each
14684 new frame.
14685 n The ordinal index of the main input frame, starting from 0.
14687 pos The byte offset position in the file of the main input frame,
14689 NAN if unknown.
14690 t The timestamp of the main input frame, expressed in seconds,
14692 NAN if unknown.
14693 Default value is "0" for both expressions.
14695 eval
14697 Set when the expressions for x and y are evaluated.
14698 It accepts the following values:
14700 init
14702 Evaluate expressions once during filter initialization or when
14703 a command is processed.
14704 frame
14706 Evaluate expressions for each incoming frame
14707 Default value is frame.
14709 eof_action
14711 See framesync.
14712 shortest
14714 See framesync.
14715 repeatlast
14717 See framesync.
14718 This filter also supports the framesync options.
14720 owdenoise
14722 Apply Overcomplete Wavelet denoiser.
14723 The filter accepts the following options:
14725 depth
14727 Set depth.
14728 Larger depth values will denoise lower frequency components more,
14730 but slow down filtering.
14731 Must be an int in the range 8-16, default is 8.
14733 luma_strength, ls
14735 Set luma strength.
14736 Must be a double value in the range 0-1000, default is 1.0.
14738 chroma_strength, cs
14740 Set chroma strength.
14741 Must be a double value in the range 0-1000, default is 1.0.
14743 pad
14745 Add paddings to the input image, and place the original input at the
14746 provided x, y coordinates.
14747 It accepts the following parameters:
14749 width, w
14751 height, h
14752 Specify an expression for the size of the output image with the
14753 paddings added. If the value for width or height is 0, the
14754 corresponding input size is used for the output.
14755 The width expression can reference the value set by the height
14757 expression, and vice versa.
14758 The default value of width and height is 0.
14760 x
14762 y Specify the offsets to place the input image at within the padded
14763 area, with respect to the top/left border of the output image.
14764 The x expression can reference the value set by the y expression,
14766 and vice versa.
14767 The default value of x and y is 0.
14769 If x or y evaluate to a negative number, they'll be changed so the
14771 input image is centered on the padded area.
14772 color
14774 Specify the color of the padded area. For the syntax of this
14775 option, check the "Color" section in the ffmpeg-utils manual.
14776 The default value of color is "black".
14778 eval
14780 Specify when to evaluate width, height, x and y expression.
14781 It accepts the following values:
14783 init
14785 Only evaluate expressions once during the filter initialization
14786 or when a command is processed.
14787 frame
14789 Evaluate expressions for each incoming frame.
14790 Default value is init.
14792 aspect
14794 Pad to aspect instead to a resolution.
14795 The value for the width, height, x, and y options are expressions
14797 containing the following constants:
14798 in_w
14800 in_h
14801 The input video width and height.
14802 iw
14804 ih These are the same as in_w and in_h.
14805 out_w
14807 out_h
14808 The output width and height (the size of the padded area), as
14809 specified by the width and height expressions.
14810 ow
14812 oh These are the same as out_w and out_h.
14813 x
14815 y The x and y offsets as specified by the x and y expressions, or NAN
14816 if not yet specified.
14817 a same as iw / ih
14819 sar input sample aspect ratio
14821 dar input display aspect ratio, it is the same as (iw / ih) * sar
14823 hsub
14825 vsub
14826 The horizontal and vertical chroma subsample values. For example
14827 for the pixel format "yuv422p" hsub is 2 and vsub is 1.
14828 Examples
14830 • Add paddings with the color "violet" to the input video. The output
14832 video size is 640x480, and the top-left corner of the input video
14833 is placed at column 0, row 40
14834 pad=640:480:0:40:violet
14836 The example above is equivalent to the following command:
14838 pad=width=640:height=480:x=0:y=40:color=violet
14840 • Pad the input to get an output with dimensions increased by 3/2,
14842 and put the input video at the center of the padded area:
14843 pad="3/2*iw:3/2*ih:(ow-iw)/2:(oh-ih)/2"
14845 • Pad the input to get a squared output with size equal to the
14847 maximum value between the input width and height, and put the input
14848 video at the center of the padded area:
14849 pad="max(iw\,ih):ow:(ow-iw)/2:(oh-ih)/2"
14851 • Pad the input to get a final w/h ratio of 16:9:
14853 pad="ih*16/9:ih:(ow-iw)/2:(oh-ih)/2"
14855 • In case of anamorphic video, in order to set the output display
14857 aspect correctly, it is necessary to use sar in the expression,
14858 according to the relation:
14859 (ih * X / ih) * sar = output_dar
14861 X = output_dar / sar
14862 Thus the previous example needs to be modified to:
14864 pad="ih*16/9/sar:ih:(ow-iw)/2:(oh-ih)/2"
14866 • Double the output size and put the input video in the bottom-right
14868 corner of the output padded area:
14869 pad="2*iw:2*ih:ow-iw:oh-ih"
14871 palettegen
14873 Generate one palette for a whole video stream.
14874 It accepts the following options:
14876 max_colors
14878 Set the maximum number of colors to quantize in the palette. Note:
14879 the palette will still contain 256 colors; the unused palette
14880 entries will be black.
14881 reserve_transparent
14883 Create a palette of 255 colors maximum and reserve the last one for
14884 transparency. Reserving the transparency color is useful for GIF
14885 optimization. If not set, the maximum of colors in the palette
14886 will be 256. You probably want to disable this option for a
14887 standalone image. Set by default.
14888 transparency_color
14890 Set the color that will be used as background for transparency.
14891 stats_mode
14893 Set statistics mode.
14894 It accepts the following values:
14896 full
14898 Compute full frame histograms.
14899 diff
14901 Compute histograms only for the part that differs from previous
14902 frame. This might be relevant to give more importance to the
14903 moving part of your input if the background is static.
14904 single
14906 Compute new histogram for each frame.
14907 Default value is full.
14909 use_alpha
14911 Create a palette of colors with alpha components. Setting this,
14912 will automatically disable 'reserve_transparent'.
14913 The filter also exports the frame metadata "lavfi.color_quant_ratio"
14915 ("nb_color_in / nb_color_out") which you can use to evaluate the degree
14916 of color quantization of the palette. This information is also visible
14917 at info logging level.
14918 Examples
14920 • Generate a representative palette of a given video using ffmpeg:
14922 ffmpeg -i input.mkv -vf palettegen palette.png
14924 paletteuse
14926 Use a palette to downsample an input video stream.
14927 The filter takes two inputs: one video stream and a palette. The
14929 palette must be a 256 pixels image.
14930 It accepts the following options:
14932 dither
14934 Select dithering mode. Available algorithms are:
14935 bayer
14937 Ordered 8x8 bayer dithering (deterministic)
14938 heckbert
14940 Dithering as defined by Paul Heckbert in 1982 (simple error
14941 diffusion). Note: this dithering is sometimes considered
14942 "wrong" and is included as a reference.
14943 floyd_steinberg
14945 Floyd and Steingberg dithering (error diffusion)
14946 sierra2
14948 Frankie Sierra dithering v2 (error diffusion)
14949 sierra2_4a
14951 Frankie Sierra dithering v2 "Lite" (error diffusion)
14952 Default is sierra2_4a.
14954 bayer_scale
14956 When bayer dithering is selected, this option defines the scale of
14957 the pattern (how much the crosshatch pattern is visible). A low
14958 value means more visible pattern for less banding, and higher value
14959 means less visible pattern at the cost of more banding.
14960 The option must be an integer value in the range [0,5]. Default is
14962 2.
14963 diff_mode
14965 If set, define the zone to process
14966 rectangle
14968 Only the changing rectangle will be reprocessed. This is
14969 similar to GIF cropping/offsetting compression mechanism. This
14970 option can be useful for speed if only a part of the image is
14971 changing, and has use cases such as limiting the scope of the
14972 error diffusal dither to the rectangle that bounds the moving
14973 scene (it leads to more deterministic output if the scene
14974 doesn't change much, and as a result less moving noise and
14975 better GIF compression).
14976 Default is none.
14978 new Take new palette for each output frame.
14980 alpha_threshold
14982 Sets the alpha threshold for transparency. Alpha values above this
14983 threshold will be treated as completely opaque, and values below
14984 this threshold will be treated as completely transparent.
14985 The option must be an integer value in the range [0,255]. Default
14987 is 128.
14988 use_alpha
14990 Apply the palette by taking alpha values into account. Only useful
14991 with palettes that are containing multiple colors with alpha
14992 components. Setting this will automatically disable
14993 'alpha_treshold'.
14994 Examples
14996 • Use a palette (generated for example with palettegen) to encode a
14998 GIF using ffmpeg:
14999 ffmpeg -i input.mkv -i palette.png -lavfi paletteuse output.gif
15001 perspective
15003 Correct perspective of video not recorded perpendicular to the screen.
15004 A description of the accepted parameters follows.
15006 x0
15008 y0
15009 x1
15010 y1
15011 x2
15012 y2
15013 x3
15014 y3 Set coordinates expression for top left, top right, bottom left and
15015 bottom right corners. Default values are "0:0:W:0:0:H:W:H" with
15016 which perspective will remain unchanged. If the "sense" option is
15017 set to "source", then the specified points will be sent to the
15018 corners of the destination. If the "sense" option is set to
15019 "destination", then the corners of the source will be sent to the
15020 specified coordinates.
15021 The expressions can use the following variables:
15023 W
15025 H the width and height of video frame.
15026 in Input frame count.
15028 on Output frame count.
15030 interpolation
15032 Set interpolation for perspective correction.
15033 It accepts the following values:
15035 linear
15037 cubic
15038 Default value is linear.
15040 sense
15042 Set interpretation of coordinate options.
15043 It accepts the following values:
15045 0, source
15047 Send point in the source specified by the given coordinates to
15048 the corners of the destination.
15049 1, destination
15051 Send the corners of the source to the point in the destination
15052 specified by the given coordinates.
15053 Default value is source.
15055 eval
15057 Set when the expressions for coordinates x0,y0,...x3,y3 are
15058 evaluated.
15059 It accepts the following values:
15061 init
15063 only evaluate expressions once during the filter initialization
15064 or when a command is processed
15065 frame
15067 evaluate expressions for each incoming frame
15068 Default value is init.
15070 phase
15072 Delay interlaced video by one field time so that the field order
15073 changes.
15074 The intended use is to fix PAL movies that have been captured with the
15076 opposite field order to the film-to-video transfer.
15077 A description of the accepted parameters follows.
15079 mode
15081 Set phase mode.
15082 It accepts the following values:
15084 t Capture field order top-first, transfer bottom-first. Filter
15086 will delay the bottom field.
15087 b Capture field order bottom-first, transfer top-first. Filter
15089 will delay the top field.
15090 p Capture and transfer with the same field order. This mode only
15092 exists for the documentation of the other options to refer to,
15093 but if you actually select it, the filter will faithfully do
15094 nothing.
15095 a Capture field order determined automatically by field flags,
15097 transfer opposite. Filter selects among t and b modes on a
15098 frame by frame basis using field flags. If no field information
15099 is available, then this works just like u.
15100 u Capture unknown or varying, transfer opposite. Filter selects
15102 among t and b on a frame by frame basis by analyzing the images
15103 and selecting the alternative that produces best match between
15104 the fields.
15105 T Capture top-first, transfer unknown or varying. Filter selects
15107 among t and p using image analysis.
15108 B Capture bottom-first, transfer unknown or varying. Filter
15110 selects among b and p using image analysis.
15111 A Capture determined by field flags, transfer unknown or varying.
15113 Filter selects among t, b and p using field flags and image
15114 analysis. If no field information is available, then this works
15115 just like U. This is the default mode.
15116 U Both capture and transfer unknown or varying. Filter selects
15118 among t, b and p using image analysis only.
15119 Commands
15121 This filter supports the all above options as commands.
15123 photosensitivity
15125 Reduce various flashes in video, so to help users with epilepsy.
15126 It accepts the following options:
15128 frames, f
15130 Set how many frames to use when filtering. Default is 30.
15131 threshold, t
15133 Set detection threshold factor. Default is 1. Lower is stricter.
15134 skip
15136 Set how many pixels to skip when sampling frames. Default is 1.
15137 Allowed range is from 1 to 1024.
15138 bypass
15140 Leave frames unchanged. Default is disabled.
15141 pixdesctest
15143 Pixel format descriptor test filter, mainly useful for internal
15144 testing. The output video should be equal to the input video.
15145 For example:
15147 format=monow, pixdesctest
15149 can be used to test the monowhite pixel format descriptor definition.
15151 pixscope
15153 Display sample values of color channels. Mainly useful for checking
15154 color and levels. Minimum supported resolution is 640x480.
15155 The filters accept the following options:
15157 x Set scope X position, relative offset on X axis.
15159 y Set scope Y position, relative offset on Y axis.
15161 w Set scope width.
15163 h Set scope height.
15165 o Set window opacity. This window also holds statistics about pixel
15167 area.
15168 wx Set window X position, relative offset on X axis.
15170 wy Set window Y position, relative offset on Y axis.
15172 Commands
15174 This filter supports same commands as options.
15176 pp
15178 Enable the specified chain of postprocessing subfilters using
15179 libpostproc. This library should be automatically selected with a GPL
15180 build ("--enable-gpl"). Subfilters must be separated by '/' and can be
15181 disabled by prepending a '-'. Each subfilter and some options have a
15182 short and a long name that can be used interchangeably, i.e. dr/dering
15183 are the same.
15184 The filters accept the following options:
15186 subfilters
15188 Set postprocessing subfilters string.
15189 All subfilters share common options to determine their scope:
15191 a/autoq
15193 Honor the quality commands for this subfilter.
15194 c/chrom
15196 Do chrominance filtering, too (default).
15197 y/nochrom
15199 Do luminance filtering only (no chrominance).
15200 n/noluma
15202 Do chrominance filtering only (no luminance).
15203 These options can be appended after the subfilter name, separated by a
15205 '|'.
15206 Available subfilters are:
15208 hb/hdeblock[|difference[|flatness]]
15210 Horizontal deblocking filter
15211 difference
15213 Difference factor where higher values mean more deblocking
15214 (default: 32).
15215 flatness
15217 Flatness threshold where lower values mean more deblocking
15218 (default: 39).
15219 vb/vdeblock[|difference[|flatness]]
15221 Vertical deblocking filter
15222 difference
15224 Difference factor where higher values mean more deblocking
15225 (default: 32).
15226 flatness
15228 Flatness threshold where lower values mean more deblocking
15229 (default: 39).
15230 ha/hadeblock[|difference[|flatness]]
15232 Accurate horizontal deblocking filter
15233 difference
15235 Difference factor where higher values mean more deblocking
15236 (default: 32).
15237 flatness
15239 Flatness threshold where lower values mean more deblocking
15240 (default: 39).
15241 va/vadeblock[|difference[|flatness]]
15243 Accurate vertical deblocking filter
15244 difference
15246 Difference factor where higher values mean more deblocking
15247 (default: 32).
15248 flatness
15250 Flatness threshold where lower values mean more deblocking
15251 (default: 39).
15252 The horizontal and vertical deblocking filters share the difference and
15254 flatness values so you cannot set different horizontal and vertical
15255 thresholds.
15256 h1/x1hdeblock
15258 Experimental horizontal deblocking filter
15259 v1/x1vdeblock
15261 Experimental vertical deblocking filter
15262 dr/dering
15264 Deringing filter
15265 tn/tmpnoise[|threshold1[|threshold2[|threshold3]]], temporal noise
15267 reducer
15268 threshold1
15269 larger -> stronger filtering
15270 threshold2
15272 larger -> stronger filtering
15273 threshold3
15275 larger -> stronger filtering
15276 al/autolevels[:f/fullyrange], automatic brightness / contrast
15278 correction
15279 f/fullyrange
15280 Stretch luminance to "0-255".
15281 lb/linblenddeint
15283 Linear blend deinterlacing filter that deinterlaces the given block
15284 by filtering all lines with a "(1 2 1)" filter.
15285 li/linipoldeint
15287 Linear interpolating deinterlacing filter that deinterlaces the
15288 given block by linearly interpolating every second line.
15289 ci/cubicipoldeint
15291 Cubic interpolating deinterlacing filter deinterlaces the given
15292 block by cubically interpolating every second line.
15293 md/mediandeint
15295 Median deinterlacing filter that deinterlaces the given block by
15296 applying a median filter to every second line.
15297 fd/ffmpegdeint
15299 FFmpeg deinterlacing filter that deinterlaces the given block by
15300 filtering every second line with a "(-1 4 2 4 -1)" filter.
15301 l5/lowpass5
15303 Vertically applied FIR lowpass deinterlacing filter that
15304 deinterlaces the given block by filtering all lines with a "(-1 2 6
15305 2 -1)" filter.
15306 fq/forceQuant[|quantizer]
15308 Overrides the quantizer table from the input with the constant
15309 quantizer you specify.
15310 quantizer
15312 Quantizer to use
15313 de/default
15315 Default pp filter combination ("hb|a,vb|a,dr|a")
15316 fa/fast
15318 Fast pp filter combination ("h1|a,v1|a,dr|a")
15319 ac High quality pp filter combination ("ha|a|128|7,va|a,dr|a")
15321 Examples
15323 • Apply horizontal and vertical deblocking, deringing and automatic
15325 brightness/contrast:
15326 pp=hb/vb/dr/al
15328 • Apply default filters without brightness/contrast correction:
15330 pp=de/-al
15332 • Apply default filters and temporal denoiser:
15334 pp=default/tmpnoise|1|2|3
15336 • Apply deblocking on luminance only, and switch vertical deblocking
15338 on or off automatically depending on available CPU time:
15339 pp=hb|y/vb|a
15341 pp7
15343 Apply Postprocessing filter 7. It is variant of the spp filter, similar
15344 to spp = 6 with 7 point DCT, where only the center sample is used after
15345 IDCT.
15346 The filter accepts the following options:
15348 qp Force a constant quantization parameter. It accepts an integer in
15350 range 0 to 63. If not set, the filter will use the QP from the
15351 video stream (if available).
15352 mode
15354 Set thresholding mode. Available modes are:
15355 hard
15357 Set hard thresholding.
15358 soft
15360 Set soft thresholding (better de-ringing effect, but likely
15361 blurrier).
15362 medium
15364 Set medium thresholding (good results, default).
15365 premultiply
15367 Apply alpha premultiply effect to input video stream using first plane
15368 of second stream as alpha.
15369 Both streams must have same dimensions and same pixel format.
15371 The filter accepts the following option:
15373 planes
15375 Set which planes will be processed, unprocessed planes will be
15376 copied. By default value 0xf, all planes will be processed.
15377 inplace
15379 Do not require 2nd input for processing, instead use alpha plane
15380 from input stream.
15381 prewitt
15383 Apply prewitt operator to input video stream.
15384 The filter accepts the following option:
15386 planes
15388 Set which planes will be processed, unprocessed planes will be
15389 copied. By default value 0xf, all planes will be processed.
15390 scale
15392 Set value which will be multiplied with filtered result.
15393 delta
15395 Set value which will be added to filtered result.
15396 Commands
15398 This filter supports the all above options as commands.
15400 pseudocolor
15402 Alter frame colors in video with pseudocolors.
15403 This filter accepts the following options:
15405 c0 set pixel first component expression
15407 c1 set pixel second component expression
15409 c2 set pixel third component expression
15411 c3 set pixel fourth component expression, corresponds to the alpha
15413 component
15414 index, i
15416 set component to use as base for altering colors
15417 preset, p
15419 Pick one of built-in LUTs. By default is set to none.
15420 Available LUTs:
15422 magma
15424 inferno
15425 plasma
15426 viridis
15427 turbo
15428 cividis
15429 range1
15430 range2
15431 shadows
15432 highlights
15433 solar
15434 nominal
15435 preferred
15436 total
15437 opacity
15438 Set opacity of output colors. Allowed range is from 0 to 1.
15439 Default value is set to 1.
15440 Each of the expression options specifies the expression to use for
15442 computing the lookup table for the corresponding pixel component
15443 values.
15444 The expressions can contain the following constants and functions:
15446 w
15448 h The input width and height.
15449 val The input value for the pixel component.
15451 ymin, umin, vmin, amin
15453 The minimum allowed component value.
15454 ymax, umax, vmax, amax
15456 The maximum allowed component value.
15457 All expressions default to "val".
15459 Commands
15461 This filter supports the all above options as commands.
15463 Examples
15465 • Change too high luma values to gradient:
15467 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'"
15469 psnr
15471 Obtain the average, maximum and minimum PSNR (Peak Signal to Noise
15472 Ratio) between two input videos.
15473 This filter takes in input two input videos, the first input is
15475 considered the "main" source and is passed unchanged to the output. The
15476 second input is used as a "reference" video for computing the PSNR.
15477 Both video inputs must have the same resolution and pixel format for
15479 this filter to work correctly. Also it assumes that both inputs have
15480 the same number of frames, which are compared one by one.
15481 The obtained average PSNR is printed through the logging system.
15483 The filter stores the accumulated MSE (mean squared error) of each
15485 frame, and at the end of the processing it is averaged across all
15486 frames equally, and the following formula is applied to obtain the
15487 PSNR:
15488 PSNR = 10*log10(MAX^2/MSE)
15490 Where MAX is the average of the maximum values of each component of the
15492 image.
15493 The description of the accepted parameters follows.
15495 stats_file, f
15497 If specified the filter will use the named file to save the PSNR of
15498 each individual frame. When filename equals "-" the data is sent to
15499 standard output.
15500 stats_version
15502 Specifies which version of the stats file format to use. Details of
15503 each format are written below. Default value is 1.
15504 stats_add_max
15506 Determines whether the max value is output to the stats log.
15507 Default value is 0. Requires stats_version >= 2. If this is set
15508 and stats_version < 2, the filter will return an error.
15509 This filter also supports the framesync options.
15511 The file printed if stats_file is selected, contains a sequence of
15513 key/value pairs of the form key:value for each compared couple of
15514 frames.
15515 If a stats_version greater than 1 is specified, a header line precedes
15517 the list of per-frame-pair stats, with key value pairs following the
15518 frame format with the following parameters:
15519 psnr_log_version
15521 The version of the log file format. Will match stats_version.
15522 fields
15524 A comma separated list of the per-frame-pair parameters included in
15525 the log.
15526 A description of each shown per-frame-pair parameter follows:
15528 n sequential number of the input frame, starting from 1
15530 mse_avg
15532 Mean Square Error pixel-by-pixel average difference of the compared
15533 frames, averaged over all the image components.
15534 mse_y, mse_u, mse_v, mse_r, mse_g, mse_b, mse_a
15536 Mean Square Error pixel-by-pixel average difference of the compared
15537 frames for the component specified by the suffix.
15538 psnr_y, psnr_u, psnr_v, psnr_r, psnr_g, psnr_b, psnr_a
15540 Peak Signal to Noise ratio of the compared frames for the component
15541 specified by the suffix.
15542 max_avg, max_y, max_u, max_v
15544 Maximum allowed value for each channel, and average over all
15545 channels.
15546 Examples
15548 • For example:
15550 movie=ref_movie.mpg, setpts=PTS-STARTPTS [main];
15552 [main][ref] psnr="stats_file=stats.log" [out]
15553 On this example the input file being processed is compared with the
15555 reference file ref_movie.mpg. The PSNR of each individual frame is
15556 stored in stats.log.
15557 • Another example with different containers:
15559 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 -
15561 pullup
15563 Pulldown reversal (inverse telecine) filter, capable of handling mixed
15564 hard-telecine, 24000/1001 fps progressive, and 30000/1001 fps
15565 progressive content.
15566 The pullup filter is designed to take advantage of future context in
15568 making its decisions. This filter is stateless in the sense that it
15569 does not lock onto a pattern to follow, but it instead looks forward to
15570 the following fields in order to identify matches and rebuild
15571 progressive frames.
15572 To produce content with an even framerate, insert the fps filter after
15574 pullup, use "fps=24000/1001" if the input frame rate is 29.97fps,
15575 "fps=24" for 30fps and the (rare) telecined 25fps input.
15576 The filter accepts the following options:
15578 jl
15580 jr
15581 jt
15582 jb These options set the amount of "junk" to ignore at the left,
15583 right, top, and bottom of the image, respectively. Left and right
15584 are in units of 8 pixels, while top and bottom are in units of 2
15585 lines. The default is 8 pixels on each side.
15586 sb Set the strict breaks. Setting this option to 1 will reduce the
15588 chances of filter generating an occasional mismatched frame, but it
15589 may also cause an excessive number of frames to be dropped during
15590 high motion sequences. Conversely, setting it to -1 will make
15591 filter match fields more easily. This may help processing of video
15592 where there is slight blurring between the fields, but may also
15593 cause there to be interlaced frames in the output. Default value
15594 is 0.
15595 mp Set the metric plane to use. It accepts the following values:
15597 l Use luma plane.
15599 u Use chroma blue plane.
15601 v Use chroma red plane.
15603 This option may be set to use chroma plane instead of the default
15605 luma plane for doing filter's computations. This may improve
15606 accuracy on very clean source material, but more likely will
15607 decrease accuracy, especially if there is chroma noise (rainbow
15608 effect) or any grayscale video. The main purpose of setting mp to
15609 a chroma plane is to reduce CPU load and make pullup usable in
15610 realtime on slow machines.
15611 For best results (without duplicated frames in the output file) it is
15613 necessary to change the output frame rate. For example, to inverse
15614 telecine NTSC input:
15615 ffmpeg -i input -vf pullup -r 24000/1001 ...
15617 qp
15619 Change video quantization parameters (QP).
15620 The filter accepts the following option:
15622 qp Set expression for quantization parameter.
15624 The expression is evaluated through the eval API and can contain, among
15626 others, the following constants:
15627 known
15629 1 if index is not 129, 0 otherwise.
15630 qp Sequential index starting from -129 to 128.
15632 Examples
15634 • Some equation like:
15636 qp=2+2*sin(PI*qp)
15638 random
15640 Flush video frames from internal cache of frames into a random order.
15641 No frame is discarded. Inspired by frei0r nervous filter.
15642 frames
15644 Set size in number of frames of internal cache, in range from 2 to
15645 512. Default is 30.
15646 seed
15648 Set seed for random number generator, must be an integer included
15649 between 0 and "UINT32_MAX". If not specified, or if explicitly set
15650 to less than 0, the filter will try to use a good random seed on a
15651 best effort basis.
15652 readeia608
15654 Read closed captioning (EIA-608) information from the top lines of a
15655 video frame.
15656 This filter adds frame metadata for "lavfi.readeia608.X.cc" and
15658 "lavfi.readeia608.X.line", where "X" is the number of the identified
15659 line with EIA-608 data (starting from 0). A description of each
15660 metadata value follows:
15661 lavfi.readeia608.X.cc
15663 The two bytes stored as EIA-608 data (printed in hexadecimal).
15664 lavfi.readeia608.X.line
15666 The number of the line on which the EIA-608 data was identified and
15667 read.
15668 This filter accepts the following options:
15670 scan_min
15672 Set the line to start scanning for EIA-608 data. Default is 0.
15673 scan_max
15675 Set the line to end scanning for EIA-608 data. Default is 29.
15676 spw Set the ratio of width reserved for sync code detection. Default
15678 is 0.27. Allowed range is "[0.1 - 0.7]".
15679 chp Enable checking the parity bit. In the event of a parity error, the
15681 filter will output 0x00 for that character. Default is false.
15682 lp Lowpass lines prior to further processing. Default is enabled.
15684 Commands
15686 This filter supports the all above options as commands.
15688 Examples
15690 • Output a csv with presentation time and the first two lines of
15692 identified EIA-608 captioning data.
15693 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
15695 readvitc
15697 Read vertical interval timecode (VITC) information from the top lines
15698 of a video frame.
15699 The filter adds frame metadata key "lavfi.readvitc.tc_str" with the
15701 timecode value, if a valid timecode has been detected. Further metadata
15702 key "lavfi.readvitc.found" is set to 0/1 depending on whether timecode
15703 data has been found or not.
15704 This filter accepts the following options:
15706 scan_max
15708 Set the maximum number of lines to scan for VITC data. If the value
15709 is set to "-1" the full video frame is scanned. Default is 45.
15710 thr_b
15712 Set the luma threshold for black. Accepts float numbers in the
15713 range [0.0,1.0], default value is 0.2. The value must be equal or
15714 less than "thr_w".
15715 thr_w
15717 Set the luma threshold for white. Accepts float numbers in the
15718 range [0.0,1.0], default value is 0.6. The value must be equal or
15719 greater than "thr_b".
15720 Examples
15722 • Detect and draw VITC data onto the video frame; if no valid VITC is
15724 detected, draw "--:--:--:--" as a placeholder:
15725 ffmpeg -i input.avi -filter:v 'readvitc,drawtext=fontfile=FreeMono.ttf:text=%{metadata\\:lavfi.readvitc.tc_str\\:--\\\\\\:--\\\\\\:--\\\\\\:--}:x=(w-tw)/2:y=400-ascent'
15727 remap
15729 Remap pixels using 2nd: Xmap and 3rd: Ymap input video stream.
15730 Destination pixel at position (X, Y) will be picked from source (x, y)
15732 position where x = Xmap(X, Y) and y = Ymap(X, Y). If mapping values are
15733 out of range, zero value for pixel will be used for destination pixel.
15734 Xmap and Ymap input video streams must be of same dimensions. Output
15736 video stream will have Xmap/Ymap video stream dimensions. Xmap and
15737 Ymap input video streams are 16bit depth, single channel.
15738 format
15740 Specify pixel format of output from this filter. Can be "color" or
15741 "gray". Default is "color".
15742 fill
15744 Specify the color of the unmapped pixels. For the syntax of this
15745 option, check the "Color" section in the ffmpeg-utils manual.
15746 Default color is "black".
15747 removegrain
15749 The removegrain filter is a spatial denoiser for progressive video.
15750 m0 Set mode for the first plane.
15752 m1 Set mode for the second plane.
15754 m2 Set mode for the third plane.
15756 m3 Set mode for the fourth plane.
15758 Range of mode is from 0 to 24. Description of each mode follows:
15760 0 Leave input plane unchanged. Default.
15762 1 Clips the pixel with the minimum and maximum of the 8 neighbour
15764 pixels.
15765 2 Clips the pixel with the second minimum and maximum of the 8
15767 neighbour pixels.
15768 3 Clips the pixel with the third minimum and maximum of the 8
15770 neighbour pixels.
15771 4 Clips the pixel with the fourth minimum and maximum of the 8
15773 neighbour pixels. This is equivalent to a median filter.
15774 5 Line-sensitive clipping giving the minimal change.
15776 6 Line-sensitive clipping, intermediate.
15778 7 Line-sensitive clipping, intermediate.
15780 8 Line-sensitive clipping, intermediate.
15782 9 Line-sensitive clipping on a line where the neighbours pixels are
15784 the closest.
15785 10 Replaces the target pixel with the closest neighbour.
15787 11 [1 2 1] horizontal and vertical kernel blur.
15789 12 Same as mode 11.
15791 13 Bob mode, interpolates top field from the line where the neighbours
15793 pixels are the closest.
15794 14 Bob mode, interpolates bottom field from the line where the
15796 neighbours pixels are the closest.
15797 15 Bob mode, interpolates top field. Same as 13 but with a more
15799 complicated interpolation formula.
15800 16 Bob mode, interpolates bottom field. Same as 14 but with a more
15802 complicated interpolation formula.
15803 17 Clips the pixel with the minimum and maximum of respectively the
15805 maximum and minimum of each pair of opposite neighbour pixels.
15806 18 Line-sensitive clipping using opposite neighbours whose greatest
15808 distance from the current pixel is minimal.
15809 19 Replaces the pixel with the average of its 8 neighbours.
15811 20 Averages the 9 pixels ([1 1 1] horizontal and vertical blur).
15813 21 Clips pixels using the averages of opposite neighbour.
15815 22 Same as mode 21 but simpler and faster.
15817 23 Small edge and halo removal, but reputed useless.
15819 24 Similar as 23.
15821 removelogo
15823 Suppress a TV station logo, using an image file to determine which
15824 pixels comprise the logo. It works by filling in the pixels that
15825 comprise the logo with neighboring pixels.
15826 The filter accepts the following options:
15828 filename, f
15830 Set the filter bitmap file, which can be any image format supported
15831 by libavformat. The width and height of the image file must match
15832 those of the video stream being processed.
15833 Pixels in the provided bitmap image with a value of zero are not
15835 considered part of the logo, non-zero pixels are considered part of the
15836 logo. If you use white (255) for the logo and black (0) for the rest,
15837 you will be safe. For making the filter bitmap, it is recommended to
15838 take a screen capture of a black frame with the logo visible, and then
15839 using a threshold filter followed by the erode filter once or twice.
15840 If needed, little splotches can be fixed manually. Remember that if
15842 logo pixels are not covered, the filter quality will be much reduced.
15843 Marking too many pixels as part of the logo does not hurt as much, but
15844 it will increase the amount of blurring needed to cover over the image
15845 and will destroy more information than necessary, and extra pixels will
15846 slow things down on a large logo.
15847 repeatfields
15849 This filter uses the repeat_field flag from the Video ES headers and
15850 hard repeats fields based on its value.
15851 reverse
15853 Reverse a video clip.
15854 Warning: This filter requires memory to buffer the entire clip, so
15856 trimming is suggested.
15857 Examples
15859 • Take the first 5 seconds of a clip, and reverse it.
15861 trim=end=5,reverse
15863 rgbashift
15865 Shift R/G/B/A pixels horizontally and/or vertically.
15866 The filter accepts the following options:
15868 rh Set amount to shift red horizontally.
15870 rv Set amount to shift red vertically.
15872 gh Set amount to shift green horizontally.
15874 gv Set amount to shift green vertically.
15876 bh Set amount to shift blue horizontally.
15878 bv Set amount to shift blue vertically.
15880 ah Set amount to shift alpha horizontally.
15882 av Set amount to shift alpha vertically.
15884 edge
15886 Set edge mode, can be smear, default, or warp.
15887 Commands
15889 This filter supports the all above options as commands.
15891 roberts
15893 Apply roberts cross operator to input video stream.
15894 The filter accepts the following option:
15896 planes
15898 Set which planes will be processed, unprocessed planes will be
15899 copied. By default value 0xf, all planes will be processed.
15900 scale
15902 Set value which will be multiplied with filtered result.
15903 delta
15905 Set value which will be added to filtered result.
15906 Commands
15908 This filter supports the all above options as commands.
15910 rotate
15912 Rotate video by an arbitrary angle expressed in radians.
15913 The filter accepts the following options:
15915 A description of the optional parameters follows.
15917 angle, a
15919 Set an expression for the angle by which to rotate the input video
15920 clockwise, expressed as a number of radians. A negative value will
15921 result in a counter-clockwise rotation. By default it is set to
15922 "0".
15923 This expression is evaluated for each frame.
15925 out_w, ow
15927 Set the output width expression, default value is "iw". This
15928 expression is evaluated just once during configuration.
15929 out_h, oh
15931 Set the output height expression, default value is "ih". This
15932 expression is evaluated just once during configuration.
15933 bilinear
15935 Enable bilinear interpolation if set to 1, a value of 0 disables
15936 it. Default value is 1.
15937 fillcolor, c
15939 Set the color used to fill the output area not covered by the
15940 rotated image. For the general syntax of this option, check the
15941 "Color" section in the ffmpeg-utils manual. If the special value
15942 "none" is selected then no background is printed (useful for
15943 example if the background is never shown).
15944 Default value is "black".
15946 The expressions for the angle and the output size can contain the
15948 following constants and functions:
15949 n sequential number of the input frame, starting from 0. It is always
15951 NAN before the first frame is filtered.
15952 t time in seconds of the input frame, it is set to 0 when the filter
15954 is configured. It is always NAN before the first frame is filtered.
15955 hsub
15957 vsub
15958 horizontal and vertical chroma subsample values. For example for
15959 the pixel format "yuv422p" hsub is 2 and vsub is 1.
15960 in_w, iw
15962 in_h, ih
15963 the input video width and height
15964 out_w, ow
15966 out_h, oh
15967 the output width and height, that is the size of the padded area as
15968 specified by the width and height expressions
15969 rotw(a)
15971 roth(a)
15972 the minimal width/height required for completely containing the
15973 input video rotated by a radians.
15974 These are only available when computing the out_w and out_h
15976 expressions.
15977 Examples
15979 • Rotate the input by PI/6 radians clockwise:
15981 rotate=PI/6
15983 • Rotate the input by PI/6 radians counter-clockwise:
15985 rotate=-PI/6
15987 • Rotate the input by 45 degrees clockwise:
15989 rotate=45*PI/180
15991 • Apply a constant rotation with period T, starting from an angle of
15993 PI/3:
15994 rotate=PI/3+2*PI*t/T
15996 • Make the input video rotation oscillating with a period of T
15998 seconds and an amplitude of A radians:
15999 rotate=A*sin(2*PI/T*t)
16001 • Rotate the video, output size is chosen so that the whole rotating
16003 input video is always completely contained in the output:
16004 rotate='2*PI*t:ow=hypot(iw,ih):oh=ow'
16006 • Rotate the video, reduce the output size so that no background is
16008 ever shown:
16009 rotate=2*PI*t:ow='min(iw,ih)/sqrt(2)':oh=ow:c=none
16011 Commands
16013 The filter supports the following commands:
16015 a, angle
16017 Set the angle expression. The command accepts the same syntax of
16018 the corresponding option.
16019 If the specified expression is not valid, it is kept at its current
16021 value.
16022 sab
16024 Apply Shape Adaptive Blur.
16025 The filter accepts the following options:
16027 luma_radius, lr
16029 Set luma blur filter strength, must be a value in range 0.1-4.0,
16030 default value is 1.0. A greater value will result in a more blurred
16031 image, and in slower processing.
16032 luma_pre_filter_radius, lpfr
16034 Set luma pre-filter radius, must be a value in the 0.1-2.0 range,
16035 default value is 1.0.
16036 luma_strength, ls
16038 Set luma maximum difference between pixels to still be considered,
16039 must be a value in the 0.1-100.0 range, default value is 1.0.
16040 chroma_radius, cr
16042 Set chroma blur filter strength, must be a value in range -0.9-4.0.
16043 A greater value will result in a more blurred image, and in slower
16044 processing.
16045 chroma_pre_filter_radius, cpfr
16047 Set chroma pre-filter radius, must be a value in the -0.9-2.0
16048 range.
16049 chroma_strength, cs
16051 Set chroma maximum difference between pixels to still be
16052 considered, must be a value in the -0.9-100.0 range.
16053 Each chroma option value, if not explicitly specified, is set to the
16055 corresponding luma option value.
16056 scale
16058 Scale (resize) the input video, using the libswscale library.
16059 The scale filter forces the output display aspect ratio to be the same
16061 of the input, by changing the output sample aspect ratio.
16062 If the input image format is different from the format requested by the
16064 next filter, the scale filter will convert the input to the requested
16065 format.
16066 Options
16068 The filter accepts the following options, or any of the options
16070 supported by the libswscale scaler.
16071 See the ffmpeg-scaler manual for the complete list of scaler options.
16073 width, w
16075 height, h
16076 Set the output video dimension expression. Default value is the
16077 input dimension.
16078 If the width or w value is 0, the input width is used for the
16080 output. If the height or h value is 0, the input height is used for
16081 the output.
16082 If one and only one of the values is -n with n >= 1, the scale
16084 filter will use a value that maintains the aspect ratio of the
16085 input image, calculated from the other specified dimension. After
16086 that it will, however, make sure that the calculated dimension is
16087 divisible by n and adjust the value if necessary.
16088 If both values are -n with n >= 1, the behavior will be identical
16090 to both values being set to 0 as previously detailed.
16091 See below for the list of accepted constants for use in the
16093 dimension expression.
16094 eval
16096 Specify when to evaluate width and height expression. It accepts
16097 the following values:
16098 init
16100 Only evaluate expressions once during the filter initialization
16101 or when a command is processed.
16102 frame
16104 Evaluate expressions for each incoming frame.
16105 Default value is init.
16107 interl
16109 Set the interlacing mode. It accepts the following values:
16110 1 Force interlaced aware scaling.
16112 0 Do not apply interlaced scaling.
16114 -1 Select interlaced aware scaling depending on whether the source
16116 frames are flagged as interlaced or not.
16117 Default value is 0.
16119 flags
16121 Set libswscale scaling flags. See the ffmpeg-scaler manual for the
16122 complete list of values. If not explicitly specified the filter
16123 applies the default flags.
16124 param0, param1
16126 Set libswscale input parameters for scaling algorithms that need
16127 them. See the ffmpeg-scaler manual for the complete documentation.
16128 If not explicitly specified the filter applies empty parameters.
16129 size, s
16131 Set the video size. For the syntax of this option, check the "Video
16132 size" section in the ffmpeg-utils manual.
16133 in_color_matrix
16135 out_color_matrix
16136 Set in/output YCbCr color space type.
16137 This allows the autodetected value to be overridden as well as
16139 allows forcing a specific value used for the output and encoder.
16140 If not specified, the color space type depends on the pixel format.
16142 Possible values:
16144 auto
16146 Choose automatically.
16147 bt709
16149 Format conforming to International Telecommunication Union
16150 (ITU) Recommendation BT.709.
16151 fcc Set color space conforming to the United States Federal
16153 Communications Commission (FCC) Code of Federal Regulations
16154 (CFR) Title 47 (2003) 73.682 (a).
16155 bt601
16157 bt470
16158 smpte170m
16159 Set color space conforming to:
16160 • ITU Radiocommunication Sector (ITU-R) Recommendation BT.601
16162 • ITU-R Rec. BT.470-6 (1998) Systems B, B1, and G
16164 • Society of Motion Picture and Television Engineers (SMPTE)
16166 ST 170:2004
16167 smpte240m
16169 Set color space conforming to SMPTE ST 240:1999.
16170 bt2020
16172 Set color space conforming to ITU-R BT.2020 non-constant
16173 luminance system.
16174 in_range
16176 out_range
16177 Set in/output YCbCr sample range.
16178 This allows the autodetected value to be overridden as well as
16180 allows forcing a specific value used for the output and encoder. If
16181 not specified, the range depends on the pixel format. Possible
16182 values:
16183 auto/unknown
16185 Choose automatically.
16186 jpeg/full/pc
16188 Set full range (0-255 in case of 8-bit luma).
16189 mpeg/limited/tv
16191 Set "MPEG" range (16-235 in case of 8-bit luma).
16192 force_original_aspect_ratio
16194 Enable decreasing or increasing output video width or height if
16195 necessary to keep the original aspect ratio. Possible values:
16196 disable
16198 Scale the video as specified and disable this feature.
16199 decrease
16201 The output video dimensions will automatically be decreased if
16202 needed.
16203 increase
16205 The output video dimensions will automatically be increased if
16206 needed.
16207 One useful instance of this option is that when you know a specific
16209 device's maximum allowed resolution, you can use this to limit the
16210 output video to that, while retaining the aspect ratio. For
16211 example, device A allows 1280x720 playback, and your video is
16212 1920x800. Using this option (set it to decrease) and specifying
16213 1280x720 to the command line makes the output 1280x533.
16214 Please note that this is a different thing than specifying -1 for w
16216 or h, you still need to specify the output resolution for this
16217 option to work.
16218 force_divisible_by
16220 Ensures that both the output dimensions, width and height, are
16221 divisible by the given integer when used together with
16222 force_original_aspect_ratio. This works similar to using "-n" in
16223 the w and h options.
16224 This option respects the value set for force_original_aspect_ratio,
16226 increasing or decreasing the resolution accordingly. The video's
16227 aspect ratio may be slightly modified.
16228 This option can be handy if you need to have a video fit within or
16230 exceed a defined resolution using force_original_aspect_ratio but
16231 also have encoder restrictions on width or height divisibility.
16232 The values of the w and h options are expressions containing the
16234 following constants:
16235 in_w
16237 in_h
16238 The input width and height
16239 iw
16241 ih These are the same as in_w and in_h.
16242 out_w
16244 out_h
16245 The output (scaled) width and height
16246 ow
16248 oh These are the same as out_w and out_h
16249 a The same as iw / ih
16251 sar input sample aspect ratio
16253 dar The input display aspect ratio. Calculated from "(iw / ih) * sar".
16255 hsub
16257 vsub
16258 horizontal and vertical input chroma subsample values. For example
16259 for the pixel format "yuv422p" hsub is 2 and vsub is 1.
16260 ohsub
16262 ovsub
16263 horizontal and vertical output chroma subsample values. For example
16264 for the pixel format "yuv422p" hsub is 2 and vsub is 1.
16265 n The (sequential) number of the input frame, starting from 0. Only
16267 available with "eval=frame".
16268 t The presentation timestamp of the input frame, expressed as a
16270 number of seconds. Only available with "eval=frame".
16271 pos The position (byte offset) of the frame in the input stream, or NaN
16273 if this information is unavailable and/or meaningless (for example
16274 in case of synthetic video). Only available with "eval=frame".
16275 Examples
16277 • Scale the input video to a size of 200x100
16279 scale=w=200:h=100
16281 This is equivalent to:
16283 scale=200:100
16285 or:
16287 scale=200x100
16289 • Specify a size abbreviation for the output size:
16291 scale=qcif
16293 which can also be written as:
16295 scale=size=qcif
16297 • Scale the input to 2x:
16299 scale=w=2*iw:h=2*ih
16301 • The above is the same as:
16303 scale=2*in_w:2*in_h
16305 • Scale the input to 2x with forced interlaced scaling:
16307 scale=2*iw:2*ih:interl=1
16309 • Scale the input to half size:
16311 scale=w=iw/2:h=ih/2
16313 • Increase the width, and set the height to the same size:
16315 scale=3/2*iw:ow
16317 • Seek Greek harmony:
16319 scale=iw:1/PHI*iw
16321 scale=ih*PHI:ih
16322 • Increase the height, and set the width to 3/2 of the height:
16324 scale=w=3/2*oh:h=3/5*ih
16326 • Increase the size, making the size a multiple of the chroma
16328 subsample values:
16329 scale="trunc(3/2*iw/hsub)*hsub:trunc(3/2*ih/vsub)*vsub"
16331 • Increase the width to a maximum of 500 pixels, keeping the same
16333 aspect ratio as the input:
16334 scale=w='min(500\, iw*3/2):h=-1'
16336 • Make pixels square by combining scale and setsar:
16338 scale='trunc(ih*dar):ih',setsar=1/1
16340 • Make pixels square by combining scale and setsar, making sure the
16342 resulting resolution is even (required by some codecs):
16343 scale='trunc(ih*dar/2)*2:trunc(ih/2)*2',setsar=1/1
16345 Commands
16347 This filter supports the following commands:
16349 width, w
16351 height, h
16352 Set the output video dimension expression. The command accepts the
16353 same syntax of the corresponding option.
16354 If the specified expression is not valid, it is kept at its current
16356 value.
16357 scale_cuda
16359 Scale (resize) and convert (pixel format) the input video, using
16360 accelerated CUDA kernels. Setting the output width and height works in
16361 the same way as for the scale filter.
16362 The filter accepts the following options:
16364 w
16366 h Set the output video dimension expression. Default value is the
16367 input dimension.
16368 Allows for the same expressions as the scale filter.
16370 interp_algo
16372 Sets the algorithm used for scaling:
16373 nearest
16375 Nearest neighbour
16376 Used by default if input parameters match the desired output.
16378 bilinear
16380 Bilinear
16381 bicubic
16383 Bicubic
16384 This is the default.
16386 lanczos
16388 Lanczos
16389 format
16391 Controls the output pixel format. By default, or if none is
16392 specified, the input pixel format is used.
16393 The filter does not support converting between YUV and RGB pixel
16395 formats.
16396 passthrough
16398 If set to 0, every frame is processed, even if no conversion is
16399 neccesary. This mode can be useful to use the filter as a buffer
16400 for a downstream frame-consumer that exhausts the limited decoder
16401 frame pool.
16402 If set to 1, frames are passed through as-is if they match the
16404 desired output parameters. This is the default behaviour.
16405 param
16407 Algorithm-Specific parameter.
16408 Affects the curves of the bicubic algorithm.
16410 force_original_aspect_ratio
16412 force_divisible_by
16413 Work the same as the identical scale filter options.
16414 Examples
16416 • Scale input to 720p, keeping aspect ratio and ensuring the output
16418 is yuv420p.
16419 scale_cuda=-2:720:format=yuv420p
16421 • Upscale to 4K using nearest neighbour algorithm.
16423 scale_cuda=4096:2160:interp_algo=nearest
16425 • Don't do any conversion or scaling, but copy all input frames into
16427 newly allocated ones. This can be useful to deal with a filter and
16428 encode chain that otherwise exhausts the decoders frame pool.
16429 scale_cuda=passthrough=0
16431 scale_npp
16433 Use the NVIDIA Performance Primitives (libnpp) to perform scaling
16434 and/or pixel format conversion on CUDA video frames. Setting the output
16435 width and height works in the same way as for the scale filter.
16436 The following additional options are accepted:
16438 format
16440 The pixel format of the output CUDA frames. If set to the string
16441 "same" (the default), the input format will be kept. Note that
16442 automatic format negotiation and conversion is not yet supported
16443 for hardware frames
16444 interp_algo
16446 The interpolation algorithm used for resizing. One of the
16447 following:
16448 nn Nearest neighbour.
16450 linear
16452 cubic
16453 cubic2p_bspline
16454 2-parameter cubic (B=1, C=0)
16455 cubic2p_catmullrom
16457 2-parameter cubic (B=0, C=1/2)
16458 cubic2p_b05c03
16460 2-parameter cubic (B=1/2, C=3/10)
16461 super
16463 Supersampling
16464 lanczos
16466 force_original_aspect_ratio
16467 Enable decreasing or increasing output video width or height if
16468 necessary to keep the original aspect ratio. Possible values:
16469 disable
16471 Scale the video as specified and disable this feature.
16472 decrease
16474 The output video dimensions will automatically be decreased if
16475 needed.
16476 increase
16478 The output video dimensions will automatically be increased if
16479 needed.
16480 One useful instance of this option is that when you know a specific
16482 device's maximum allowed resolution, you can use this to limit the
16483 output video to that, while retaining the aspect ratio. For
16484 example, device A allows 1280x720 playback, and your video is
16485 1920x800. Using this option (set it to decrease) and specifying
16486 1280x720 to the command line makes the output 1280x533.
16487 Please note that this is a different thing than specifying -1 for w
16489 or h, you still need to specify the output resolution for this
16490 option to work.
16491 force_divisible_by
16493 Ensures that both the output dimensions, width and height, are
16494 divisible by the given integer when used together with
16495 force_original_aspect_ratio. This works similar to using "-n" in
16496 the w and h options.
16497 This option respects the value set for force_original_aspect_ratio,
16499 increasing or decreasing the resolution accordingly. The video's
16500 aspect ratio may be slightly modified.
16501 This option can be handy if you need to have a video fit within or
16503 exceed a defined resolution using force_original_aspect_ratio but
16504 also have encoder restrictions on width or height divisibility.
16505 eval
16507 Specify when to evaluate width and height expression. It accepts
16508 the following values:
16509 init
16511 Only evaluate expressions once during the filter initialization
16512 or when a command is processed.
16513 frame
16515 Evaluate expressions for each incoming frame.
16516 The values of the w and h options are expressions containing the
16518 following constants:
16519 in_w
16521 in_h
16522 The input width and height
16523 iw
16525 ih These are the same as in_w and in_h.
16526 out_w
16528 out_h
16529 The output (scaled) width and height
16530 ow
16532 oh These are the same as out_w and out_h
16533 a The same as iw / ih
16535 sar input sample aspect ratio
16537 dar The input display aspect ratio. Calculated from "(iw / ih) * sar".
16539 n The (sequential) number of the input frame, starting from 0. Only
16541 available with "eval=frame".
16542 t The presentation timestamp of the input frame, expressed as a
16544 number of seconds. Only available with "eval=frame".
16545 pos The position (byte offset) of the frame in the input stream, or NaN
16547 if this information is unavailable and/or meaningless (for example
16548 in case of synthetic video). Only available with "eval=frame".
16549 scale2ref
16551 Scale (resize) the input video, based on a reference video.
16552 See the scale filter for available options, scale2ref supports the same
16554 but uses the reference video instead of the main input as basis.
16555 scale2ref also supports the following additional constants for the w
16556 and h options:
16557 main_w
16559 main_h
16560 The main input video's width and height
16561 main_a
16563 The same as main_w / main_h
16564 main_sar
16566 The main input video's sample aspect ratio
16567 main_dar, mdar
16569 The main input video's display aspect ratio. Calculated from
16570 "(main_w / main_h) * main_sar".
16571 main_hsub
16573 main_vsub
16574 The main input video's horizontal and vertical chroma subsample
16575 values. For example for the pixel format "yuv422p" hsub is 2 and
16576 vsub is 1.
16577 main_n
16579 The (sequential) number of the main input frame, starting from 0.
16580 Only available with "eval=frame".
16581 main_t
16583 The presentation timestamp of the main input frame, expressed as a
16584 number of seconds. Only available with "eval=frame".
16585 main_pos
16587 The position (byte offset) of the frame in the main input stream,
16588 or NaN if this information is unavailable and/or meaningless (for
16589 example in case of synthetic video). Only available with
16590 "eval=frame".
16591 Examples
16593 • Scale a subtitle stream (b) to match the main video (a) in size
16595 before overlaying
16596 'scale2ref[b][a];[a][b]overlay'
16598 • Scale a logo to 1/10th the height of a video, while preserving its
16600 display aspect ratio.
16601 [logo-in][video-in]scale2ref=w=oh*mdar:h=ih/10[logo-out][video-out]
16603 Commands
16605 This filter supports the following commands:
16607 width, w
16609 height, h
16610 Set the output video dimension expression. The command accepts the
16611 same syntax of the corresponding option.
16612 If the specified expression is not valid, it is kept at its current
16614 value.
16615 scale2ref_npp
16617 Use the NVIDIA Performance Primitives (libnpp) to scale (resize) the
16618 input video, based on a reference video.
16619 See the scale_npp filter for available options, scale2ref_npp supports
16621 the same but uses the reference video instead of the main input as
16622 basis. scale2ref_npp also supports the following additional constants
16623 for the w and h options:
16624 main_w
16626 main_h
16627 The main input video's width and height
16628 main_a
16630 The same as main_w / main_h
16631 main_sar
16633 The main input video's sample aspect ratio
16634 main_dar, mdar
16636 The main input video's display aspect ratio. Calculated from
16637 "(main_w / main_h) * main_sar".
16638 main_n
16640 The (sequential) number of the main input frame, starting from 0.
16641 Only available with "eval=frame".
16642 main_t
16644 The presentation timestamp of the main input frame, expressed as a
16645 number of seconds. Only available with "eval=frame".
16646 main_pos
16648 The position (byte offset) of the frame in the main input stream,
16649 or NaN if this information is unavailable and/or meaningless (for
16650 example in case of synthetic video). Only available with
16651 "eval=frame".
16652 Examples
16654 • Scale a subtitle stream (b) to match the main video (a) in size
16656 before overlaying
16657 'scale2ref_npp[b][a];[a][b]overlay_cuda'
16659 • Scale a logo to 1/10th the height of a video, while preserving its
16661 display aspect ratio.
16662 [logo-in][video-in]scale2ref_npp=w=oh*mdar:h=ih/10[logo-out][video-out]
16664 scharr
16666 Apply scharr operator to input video stream.
16667 The filter accepts the following option:
16669 planes
16671 Set which planes will be processed, unprocessed planes will be
16672 copied. By default value 0xf, all planes will be processed.
16673 scale
16675 Set value which will be multiplied with filtered result.
16676 delta
16678 Set value which will be added to filtered result.
16679 Commands
16681 This filter supports the all above options as commands.
16683 scroll
16685 Scroll input video horizontally and/or vertically by constant speed.
16686 The filter accepts the following options:
16688 horizontal, h
16690 Set the horizontal scrolling speed. Default is 0. Allowed range is
16691 from -1 to 1. Negative values changes scrolling direction.
16692 vertical, v
16694 Set the vertical scrolling speed. Default is 0. Allowed range is
16695 from -1 to 1. Negative values changes scrolling direction.
16696 hpos
16698 Set the initial horizontal scrolling position. Default is 0.
16699 Allowed range is from 0 to 1.
16700 vpos
16702 Set the initial vertical scrolling position. Default is 0. Allowed
16703 range is from 0 to 1.
16704 Commands
16706 This filter supports the following commands:
16708 horizontal, h
16710 Set the horizontal scrolling speed.
16711 vertical, v
16713 Set the vertical scrolling speed.
16714 scdet
16716 Detect video scene change.
16717 This filter sets frame metadata with mafd between frame, the scene
16719 score, and forward the frame to the next filter, so they can use these
16720 metadata to detect scene change or others.
16721 In addition, this filter logs a message and sets frame metadata when it
16723 detects a scene change by threshold.
16724 "lavfi.scd.mafd" metadata keys are set with mafd for every frame.
16726 "lavfi.scd.score" metadata keys are set with scene change score for
16728 every frame to detect scene change.
16729 "lavfi.scd.time" metadata keys are set with current filtered frame time
16731 which detect scene change with threshold.
16732 The filter accepts the following options:
16734 threshold, t
16736 Set the scene change detection threshold as a percentage of maximum
16737 change. Good values are in the "[8.0, 14.0]" range. The range for
16738 threshold is "[0., 100.]".
16739 Default value is 10..
16741 sc_pass, s
16743 Set the flag to pass scene change frames to the next filter.
16744 Default value is 0 You can enable it if you want to get snapshot of
16745 scene change frames only.
16746 selectivecolor
16748 Adjust cyan, magenta, yellow and black (CMYK) to certain ranges of
16749 colors (such as "reds", "yellows", "greens", "cyans", ...). The
16750 adjustment range is defined by the "purity" of the color (that is, how
16751 saturated it already is).
16752 This filter is similar to the Adobe Photoshop Selective Color tool.
16754 The filter accepts the following options:
16756 correction_method
16758 Select color correction method.
16759 Available values are:
16761 absolute
16763 Specified adjustments are applied "as-is" (added/subtracted to
16764 original pixel component value).
16765 relative
16767 Specified adjustments are relative to the original component
16768 value.
16769 Default is "absolute".
16771 reds
16773 Adjustments for red pixels (pixels where the red component is the
16774 maximum)
16775 yellows
16777 Adjustments for yellow pixels (pixels where the blue component is
16778 the minimum)
16779 greens
16781 Adjustments for green pixels (pixels where the green component is
16782 the maximum)
16783 cyans
16785 Adjustments for cyan pixels (pixels where the red component is the
16786 minimum)
16787 blues
16789 Adjustments for blue pixels (pixels where the blue component is the
16790 maximum)
16791 magentas
16793 Adjustments for magenta pixels (pixels where the green component is
16794 the minimum)
16795 whites
16797 Adjustments for white pixels (pixels where all components are
16798 greater than 128)
16799 neutrals
16801 Adjustments for all pixels except pure black and pure white
16802 blacks
16804 Adjustments for black pixels (pixels where all components are
16805 lesser than 128)
16806 psfile
16808 Specify a Photoshop selective color file (".asv") to import the
16809 settings from.
16810 All the adjustment settings (reds, yellows, ...) accept up to 4 space
16812 separated floating point adjustment values in the [-1,1] range,
16813 respectively to adjust the amount of cyan, magenta, yellow and black
16814 for the pixels of its range.
16815 Examples
16817 • Increase cyan by 50% and reduce yellow by 33% in every green areas,
16819 and increase magenta by 27% in blue areas:
16820 selectivecolor=greens=.5 0 -.33 0:blues=0 .27
16822 • Use a Photoshop selective color preset:
16824 selectivecolor=psfile=MySelectiveColorPresets/Misty.asv
16826 separatefields
16828 The "separatefields" takes a frame-based video input and splits each
16829 frame into its components fields, producing a new half height clip with
16830 twice the frame rate and twice the frame count.
16831 This filter use field-dominance information in frame to decide which of
16833 each pair of fields to place first in the output. If it gets it wrong
16834 use setfield filter before "separatefields" filter.
16835 setdar, setsar
16837 The "setdar" filter sets the Display Aspect Ratio for the filter output
16838 video.
16839 This is done by changing the specified Sample (aka Pixel) Aspect Ratio,
16841 according to the following equation:
16842 <DAR> = <HORIZONTAL_RESOLUTION> / <VERTICAL_RESOLUTION> * <SAR>
16844 Keep in mind that the "setdar" filter does not modify the pixel
16846 dimensions of the video frame. Also, the display aspect ratio set by
16847 this filter may be changed by later filters in the filterchain, e.g. in
16848 case of scaling or if another "setdar" or a "setsar" filter is applied.
16849 The "setsar" filter sets the Sample (aka Pixel) Aspect Ratio for the
16851 filter output video.
16852 Note that as a consequence of the application of this filter, the
16854 output display aspect ratio will change according to the equation
16855 above.
16856 Keep in mind that the sample aspect ratio set by the "setsar" filter
16858 may be changed by later filters in the filterchain, e.g. if another
16859 "setsar" or a "setdar" filter is applied.
16860 It accepts the following parameters:
16862 r, ratio, dar ("setdar" only), sar ("setsar" only)
16864 Set the aspect ratio used by the filter.
16865 The parameter can be a floating point number string, an expression,
16867 or a string of the form num:den, where num and den are the
16868 numerator and denominator of the aspect ratio. If the parameter is
16869 not specified, it is assumed the value "0". In case the form
16870 "num:den" is used, the ":" character should be escaped.
16871 max Set the maximum integer value to use for expressing numerator and
16873 denominator when reducing the expressed aspect ratio to a rational.
16874 Default value is 100.
16875 The parameter sar is an expression containing the following constants:
16877 E, PI, PHI
16879 These are approximated values for the mathematical constants e
16880 (Euler's number), pi (Greek pi), and phi (the golden ratio).
16881 w, h
16883 The input width and height.
16884 a These are the same as w / h.
16886 sar The input sample aspect ratio.
16888 dar The input display aspect ratio. It is the same as (w / h) * sar.
16890 hsub, vsub
16892 Horizontal and vertical chroma subsample values. For example, for
16893 the pixel format "yuv422p" hsub is 2 and vsub is 1.
16894 Examples
16896 • To change the display aspect ratio to 16:9, specify one of the
16898 following:
16899 setdar=dar=1.77777
16901 setdar=dar=16/9
16902 • To change the sample aspect ratio to 10:11, specify:
16904 setsar=sar=10/11
16906 • To set a display aspect ratio of 16:9, and specify a maximum
16908 integer value of 1000 in the aspect ratio reduction, use the
16909 command:
16910 setdar=ratio=16/9:max=1000
16912 setfield
16914 Force field for the output video frame.
16915 The "setfield" filter marks the interlace type field for the output
16917 frames. It does not change the input frame, but only sets the
16918 corresponding property, which affects how the frame is treated by
16919 following filters (e.g. "fieldorder" or "yadif").
16920 The filter accepts the following options:
16922 mode
16924 Available values are:
16925 auto
16927 Keep the same field property.
16928 bff Mark the frame as bottom-field-first.
16930 tff Mark the frame as top-field-first.
16932 prog
16934 Mark the frame as progressive.
16935 setparams
16937 Force frame parameter for the output video frame.
16938 The "setparams" filter marks interlace and color range for the output
16940 frames. It does not change the input frame, but only sets the
16941 corresponding property, which affects how the frame is treated by
16942 filters/encoders.
16943 field_mode
16945 Available values are:
16946 auto
16948 Keep the same field property (default).
16949 bff Mark the frame as bottom-field-first.
16951 tff Mark the frame as top-field-first.
16953 prog
16955 Mark the frame as progressive.
16956 range
16958 Available values are:
16959 auto
16961 Keep the same color range property (default).
16962 unspecified, unknown
16964 Mark the frame as unspecified color range.
16965 limited, tv, mpeg
16967 Mark the frame as limited range.
16968 full, pc, jpeg
16970 Mark the frame as full range.
16971 color_primaries
16973 Set the color primaries. Available values are:
16974 auto
16976 Keep the same color primaries property (default).
16977 bt709
16979 unknown
16980 bt470m
16981 bt470bg
16982 smpte170m
16983 smpte240m
16984 film
16985 bt2020
16986 smpte428
16987 smpte431
16988 smpte432
16989 jedec-p22
16990 color_trc
16991 Set the color transfer. Available values are:
16992 auto
16994 Keep the same color trc property (default).
16995 bt709
16997 unknown
16998 bt470m
16999 bt470bg
17000 smpte170m
17001 smpte240m
17002 linear
17003 log100
17004 log316
17005 iec61966-2-4
17006 bt1361e
17007 iec61966-2-1
17008 bt2020-10
17009 bt2020-12
17010 smpte2084
17011 smpte428
17012 arib-std-b67
17013 colorspace
17014 Set the colorspace. Available values are:
17015 auto
17017 Keep the same colorspace property (default).
17018 gbr
17020 bt709
17021 unknown
17022 fcc
17023 bt470bg
17024 smpte170m
17025 smpte240m
17026 ycgco
17027 bt2020nc
17028 bt2020c
17029 smpte2085
17030 chroma-derived-nc
17031 chroma-derived-c
17032 ictcp
17033 sharpen_npp
17035 Use the NVIDIA Performance Primitives (libnpp) to perform image
17036 sharpening with border control.
17037 The following additional options are accepted:
17039 border_type
17041 Type of sampling to be used ad frame borders. One of the following:
17042 replicate
17044 Replicate pixel values.
17045 shear
17047 Apply shear transform to input video.
17048 This filter supports the following options:
17050 shx Shear factor in X-direction. Default value is 0. Allowed range is
17052 from -2 to 2.
17053 shy Shear factor in Y-direction. Default value is 0. Allowed range is
17055 from -2 to 2.
17056 fillcolor, c
17058 Set the color used to fill the output area not covered by the
17059 transformed video. For the general syntax of this option, check the
17060 "Color" section in the ffmpeg-utils manual. If the special value
17061 "none" is selected then no background is printed (useful for
17062 example if the background is never shown).
17063 Default value is "black".
17065 interp
17067 Set interpolation type. Can be "bilinear" or "nearest". Default is
17068 "bilinear".
17069 Commands
17071 This filter supports the all above options as commands.
17073 showinfo
17075 Show a line containing various information for each input video frame.
17076 The input video is not modified.
17077 This filter supports the following options:
17079 checksum
17081 Calculate checksums of each plane. By default enabled.
17082 The shown line contains a sequence of key/value pairs of the form
17084 key:value.
17085 The following values are shown in the output:
17087 n The (sequential) number of the input frame, starting from 0.
17089 pts The Presentation TimeStamp of the input frame, expressed as a
17091 number of time base units. The time base unit depends on the filter
17092 input pad.
17093 pts_time
17095 The Presentation TimeStamp of the input frame, expressed as a
17096 number of seconds.
17097 pos The position of the frame in the input stream, or -1 if this
17099 information is unavailable and/or meaningless (for example in case
17100 of synthetic video).
17101 fmt The pixel format name.
17103 sar The sample aspect ratio of the input frame, expressed in the form
17105 num/den.
17106 s The size of the input frame. For the syntax of this option, check
17108 the "Video size" section in the ffmpeg-utils manual.
17109 i The type of interlaced mode ("P" for "progressive", "T" for top
17111 field first, "B" for bottom field first).
17112 iskey
17114 This is 1 if the frame is a key frame, 0 otherwise.
17115 type
17117 The picture type of the input frame ("I" for an I-frame, "P" for a
17118 P-frame, "B" for a B-frame, or "?" for an unknown type). Also
17119 refer to the documentation of the "AVPictureType" enum and of the
17120 "av_get_picture_type_char" function defined in libavutil/avutil.h.
17121 checksum
17123 The Adler-32 checksum (printed in hexadecimal) of all the planes of
17124 the input frame.
17125 plane_checksum
17127 The Adler-32 checksum (printed in hexadecimal) of each plane of the
17128 input frame, expressed in the form "[c0 c1 c2 c3]".
17129 mean
17131 The mean value of pixels in each plane of the input frame,
17132 expressed in the form "[mean0 mean1 mean2 mean3]".
17133 stdev
17135 The standard deviation of pixel values in each plane of the input
17136 frame, expressed in the form "[stdev0 stdev1 stdev2 stdev3]".
17137 showpalette
17139 Displays the 256 colors palette of each frame. This filter is only
17140 relevant for pal8 pixel format frames.
17141 It accepts the following option:
17143 s Set the size of the box used to represent one palette color entry.
17145 Default is 30 (for a "30x30" pixel box).
17146 shuffleframes
17148 Reorder and/or duplicate and/or drop video frames.
17149 It accepts the following parameters:
17151 mapping
17153 Set the destination indexes of input frames. This is space or '|'
17154 separated list of indexes that maps input frames to output frames.
17155 Number of indexes also sets maximal value that each index may have.
17156 '-1' index have special meaning and that is to drop frame.
17157 The first frame has the index 0. The default is to keep the input
17159 unchanged.
17160 Examples
17162 • Swap second and third frame of every three frames of the input:
17164 ffmpeg -i INPUT -vf "shuffleframes=0 2 1" OUTPUT
17166 • Swap 10th and 1st frame of every ten frames of the input:
17168 ffmpeg -i INPUT -vf "shuffleframes=9 1 2 3 4 5 6 7 8 0" OUTPUT
17170 shufflepixels
17172 Reorder pixels in video frames.
17173 This filter accepts the following options:
17175 direction, d
17177 Set shuffle direction. Can be forward or inverse direction.
17178 Default direction is forward.
17179 mode, m
17181 Set shuffle mode. Can be horizontal, vertical or block mode.
17182 width, w
17184 height, h
17185 Set shuffle block_size. In case of horizontal shuffle mode only
17186 width part of size is used, and in case of vertical shuffle mode
17187 only height part of size is used.
17188 seed, s
17190 Set random seed used with shuffling pixels. Mainly useful to set to
17191 be able to reverse filtering process to get original input. For
17192 example, to reverse forward shuffle you need to use same parameters
17193 and exact same seed and to set direction to inverse.
17194 shuffleplanes
17196 Reorder and/or duplicate video planes.
17197 It accepts the following parameters:
17199 map0
17201 The index of the input plane to be used as the first output plane.
17202 map1
17204 The index of the input plane to be used as the second output plane.
17205 map2
17207 The index of the input plane to be used as the third output plane.
17208 map3
17210 The index of the input plane to be used as the fourth output plane.
17211 The first plane has the index 0. The default is to keep the input
17213 unchanged.
17214 Examples
17216 • Swap the second and third planes of the input:
17218 ffmpeg -i INPUT -vf shuffleplanes=0:2:1:3 OUTPUT
17220 signalstats
17222 Evaluate various visual metrics that assist in determining issues
17223 associated with the digitization of analog video media.
17224 By default the filter will log these metadata values:
17226 YMIN
17228 Display the minimal Y value contained within the input frame.
17229 Expressed in range of [0-255].
17230 YLOW
17232 Display the Y value at the 10% percentile within the input frame.
17233 Expressed in range of [0-255].
17234 YAVG
17236 Display the average Y value within the input frame. Expressed in
17237 range of [0-255].
17238 YHIGH
17240 Display the Y value at the 90% percentile within the input frame.
17241 Expressed in range of [0-255].
17242 YMAX
17244 Display the maximum Y value contained within the input frame.
17245 Expressed in range of [0-255].
17246 UMIN
17248 Display the minimal U value contained within the input frame.
17249 Expressed in range of [0-255].
17250 ULOW
17252 Display the U value at the 10% percentile within the input frame.
17253 Expressed in range of [0-255].
17254 UAVG
17256 Display the average U value within the input frame. Expressed in
17257 range of [0-255].
17258 UHIGH
17260 Display the U value at the 90% percentile within the input frame.
17261 Expressed in range of [0-255].
17262 UMAX
17264 Display the maximum U value contained within the input frame.
17265 Expressed in range of [0-255].
17266 VMIN
17268 Display the minimal V value contained within the input frame.
17269 Expressed in range of [0-255].
17270 VLOW
17272 Display the V value at the 10% percentile within the input frame.
17273 Expressed in range of [0-255].
17274 VAVG
17276 Display the average V value within the input frame. Expressed in
17277 range of [0-255].
17278 VHIGH
17280 Display the V value at the 90% percentile within the input frame.
17281 Expressed in range of [0-255].
17282 VMAX
17284 Display the maximum V value contained within the input frame.
17285 Expressed in range of [0-255].
17286 SATMIN
17288 Display the minimal saturation value contained within the input
17289 frame. Expressed in range of [0-~181.02].
17290 SATLOW
17292 Display the saturation value at the 10% percentile within the input
17293 frame. Expressed in range of [0-~181.02].
17294 SATAVG
17296 Display the average saturation value within the input frame.
17297 Expressed in range of [0-~181.02].
17298 SATHIGH
17300 Display the saturation value at the 90% percentile within the input
17301 frame. Expressed in range of [0-~181.02].
17302 SATMAX
17304 Display the maximum saturation value contained within the input
17305 frame. Expressed in range of [0-~181.02].
17306 HUEMED
17308 Display the median value for hue within the input frame. Expressed
17309 in range of [0-360].
17310 HUEAVG
17312 Display the average value for hue within the input frame. Expressed
17313 in range of [0-360].
17314 YDIF
17316 Display the average of sample value difference between all values
17317 of the Y plane in the current frame and corresponding values of the
17318 previous input frame. Expressed in range of [0-255].
17319 UDIF
17321 Display the average of sample value difference between all values
17322 of the U plane in the current frame and corresponding values of the
17323 previous input frame. Expressed in range of [0-255].
17324 VDIF
17326 Display the average of sample value difference between all values
17327 of the V plane in the current frame and corresponding values of the
17328 previous input frame. Expressed in range of [0-255].
17329 YBITDEPTH
17331 Display bit depth of Y plane in current frame. Expressed in range
17332 of [0-16].
17333 UBITDEPTH
17335 Display bit depth of U plane in current frame. Expressed in range
17336 of [0-16].
17337 VBITDEPTH
17339 Display bit depth of V plane in current frame. Expressed in range
17340 of [0-16].
17341 The filter accepts the following options:
17343 stat
17345 out stat specify an additional form of image analysis. out output
17346 video with the specified type of pixel highlighted.
17347 Both options accept the following values:
17349 tout
17351 Identify temporal outliers pixels. A temporal outlier is a
17352 pixel unlike the neighboring pixels of the same field. Examples
17353 of temporal outliers include the results of video dropouts,
17354 head clogs, or tape tracking issues.
17355 vrep
17357 Identify vertical line repetition. Vertical line repetition
17358 includes similar rows of pixels within a frame. In born-digital
17359 video vertical line repetition is common, but this pattern is
17360 uncommon in video digitized from an analog source. When it
17361 occurs in video that results from the digitization of an analog
17362 source it can indicate concealment from a dropout compensator.
17363 brng
17365 Identify pixels that fall outside of legal broadcast range.
17366 color, c
17368 Set the highlight color for the out option. The default color is
17369 yellow.
17370 Examples
17372 • Output data of various video metrics:
17374 ffprobe -f lavfi movie=example.mov,signalstats="stat=tout+vrep+brng" -show_frames
17376 • Output specific data about the minimum and maximum values of the Y
17378 plane per frame:
17379 ffprobe -f lavfi movie=example.mov,signalstats -show_entries frame_tags=lavfi.signalstats.YMAX,lavfi.signalstats.YMIN
17381 • Playback video while highlighting pixels that are outside of
17383 broadcast range in red.
17384 ffplay example.mov -vf signalstats="out=brng:color=red"
17386 • Playback video with signalstats metadata drawn over the frame.
17388 ffplay example.mov -vf signalstats=stat=brng+vrep+tout,drawtext=fontfile=FreeSerif.ttf:textfile=signalstat_drawtext.txt
17390 The contents of signalstat_drawtext.txt used in the command are:
17392 time %{pts:hms}
17394 Y (%{metadata:lavfi.signalstats.YMIN}-%{metadata:lavfi.signalstats.YMAX})
17395 U (%{metadata:lavfi.signalstats.UMIN}-%{metadata:lavfi.signalstats.UMAX})
17396 V (%{metadata:lavfi.signalstats.VMIN}-%{metadata:lavfi.signalstats.VMAX})
17397 saturation maximum: %{metadata:lavfi.signalstats.SATMAX}
17398 signature
17400 Calculates the MPEG-7 Video Signature. The filter can handle more than
17401 one input. In this case the matching between the inputs can be
17402 calculated additionally. The filter always passes through the first
17403 input. The signature of each stream can be written into a file.
17404 It accepts the following options:
17406 detectmode
17408 Enable or disable the matching process.
17409 Available values are:
17411 off Disable the calculation of a matching (default).
17413 full
17415 Calculate the matching for the whole video and output whether
17416 the whole video matches or only parts.
17417 fast
17419 Calculate only until a matching is found or the video ends.
17420 Should be faster in some cases.
17421 nb_inputs
17423 Set the number of inputs. The option value must be a non negative
17424 integer. Default value is 1.
17425 filename
17427 Set the path to which the output is written. If there is more than
17428 one input, the path must be a prototype, i.e. must contain %d or
17429 %0nd (where n is a positive integer), that will be replaced with
17430 the input number. If no filename is specified, no output will be
17431 written. This is the default.
17432 format
17434 Choose the output format.
17435 Available values are:
17437 binary
17439 Use the specified binary representation (default).
17440 xml Use the specified xml representation.
17442 th_d
17444 Set threshold to detect one word as similar. The option value must
17445 be an integer greater than zero. The default value is 9000.
17446 th_dc
17448 Set threshold to detect all words as similar. The option value must
17449 be an integer greater than zero. The default value is 60000.
17450 th_xh
17452 Set threshold to detect frames as similar. The option value must be
17453 an integer greater than zero. The default value is 116.
17454 th_di
17456 Set the minimum length of a sequence in frames to recognize it as
17457 matching sequence. The option value must be a non negative integer
17458 value. The default value is 0.
17459 th_it
17461 Set the minimum relation, that matching frames to all frames must
17462 have. The option value must be a double value between 0 and 1. The
17463 default value is 0.5.
17464 Examples
17466 • To calculate the signature of an input video and store it in
17468 signature.bin:
17469 ffmpeg -i input.mkv -vf signature=filename=signature.bin -map 0:v -f null -
17471 • To detect whether two videos match and store the signatures in XML
17473 format in signature0.xml and signature1.xml:
17474 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 -
17476 smartblur
17478 Blur the input video without impacting the outlines.
17479 It accepts the following options:
17481 luma_radius, lr
17483 Set the luma radius. The option value must be a float number in the
17484 range [0.1,5.0] that specifies the variance of the gaussian filter
17485 used to blur the image (slower if larger). Default value is 1.0.
17486 luma_strength, ls
17488 Set the luma strength. The option value must be a float number in
17489 the range [-1.0,1.0] that configures the blurring. A value included
17490 in [0.0,1.0] will blur the image whereas a value included in
17491 [-1.0,0.0] will sharpen the image. Default value is 1.0.
17492 luma_threshold, lt
17494 Set the luma threshold used as a coefficient to determine whether a
17495 pixel should be blurred or not. The option value must be an integer
17496 in the range [-30,30]. A value of 0 will filter all the image, a
17497 value included in [0,30] will filter flat areas and a value
17498 included in [-30,0] will filter edges. Default value is 0.
17499 chroma_radius, cr
17501 Set the chroma radius. The option value must be a float number in
17502 the range [0.1,5.0] that specifies the variance of the gaussian
17503 filter used to blur the image (slower if larger). Default value is
17504 luma_radius.
17505 chroma_strength, cs
17507 Set the chroma strength. The option value must be a float number in
17508 the range [-1.0,1.0] that configures the blurring. A value included
17509 in [0.0,1.0] will blur the image whereas a value included in
17510 [-1.0,0.0] will sharpen the image. Default value is luma_strength.
17511 chroma_threshold, ct
17513 Set the chroma threshold used as a coefficient to determine whether
17514 a pixel should be blurred or not. The option value must be an
17515 integer in the range [-30,30]. A value of 0 will filter all the
17516 image, a value included in [0,30] will filter flat areas and a
17517 value included in [-30,0] will filter edges. Default value is
17518 luma_threshold.
17519 If a chroma option is not explicitly set, the corresponding luma value
17521 is set.
17522 sobel
17524 Apply sobel operator to input video stream.
17525 The filter accepts the following option:
17527 planes
17529 Set which planes will be processed, unprocessed planes will be
17530 copied. By default value 0xf, all planes will be processed.
17531 scale
17533 Set value which will be multiplied with filtered result.
17534 delta
17536 Set value which will be added to filtered result.
17537 Commands
17539 This filter supports the all above options as commands.
17541 spp
17543 Apply a simple postprocessing filter that compresses and decompresses
17544 the image at several (or - in the case of quality level 6 - all) shifts
17545 and average the results.
17546 The filter accepts the following options:
17548 quality
17550 Set quality. This option defines the number of levels for
17551 averaging. It accepts an integer in the range 0-6. If set to 0, the
17552 filter will have no effect. A value of 6 means the higher quality.
17553 For each increment of that value the speed drops by a factor of
17554 approximately 2. Default value is 3.
17555 qp Force a constant quantization parameter. If not set, the filter
17557 will use the QP from the video stream (if available).
17558 mode
17560 Set thresholding mode. Available modes are:
17561 hard
17563 Set hard thresholding (default).
17564 soft
17566 Set soft thresholding (better de-ringing effect, but likely
17567 blurrier).
17568 use_bframe_qp
17570 Enable the use of the QP from the B-Frames if set to 1. Using this
17571 option may cause flicker since the B-Frames have often larger QP.
17572 Default is 0 (not enabled).
17573 Commands
17575 This filter supports the following commands:
17577 quality, level
17579 Set quality level. The value "max" can be used to set the maximum
17580 level, currently 6.
17581 sr
17583 Scale the input by applying one of the super-resolution methods based
17584 on convolutional neural networks. Supported models:
17585 • Super-Resolution Convolutional Neural Network model (SRCNN). See
17587 <https://arxiv.org/abs/1501.00092>.
17588 • Efficient Sub-Pixel Convolutional Neural Network model (ESPCN).
17590 See <https://arxiv.org/abs/1609.05158>.
17591 Training scripts as well as scripts for model file (.pb) saving can be
17593 found at <https://github.com/XueweiMeng/sr/tree/sr_dnn_native>.
17594 Original repository is at
17595 <https://github.com/HighVoltageRocknRoll/sr.git>.
17596 Native model files (.model) can be generated from TensorFlow model
17598 files (.pb) by using tools/python/convert.py
17599 The filter accepts the following options:
17601 dnn_backend
17603 Specify which DNN backend to use for model loading and execution.
17604 This option accepts the following values:
17605 native
17607 Native implementation of DNN loading and execution.
17608 tensorflow
17610 TensorFlow backend. To enable this backend you need to install
17611 the TensorFlow for C library (see
17612 <https://www.tensorflow.org/install/lang_c>) and configure
17613 FFmpeg with "--enable-libtensorflow"
17614 Default value is native.
17616 model
17618 Set path to model file specifying network architecture and its
17619 parameters. Note that different backends use different file
17620 formats. TensorFlow backend can load files for both formats, while
17621 native backend can load files for only its format.
17622 scale_factor
17624 Set scale factor for SRCNN model. Allowed values are 2, 3 and 4.
17625 Default value is 2. Scale factor is necessary for SRCNN model,
17626 because it accepts input upscaled using bicubic upscaling with
17627 proper scale factor.
17628 To get full functionality (such as async execution), please use the
17630 dnn_processing filter.
17631 ssim
17633 Obtain the SSIM (Structural SImilarity Metric) between two input
17634 videos.
17635 This filter takes in input two input videos, the first input is
17637 considered the "main" source and is passed unchanged to the output. The
17638 second input is used as a "reference" video for computing the SSIM.
17639 Both video inputs must have the same resolution and pixel format for
17641 this filter to work correctly. Also it assumes that both inputs have
17642 the same number of frames, which are compared one by one.
17643 The filter stores the calculated SSIM of each frame.
17645 The description of the accepted parameters follows.
17647 stats_file, f
17649 If specified the filter will use the named file to save the SSIM of
17650 each individual frame. When filename equals "-" the data is sent to
17651 standard output.
17652 The file printed if stats_file is selected, contains a sequence of
17654 key/value pairs of the form key:value for each compared couple of
17655 frames.
17656 A description of each shown parameter follows:
17658 n sequential number of the input frame, starting from 1
17660 Y, U, V, R, G, B
17662 SSIM of the compared frames for the component specified by the
17663 suffix.
17664 All SSIM of the compared frames for the whole frame.
17666 dB Same as above but in dB representation.
17668 This filter also supports the framesync options.
17670 Examples
17672 • For example:
17674 movie=ref_movie.mpg, setpts=PTS-STARTPTS [main];
17676 [main][ref] ssim="stats_file=stats.log" [out]
17677 On this example the input file being processed is compared with the
17679 reference file ref_movie.mpg. The SSIM of each individual frame is
17680 stored in stats.log.
17681 • Another example with both psnr and ssim at same time:
17683 ffmpeg -i main.mpg -i ref.mpg -lavfi "ssim;[0:v][1:v]psnr" -f null -
17685 • Another example with different containers:
17687 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 -
17689 stereo3d
17691 Convert between different stereoscopic image formats.
17692 The filters accept the following options:
17694 in Set stereoscopic image format of input.
17696 Available values for input image formats are:
17698 sbsl
17700 side by side parallel (left eye left, right eye right)
17701 sbsr
17703 side by side crosseye (right eye left, left eye right)
17704 sbs2l
17706 side by side parallel with half width resolution (left eye
17707 left, right eye right)
17708 sbs2r
17710 side by side crosseye with half width resolution (right eye
17711 left, left eye right)
17712 abl
17714 tbl above-below (left eye above, right eye below)
17715 abr
17717 tbr above-below (right eye above, left eye below)
17718 ab2l
17720 tb2l
17721 above-below with half height resolution (left eye above, right
17722 eye below)
17723 ab2r
17725 tb2r
17726 above-below with half height resolution (right eye above, left
17727 eye below)
17728 al alternating frames (left eye first, right eye second)
17730 ar alternating frames (right eye first, left eye second)
17732 irl interleaved rows (left eye has top row, right eye starts on
17734 next row)
17735 irr interleaved rows (right eye has top row, left eye starts on
17737 next row)
17738 icl interleaved columns, left eye first
17740 icr interleaved columns, right eye first
17742 Default value is sbsl.
17744 out Set stereoscopic image format of output.
17746 sbsl
17748 side by side parallel (left eye left, right eye right)
17749 sbsr
17751 side by side crosseye (right eye left, left eye right)
17752 sbs2l
17754 side by side parallel with half width resolution (left eye
17755 left, right eye right)
17756 sbs2r
17758 side by side crosseye with half width resolution (right eye
17759 left, left eye right)
17760 abl
17762 tbl above-below (left eye above, right eye below)
17763 abr
17765 tbr above-below (right eye above, left eye below)
17766 ab2l
17768 tb2l
17769 above-below with half height resolution (left eye above, right
17770 eye below)
17771 ab2r
17773 tb2r
17774 above-below with half height resolution (right eye above, left
17775 eye below)
17776 al alternating frames (left eye first, right eye second)
17778 ar alternating frames (right eye first, left eye second)
17780 irl interleaved rows (left eye has top row, right eye starts on
17782 next row)
17783 irr interleaved rows (right eye has top row, left eye starts on
17785 next row)
17786 arbg
17788 anaglyph red/blue gray (red filter on left eye, blue filter on
17789 right eye)
17790 argg
17792 anaglyph red/green gray (red filter on left eye, green filter
17793 on right eye)
17794 arcg
17796 anaglyph red/cyan gray (red filter on left eye, cyan filter on
17797 right eye)
17798 arch
17800 anaglyph red/cyan half colored (red filter on left eye, cyan
17801 filter on right eye)
17802 arcc
17804 anaglyph red/cyan color (red filter on left eye, cyan filter on
17805 right eye)
17806 arcd
17808 anaglyph red/cyan color optimized with the least squares
17809 projection of dubois (red filter on left eye, cyan filter on
17810 right eye)
17811 agmg
17813 anaglyph green/magenta gray (green filter on left eye, magenta
17814 filter on right eye)
17815 agmh
17817 anaglyph green/magenta half colored (green filter on left eye,
17818 magenta filter on right eye)
17819 agmc
17821 anaglyph green/magenta colored (green filter on left eye,
17822 magenta filter on right eye)
17823 agmd
17825 anaglyph green/magenta color optimized with the least squares
17826 projection of dubois (green filter on left eye, magenta filter
17827 on right eye)
17828 aybg
17830 anaglyph yellow/blue gray (yellow filter on left eye, blue
17831 filter on right eye)
17832 aybh
17834 anaglyph yellow/blue half colored (yellow filter on left eye,
17835 blue filter on right eye)
17836 aybc
17838 anaglyph yellow/blue colored (yellow filter on left eye, blue
17839 filter on right eye)
17840 aybd
17842 anaglyph yellow/blue color optimized with the least squares
17843 projection of dubois (yellow filter on left eye, blue filter on
17844 right eye)
17845 ml mono output (left eye only)
17847 mr mono output (right eye only)
17849 chl checkerboard, left eye first
17851 chr checkerboard, right eye first
17853 icl interleaved columns, left eye first
17855 icr interleaved columns, right eye first
17857 hdmi
17859 HDMI frame pack
17860 Default value is arcd.
17862 Examples
17864 • Convert input video from side by side parallel to anaglyph
17866 yellow/blue dubois:
17867 stereo3d=sbsl:aybd
17869 • Convert input video from above below (left eye above, right eye
17871 below) to side by side crosseye.
17872 stereo3d=abl:sbsr
17874 streamselect, astreamselect
17876 Select video or audio streams.
17877 The filter accepts the following options:
17879 inputs
17881 Set number of inputs. Default is 2.
17882 map Set input indexes to remap to outputs.
17884 Commands
17886 The "streamselect" and "astreamselect" filter supports the following
17888 commands:
17889 map Set input indexes to remap to outputs.
17891 Examples
17893 • Select first 5 seconds 1st stream and rest of time 2nd stream:
17895 sendcmd='5.0 streamselect map 1',streamselect=inputs=2:map=0
17897 • Same as above, but for audio:
17899 asendcmd='5.0 astreamselect map 1',astreamselect=inputs=2:map=0
17901 subtitles
17903 Draw subtitles on top of input video using the libass library.
17904 To enable compilation of this filter you need to configure FFmpeg with
17906 "--enable-libass". This filter also requires a build with libavcodec
17907 and libavformat to convert the passed subtitles file to ASS (Advanced
17908 Substation Alpha) subtitles format.
17909 The filter accepts the following options:
17911 filename, f
17913 Set the filename of the subtitle file to read. It must be
17914 specified.
17915 original_size
17917 Specify the size of the original video, the video for which the ASS
17918 file was composed. For the syntax of this option, check the "Video
17919 size" section in the ffmpeg-utils manual. Due to a misdesign in
17920 ASS aspect ratio arithmetic, this is necessary to correctly scale
17921 the fonts if the aspect ratio has been changed.
17922 fontsdir
17924 Set a directory path containing fonts that can be used by the
17925 filter. These fonts will be used in addition to whatever the font
17926 provider uses.
17927 alpha
17929 Process alpha channel, by default alpha channel is untouched.
17930 charenc
17932 Set subtitles input character encoding. "subtitles" filter only.
17933 Only useful if not UTF-8.
17934 stream_index, si
17936 Set subtitles stream index. "subtitles" filter only.
17937 force_style
17939 Override default style or script info parameters of the subtitles.
17940 It accepts a string containing ASS style format "KEY=VALUE" couples
17941 separated by ",".
17942 If the first key is not specified, it is assumed that the first value
17944 specifies the filename.
17945 For example, to render the file sub.srt on top of the input video, use
17947 the command:
17948 subtitles=sub.srt
17950 which is equivalent to:
17952 subtitles=filename=sub.srt
17954 To render the default subtitles stream from file video.mkv, use:
17956 subtitles=video.mkv
17958 To render the second subtitles stream from that file, use:
17960 subtitles=video.mkv:si=1
17962 To make the subtitles stream from sub.srt appear in 80% transparent
17964 blue "DejaVu Serif", use:
17965 subtitles=sub.srt:force_style='Fontname=DejaVu Serif,PrimaryColour=&HCCFF0000'
17967 super2xsai
17969 Scale the input by 2x and smooth using the Super2xSaI (Scale and
17970 Interpolate) pixel art scaling algorithm.
17971 Useful for enlarging pixel art images without reducing sharpness.
17973 swaprect
17975 Swap two rectangular objects in video.
17976 This filter accepts the following options:
17978 w Set object width.
17980 h Set object height.
17982 x1 Set 1st rect x coordinate.
17984 y1 Set 1st rect y coordinate.
17986 x2 Set 2nd rect x coordinate.
17988 y2 Set 2nd rect y coordinate.
17990 All expressions are evaluated once for each frame.
17992 The all options are expressions containing the following constants:
17994 w
17996 h The input width and height.
17997 a same as w / h
17999 sar input sample aspect ratio
18001 dar input display aspect ratio, it is the same as (w / h) * sar
18003 n The number of the input frame, starting from 0.
18005 t The timestamp expressed in seconds. It's NAN if the input timestamp
18007 is unknown.
18008 pos the position in the file of the input frame, NAN if unknown
18010 Commands
18012 This filter supports the all above options as commands.
18014 swapuv
18016 Swap U & V plane.
18017 tblend
18019 Blend successive video frames.
18020 See blend
18022 telecine
18024 Apply telecine process to the video.
18025 This filter accepts the following options:
18027 first_field
18029 top, t
18030 top field first
18031 bottom, b
18033 bottom field first The default value is "top".
18034 pattern
18036 A string of numbers representing the pulldown pattern you wish to
18037 apply. The default value is 23.
18038 Some typical patterns:
18040 NTSC output (30i):
18042 27.5p: 32222
18043 24p: 23 (classic)
18044 24p: 2332 (preferred)
18045 20p: 33
18046 18p: 334
18047 16p: 3444
18048 PAL output (25i):
18050 27.5p: 12222
18051 24p: 222222222223 ("Euro pulldown")
18052 16.67p: 33
18053 16p: 33333334
18054 thistogram
18056 Compute and draw a color distribution histogram for the input video
18057 across time.
18058 Unlike histogram video filter which only shows histogram of single
18060 input frame at certain time, this filter shows also past histograms of
18061 number of frames defined by "width" option.
18062 The computed histogram is a representation of the color component
18064 distribution in an image.
18065 The filter accepts the following options:
18067 width, w
18069 Set width of single color component output. Default value is 0.
18070 Value of 0 means width will be picked from input video. This also
18071 set number of passed histograms to keep. Allowed range is [0,
18072 8192].
18073 display_mode, d
18075 Set display mode. It accepts the following values:
18076 stack
18078 Per color component graphs are placed below each other.
18079 parade
18081 Per color component graphs are placed side by side.
18082 overlay
18084 Presents information identical to that in the "parade", except
18085 that the graphs representing color components are superimposed
18086 directly over one another.
18087 Default is "stack".
18089 levels_mode, m
18091 Set mode. Can be either "linear", or "logarithmic". Default is
18092 "linear".
18093 components, c
18095 Set what color components to display. Default is 7.
18096 bgopacity, b
18098 Set background opacity. Default is 0.9.
18099 envelope, e
18101 Show envelope. Default is disabled.
18102 ecolor, ec
18104 Set envelope color. Default is "gold".
18105 slide
18107 Set slide mode.
18108 Available values for slide is:
18110 frame
18112 Draw new frame when right border is reached.
18113 replace
18115 Replace old columns with new ones.
18116 scroll
18118 Scroll from right to left.
18119 rscroll
18121 Scroll from left to right.
18122 picture
18124 Draw single picture.
18125 Default is "replace".
18127 threshold
18129 Apply threshold effect to video stream.
18130 This filter needs four video streams to perform thresholding. First
18132 stream is stream we are filtering. Second stream is holding threshold
18133 values, third stream is holding min values, and last, fourth stream is
18134 holding max values.
18135 The filter accepts the following option:
18137 planes
18139 Set which planes will be processed, unprocessed planes will be
18140 copied. By default value 0xf, all planes will be processed.
18141 For example if first stream pixel's component value is less then
18143 threshold value of pixel component from 2nd threshold stream, third
18144 stream value will picked, otherwise fourth stream pixel component value
18145 will be picked.
18146 Using color source filter one can perform various types of
18148 thresholding:
18149 Commands
18151 This filter supports the all options as commands.
18153 Examples
18155 • Binary threshold, using gray color as threshold:
18157 ffmpeg -i 320x240.avi -f lavfi -i color=gray -f lavfi -i color=black -f lavfi -i color=white -lavfi threshold output.avi
18159 • Inverted binary threshold, using gray color as threshold:
18161 ffmpeg -i 320x240.avi -f lavfi -i color=gray -f lavfi -i color=white -f lavfi -i color=black -lavfi threshold output.avi
18163 • Truncate binary threshold, using gray color as threshold:
18165 ffmpeg -i 320x240.avi -f lavfi -i color=gray -i 320x240.avi -f lavfi -i color=gray -lavfi threshold output.avi
18167 • Threshold to zero, using gray color as threshold:
18169 ffmpeg -i 320x240.avi -f lavfi -i color=gray -f lavfi -i color=white -i 320x240.avi -lavfi threshold output.avi
18171 • Inverted threshold to zero, using gray color as threshold:
18173 ffmpeg -i 320x240.avi -f lavfi -i color=gray -i 320x240.avi -f lavfi -i color=white -lavfi threshold output.avi
18175 thumbnail
18177 Select the most representative frame in a given sequence of consecutive
18178 frames.
18179 The filter accepts the following options:
18181 n Set the frames batch size to analyze; in a set of n frames, the
18183 filter will pick one of them, and then handle the next batch of n
18184 frames until the end. Default is 100.
18185 Since the filter keeps track of the whole frames sequence, a bigger n
18187 value will result in a higher memory usage, so a high value is not
18188 recommended.
18189 Examples
18191 • Extract one picture each 50 frames:
18193 thumbnail=50
18195 • Complete example of a thumbnail creation with ffmpeg:
18197 ffmpeg -i in.avi -vf thumbnail,scale=300:200 -frames:v 1 out.png
18199 tile
18201 Tile several successive frames together.
18202 The untile filter can do the reverse.
18204 The filter accepts the following options:
18206 layout
18208 Set the grid size (i.e. the number of lines and columns). For the
18209 syntax of this option, check the "Video size" section in the
18210 ffmpeg-utils manual.
18211 nb_frames
18213 Set the maximum number of frames to render in the given area. It
18214 must be less than or equal to wxh. The default value is 0, meaning
18215 all the area will be used.
18216 margin
18218 Set the outer border margin in pixels.
18219 padding
18221 Set the inner border thickness (i.e. the number of pixels between
18222 frames). For more advanced padding options (such as having
18223 different values for the edges), refer to the pad video filter.
18224 color
18226 Specify the color of the unused area. For the syntax of this
18227 option, check the "Color" section in the ffmpeg-utils manual. The
18228 default value of color is "black".
18229 overlap
18231 Set the number of frames to overlap when tiling several successive
18232 frames together. The value must be between 0 and nb_frames - 1.
18233 init_padding
18235 Set the number of frames to initially be empty before displaying
18236 first output frame. This controls how soon will one get first
18237 output frame. The value must be between 0 and nb_frames - 1.
18238 Examples
18240 • Produce 8x8 PNG tiles of all keyframes (-skip_frame nokey) in a
18242 movie:
18243 ffmpeg -skip_frame nokey -i file.avi -vf 'scale=128:72,tile=8x8' -an -vsync 0 keyframes%03d.png
18245 The -vsync 0 is necessary to prevent ffmpeg from duplicating each
18247 output frame to accommodate the originally detected frame rate.
18248 • Display 5 pictures in an area of "3x2" frames, with 7 pixels
18250 between them, and 2 pixels of initial margin, using mixed flat and
18251 named options:
18252 tile=3x2:nb_frames=5:padding=7:margin=2
18254 tinterlace
18256 Perform various types of temporal field interlacing.
18257 Frames are counted starting from 1, so the first input frame is
18259 considered odd.
18260 The filter accepts the following options:
18262 mode
18264 Specify the mode of the interlacing. This option can also be
18265 specified as a value alone. See below for a list of values for this
18266 option.
18267 Available values are:
18269 merge, 0
18271 Move odd frames into the upper field, even into the lower
18272 field, generating a double height frame at half frame rate.
18273 ------> time
18275 Input:
18276 Frame 1 Frame 2 Frame 3 Frame 4
18277 11111 22222 33333 44444
18279 11111 22222 33333 44444
18280 11111 22222 33333 44444
18281 11111 22222 33333 44444
18282 Output:
18284 11111 33333
18285 22222 44444
18286 11111 33333
18287 22222 44444
18288 11111 33333
18289 22222 44444
18290 11111 33333
18291 22222 44444
18292 drop_even, 1
18294 Only output odd frames, even frames are dropped, generating a
18295 frame with unchanged height at half frame rate.
18296 ------> time
18298 Input:
18299 Frame 1 Frame 2 Frame 3 Frame 4
18300 11111 22222 33333 44444
18302 11111 22222 33333 44444
18303 11111 22222 33333 44444
18304 11111 22222 33333 44444
18305 Output:
18307 11111 33333
18308 11111 33333
18309 11111 33333
18310 11111 33333
18311 drop_odd, 2
18313 Only output even frames, odd frames are dropped, generating a
18314 frame with unchanged height at half frame rate.
18315 ------> time
18317 Input:
18318 Frame 1 Frame 2 Frame 3 Frame 4
18319 11111 22222 33333 44444
18321 11111 22222 33333 44444
18322 11111 22222 33333 44444
18323 11111 22222 33333 44444
18324 Output:
18326 22222 44444
18327 22222 44444
18328 22222 44444
18329 22222 44444
18330 pad, 3
18332 Expand each frame to full height, but pad alternate lines with
18333 black, generating a frame with double height at the same input
18334 frame rate.
18335 ------> time
18337 Input:
18338 Frame 1 Frame 2 Frame 3 Frame 4
18339 11111 22222 33333 44444
18341 11111 22222 33333 44444
18342 11111 22222 33333 44444
18343 11111 22222 33333 44444
18344 Output:
18346 11111 ..... 33333 .....
18347 ..... 22222 ..... 44444
18348 11111 ..... 33333 .....
18349 ..... 22222 ..... 44444
18350 11111 ..... 33333 .....
18351 ..... 22222 ..... 44444
18352 11111 ..... 33333 .....
18353 ..... 22222 ..... 44444
18354 interleave_top, 4
18356 Interleave the upper field from odd frames with the lower field
18357 from even frames, generating a frame with unchanged height at
18358 half frame rate.
18359 ------> time
18361 Input:
18362 Frame 1 Frame 2 Frame 3 Frame 4
18363 11111<- 22222 33333<- 44444
18365 11111 22222<- 33333 44444<-
18366 11111<- 22222 33333<- 44444
18367 11111 22222<- 33333 44444<-
18368 Output:
18370 11111 33333
18371 22222 44444
18372 11111 33333
18373 22222 44444
18374 interleave_bottom, 5
18376 Interleave the lower field from odd frames with the upper field
18377 from even frames, generating a frame with unchanged height at
18378 half frame rate.
18379 ------> time
18381 Input:
18382 Frame 1 Frame 2 Frame 3 Frame 4
18383 11111 22222<- 33333 44444<-
18385 11111<- 22222 33333<- 44444
18386 11111 22222<- 33333 44444<-
18387 11111<- 22222 33333<- 44444
18388 Output:
18390 22222 44444
18391 11111 33333
18392 22222 44444
18393 11111 33333
18394 interlacex2, 6
18396 Double frame rate with unchanged height. Frames are inserted
18397 each containing the second temporal field from the previous
18398 input frame and the first temporal field from the next input
18399 frame. This mode relies on the top_field_first flag. Useful for
18400 interlaced video displays with no field synchronisation.
18401 ------> time
18403 Input:
18404 Frame 1 Frame 2 Frame 3 Frame 4
18405 11111 22222 33333 44444
18407 11111 22222 33333 44444
18408 11111 22222 33333 44444
18409 11111 22222 33333 44444
18410 Output:
18412 11111 22222 22222 33333 33333 44444 44444
18413 11111 11111 22222 22222 33333 33333 44444
18414 11111 22222 22222 33333 33333 44444 44444
18415 11111 11111 22222 22222 33333 33333 44444
18416 mergex2, 7
18418 Move odd frames into the upper field, even into the lower
18419 field, generating a double height frame at same frame rate.
18420 ------> time
18422 Input:
18423 Frame 1 Frame 2 Frame 3 Frame 4
18424 11111 22222 33333 44444
18426 11111 22222 33333 44444
18427 11111 22222 33333 44444
18428 11111 22222 33333 44444
18429 Output:
18431 11111 33333 33333 55555
18432 22222 22222 44444 44444
18433 11111 33333 33333 55555
18434 22222 22222 44444 44444
18435 11111 33333 33333 55555
18436 22222 22222 44444 44444
18437 11111 33333 33333 55555
18438 22222 22222 44444 44444
18439 Numeric values are deprecated but are accepted for backward
18441 compatibility reasons.
18442 Default mode is "merge".
18444 flags
18446 Specify flags influencing the filter process.
18447 Available value for flags is:
18449 low_pass_filter, vlpf
18451 Enable linear vertical low-pass filtering in the filter.
18452 Vertical low-pass filtering is required when creating an
18453 interlaced destination from a progressive source which contains
18454 high-frequency vertical detail. Filtering will reduce interlace
18455 'twitter' and Moire patterning.
18456 complex_filter, cvlpf
18458 Enable complex vertical low-pass filtering. This will slightly
18459 less reduce interlace 'twitter' and Moire patterning but better
18460 retain detail and subjective sharpness impression.
18461 bypass_il
18463 Bypass already interlaced frames, only adjust the frame rate.
18464 Vertical low-pass filtering and bypassing already interlaced frames
18466 can only be enabled for mode interleave_top and interleave_bottom.
18467 tmedian
18469 Pick median pixels from several successive input video frames.
18470 The filter accepts the following options:
18472 radius
18474 Set radius of median filter. Default is 1. Allowed range is from 1
18475 to 127.
18476 planes
18478 Set which planes to filter. Default value is 15, by which all
18479 planes are processed.
18480 percentile
18482 Set median percentile. Default value is 0.5. Default value of 0.5
18483 will pick always median values, while 0 will pick minimum values,
18484 and 1 maximum values.
18485 Commands
18487 This filter supports all above options as commands, excluding option
18489 "radius".
18490 tmidequalizer
18492 Apply Temporal Midway Video Equalization effect.
18493 Midway Video Equalization adjusts a sequence of video frames to have
18495 the same histograms, while maintaining their dynamics as much as
18496 possible. It's useful for e.g. matching exposures from a video frames
18497 sequence.
18498 This filter accepts the following option:
18500 radius
18502 Set filtering radius. Default is 5. Allowed range is from 1 to 127.
18503 sigma
18505 Set filtering sigma. Default is 0.5. This controls strength of
18506 filtering. Setting this option to 0 effectively does nothing.
18507 planes
18509 Set which planes to process. Default is 15, which is all available
18510 planes.
18511 tmix
18513 Mix successive video frames.
18514 A description of the accepted options follows.
18516 frames
18518 The number of successive frames to mix. If unspecified, it defaults
18519 to 3.
18520 weights
18522 Specify weight of each input video frame. Each weight is separated
18523 by space. If number of weights is smaller than number of frames
18524 last specified weight will be used for all remaining unset weights.
18525 scale
18527 Specify scale, if it is set it will be multiplied with sum of each
18528 weight multiplied with pixel values to give final destination pixel
18529 value. By default scale is auto scaled to sum of weights.
18530 Examples
18532 • Average 7 successive frames:
18534 tmix=frames=7:weights="1 1 1 1 1 1 1"
18536 • Apply simple temporal convolution:
18538 tmix=frames=3:weights="-1 3 -1"
18540 • Similar as above but only showing temporal differences:
18542 tmix=frames=3:weights="-1 2 -1":scale=1
18544 Commands
18546 This filter supports the following commands:
18548 weights
18550 scale
18551 Syntax is same as option with same name.
18552 tonemap
18554 Tone map colors from different dynamic ranges.
18555 This filter expects data in single precision floating point, as it
18557 needs to operate on (and can output) out-of-range values. Another
18558 filter, such as zscale, is needed to convert the resulting frame to a
18559 usable format.
18560 The tonemapping algorithms implemented only work on linear light, so
18562 input data should be linearized beforehand (and possibly correctly
18563 tagged).
18564 ffmpeg -i INPUT -vf zscale=transfer=linear,tonemap=clip,zscale=transfer=bt709,format=yuv420p OUTPUT
18566 Options
18568 The filter accepts the following options.
18570 tonemap
18572 Set the tone map algorithm to use.
18573 Possible values are:
18575 none
18577 Do not apply any tone map, only desaturate overbright pixels.
18578 clip
18580 Hard-clip any out-of-range values. Use it for perfect color
18581 accuracy for in-range values, while distorting out-of-range
18582 values.
18583 linear
18585 Stretch the entire reference gamut to a linear multiple of the
18586 display.
18587 gamma
18589 Fit a logarithmic transfer between the tone curves.
18590 reinhard
18592 Preserve overall image brightness with a simple curve, using
18593 nonlinear contrast, which results in flattening details and
18594 degrading color accuracy.
18595 hable
18597 Preserve both dark and bright details better than reinhard, at
18598 the cost of slightly darkening everything. Use it when detail
18599 preservation is more important than color and brightness
18600 accuracy.
18601 mobius
18603 Smoothly map out-of-range values, while retaining contrast and
18604 colors for in-range material as much as possible. Use it when
18605 color accuracy is more important than detail preservation.
18606 Default is none.
18608 param
18610 Tune the tone mapping algorithm.
18611 This affects the following algorithms:
18613 none
18615 Ignored.
18616 linear
18618 Specifies the scale factor to use while stretching. Default to
18619 1.0.
18620 gamma
18622 Specifies the exponent of the function. Default to 1.8.
18623 clip
18625 Specify an extra linear coefficient to multiply into the signal
18626 before clipping. Default to 1.0.
18627 reinhard
18629 Specify the local contrast coefficient at the display peak.
18630 Default to 0.5, which means that in-gamut values will be about
18631 half as bright as when clipping.
18632 hable
18634 Ignored.
18635 mobius
18637 Specify the transition point from linear to mobius transform.
18638 Every value below this point is guaranteed to be mapped 1:1.
18639 The higher the value, the more accurate the result will be, at
18640 the cost of losing bright details. Default to 0.3, which due
18641 to the steep initial slope still preserves in-range colors
18642 fairly accurately.
18643 desat
18645 Apply desaturation for highlights that exceed this level of
18646 brightness. The higher the parameter, the more color information
18647 will be preserved. This setting helps prevent unnaturally blown-out
18648 colors for super-highlights, by (smoothly) turning into white
18649 instead. This makes images feel more natural, at the cost of
18650 reducing information about out-of-range colors.
18651 The default of 2.0 is somewhat conservative and will mostly just
18653 apply to skies or directly sunlit surfaces. A setting of 0.0
18654 disables this option.
18655 This option works only if the input frame has a supported color
18657 tag.
18658 peak
18660 Override signal/nominal/reference peak with this value. Useful when
18661 the embedded peak information in display metadata is not reliable
18662 or when tone mapping from a lower range to a higher range.
18663 tpad
18665 Temporarily pad video frames.
18666 The filter accepts the following options:
18668 start
18670 Specify number of delay frames before input video stream. Default
18671 is 0.
18672 stop
18674 Specify number of padding frames after input video stream. Set to
18675 -1 to pad indefinitely. Default is 0.
18676 start_mode
18678 Set kind of frames added to beginning of stream. Can be either add
18679 or clone. With add frames of solid-color are added. With clone
18680 frames are clones of first frame. Default is add.
18681 stop_mode
18683 Set kind of frames added to end of stream. Can be either add or
18684 clone. With add frames of solid-color are added. With clone
18685 frames are clones of last frame. Default is add.
18686 start_duration, stop_duration
18688 Specify the duration of the start/stop delay. See the Time duration
18689 section in the ffmpeg-utils(1) manual for the accepted syntax.
18690 These options override start and stop. Default is 0.
18691 color
18693 Specify the color of the padded area. For the syntax of this
18694 option, check the "Color" section in the ffmpeg-utils manual.
18695 The default value of color is "black".
18697 transpose
18699 Transpose rows with columns in the input video and optionally flip it.
18700 It accepts the following parameters:
18702 dir Specify the transposition direction.
18704 Can assume the following values:
18706 0, 4, cclock_flip
18708 Rotate by 90 degrees counterclockwise and vertically flip
18709 (default), that is:
18710 L.R L.l
18712 . . -> . .
18713 l.r R.r
18714 1, 5, clock
18716 Rotate by 90 degrees clockwise, that is:
18717 L.R l.L
18719 . . -> . .
18720 l.r r.R
18721 2, 6, cclock
18723 Rotate by 90 degrees counterclockwise, that is:
18724 L.R R.r
18726 . . -> . .
18727 l.r L.l
18728 3, 7, clock_flip
18730 Rotate by 90 degrees clockwise and vertically flip, that is:
18731 L.R r.R
18733 . . -> . .
18734 l.r l.L
18735 For values between 4-7, the transposition is only done if the input
18737 video geometry is portrait and not landscape. These values are
18738 deprecated, the "passthrough" option should be used instead.
18739 Numerical values are deprecated, and should be dropped in favor of
18741 symbolic constants.
18742 passthrough
18744 Do not apply the transposition if the input geometry matches the
18745 one specified by the specified value. It accepts the following
18746 values:
18747 none
18749 Always apply transposition.
18750 portrait
18752 Preserve portrait geometry (when height >= width).
18753 landscape
18755 Preserve landscape geometry (when width >= height).
18756 Default value is "none".
18758 For example to rotate by 90 degrees clockwise and preserve portrait
18760 layout:
18761 transpose=dir=1:passthrough=portrait
18763 The command above can also be specified as:
18765 transpose=1:portrait
18767 transpose_npp
18769 Transpose rows with columns in the input video and optionally flip it.
18770 For more in depth examples see the transpose video filter, which shares
18771 mostly the same options.
18772 It accepts the following parameters:
18774 dir Specify the transposition direction.
18776 Can assume the following values:
18778 cclock_flip
18780 Rotate by 90 degrees counterclockwise and vertically flip.
18781 (default)
18782 clock
18784 Rotate by 90 degrees clockwise.
18785 cclock
18787 Rotate by 90 degrees counterclockwise.
18788 clock_flip
18790 Rotate by 90 degrees clockwise and vertically flip.
18791 passthrough
18793 Do not apply the transposition if the input geometry matches the
18794 one specified by the specified value. It accepts the following
18795 values:
18796 none
18798 Always apply transposition. (default)
18799 portrait
18801 Preserve portrait geometry (when height >= width).
18802 landscape
18804 Preserve landscape geometry (when width >= height).
18805 trim
18807 Trim the input so that the output contains one continuous subpart of
18808 the input.
18809 It accepts the following parameters:
18811 start
18813 Specify the time of the start of the kept section, i.e. the frame
18814 with the timestamp start will be the first frame in the output.
18815 end Specify the time of the first frame that will be dropped, i.e. the
18817 frame immediately preceding the one with the timestamp end will be
18818 the last frame in the output.
18819 start_pts
18821 This is the same as start, except this option sets the start
18822 timestamp in timebase units instead of seconds.
18823 end_pts
18825 This is the same as end, except this option sets the end timestamp
18826 in timebase units instead of seconds.
18827 duration
18829 The maximum duration of the output in seconds.
18830 start_frame
18832 The number of the first frame that should be passed to the output.
18833 end_frame
18835 The number of the first frame that should be dropped.
18836 start, end, and duration are expressed as time duration specifications;
18838 see the Time duration section in the ffmpeg-utils(1) manual for the
18839 accepted syntax.
18840 Note that the first two sets of the start/end options and the duration
18842 option look at the frame timestamp, while the _frame variants simply
18843 count the frames that pass through the filter. Also note that this
18844 filter does not modify the timestamps. If you wish for the output
18845 timestamps to start at zero, insert a setpts filter after the trim
18846 filter.
18847 If multiple start or end options are set, this filter tries to be
18849 greedy and keep all the frames that match at least one of the specified
18850 constraints. To keep only the part that matches all the constraints at
18851 once, chain multiple trim filters.
18852 The defaults are such that all the input is kept. So it is possible to
18854 set e.g. just the end values to keep everything before the specified
18855 time.
18856 Examples:
18858 • Drop everything except the second minute of input:
18860 ffmpeg -i INPUT -vf trim=60:120
18862 • Keep only the first second:
18864 ffmpeg -i INPUT -vf trim=duration=1
18866 unpremultiply
18868 Apply alpha unpremultiply effect to input video stream using first
18869 plane of second stream as alpha.
18870 Both streams must have same dimensions and same pixel format.
18872 The filter accepts the following option:
18874 planes
18876 Set which planes will be processed, unprocessed planes will be
18877 copied. By default value 0xf, all planes will be processed.
18878 If the format has 1 or 2 components, then luma is bit 0. If the
18880 format has 3 or 4 components: for RGB formats bit 0 is green, bit 1
18881 is blue and bit 2 is red; for YUV formats bit 0 is luma, bit 1 is
18882 chroma-U and bit 2 is chroma-V. If present, the alpha channel is
18883 always the last bit.
18884 inplace
18886 Do not require 2nd input for processing, instead use alpha plane
18887 from input stream.
18888 unsharp
18890 Sharpen or blur the input video.
18891 It accepts the following parameters:
18893 luma_msize_x, lx
18895 Set the luma matrix horizontal size. It must be an odd integer
18896 between 3 and 23. The default value is 5.
18897 luma_msize_y, ly
18899 Set the luma matrix vertical size. It must be an odd integer
18900 between 3 and 23. The default value is 5.
18901 luma_amount, la
18903 Set the luma effect strength. It must be a floating point number,
18904 reasonable values lay between -1.5 and 1.5.
18905 Negative values will blur the input video, while positive values
18907 will sharpen it, a value of zero will disable the effect.
18908 Default value is 1.0.
18910 chroma_msize_x, cx
18912 Set the chroma matrix horizontal size. It must be an odd integer
18913 between 3 and 23. The default value is 5.
18914 chroma_msize_y, cy
18916 Set the chroma matrix vertical size. It must be an odd integer
18917 between 3 and 23. The default value is 5.
18918 chroma_amount, ca
18920 Set the chroma effect strength. It must be a floating point number,
18921 reasonable values lay between -1.5 and 1.5.
18922 Negative values will blur the input video, while positive values
18924 will sharpen it, a value of zero will disable the effect.
18925 Default value is 0.0.
18927 All parameters are optional and default to the equivalent of the string
18929 '5:5:1.0:5:5:0.0'.
18930 Examples
18932 • Apply strong luma sharpen effect:
18934 unsharp=luma_msize_x=7:luma_msize_y=7:luma_amount=2.5
18936 • Apply a strong blur of both luma and chroma parameters:
18938 unsharp=7:7:-2:7:7:-2
18940 untile
18942 Decompose a video made of tiled images into the individual images.
18943 The frame rate of the output video is the frame rate of the input video
18945 multiplied by the number of tiles.
18946 This filter does the reverse of tile.
18948 The filter accepts the following options:
18950 layout
18952 Set the grid size (i.e. the number of lines and columns). For the
18953 syntax of this option, check the "Video size" section in the
18954 ffmpeg-utils manual.
18955 Examples
18957 • Produce a 1-second video from a still image file made of 25 frames
18959 stacked vertically, like an analogic film reel:
18960 ffmpeg -r 1 -i image.jpg -vf untile=1x25 movie.mkv
18962 uspp
18964 Apply ultra slow/simple postprocessing filter that compresses and
18965 decompresses the image at several (or - in the case of quality level 8
18966 - all) shifts and average the results.
18967 The way this differs from the behavior of spp is that uspp actually
18969 encodes & decodes each case with libavcodec Snow, whereas spp uses a
18970 simplified intra only 8x8 DCT similar to MJPEG.
18971 This filter is only available in ffmpeg version 4.4 or earlier.
18973 The filter accepts the following options:
18975 quality
18977 Set quality. This option defines the number of levels for
18978 averaging. It accepts an integer in the range 0-8. If set to 0, the
18979 filter will have no effect. A value of 8 means the higher quality.
18980 For each increment of that value the speed drops by a factor of
18981 approximately 2. Default value is 3.
18982 qp Force a constant quantization parameter. If not set, the filter
18984 will use the QP from the video stream (if available).
18985 v360
18987 Convert 360 videos between various formats.
18988 The filter accepts the following options:
18990 input
18992 output
18993 Set format of the input/output video.
18994 Available formats:
18996 e
18998 equirect
18999 Equirectangular projection.
19000 c3x2
19002 c6x1
19003 c1x6
19004 Cubemap with 3x2/6x1/1x6 layout.
19005 Format specific options:
19007 in_pad
19009 out_pad
19010 Set padding proportion for the input/output cubemap. Values
19011 in decimals.
19012 Example values:
19014 0 No padding.
19016 0.01
19018 1% of face is padding. For example, with 1920x1280
19019 resolution face size would be 640x640 and padding would
19020 be 3 pixels from each side. (640 * 0.01 = 6 pixels)
19021 Default value is @samp{0}. Maximum value is @samp{0.1}.
19023 fin_pad
19025 fout_pad
19026 Set fixed padding for the input/output cubemap. Values in
19027 pixels.
19028 Default value is @samp{0}. If greater than zero it
19030 overrides other padding options.
19031 in_forder
19033 out_forder
19034 Set order of faces for the input/output cubemap. Choose one
19035 direction for each position.
19036 Designation of directions:
19038 r right
19040 l left
19042 u up
19044 d down
19046 f forward
19048 b back
19050 Default value is @samp{rludfb}.
19052 in_frot
19054 out_frot
19055 Set rotation of faces for the input/output cubemap. Choose
19056 one angle for each position.
19057 Designation of angles:
19059 0 0 degrees clockwise
19061 1 90 degrees clockwise
19063 2 180 degrees clockwise
19065 3 270 degrees clockwise
19067 Default value is @samp{000000}.
19069 eac Equi-Angular Cubemap.
19071 flat
19073 gnomonic
19074 rectilinear
19075 Regular video.
19076 Format specific options:
19078 h_fov
19080 v_fov
19081 d_fov
19082 Set output horizontal/vertical/diagonal field of view.
19083 Values in degrees.
19084 If diagonal field of view is set it overrides horizontal
19086 and vertical field of view.
19087 ih_fov
19089 iv_fov
19090 id_fov
19091 Set input horizontal/vertical/diagonal field of view.
19092 Values in degrees.
19093 If diagonal field of view is set it overrides horizontal
19095 and vertical field of view.
19096 dfisheye
19098 Dual fisheye.
19099 Format specific options:
19101 h_fov
19103 v_fov
19104 d_fov
19105 Set output horizontal/vertical/diagonal field of view.
19106 Values in degrees.
19107 If diagonal field of view is set it overrides horizontal
19109 and vertical field of view.
19110 ih_fov
19112 iv_fov
19113 id_fov
19114 Set input horizontal/vertical/diagonal field of view.
19115 Values in degrees.
19116 If diagonal field of view is set it overrides horizontal
19118 and vertical field of view.
19119 barrel
19121 fb
19122 barrelsplit
19123 Facebook's 360 formats.
19124 sg Stereographic format.
19126 Format specific options:
19128 h_fov
19130 v_fov
19131 d_fov
19132 Set output horizontal/vertical/diagonal field of view.
19133 Values in degrees.
19134 If diagonal field of view is set it overrides horizontal
19136 and vertical field of view.
19137 ih_fov
19139 iv_fov
19140 id_fov
19141 Set input horizontal/vertical/diagonal field of view.
19142 Values in degrees.
19143 If diagonal field of view is set it overrides horizontal
19145 and vertical field of view.
19146 mercator
19148 Mercator format.
19149 ball
19151 Ball format, gives significant distortion toward the back.
19152 hammer
19154 Hammer-Aitoff map projection format.
19155 sinusoidal
19157 Sinusoidal map projection format.
19158 fisheye
19160 Fisheye projection.
19161 Format specific options:
19163 h_fov
19165 v_fov
19166 d_fov
19167 Set output horizontal/vertical/diagonal field of view.
19168 Values in degrees.
19169 If diagonal field of view is set it overrides horizontal
19171 and vertical field of view.
19172 ih_fov
19174 iv_fov
19175 id_fov
19176 Set input horizontal/vertical/diagonal field of view.
19177 Values in degrees.
19178 If diagonal field of view is set it overrides horizontal
19180 and vertical field of view.
19181 pannini
19183 Pannini projection.
19184 Format specific options:
19186 h_fov
19188 Set output pannini parameter.
19189 ih_fov
19191 Set input pannini parameter.
19192 cylindrical
19194 Cylindrical projection.
19195 Format specific options:
19197 h_fov
19199 v_fov
19200 d_fov
19201 Set output horizontal/vertical/diagonal field of view.
19202 Values in degrees.
19203 If diagonal field of view is set it overrides horizontal
19205 and vertical field of view.
19206 ih_fov
19208 iv_fov
19209 id_fov
19210 Set input horizontal/vertical/diagonal field of view.
19211 Values in degrees.
19212 If diagonal field of view is set it overrides horizontal
19214 and vertical field of view.
19215 perspective
19217 Perspective projection. (output only)
19218 Format specific options:
19220 v_fov
19222 Set perspective parameter.
19223 tetrahedron
19225 Tetrahedron projection.
19226 tsp Truncated square pyramid projection.
19228 he
19230 hequirect
19231 Half equirectangular projection.
19232 equisolid
19234 Equisolid format.
19235 Format specific options:
19237 h_fov
19239 v_fov
19240 d_fov
19241 Set output horizontal/vertical/diagonal field of view.
19242 Values in degrees.
19243 If diagonal field of view is set it overrides horizontal
19245 and vertical field of view.
19246 ih_fov
19248 iv_fov
19249 id_fov
19250 Set input horizontal/vertical/diagonal field of view.
19251 Values in degrees.
19252 If diagonal field of view is set it overrides horizontal
19254 and vertical field of view.
19255 og Orthographic format.
19257 Format specific options:
19259 h_fov
19261 v_fov
19262 d_fov
19263 Set output horizontal/vertical/diagonal field of view.
19264 Values in degrees.
19265 If diagonal field of view is set it overrides horizontal
19267 and vertical field of view.
19268 ih_fov
19270 iv_fov
19271 id_fov
19272 Set input horizontal/vertical/diagonal field of view.
19273 Values in degrees.
19274 If diagonal field of view is set it overrides horizontal
19276 and vertical field of view.
19277 octahedron
19279 Octahedron projection.
19280 cylindricalea
19282 Cylindrical Equal Area projection.
19283 interp
19285 Set interpolation method.Note: more complex interpolation methods
19286 require much more memory to run.
19287 Available methods:
19289 near
19291 nearest
19292 Nearest neighbour.
19293 line
19295 linear
19296 Bilinear interpolation.
19297 lagrange9
19299 Lagrange9 interpolation.
19300 cube
19302 cubic
19303 Bicubic interpolation.
19304 lanc
19306 lanczos
19307 Lanczos interpolation.
19308 sp16
19310 spline16
19311 Spline16 interpolation.
19312 gauss
19314 gaussian
19315 Gaussian interpolation.
19316 mitchell
19318 Mitchell interpolation.
19319 Default value is @samp{line}.
19321 w
19323 h Set the output video resolution.
19324 Default resolution depends on formats.
19326 in_stereo
19328 out_stereo
19329 Set the input/output stereo format.
19330 2d 2D mono
19332 sbs Side by side
19334 tb Top bottom
19336 Default value is @samp{2d} for input and output format.
19338 yaw
19340 pitch
19341 roll
19342 Set rotation for the output video. Values in degrees.
19343 rorder
19345 Set rotation order for the output video. Choose one item for each
19346 position.
19347 y, Y
19349 yaw
19350 p, P
19352 pitch
19353 r, R
19355 roll
19356 Default value is @samp{ypr}.
19358 h_flip
19360 v_flip
19361 d_flip
19362 Flip the output video horizontally(swaps
19363 left-right)/vertically(swaps up-down)/in-depth(swaps back-forward).
19364 Boolean values.
19365 ih_flip
19367 iv_flip
19368 Set if input video is flipped horizontally/vertically. Boolean
19369 values.
19370 in_trans
19372 Set if input video is transposed. Boolean value, by default
19373 disabled.
19374 out_trans
19376 Set if output video needs to be transposed. Boolean value, by
19377 default disabled.
19378 h_offset
19380 v_offset
19381 Set output horizontal/vertical off-axis offset. Default is set to
19382 0. Allowed range is from -1 to 1.
19383 alpha_mask
19385 Build mask in alpha plane for all unmapped pixels by marking them
19386 fully transparent. Boolean value, by default disabled.
19387 reset_rot
19389 Reset rotation of output video. Boolean value, by default disabled.
19390 Examples
19392 • Convert equirectangular video to cubemap with 3x2 layout and 1%
19394 padding using bicubic interpolation:
19395 ffmpeg -i input.mkv -vf v360=e:c3x2:cubic:out_pad=0.01 output.mkv
19397 • Extract back view of Equi-Angular Cubemap:
19399 ffmpeg -i input.mkv -vf v360=eac:flat:yaw=180 output.mkv
19401 • Convert transposed and horizontally flipped Equi-Angular Cubemap in
19403 side-by-side stereo format to equirectangular top-bottom stereo
19404 format:
19405 v360=eac:equirect:in_stereo=sbs:in_trans=1:ih_flip=1:out_stereo=tb
19407 Commands
19409 This filter supports subset of above options as commands.
19411 vaguedenoiser
19413 Apply a wavelet based denoiser.
19414 It transforms each frame from the video input into the wavelet domain,
19416 using Cohen-Daubechies-Feauveau 9/7. Then it applies some filtering to
19417 the obtained coefficients. It does an inverse wavelet transform after.
19418 Due to wavelet properties, it should give a nice smoothed result, and
19419 reduced noise, without blurring picture features.
19420 This filter accepts the following options:
19422 threshold
19424 The filtering strength. The higher, the more filtered the video
19425 will be. Hard thresholding can use a higher threshold than soft
19426 thresholding before the video looks overfiltered. Default value is
19427 2.
19428 method
19430 The filtering method the filter will use.
19431 It accepts the following values:
19433 hard
19435 All values under the threshold will be zeroed.
19436 soft
19438 All values under the threshold will be zeroed. All values above
19439 will be reduced by the threshold.
19440 garrote
19442 Scales or nullifies coefficients - intermediary between (more)
19443 soft and (less) hard thresholding.
19444 Default is garrote.
19446 nsteps
19448 Number of times, the wavelet will decompose the picture. Picture
19449 can't be decomposed beyond a particular point (typically, 8 for a
19450 640x480 frame - as 2^9 = 512 > 480). Valid values are integers
19451 between 1 and 32. Default value is 6.
19452 percent
19454 Partial of full denoising (limited coefficients shrinking), from 0
19455 to 100. Default value is 85.
19456 planes
19458 A list of the planes to process. By default all planes are
19459 processed.
19460 type
19462 The threshold type the filter will use.
19463 It accepts the following values:
19465 universal
19467 Threshold used is same for all decompositions.
19468 bayes
19470 Threshold used depends also on each decomposition coefficients.
19471 Default is universal.
19473 varblur
19475 Apply variable blur filter by using 2nd video stream to set blur
19476 radius. The 2nd stream must have the same dimensions.
19477 This filter accepts the following options:
19479 min_r
19481 Set min allowed radius. Allowed range is from 0 to 254. Default is
19482 0.
19483 max_r
19485 Set max allowed radius. Allowed range is from 1 to 255. Default is
19486 8.
19487 planes
19489 Set which planes to process. By default, all are used.
19490 The "varblur" filter also supports the framesync options.
19492 Commands
19494 This filter supports all the above options as commands.
19496 vectorscope
19498 Display 2 color component values in the two dimensional graph (which is
19499 called a vectorscope).
19500 This filter accepts the following options:
19502 mode, m
19504 Set vectorscope mode.
19505 It accepts the following values:
19507 gray
19509 tint
19510 Gray values are displayed on graph, higher brightness means
19511 more pixels have same component color value on location in
19512 graph. This is the default mode.
19513 color
19515 Gray values are displayed on graph. Surrounding pixels values
19516 which are not present in video frame are drawn in gradient of 2
19517 color components which are set by option "x" and "y". The 3rd
19518 color component is static.
19519 color2
19521 Actual color components values present in video frame are
19522 displayed on graph.
19523 color3
19525 Similar as color2 but higher frequency of same values "x" and
19526 "y" on graph increases value of another color component, which
19527 is luminance by default values of "x" and "y".
19528 color4
19530 Actual colors present in video frame are displayed on graph. If
19531 two different colors map to same position on graph then color
19532 with higher value of component not present in graph is picked.
19533 color5
19535 Gray values are displayed on graph. Similar to "color" but with
19536 3rd color component picked from radial gradient.
19537 x Set which color component will be represented on X-axis. Default is
19539 1.
19540 y Set which color component will be represented on Y-axis. Default is
19542 2.
19543 intensity, i
19545 Set intensity, used by modes: gray, color, color3 and color5 for
19546 increasing brightness of color component which represents frequency
19547 of (X, Y) location in graph.
19548 envelope, e
19550 none
19551 No envelope, this is default.
19552 instant
19554 Instant envelope, even darkest single pixel will be clearly
19555 highlighted.
19556 peak
19558 Hold maximum and minimum values presented in graph over time.
19559 This way you can still spot out of range values without
19560 constantly looking at vectorscope.
19561 peak+instant
19563 Peak and instant envelope combined together.
19564 graticule, g
19566 Set what kind of graticule to draw.
19567 none
19569 green
19570 color
19571 invert
19572 opacity, o
19573 Set graticule opacity.
19574 flags, f
19576 Set graticule flags.
19577 white
19579 Draw graticule for white point.
19580 black
19582 Draw graticule for black point.
19583 name
19585 Draw color points short names.
19586 bgopacity, b
19588 Set background opacity.
19589 lthreshold, l
19591 Set low threshold for color component not represented on X or Y
19592 axis. Values lower than this value will be ignored. Default is 0.
19593 Note this value is multiplied with actual max possible value one
19594 pixel component can have. So for 8-bit input and low threshold
19595 value of 0.1 actual threshold is 0.1 * 255 = 25.
19596 hthreshold, h
19598 Set high threshold for color component not represented on X or Y
19599 axis. Values higher than this value will be ignored. Default is 1.
19600 Note this value is multiplied with actual max possible value one
19601 pixel component can have. So for 8-bit input and high threshold
19602 value of 0.9 actual threshold is 0.9 * 255 = 230.
19603 colorspace, c
19605 Set what kind of colorspace to use when drawing graticule.
19606 auto
19608 601
19609 709
19610 Default is auto.
19612 tint0, t0
19614 tint1, t1
19615 Set color tint for gray/tint vectorscope mode. By default both
19616 options are zero. This means no tint, and output will remain gray.
19617 vidstabdetect
19619 Analyze video stabilization/deshaking. Perform pass 1 of 2, see
19620 vidstabtransform for pass 2.
19621 This filter generates a file with relative translation and rotation
19623 transform information about subsequent frames, which is then used by
19624 the vidstabtransform filter.
19625 To enable compilation of this filter you need to configure FFmpeg with
19627 "--enable-libvidstab".
19628 This filter accepts the following options:
19630 result
19632 Set the path to the file used to write the transforms information.
19633 Default value is transforms.trf.
19634 shakiness
19636 Set how shaky the video is and how quick the camera is. It accepts
19637 an integer in the range 1-10, a value of 1 means little shakiness,
19638 a value of 10 means strong shakiness. Default value is 5.
19639 accuracy
19641 Set the accuracy of the detection process. It must be a value in
19642 the range 1-15. A value of 1 means low accuracy, a value of 15
19643 means high accuracy. Default value is 15.
19644 stepsize
19646 Set stepsize of the search process. The region around minimum is
19647 scanned with 1 pixel resolution. Default value is 6.
19648 mincontrast
19650 Set minimum contrast. Below this value a local measurement field is
19651 discarded. Must be a floating point value in the range 0-1. Default
19652 value is 0.3.
19653 tripod
19655 Set reference frame number for tripod mode.
19656 If enabled, the motion of the frames is compared to a reference
19658 frame in the filtered stream, identified by the specified number.
19659 The idea is to compensate all movements in a more-or-less static
19660 scene and keep the camera view absolutely still.
19661 If set to 0, it is disabled. The frames are counted starting from
19663 1.
19664 show
19666 Show fields and transforms in the resulting frames. It accepts an
19667 integer in the range 0-2. Default value is 0, which disables any
19668 visualization.
19669 Examples
19671 • Use default values:
19673 vidstabdetect
19675 • Analyze strongly shaky movie and put the results in file
19677 mytransforms.trf:
19678 vidstabdetect=shakiness=10:accuracy=15:result="mytransforms.trf"
19680 • Visualize the result of internal transformations in the resulting
19682 video:
19683 vidstabdetect=show=1
19685 • Analyze a video with medium shakiness using ffmpeg:
19687 ffmpeg -i input -vf vidstabdetect=shakiness=5:show=1 dummy.avi
19689 vidstabtransform
19691 Video stabilization/deshaking: pass 2 of 2, see vidstabdetect for pass
19692 1.
19693 Read a file with transform information for each frame and
19695 apply/compensate them. Together with the vidstabdetect filter this can
19696 be used to deshake videos. See also
19697 <http://public.hronopik.de/vid.stab>. It is important to also use the
19698 unsharp filter, see below.
19699 To enable compilation of this filter you need to configure FFmpeg with
19701 "--enable-libvidstab".
19702 Options
19704 input
19706 Set path to the file used to read the transforms. Default value is
19707 transforms.trf.
19708 smoothing
19710 Set the number of frames (value*2 + 1) used for lowpass filtering
19711 the camera movements. Default value is 10.
19712 For example a number of 10 means that 21 frames are used (10 in the
19714 past and 10 in the future) to smoothen the motion in the video. A
19715 larger value leads to a smoother video, but limits the acceleration
19716 of the camera (pan/tilt movements). 0 is a special case where a
19717 static camera is simulated.
19718 optalgo
19720 Set the camera path optimization algorithm.
19721 Accepted values are:
19723 gauss
19725 gaussian kernel low-pass filter on camera motion (default)
19726 avg averaging on transformations
19728 maxshift
19730 Set maximal number of pixels to translate frames. Default value is
19731 -1, meaning no limit.
19732 maxangle
19734 Set maximal angle in radians (degree*PI/180) to rotate frames.
19735 Default value is -1, meaning no limit.
19736 crop
19738 Specify how to deal with borders that may be visible due to
19739 movement compensation.
19740 Available values are:
19742 keep
19744 keep image information from previous frame (default)
19745 black
19747 fill the border black
19748 invert
19750 Invert transforms if set to 1. Default value is 0.
19751 relative
19753 Consider transforms as relative to previous frame if set to 1,
19754 absolute if set to 0. Default value is 0.
19755 zoom
19757 Set percentage to zoom. A positive value will result in a zoom-in
19758 effect, a negative value in a zoom-out effect. Default value is 0
19759 (no zoom).
19760 optzoom
19762 Set optimal zooming to avoid borders.
19763 Accepted values are:
19765 0 disabled
19767 1 optimal static zoom value is determined (only very strong
19769 movements will lead to visible borders) (default)
19770 2 optimal adaptive zoom value is determined (no borders will be
19772 visible), see zoomspeed
19773 Note that the value given at zoom is added to the one calculated
19775 here.
19776 zoomspeed
19778 Set percent to zoom maximally each frame (enabled when optzoom is
19779 set to 2). Range is from 0 to 5, default value is 0.25.
19780 interpol
19782 Specify type of interpolation.
19783 Available values are:
19785 no no interpolation
19787 linear
19789 linear only horizontal
19790 bilinear
19792 linear in both directions (default)
19793 bicubic
19795 cubic in both directions (slow)
19796 tripod
19798 Enable virtual tripod mode if set to 1, which is equivalent to
19799 "relative=0:smoothing=0". Default value is 0.
19800 Use also "tripod" option of vidstabdetect.
19802 debug
19804 Increase log verbosity if set to 1. Also the detected global
19805 motions are written to the temporary file global_motions.trf.
19806 Default value is 0.
19807 Examples
19809 • Use ffmpeg for a typical stabilization with default values:
19811 ffmpeg -i inp.mpeg -vf vidstabtransform,unsharp=5:5:0.8:3:3:0.4 inp_stabilized.mpeg
19813 Note the use of the unsharp filter which is always recommended.
19815 • Zoom in a bit more and load transform data from a given file:
19817 vidstabtransform=zoom=5:input="mytransforms.trf"
19819 • Smoothen the video even more:
19821 vidstabtransform=smoothing=30
19823 vflip
19825 Flip the input video vertically.
19826 For example, to vertically flip a video with ffmpeg:
19828 ffmpeg -i in.avi -vf "vflip" out.avi
19830 vfrdet
19832 Detect variable frame rate video.
19833 This filter tries to detect if the input is variable or constant frame
19835 rate.
19836 At end it will output number of frames detected as having variable
19838 delta pts, and ones with constant delta pts. If there was frames with
19839 variable delta, than it will also show min, max and average delta
19840 encountered.
19841 vibrance
19843 Boost or alter saturation.
19844 The filter accepts the following options:
19846 intensity
19848 Set strength of boost if positive value or strength of alter if
19849 negative value. Default is 0. Allowed range is from -2 to 2.
19850 rbal
19852 Set the red balance. Default is 1. Allowed range is from -10 to 10.
19853 gbal
19855 Set the green balance. Default is 1. Allowed range is from -10 to
19856 10.
19857 bbal
19859 Set the blue balance. Default is 1. Allowed range is from -10 to
19860 10.
19861 rlum
19863 Set the red luma coefficient.
19864 glum
19866 Set the green luma coefficient.
19867 blum
19869 Set the blue luma coefficient.
19870 alternate
19872 If "intensity" is negative and this is set to 1, colors will
19873 change, otherwise colors will be less saturated, more towards gray.
19874 Commands
19876 This filter supports the all above options as commands.
19878 vif
19880 Obtain the average VIF (Visual Information Fidelity) between two input
19881 videos.
19882 This filter takes two input videos.
19884 Both input videos must have the same resolution and pixel format for
19886 this filter to work correctly. Also it assumes that both inputs have
19887 the same number of frames, which are compared one by one.
19888 The obtained average VIF score is printed through the logging system.
19890 The filter stores the calculated VIF score of each frame.
19892 In the below example the input file main.mpg being processed is
19894 compared with the reference file ref.mpg.
19895 ffmpeg -i main.mpg -i ref.mpg -lavfi vif -f null -
19897 vignette
19899 Make or reverse a natural vignetting effect.
19900 The filter accepts the following options:
19902 angle, a
19904 Set lens angle expression as a number of radians.
19905 The value is clipped in the "[0,PI/2]" range.
19907 Default value: "PI/5"
19909 x0
19911 y0 Set center coordinates expressions. Respectively "w/2" and "h/2" by
19912 default.
19913 mode
19915 Set forward/backward mode.
19916 Available modes are:
19918 forward
19920 The larger the distance from the central point, the darker the
19921 image becomes.
19922 backward
19924 The larger the distance from the central point, the brighter
19925 the image becomes. This can be used to reverse a vignette
19926 effect, though there is no automatic detection to extract the
19927 lens angle and other settings (yet). It can also be used to
19928 create a burning effect.
19929 Default value is forward.
19931 eval
19933 Set evaluation mode for the expressions (angle, x0, y0).
19934 It accepts the following values:
19936 init
19938 Evaluate expressions only once during the filter
19939 initialization.
19940 frame
19942 Evaluate expressions for each incoming frame. This is way
19943 slower than the init mode since it requires all the scalers to
19944 be re-computed, but it allows advanced dynamic expressions.
19945 Default value is init.
19947 dither
19949 Set dithering to reduce the circular banding effects. Default is 1
19950 (enabled).
19951 aspect
19953 Set vignette aspect. This setting allows one to adjust the shape of
19954 the vignette. Setting this value to the SAR of the input will make
19955 a rectangular vignetting following the dimensions of the video.
19956 Default is "1/1".
19958 Expressions
19960 The alpha, x0 and y0 expressions can contain the following parameters.
19962 w
19964 h input width and height
19965 n the number of input frame, starting from 0
19967 pts the PTS (Presentation TimeStamp) time of the filtered video frame,
19969 expressed in TB units, NAN if undefined
19970 r frame rate of the input video, NAN if the input frame rate is
19972 unknown
19973 t the PTS (Presentation TimeStamp) of the filtered video frame,
19975 expressed in seconds, NAN if undefined
19976 tb time base of the input video
19978 Examples
19980 • Apply simple strong vignetting effect:
19982 vignette=PI/4
19984 • Make a flickering vignetting:
19986 vignette='PI/4+random(1)*PI/50':eval=frame
19988 vmafmotion
19990 Obtain the average VMAF motion score of a video. It is one of the
19991 component metrics of VMAF.
19992 The obtained average motion score is printed through the logging
19994 system.
19995 The filter accepts the following options:
19997 stats_file
19999 If specified, the filter will use the named file to save the motion
20000 score of each frame with respect to the previous frame. When
20001 filename equals "-" the data is sent to standard output.
20002 Example:
20004 ffmpeg -i ref.mpg -vf vmafmotion -f null -
20006 vstack
20008 Stack input videos vertically.
20009 All streams must be of same pixel format and of same width.
20011 Note that this filter is faster than using overlay and pad filter to
20013 create same output.
20014 The filter accepts the following options:
20016 inputs
20018 Set number of input streams. Default is 2.
20019 shortest
20021 If set to 1, force the output to terminate when the shortest input
20022 terminates. Default value is 0.
20023 w3fdif
20025 Deinterlace the input video ("w3fdif" stands for "Weston 3 Field
20026 Deinterlacing Filter").
20027 Based on the process described by Martin Weston for BBC R&D, and
20029 implemented based on the de-interlace algorithm written by Jim
20030 Easterbrook for BBC R&D, the Weston 3 field deinterlacing filter uses
20031 filter coefficients calculated by BBC R&D.
20032 This filter uses field-dominance information in frame to decide which
20034 of each pair of fields to place first in the output. If it gets it
20035 wrong use setfield filter before "w3fdif" filter.
20036 There are two sets of filter coefficients, so called "simple" and
20038 "complex". Which set of filter coefficients is used can be set by
20039 passing an optional parameter:
20040 filter
20042 Set the interlacing filter coefficients. Accepts one of the
20043 following values:
20044 simple
20046 Simple filter coefficient set.
20047 complex
20049 More-complex filter coefficient set.
20050 Default value is complex.
20052 mode
20054 The interlacing mode to adopt. It accepts one of the following
20055 values:
20056 frame
20058 Output one frame for each frame.
20059 field
20061 Output one frame for each field.
20062 The default value is "field".
20064 parity
20066 The picture field parity assumed for the input interlaced video. It
20067 accepts one of the following values:
20068 tff Assume the top field is first.
20070 bff Assume the bottom field is first.
20072 auto
20074 Enable automatic detection of field parity.
20075 The default value is "auto". If the interlacing is unknown or the
20077 decoder does not export this information, top field first will be
20078 assumed.
20079 deint
20081 Specify which frames to deinterlace. Accepts one of the following
20082 values:
20083 all Deinterlace all frames,
20085 interlaced
20087 Only deinterlace frames marked as interlaced.
20088 Default value is all.
20090 Commands
20092 This filter supports same commands as options.
20094 waveform
20096 Video waveform monitor.
20097 The waveform monitor plots color component intensity. By default
20099 luminance only. Each column of the waveform corresponds to a column of
20100 pixels in the source video.
20101 It accepts the following options:
20103 mode, m
20105 Can be either "row", or "column". Default is "column". In row
20106 mode, the graph on the left side represents color component value 0
20107 and the right side represents value = 255. In column mode, the top
20108 side represents color component value = 0 and bottom side
20109 represents value = 255.
20110 intensity, i
20112 Set intensity. Smaller values are useful to find out how many
20113 values of the same luminance are distributed across input
20114 rows/columns. Default value is 0.04. Allowed range is [0, 1].
20115 mirror, r
20117 Set mirroring mode. 0 means unmirrored, 1 means mirrored. In
20118 mirrored mode, higher values will be represented on the left side
20119 for "row" mode and at the top for "column" mode. Default is 1
20120 (mirrored).
20121 display, d
20123 Set display mode. It accepts the following values:
20124 overlay
20126 Presents information identical to that in the "parade", except
20127 that the graphs representing color components are superimposed
20128 directly over one another.
20129 This display mode makes it easier to spot relative differences
20131 or similarities in overlapping areas of the color components
20132 that are supposed to be identical, such as neutral whites,
20133 grays, or blacks.
20134 stack
20136 Display separate graph for the color components side by side in
20137 "row" mode or one below the other in "column" mode.
20138 parade
20140 Display separate graph for the color components side by side in
20141 "column" mode or one below the other in "row" mode.
20142 Using this display mode makes it easy to spot color casts in
20144 the highlights and shadows of an image, by comparing the
20145 contours of the top and the bottom graphs of each waveform.
20146 Since whites, grays, and blacks are characterized by exactly
20147 equal amounts of red, green, and blue, neutral areas of the
20148 picture should display three waveforms of roughly equal
20149 width/height. If not, the correction is easy to perform by
20150 making level adjustments the three waveforms.
20151 Default is "stack".
20153 components, c
20155 Set which color components to display. Default is 1, which means
20156 only luminance or red color component if input is in RGB
20157 colorspace. If is set for example to 7 it will display all 3 (if)
20158 available color components.
20159 envelope, e
20161 none
20162 No envelope, this is default.
20163 instant
20165 Instant envelope, minimum and maximum values presented in graph
20166 will be easily visible even with small "step" value.
20167 peak
20169 Hold minimum and maximum values presented in graph across time.
20170 This way you can still spot out of range values without
20171 constantly looking at waveforms.
20172 peak+instant
20174 Peak and instant envelope combined together.
20175 filter, f
20177 lowpass
20178 No filtering, this is default.
20179 flat
20181 Luma and chroma combined together.
20182 aflat
20184 Similar as above, but shows difference between blue and red
20185 chroma.
20186 xflat
20188 Similar as above, but use different colors.
20189 yflat
20191 Similar as above, but again with different colors.
20192 chroma
20194 Displays only chroma.
20195 color
20197 Displays actual color value on waveform.
20198 acolor
20200 Similar as above, but with luma showing frequency of chroma
20201 values.
20202 graticule, g
20204 Set which graticule to display.
20205 none
20207 Do not display graticule.
20208 green
20210 Display green graticule showing legal broadcast ranges.
20211 orange
20213 Display orange graticule showing legal broadcast ranges.
20214 invert
20216 Display invert graticule showing legal broadcast ranges.
20217 opacity, o
20219 Set graticule opacity.
20220 flags, fl
20222 Set graticule flags.
20223 numbers
20225 Draw numbers above lines. By default enabled.
20226 dots
20228 Draw dots instead of lines.
20229 scale, s
20231 Set scale used for displaying graticule.
20232 digital
20234 millivolts
20235 ire
20236 Default is digital.
20238 bgopacity, b
20240 Set background opacity.
20241 tint0, t0
20243 tint1, t1
20244 Set tint for output. Only used with lowpass filter and when
20245 display is not overlay and input pixel formats are not RGB.
20246 fitmode, fm
20248 Set sample aspect ratio of video output frames. Can be used to
20249 configure waveform so it is not streched too much in one of
20250 directions.
20251 none
20253 Set sample aspect ration to 1/1.
20254 size
20256 Set sample aspect ratio to match input size of video
20257 Default is none.
20259 weave, doubleweave
20261 The "weave" takes a field-based video input and join each two
20262 sequential fields into single frame, producing a new double height clip
20263 with half the frame rate and half the frame count.
20264 The "doubleweave" works same as "weave" but without halving frame rate
20266 and frame count.
20267 It accepts the following option:
20269 first_field
20271 Set first field. Available values are:
20272 top, t
20274 Set the frame as top-field-first.
20275 bottom, b
20277 Set the frame as bottom-field-first.
20278 Examples
20280 • Interlace video using select and separatefields filter:
20282 separatefields,select=eq(mod(n,4),0)+eq(mod(n,4),3),weave
20284 xbr
20286 Apply the xBR high-quality magnification filter which is designed for
20287 pixel art. It follows a set of edge-detection rules, see
20288 <https://forums.libretro.com/t/xbr-algorithm-tutorial/123>.
20289 It accepts the following option:
20291 n Set the scaling dimension: 2 for "2xBR", 3 for "3xBR" and 4 for
20293 "4xBR". Default is 3.
20294 xcorrelate
20296 Apply normalized cross-correlation between first and second input video
20297 stream.
20298 Second input video stream dimensions must be lower than first input
20300 video stream.
20301 The filter accepts the following options:
20303 planes
20305 Set which planes to process.
20306 secondary
20308 Set which secondary video frames will be processed from second
20309 input video stream, can be first or all. Default is all.
20310 The "xcorrelate" filter also supports the framesync options.
20312 xfade
20314 Apply cross fade from one input video stream to another input video
20315 stream. The cross fade is applied for specified duration.
20316 Both inputs must be constant frame-rate and have the same resolution,
20318 pixel format, frame rate and timebase.
20319 The filter accepts the following options:
20321 transition
20323 Set one of available transition effects:
20324 custom
20326 fade
20327 wipeleft
20328 wiperight
20329 wipeup
20330 wipedown
20331 slideleft
20332 slideright
20333 slideup
20334 slidedown
20335 circlecrop
20336 rectcrop
20337 distance
20338 fadeblack
20339 fadewhite
20340 radial
20341 smoothleft
20342 smoothright
20343 smoothup
20344 smoothdown
20345 circleopen
20346 circleclose
20347 vertopen
20348 vertclose
20349 horzopen
20350 horzclose
20351 dissolve
20352 pixelize
20353 diagtl
20354 diagtr
20355 diagbl
20356 diagbr
20357 hlslice
20358 hrslice
20359 vuslice
20360 vdslice
20361 hblur
20362 fadegrays
20363 wipetl
20364 wipetr
20365 wipebl
20366 wipebr
20367 squeezeh
20368 squeezev
20369 zoomin
20370 Default transition effect is fade.
20372 duration
20374 Set cross fade duration in seconds. Range is 0 to 60 seconds.
20375 Default duration is 1 second.
20376 offset
20378 Set cross fade start relative to first input stream in seconds.
20379 Default offset is 0.
20380 expr
20382 Set expression for custom transition effect.
20383 The expressions can use the following variables and functions:
20385 X
20387 Y The coordinates of the current sample.
20388 W
20390 H The width and height of the image.
20391 P Progress of transition effect.
20393 PLANE
20395 Currently processed plane.
20396 A Return value of first input at current location and plane.
20398 B Return value of second input at current location and plane.
20400 a0(x, y)
20402 a1(x, y)
20403 a2(x, y)
20404 a3(x, y)
20405 Return the value of the pixel at location (x,y) of the
20406 first/second/third/fourth component of first input.
20407 b0(x, y)
20409 b1(x, y)
20410 b2(x, y)
20411 b3(x, y)
20412 Return the value of the pixel at location (x,y) of the
20413 first/second/third/fourth component of second input.
20414 Examples
20416 • Cross fade from one input video to another input video, with fade
20418 transition and duration of transition of 2 seconds starting at
20419 offset of 5 seconds:
20420 ffmpeg -i first.mp4 -i second.mp4 -filter_complex xfade=transition=fade:duration=2:offset=5 output.mp4
20422 xmedian
20424 Pick median pixels from several input videos.
20425 The filter accepts the following options:
20427 inputs
20429 Set number of inputs. Default is 3. Allowed range is from 3 to
20430 255. If number of inputs is even number, than result will be mean
20431 value between two median values.
20432 planes
20434 Set which planes to filter. Default value is 15, by which all
20435 planes are processed.
20436 percentile
20438 Set median percentile. Default value is 0.5. Default value of 0.5
20439 will pick always median values, while 0 will pick minimum values,
20440 and 1 maximum values.
20441 Commands
20443 This filter supports all above options as commands, excluding option
20445 "inputs".
20446 xstack
20448 Stack video inputs into custom layout.
20449 All streams must be of same pixel format.
20451 The filter accepts the following options:
20453 inputs
20455 Set number of input streams. Default is 2.
20456 layout
20458 Specify layout of inputs. This option requires the desired layout
20459 configuration to be explicitly set by the user. This sets position
20460 of each video input in output. Each input is separated by '|'. The
20461 first number represents the column, and the second number
20462 represents the row. Numbers start at 0 and are separated by '_'.
20463 Optionally one can use wX and hX, where X is video input from which
20464 to take width or height. Multiple values can be used when
20465 separated by '+'. In such case values are summed together.
20466 Note that if inputs are of different sizes gaps may appear, as not
20468 all of the output video frame will be filled. Similarly, videos can
20469 overlap each other if their position doesn't leave enough space for
20470 the full frame of adjoining videos.
20471 For 2 inputs, a default layout of "0_0|w0_0" is set. In all other
20473 cases, a layout must be set by the user.
20474 shortest
20476 If set to 1, force the output to terminate when the shortest input
20477 terminates. Default value is 0.
20478 fill
20480 If set to valid color, all unused pixels will be filled with that
20481 color. By default fill is set to none, so it is disabled.
20482 Examples
20484 • Display 4 inputs into 2x2 grid.
20486 Layout:
20488 input1(0, 0) | input3(w0, 0)
20490 input2(0, h0) | input4(w0, h0)
20491 xstack=inputs=4:layout=0_0|0_h0|w0_0|w0_h0
20495 Note that if inputs are of different sizes, gaps or overlaps may
20497 occur.
20498 • Display 4 inputs into 1x4 grid.
20500 Layout:
20502 input1(0, 0)
20504 input2(0, h0)
20505 input3(0, h0+h1)
20506 input4(0, h0+h1+h2)
20507 xstack=inputs=4:layout=0_0|0_h0|0_h0+h1|0_h0+h1+h2
20511 Note that if inputs are of different widths, unused space will
20513 appear.
20514 • Display 9 inputs into 3x3 grid.
20516 Layout:
20518 input1(0, 0) | input4(w0, 0) | input7(w0+w3, 0)
20520 input2(0, h0) | input5(w0, h0) | input8(w0+w3, h0)
20521 input3(0, h0+h1) | input6(w0, h0+h1) | input9(w0+w3, h0+h1)
20522 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
20526 Note that if inputs are of different sizes, gaps or overlaps may
20528 occur.
20529 • Display 16 inputs into 4x4 grid.
20531 Layout:
20533 input1(0, 0) | input5(w0, 0) | input9 (w0+w4, 0) | input13(w0+w4+w8, 0)
20535 input2(0, h0) | input6(w0, h0) | input10(w0+w4, h0) | input14(w0+w4+w8, h0)
20536 input3(0, h0+h1) | input7(w0, h0+h1) | input11(w0+w4, h0+h1) | input15(w0+w4+w8, h0+h1)
20537 input4(0, h0+h1+h2)| input8(w0, h0+h1+h2)| input12(w0+w4, h0+h1+h2)| input16(w0+w4+w8, h0+h1+h2)
20538 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|
20542 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
20543 Note that if inputs are of different sizes, gaps or overlaps may
20545 occur.
20546 yadif
20548 Deinterlace the input video ("yadif" means "yet another deinterlacing
20549 filter").
20550 It accepts the following parameters:
20552 mode
20554 The interlacing mode to adopt. It accepts one of the following
20555 values:
20556 0, send_frame
20558 Output one frame for each frame.
20559 1, send_field
20561 Output one frame for each field.
20562 2, send_frame_nospatial
20564 Like "send_frame", but it skips the spatial interlacing check.
20565 3, send_field_nospatial
20567 Like "send_field", but it skips the spatial interlacing check.
20568 The default value is "send_frame".
20570 parity
20572 The picture field parity assumed for the input interlaced video. It
20573 accepts one of the following values:
20574 0, tff
20576 Assume the top field is first.
20577 1, bff
20579 Assume the bottom field is first.
20580 -1, auto
20582 Enable automatic detection of field parity.
20583 The default value is "auto". If the interlacing is unknown or the
20585 decoder does not export this information, top field first will be
20586 assumed.
20587 deint
20589 Specify which frames to deinterlace. Accepts one of the following
20590 values:
20591 0, all
20593 Deinterlace all frames.
20594 1, interlaced
20596 Only deinterlace frames marked as interlaced.
20597 The default value is "all".
20599 yadif_cuda
20601 Deinterlace the input video using the yadif algorithm, but implemented
20602 in CUDA so that it can work as part of a GPU accelerated pipeline with
20603 nvdec and/or nvenc.
20604 It accepts the following parameters:
20606 mode
20608 The interlacing mode to adopt. It accepts one of the following
20609 values:
20610 0, send_frame
20612 Output one frame for each frame.
20613 1, send_field
20615 Output one frame for each field.
20616 2, send_frame_nospatial
20618 Like "send_frame", but it skips the spatial interlacing check.
20619 3, send_field_nospatial
20621 Like "send_field", but it skips the spatial interlacing check.
20622 The default value is "send_frame".
20624 parity
20626 The picture field parity assumed for the input interlaced video. It
20627 accepts one of the following values:
20628 0, tff
20630 Assume the top field is first.
20631 1, bff
20633 Assume the bottom field is first.
20634 -1, auto
20636 Enable automatic detection of field parity.
20637 The default value is "auto". If the interlacing is unknown or the
20639 decoder does not export this information, top field first will be
20640 assumed.
20641 deint
20643 Specify which frames to deinterlace. Accepts one of the following
20644 values:
20645 0, all
20647 Deinterlace all frames.
20648 1, interlaced
20650 Only deinterlace frames marked as interlaced.
20651 The default value is "all".
20653 yaepblur
20655 Apply blur filter while preserving edges ("yaepblur" means "yet another
20656 edge preserving blur filter"). The algorithm is described in "J. S.
20657 Lee, Digital image enhancement and noise filtering by use of local
20658 statistics, IEEE Trans. Pattern Anal. Mach. Intell. PAMI-2, 1980."
20659 It accepts the following parameters:
20661 radius, r
20663 Set the window radius. Default value is 3.
20664 planes, p
20666 Set which planes to filter. Default is only the first plane.
20667 sigma, s
20669 Set blur strength. Default value is 128.
20670 Commands
20672 This filter supports same commands as options.
20674 zoompan
20676 Apply Zoom & Pan effect.
20677 This filter accepts the following options:
20679 zoom, z
20681 Set the zoom expression. Range is 1-10. Default is 1.
20682 x
20684 y Set the x and y expression. Default is 0.
20685 d Set the duration expression in number of frames. This sets for how
20687 many number of frames effect will last for single input image.
20688 Default is 90.
20689 s Set the output image size, default is 'hd720'.
20691 fps Set the output frame rate, default is '25'.
20693 Each expression can contain the following constants:
20695 in_w, iw
20697 Input width.
20698 in_h, ih
20700 Input height.
20701 out_w, ow
20703 Output width.
20704 out_h, oh
20706 Output height.
20707 in Input frame count.
20709 on Output frame count.
20711 in_time, it
20713 The input timestamp expressed in seconds. It's NAN if the input
20714 timestamp is unknown.
20715 out_time, time, ot
20717 The output timestamp expressed in seconds.
20718 x
20720 y Last calculated 'x' and 'y' position from 'x' and 'y' expression
20721 for current input frame.
20722 px
20724 py 'x' and 'y' of last output frame of previous input frame or 0 when
20725 there was not yet such frame (first input frame).
20726 zoom
20728 Last calculated zoom from 'z' expression for current input frame.
20729 pzoom
20731 Last calculated zoom of last output frame of previous input frame.
20732 duration
20734 Number of output frames for current input frame. Calculated from
20735 'd' expression for each input frame.
20736 pduration
20738 number of output frames created for previous input frame
20739 a Rational number: input width / input height
20741 sar sample aspect ratio
20743 dar display aspect ratio
20745 Examples
20747 • Zoom in up to 1.5x and pan at same time to some spot near center of
20749 picture:
20750 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
20752 • Zoom in up to 1.5x and pan always at center of picture:
20754 zoompan=z='min(zoom+0.0015,1.5)':d=700:x='iw/2-(iw/zoom/2)':y='ih/2-(ih/zoom/2)'
20756 • Same as above but without pausing:
20758 zoompan=z='min(max(zoom,pzoom)+0.0015,1.5)':d=1:x='iw/2-(iw/zoom/2)':y='ih/2-(ih/zoom/2)'
20760 • Zoom in 2x into center of picture only for the first second of the
20762 input video:
20763 zoompan=z='if(between(in_time,0,1),2,1)':d=1:x='iw/2-(iw/zoom/2)':y='ih/2-(ih/zoom/2)'
20765 zscale
20767 Scale (resize) the input video, using the z.lib library:
20768 <https://github.com/sekrit-twc/zimg>. To enable compilation of this
20769 filter, you need to configure FFmpeg with "--enable-libzimg".
20770 The zscale filter forces the output display aspect ratio to be the same
20772 as the input, by changing the output sample aspect ratio.
20773 If the input image format is different from the format requested by the
20775 next filter, the zscale filter will convert the input to the requested
20776 format.
20777 Options
20779 The filter accepts the following options.
20781 width, w
20783 height, h
20784 Set the output video dimension expression. Default value is the
20785 input dimension.
20786 If the width or w value is 0, the input width is used for the
20788 output. If the height or h value is 0, the input height is used for
20789 the output.
20790 If one and only one of the values is -n with n >= 1, the zscale
20792 filter will use a value that maintains the aspect ratio of the
20793 input image, calculated from the other specified dimension. After
20794 that it will, however, make sure that the calculated dimension is
20795 divisible by n and adjust the value if necessary.
20796 If both values are -n with n >= 1, the behavior will be identical
20798 to both values being set to 0 as previously detailed.
20799 See below for the list of accepted constants for use in the
20801 dimension expression.
20802 size, s
20804 Set the video size. For the syntax of this option, check the "Video
20805 size" section in the ffmpeg-utils manual.
20806 dither, d
20808 Set the dither type.
20809 Possible values are:
20811 none
20813 ordered
20814 random
20815 error_diffusion
20816 Default is none.
20818 filter, f
20820 Set the resize filter type.
20821 Possible values are:
20823 point
20825 bilinear
20826 bicubic
20827 spline16
20828 spline36
20829 lanczos
20830 Default is bilinear.
20832 range, r
20834 Set the color range.
20835 Possible values are:
20837 input
20839 limited
20840 full
20841 Default is same as input.
20843 primaries, p
20845 Set the color primaries.
20846 Possible values are:
20848 input
20850 709
20851 unspecified
20852 170m
20853 240m
20854 2020
20855 Default is same as input.
20857 transfer, t
20859 Set the transfer characteristics.
20860 Possible values are:
20862 input
20864 709
20865 unspecified
20866 601
20867 linear
20868 2020_10
20869 2020_12
20870 smpte2084
20871 iec61966-2-1
20872 arib-std-b67
20873 Default is same as input.
20875 matrix, m
20877 Set the colorspace matrix.
20878 Possible value are:
20880 input
20882 709
20883 unspecified
20884 470bg
20885 170m
20886 2020_ncl
20887 2020_cl
20888 Default is same as input.
20890 rangein, rin
20892 Set the input color range.
20893 Possible values are:
20895 input
20897 limited
20898 full
20899 Default is same as input.
20901 primariesin, pin
20903 Set the input color primaries.
20904 Possible values are:
20906 input
20908 709
20909 unspecified
20910 170m
20911 240m
20912 2020
20913 Default is same as input.
20915 transferin, tin
20917 Set the input transfer characteristics.
20918 Possible values are:
20920 input
20922 709
20923 unspecified
20924 601
20925 linear
20926 2020_10
20927 2020_12
20928 Default is same as input.
20930 matrixin, min
20932 Set the input colorspace matrix.
20933 Possible value are:
20935 input
20937 709
20938 unspecified
20939 470bg
20940 170m
20941 2020_ncl
20942 2020_cl
20943 chromal, c
20944 Set the output chroma location.
20945 Possible values are:
20947 input
20949 left
20950 center
20951 topleft
20952 top
20953 bottomleft
20954 bottom
20955 chromalin, cin
20956 Set the input chroma location.
20957 Possible values are:
20959 input
20961 left
20962 center
20963 topleft
20964 top
20965 bottomleft
20966 bottom
20967 npl Set the nominal peak luminance.
20968 param_a
20970 Parameter A for scaling filters. Parameter "b" for bicubic, and the
20971 number of filter taps for lanczos.
20972 param_b
20974 Parameter B for scaling filters. Parameter "c" for bicubic.
20975 The values of the w and h options are expressions containing the
20977 following constants:
20978 in_w
20980 in_h
20981 The input width and height
20982 iw
20984 ih These are the same as in_w and in_h.
20985 out_w
20987 out_h
20988 The output (scaled) width and height
20989 ow
20991 oh These are the same as out_w and out_h
20992 a The same as iw / ih
20994 sar input sample aspect ratio
20996 dar The input display aspect ratio. Calculated from "(iw / ih) * sar".
20998 hsub
21000 vsub
21001 horizontal and vertical input chroma subsample values. For example
21002 for the pixel format "yuv422p" hsub is 2 and vsub is 1.
21003 ohsub
21005 ovsub
21006 horizontal and vertical output chroma subsample values. For example
21007 for the pixel format "yuv422p" hsub is 2 and vsub is 1.
21008 Commands
21010 This filter supports the following commands:
21012 width, w
21014 height, h
21015 Set the output video dimension expression. The command accepts the
21016 same syntax of the corresponding option.
21017 If the specified expression is not valid, it is kept at its current
21019 value.
21020
21022 Below is a description of the currently available OpenCL video filters.
21023 To enable compilation of these filters you need to configure FFmpeg
21025 with "--enable-opencl".
21026 Running OpenCL filters requires you to initialize a hardware device and
21028 to pass that device to all filters in any filter graph.
21029 -init_hw_device opencl[=name][:device[,key=value...]]
21031 Initialise a new hardware device of type opencl called name, using
21032 the given device parameters.
21033 -filter_hw_device name
21035 Pass the hardware device called name to all filters in any filter
21036 graph.
21037 For more detailed information see
21039 <https://www.ffmpeg.org/ffmpeg.html#Advanced-Video-options>
21040 • Example of choosing the first device on the second platform and
21042 running avgblur_opencl filter with default parameters on it.
21043 -init_hw_device opencl=gpu:1.0 -filter_hw_device gpu -i INPUT -vf "hwupload, avgblur_opencl, hwdownload" OUTPUT
21045 Since OpenCL filters are not able to access frame data in normal
21047 memory, all frame data needs to be uploaded(hwupload) to hardware
21048 surfaces connected to the appropriate device before being used and then
21049 downloaded(hwdownload) back to normal memory. Note that hwupload will
21050 upload to a surface with the same layout as the software frame, so it
21051 may be necessary to add a format filter immediately before to get the
21052 input into the right format and hwdownload does not support all formats
21053 on the output - it may be necessary to insert an additional format
21054 filter immediately following in the graph to get the output in a
21055 supported format.
21056 avgblur_opencl
21058 Apply average blur filter.
21059 The filter accepts the following options:
21061 sizeX
21063 Set horizontal radius size. Range is "[1, 1024]" and default value
21064 is 1.
21065 planes
21067 Set which planes to filter. Default value is 0xf, by which all
21068 planes are processed.
21069 sizeY
21071 Set vertical radius size. Range is "[1, 1024]" and default value is
21072 0. If zero, "sizeX" value will be used.
21073 Example
21075 • Apply average blur filter with horizontal and vertical size of 3,
21077 setting each pixel of the output to the average value of the 7x7
21078 region centered on it in the input. For pixels on the edges of the
21079 image, the region does not extend beyond the image boundaries, and
21080 so out-of-range coordinates are not used in the calculations.
21081 -i INPUT -vf "hwupload, avgblur_opencl=3, hwdownload" OUTPUT
21083 boxblur_opencl
21085 Apply a boxblur algorithm to the input video.
21086 It accepts the following parameters:
21088 luma_radius, lr
21090 luma_power, lp
21091 chroma_radius, cr
21092 chroma_power, cp
21093 alpha_radius, ar
21094 alpha_power, ap
21095 A description of the accepted options follows.
21097 luma_radius, lr
21099 chroma_radius, cr
21100 alpha_radius, ar
21101 Set an expression for the box radius in pixels used for blurring
21102 the corresponding input plane.
21103 The radius value must be a non-negative number, and must not be
21105 greater than the value of the expression "min(w,h)/2" for the luma
21106 and alpha planes, and of "min(cw,ch)/2" for the chroma planes.
21107 Default value for luma_radius is "2". If not specified,
21109 chroma_radius and alpha_radius default to the corresponding value
21110 set for luma_radius.
21111 The expressions can contain the following constants:
21113 w
21115 h The input width and height in pixels.
21116 cw
21118 ch The input chroma image width and height in pixels.
21119 hsub
21121 vsub
21122 The horizontal and vertical chroma subsample values. For
21123 example, for the pixel format "yuv422p", hsub is 2 and vsub is
21124 1.
21125 luma_power, lp
21127 chroma_power, cp
21128 alpha_power, ap
21129 Specify how many times the boxblur filter is applied to the
21130 corresponding plane.
21131 Default value for luma_power is 2. If not specified, chroma_power
21133 and alpha_power default to the corresponding value set for
21134 luma_power.
21135 A value of 0 will disable the effect.
21137 Examples
21139 Apply boxblur filter, setting each pixel of the output to the average
21141 value of box-radiuses luma_radius, chroma_radius, alpha_radius for each
21142 plane respectively. The filter will apply luma_power, chroma_power,
21143 alpha_power times onto the corresponding plane. For pixels on the edges
21144 of the image, the radius does not extend beyond the image boundaries,
21145 and so out-of-range coordinates are not used in the calculations.
21146 • Apply a boxblur filter with the luma, chroma, and alpha radius set
21148 to 2 and luma, chroma, and alpha power set to 3. The filter will
21149 run 3 times with box-radius set to 2 for every plane of the image.
21150 -i INPUT -vf "hwupload, boxblur_opencl=luma_radius=2:luma_power=3, hwdownload" OUTPUT
21152 -i INPUT -vf "hwupload, boxblur_opencl=2:3, hwdownload" OUTPUT
21153 • Apply a boxblur filter with luma radius set to 2, luma_power to 1,
21155 chroma_radius to 4, chroma_power to 5, alpha_radius to 3 and
21156 alpha_power to 7.
21157 For the luma plane, a 2x2 box radius will be run once.
21159 For the chroma plane, a 4x4 box radius will be run 5 times.
21161 For the alpha plane, a 3x3 box radius will be run 7 times.
21163 -i INPUT -vf "hwupload, boxblur_opencl=2:1:4:5:3:7, hwdownload" OUTPUT
21165 colorkey_opencl
21167 RGB colorspace color keying.
21168 The filter accepts the following options:
21170 color
21172 The color which will be replaced with transparency.
21173 similarity
21175 Similarity percentage with the key color.
21176 0.01 matches only the exact key color, while 1.0 matches
21178 everything.
21179 blend
21181 Blend percentage.
21182 0.0 makes pixels either fully transparent, or not transparent at
21184 all.
21185 Higher values result in semi-transparent pixels, with a higher
21187 transparency the more similar the pixels color is to the key color.
21188 Examples
21190 • Make every semi-green pixel in the input transparent with some
21192 slight blending:
21193 -i INPUT -vf "hwupload, colorkey_opencl=green:0.3:0.1, hwdownload" OUTPUT
21195 convolution_opencl
21197 Apply convolution of 3x3, 5x5, 7x7 matrix.
21198 The filter accepts the following options:
21200 0m
21202 1m
21203 2m
21204 3m Set matrix for each plane. Matrix is sequence of 9, 25 or 49
21205 signed numbers. Default value for each plane is "0 0 0 0 1 0 0 0
21206 0".
21207 0rdiv
21209 1rdiv
21210 2rdiv
21211 3rdiv
21212 Set multiplier for calculated value for each plane. If unset or 0,
21213 it will be sum of all matrix elements. The option value must be a
21214 float number greater or equal to 0.0. Default value is 1.0.
21215 0bias
21217 1bias
21218 2bias
21219 3bias
21220 Set bias for each plane. This value is added to the result of the
21221 multiplication. Useful for making the overall image brighter or
21222 darker. The option value must be a float number greater or equal
21223 to 0.0. Default value is 0.0.
21224 Examples
21226 • Apply sharpen:
21228 -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
21230 • Apply blur:
21232 -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
21234 • Apply edge enhance:
21236 -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
21238 • Apply edge detect:
21240 -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
21242 • Apply laplacian edge detector which includes diagonals:
21244 -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
21246 • Apply emboss:
21248 -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
21250 erosion_opencl
21252 Apply erosion effect to the video.
21253 This filter replaces the pixel by the local(3x3) minimum.
21255 It accepts the following options:
21257 threshold0
21259 threshold1
21260 threshold2
21261 threshold3
21262 Limit the maximum change for each plane. Range is "[0, 65535]" and
21263 default value is 65535. If 0, plane will remain unchanged.
21264 coordinates
21266 Flag which specifies the pixel to refer to. Range is "[0, 255]"
21267 and default value is 255, i.e. all eight pixels are used.
21268 Flags to local 3x3 coordinates region centered on "x":
21270 1 2 3
21272 4 x 5
21274 6 7 8
21276 Example
21278 • Apply erosion filter with threshold0 set to 30, threshold1 set 40,
21280 threshold2 set to 50 and coordinates set to 231, setting each pixel
21281 of the output to the local minimum between pixels: 1, 2, 3, 6, 7, 8
21282 of the 3x3 region centered on it in the input. If the difference
21283 between input pixel and local minimum is more then threshold of the
21284 corresponding plane, output pixel will be set to input pixel -
21285 threshold of corresponding plane.
21286 -i INPUT -vf "hwupload, erosion_opencl=30:40:50:coordinates=231, hwdownload" OUTPUT
21288 deshake_opencl
21290 Feature-point based video stabilization filter.
21291 The filter accepts the following options:
21293 tripod
21295 Simulates a tripod by preventing any camera movement whatsoever
21296 from the original frame. Defaults to 0.
21297 debug
21299 Whether or not additional debug info should be displayed, both in
21300 the processed output and in the console.
21301 Note that in order to see console debug output you will also need
21303 to pass "-v verbose" to ffmpeg.
21304 Viewing point matches in the output video is only supported for RGB
21306 input.
21307 Defaults to 0.
21309 adaptive_crop
21311 Whether or not to do a tiny bit of cropping at the borders to cut
21312 down on the amount of mirrored pixels.
21313 Defaults to 1.
21315 refine_features
21317 Whether or not feature points should be refined at a sub-pixel
21318 level.
21319 This can be turned off for a slight performance gain at the cost of
21321 precision.
21322 Defaults to 1.
21324 smooth_strength
21326 The strength of the smoothing applied to the camera path from 0.0
21327 to 1.0.
21328 1.0 is the maximum smoothing strength while values less than that
21330 result in less smoothing.
21331 0.0 causes the filter to adaptively choose a smoothing strength on
21333 a per-frame basis.
21334 Defaults to 0.0.
21336 smooth_window_multiplier
21338 Controls the size of the smoothing window (the number of frames
21339 buffered to determine motion information from).
21340 The size of the smoothing window is determined by multiplying the
21342 framerate of the video by this number.
21343 Acceptable values range from 0.1 to 10.0.
21345 Larger values increase the amount of motion data available for
21347 determining how to smooth the camera path, potentially improving
21348 smoothness, but also increase latency and memory usage.
21349 Defaults to 2.0.
21351 Examples
21353 • Stabilize a video with a fixed, medium smoothing strength:
21355 -i INPUT -vf "hwupload, deshake_opencl=smooth_strength=0.5, hwdownload" OUTPUT
21357 • Stabilize a video with debugging (both in console and in rendered
21359 video):
21360 -i INPUT -filter_complex "[0:v]format=rgba, hwupload, deshake_opencl=debug=1, hwdownload, format=rgba, format=yuv420p" -v verbose OUTPUT
21362 dilation_opencl
21364 Apply dilation effect to the video.
21365 This filter replaces the pixel by the local(3x3) maximum.
21367 It accepts the following options:
21369 threshold0
21371 threshold1
21372 threshold2
21373 threshold3
21374 Limit the maximum change for each plane. Range is "[0, 65535]" and
21375 default value is 65535. If 0, plane will remain unchanged.
21376 coordinates
21378 Flag which specifies the pixel to refer to. Range is "[0, 255]"
21379 and default value is 255, i.e. all eight pixels are used.
21380 Flags to local 3x3 coordinates region centered on "x":
21382 1 2 3
21384 4 x 5
21386 6 7 8
21388 Example
21390 • Apply dilation filter with threshold0 set to 30, threshold1 set 40,
21392 threshold2 set to 50 and coordinates set to 231, setting each pixel
21393 of the output to the local maximum between pixels: 1, 2, 3, 6, 7, 8
21394 of the 3x3 region centered on it in the input. If the difference
21395 between input pixel and local maximum is more then threshold of the
21396 corresponding plane, output pixel will be set to input pixel +
21397 threshold of corresponding plane.
21398 -i INPUT -vf "hwupload, dilation_opencl=30:40:50:coordinates=231, hwdownload" OUTPUT
21400 nlmeans_opencl
21402 Non-local Means denoise filter through OpenCL, this filter accepts same
21403 options as nlmeans.
21404 overlay_opencl
21406 Overlay one video on top of another.
21407 It takes two inputs and has one output. The first input is the "main"
21409 video on which the second input is overlaid. This filter requires same
21410 memory layout for all the inputs. So, format conversion may be needed.
21411 The filter accepts the following options:
21413 x Set the x coordinate of the overlaid video on the main video.
21415 Default value is 0.
21416 y Set the y coordinate of the overlaid video on the main video.
21418 Default value is 0.
21419 Examples
21421 • Overlay an image LOGO at the top-left corner of the INPUT video.
21423 Both inputs are yuv420p format.
21424 -i INPUT -i LOGO -filter_complex "[0:v]hwupload[a], [1:v]format=yuv420p, hwupload[b], [a][b]overlay_opencl, hwdownload" OUTPUT
21426 • The inputs have same memory layout for color channels , the overlay
21428 has additional alpha plane, like INPUT is yuv420p, and the LOGO is
21429 yuva420p.
21430 -i INPUT -i LOGO -filter_complex "[0:v]hwupload[a], [1:v]format=yuva420p, hwupload[b], [a][b]overlay_opencl, hwdownload" OUTPUT
21432 pad_opencl
21434 Add paddings to the input image, and place the original input at the
21435 provided x, y coordinates.
21436 It accepts the following options:
21438 width, w
21440 height, h
21441 Specify an expression for the size of the output image with the
21442 paddings added. If the value for width or height is 0, the
21443 corresponding input size is used for the output.
21444 The width expression can reference the value set by the height
21446 expression, and vice versa.
21447 The default value of width and height is 0.
21449 x
21451 y Specify the offsets to place the input image at within the padded
21452 area, with respect to the top/left border of the output image.
21453 The x expression can reference the value set by the y expression,
21455 and vice versa.
21456 The default value of x and y is 0.
21458 If x or y evaluate to a negative number, they'll be changed so the
21460 input image is centered on the padded area.
21461 color
21463 Specify the color of the padded area. For the syntax of this
21464 option, check the "Color" section in the ffmpeg-utils manual.
21465 aspect
21467 Pad to an aspect instead to a resolution.
21468 The value for the width, height, x, and y options are expressions
21470 containing the following constants:
21471 in_w
21473 in_h
21474 The input video width and height.
21475 iw
21477 ih These are the same as in_w and in_h.
21478 out_w
21480 out_h
21481 The output width and height (the size of the padded area), as
21482 specified by the width and height expressions.
21483 ow
21485 oh These are the same as out_w and out_h.
21486 x
21488 y The x and y offsets as specified by the x and y expressions, or NAN
21489 if not yet specified.
21490 a same as iw / ih
21492 sar input sample aspect ratio
21494 dar input display aspect ratio, it is the same as (iw / ih) * sar
21496 prewitt_opencl
21498 Apply the Prewitt operator
21499 (<https://en.wikipedia.org/wiki/Prewitt_operator>) to input video
21500 stream.
21501 The filter accepts the following option:
21503 planes
21505 Set which planes to filter. Default value is 0xf, by which all
21506 planes are processed.
21507 scale
21509 Set value which will be multiplied with filtered result. Range is
21510 "[0.0, 65535]" and default value is 1.0.
21511 delta
21513 Set value which will be added to filtered result. Range is
21514 "[-65535, 65535]" and default value is 0.0.
21515 Example
21517 • Apply the Prewitt operator with scale set to 2 and delta set to 10.
21519 -i INPUT -vf "hwupload, prewitt_opencl=scale=2:delta=10, hwdownload" OUTPUT
21521 program_opencl
21523 Filter video using an OpenCL program.
21524 source
21526 OpenCL program source file.
21527 kernel
21529 Kernel name in program.
21530 inputs
21532 Number of inputs to the filter. Defaults to 1.
21533 size, s
21535 Size of output frames. Defaults to the same as the first input.
21536 The "program_opencl" filter also supports the framesync options.
21538 The program source file must contain a kernel function with the given
21540 name, which will be run once for each plane of the output. Each run on
21541 a plane gets enqueued as a separate 2D global NDRange with one work-
21542 item for each pixel to be generated. The global ID offset for each
21543 work-item is therefore the coordinates of a pixel in the destination
21544 image.
21545 The kernel function needs to take the following arguments:
21547 • Destination image, __write_only image2d_t.
21549 This image will become the output; the kernel should write all of
21551 it.
21552 • Frame index, unsigned int.
21554 This is a counter starting from zero and increasing by one for each
21556 frame.
21557 • Source images, __read_only image2d_t.
21559 These are the most recent images on each input. The kernel may
21561 read from them to generate the output, but they can't be written
21562 to.
21563 Example programs:
21565 • Copy the input to the output (output must be the same size as the
21567 input).
21568 __kernel void copy(__write_only image2d_t destination,
21570 unsigned int index,
21571 __read_only image2d_t source)
21572 {
21573 const sampler_t sampler = CLK_NORMALIZED_COORDS_FALSE;
21574 int2 location = (int2)(get_global_id(0), get_global_id(1));
21576 float4 value = read_imagef(source, sampler, location);
21578 write_imagef(destination, location, value);
21580 }
21581 • Apply a simple transformation, rotating the input by an amount
21583 increasing with the index counter. Pixel values are linearly
21584 interpolated by the sampler, and the output need not have the same
21585 dimensions as the input.
21586 __kernel void rotate_image(__write_only image2d_t dst,
21588 unsigned int index,
21589 __read_only image2d_t src)
21590 {
21591 const sampler_t sampler = (CLK_NORMALIZED_COORDS_FALSE |
21592 CLK_FILTER_LINEAR);
21593 float angle = (float)index / 100.0f;
21595 float2 dst_dim = convert_float2(get_image_dim(dst));
21597 float2 src_dim = convert_float2(get_image_dim(src));
21598 float2 dst_cen = dst_dim / 2.0f;
21600 float2 src_cen = src_dim / 2.0f;
21601 int2 dst_loc = (int2)(get_global_id(0), get_global_id(1));
21603 float2 dst_pos = convert_float2(dst_loc) - dst_cen;
21605 float2 src_pos = {
21606 cos(angle) * dst_pos.x - sin(angle) * dst_pos.y,
21607 sin(angle) * dst_pos.x + cos(angle) * dst_pos.y
21608 };
21609 src_pos = src_pos * src_dim / dst_dim;
21610 float2 src_loc = src_pos + src_cen;
21612 if (src_loc.x < 0.0f || src_loc.y < 0.0f ||
21614 src_loc.x > src_dim.x || src_loc.y > src_dim.y)
21615 write_imagef(dst, dst_loc, 0.5f);
21616 else
21617 write_imagef(dst, dst_loc, read_imagef(src, sampler, src_loc));
21618 }
21619 • Blend two inputs together, with the amount of each input used
21621 varying with the index counter.
21622 __kernel void blend_images(__write_only image2d_t dst,
21624 unsigned int index,
21625 __read_only image2d_t src1,
21626 __read_only image2d_t src2)
21627 {
21628 const sampler_t sampler = (CLK_NORMALIZED_COORDS_FALSE |
21629 CLK_FILTER_LINEAR);
21630 float blend = (cos((float)index / 50.0f) + 1.0f) / 2.0f;
21632 int2 dst_loc = (int2)(get_global_id(0), get_global_id(1));
21634 int2 src1_loc = dst_loc * get_image_dim(src1) / get_image_dim(dst);
21635 int2 src2_loc = dst_loc * get_image_dim(src2) / get_image_dim(dst);
21636 float4 val1 = read_imagef(src1, sampler, src1_loc);
21638 float4 val2 = read_imagef(src2, sampler, src2_loc);
21639 write_imagef(dst, dst_loc, val1 * blend + val2 * (1.0f - blend));
21641 }
21642 roberts_opencl
21644 Apply the Roberts cross operator
21645 (<https://en.wikipedia.org/wiki/Roberts_cross>) to input video stream.
21646 The filter accepts the following option:
21648 planes
21650 Set which planes to filter. Default value is 0xf, by which all
21651 planes are processed.
21652 scale
21654 Set value which will be multiplied with filtered result. Range is
21655 "[0.0, 65535]" and default value is 1.0.
21656 delta
21658 Set value which will be added to filtered result. Range is
21659 "[-65535, 65535]" and default value is 0.0.
21660 Example
21662 • Apply the Roberts cross operator with scale set to 2 and delta set
21664 to 10
21665 -i INPUT -vf "hwupload, roberts_opencl=scale=2:delta=10, hwdownload" OUTPUT
21667 sobel_opencl
21669 Apply the Sobel operator
21670 (<https://en.wikipedia.org/wiki/Sobel_operator>) to input video stream.
21671 The filter accepts the following option:
21673 planes
21675 Set which planes to filter. Default value is 0xf, by which all
21676 planes are processed.
21677 scale
21679 Set value which will be multiplied with filtered result. Range is
21680 "[0.0, 65535]" and default value is 1.0.
21681 delta
21683 Set value which will be added to filtered result. Range is
21684 "[-65535, 65535]" and default value is 0.0.
21685 Example
21687 • Apply sobel operator with scale set to 2 and delta set to 10
21689 -i INPUT -vf "hwupload, sobel_opencl=scale=2:delta=10, hwdownload" OUTPUT
21691 tonemap_opencl
21693 Perform HDR(PQ/HLG) to SDR conversion with tone-mapping.
21694 It accepts the following parameters:
21696 tonemap
21698 Specify the tone-mapping operator to be used. Same as tonemap
21699 option in tonemap.
21700 param
21702 Tune the tone mapping algorithm. same as param option in tonemap.
21703 desat
21705 Apply desaturation for highlights that exceed this level of
21706 brightness. The higher the parameter, the more color information
21707 will be preserved. This setting helps prevent unnaturally blown-out
21708 colors for super-highlights, by (smoothly) turning into white
21709 instead. This makes images feel more natural, at the cost of
21710 reducing information about out-of-range colors.
21711 The default value is 0.5, and the algorithm here is a little
21713 different from the cpu version tonemap currently. A setting of 0.0
21714 disables this option.
21715 threshold
21717 The tonemapping algorithm parameters is fine-tuned per each scene.
21718 And a threshold is used to detect whether the scene has changed or
21719 not. If the distance between the current frame average brightness
21720 and the current running average exceeds a threshold value, we would
21721 re-calculate scene average and peak brightness. The default value
21722 is 0.2.
21723 format
21725 Specify the output pixel format.
21726 Currently supported formats are:
21728 p010
21730 nv12
21731 range, r
21732 Set the output color range.
21733 Possible values are:
21735 tv/mpeg
21737 pc/jpeg
21738 Default is same as input.
21740 primaries, p
21742 Set the output color primaries.
21743 Possible values are:
21745 bt709
21747 bt2020
21748 Default is same as input.
21750 transfer, t
21752 Set the output transfer characteristics.
21753 Possible values are:
21755 bt709
21757 bt2020
21758 Default is bt709.
21760 matrix, m
21762 Set the output colorspace matrix.
21763 Possible value are:
21765 bt709
21767 bt2020
21768 Default is same as input.
21770 Example
21772 • Convert HDR(PQ/HLG) video to bt2020-transfer-characteristic p010
21774 format using linear operator.
21775 -i INPUT -vf "format=p010,hwupload,tonemap_opencl=t=bt2020:tonemap=linear:format=p010,hwdownload,format=p010" OUTPUT
21777 unsharp_opencl
21779 Sharpen or blur the input video.
21780 It accepts the following parameters:
21782 luma_msize_x, lx
21784 Set the luma matrix horizontal size. Range is "[1, 23]" and
21785 default value is 5.
21786 luma_msize_y, ly
21788 Set the luma matrix vertical size. Range is "[1, 23]" and default
21789 value is 5.
21790 luma_amount, la
21792 Set the luma effect strength. Range is "[-10, 10]" and default
21793 value is 1.0.
21794 Negative values will blur the input video, while positive values
21796 will sharpen it, a value of zero will disable the effect.
21797 chroma_msize_x, cx
21799 Set the chroma matrix horizontal size. Range is "[1, 23]" and
21800 default value is 5.
21801 chroma_msize_y, cy
21803 Set the chroma matrix vertical size. Range is "[1, 23]" and
21804 default value is 5.
21805 chroma_amount, ca
21807 Set the chroma effect strength. Range is "[-10, 10]" and default
21808 value is 0.0.
21809 Negative values will blur the input video, while positive values
21811 will sharpen it, a value of zero will disable the effect.
21812 All parameters are optional and default to the equivalent of the string
21814 '5:5:1.0:5:5:0.0'.
21815 Examples
21817 • Apply strong luma sharpen effect:
21819 -i INPUT -vf "hwupload, unsharp_opencl=luma_msize_x=7:luma_msize_y=7:luma_amount=2.5, hwdownload" OUTPUT
21821 • Apply a strong blur of both luma and chroma parameters:
21823 -i INPUT -vf "hwupload, unsharp_opencl=7:7:-2:7:7:-2, hwdownload" OUTPUT
21825 xfade_opencl
21827 Cross fade two videos with custom transition effect by using OpenCL.
21828 It accepts the following options:
21830 transition
21832 Set one of possible transition effects.
21833 custom
21835 Select custom transition effect, the actual transition
21836 description will be picked from source and kernel options.
21837 fade
21839 wipeleft
21840 wiperight
21841 wipeup
21842 wipedown
21843 slideleft
21844 slideright
21845 slideup
21846 slidedown
21847 Default transition is fade.
21848 source
21850 OpenCL program source file for custom transition.
21851 kernel
21853 Set name of kernel to use for custom transition from program source
21854 file.
21855 duration
21857 Set duration of video transition.
21858 offset
21860 Set time of start of transition relative to first video.
21861 The program source file must contain a kernel function with the given
21863 name, which will be run once for each plane of the output. Each run on
21864 a plane gets enqueued as a separate 2D global NDRange with one work-
21865 item for each pixel to be generated. The global ID offset for each
21866 work-item is therefore the coordinates of a pixel in the destination
21867 image.
21868 The kernel function needs to take the following arguments:
21870 • Destination image, __write_only image2d_t.
21872 This image will become the output; the kernel should write all of
21874 it.
21875 • First Source image, __read_only image2d_t. Second Source image,
21877 __read_only image2d_t.
21878 These are the most recent images on each input. The kernel may
21880 read from them to generate the output, but they can't be written
21881 to.
21882 • Transition progress, float. This value is always between 0 and 1
21884 inclusive.
21885 Example programs:
21887 • Apply dots curtain transition effect:
21889 __kernel void blend_images(__write_only image2d_t dst,
21891 __read_only image2d_t src1,
21892 __read_only image2d_t src2,
21893 float progress)
21894 {
21895 const sampler_t sampler = (CLK_NORMALIZED_COORDS_FALSE |
21896 CLK_FILTER_LINEAR);
21897 int2 p = (int2)(get_global_id(0), get_global_id(1));
21898 float2 rp = (float2)(get_global_id(0), get_global_id(1));
21899 float2 dim = (float2)(get_image_dim(src1).x, get_image_dim(src1).y);
21900 rp = rp / dim;
21901 float2 dots = (float2)(20.0, 20.0);
21903 float2 center = (float2)(0,0);
21904 float2 unused;
21905 float4 val1 = read_imagef(src1, sampler, p);
21907 float4 val2 = read_imagef(src2, sampler, p);
21908 bool next = distance(fract(rp * dots, &unused), (float2)(0.5, 0.5)) < (progress / distance(rp, center));
21909 write_imagef(dst, p, next ? val1 : val2);
21911 }
21912
21914 VAAPI Video filters are usually used with VAAPI decoder and VAAPI
21915 encoder. Below is a description of VAAPI video filters.
21916 To enable compilation of these filters you need to configure FFmpeg
21918 with "--enable-vaapi".
21919 To use vaapi filters, you need to setup the vaapi device correctly. For
21921 more information, please read
21922 <https://trac.ffmpeg.org/wiki/Hardware/VAAPI>
21923 tonemap_vaapi
21925 Perform HDR(High Dynamic Range) to SDR(Standard Dynamic Range)
21926 conversion with tone-mapping. It maps the dynamic range of HDR10
21927 content to the SDR content. It currently only accepts HDR10 as input.
21928 It accepts the following parameters:
21930 format
21932 Specify the output pixel format.
21933 Currently supported formats are:
21935 p010
21937 nv12
21938 Default is nv12.
21940 primaries, p
21942 Set the output color primaries.
21943 Default is same as input.
21945 transfer, t
21947 Set the output transfer characteristics.
21948 Default is bt709.
21950 matrix, m
21952 Set the output colorspace matrix.
21953 Default is same as input.
21955 Example
21957 • Convert HDR(HDR10) video to bt2020-transfer-characteristic p010
21959 format
21960 tonemap_vaapi=format=p010:t=bt2020-10
21962
21964 Below is a description of the currently available video sources.
21965 buffer
21967 Buffer video frames, and make them available to the filter chain.
21968 This source is mainly intended for a programmatic use, in particular
21970 through the interface defined in libavfilter/buffersrc.h.
21971 It accepts the following parameters:
21973 video_size
21975 Specify the size (width and height) of the buffered video frames.
21976 For the syntax of this option, check the "Video size" section in
21977 the ffmpeg-utils manual.
21978 width
21980 The input video width.
21981 height
21983 The input video height.
21984 pix_fmt
21986 A string representing the pixel format of the buffered video
21987 frames. It may be a number corresponding to a pixel format, or a
21988 pixel format name.
21989 time_base
21991 Specify the timebase assumed by the timestamps of the buffered
21992 frames.
21993 frame_rate
21995 Specify the frame rate expected for the video stream.
21996 pixel_aspect, sar
21998 The sample (pixel) aspect ratio of the input video.
21999 sws_param
22001 This option is deprecated and ignored. Prepend "sws_flags=flags;"
22002 to the filtergraph description to specify swscale flags for
22003 automatically inserted scalers. See Filtergraph syntax.
22004 hw_frames_ctx
22006 When using a hardware pixel format, this should be a reference to
22007 an AVHWFramesContext describing input frames.
22008 For example:
22010 buffer=width=320:height=240:pix_fmt=yuv410p:time_base=1/24:sar=1
22012 will instruct the source to accept video frames with size 320x240 and
22014 with format "yuv410p", assuming 1/24 as the timestamps timebase and
22015 square pixels (1:1 sample aspect ratio). Since the pixel format with
22016 name "yuv410p" corresponds to the number 6 (check the enum
22017 AVPixelFormat definition in libavutil/pixfmt.h), this example
22018 corresponds to:
22019 buffer=size=320x240:pixfmt=6:time_base=1/24:pixel_aspect=1/1
22021 Alternatively, the options can be specified as a flat string, but this
22023 syntax is deprecated:
22024 width:height:pix_fmt:time_base.num:time_base.den:pixel_aspect.num:pixel_aspect.den
22026 cellauto
22028 Create a pattern generated by an elementary cellular automaton.
22029 The initial state of the cellular automaton can be defined through the
22031 filename and pattern options. If such options are not specified an
22032 initial state is created randomly.
22033 At each new frame a new row in the video is filled with the result of
22035 the cellular automaton next generation. The behavior when the whole
22036 frame is filled is defined by the scroll option.
22037 This source accepts the following options:
22039 filename, f
22041 Read the initial cellular automaton state, i.e. the starting row,
22042 from the specified file. In the file, each non-whitespace
22043 character is considered an alive cell, a newline will terminate the
22044 row, and further characters in the file will be ignored.
22045 pattern, p
22047 Read the initial cellular automaton state, i.e. the starting row,
22048 from the specified string.
22049 Each non-whitespace character in the string is considered an alive
22051 cell, a newline will terminate the row, and further characters in
22052 the string will be ignored.
22053 rate, r
22055 Set the video rate, that is the number of frames generated per
22056 second. Default is 25.
22057 random_fill_ratio, ratio
22059 Set the random fill ratio for the initial cellular automaton row.
22060 It is a floating point number value ranging from 0 to 1, defaults
22061 to 1/PHI.
22062 This option is ignored when a file or a pattern is specified.
22064 random_seed, seed
22066 Set the seed for filling randomly the initial row, must be an
22067 integer included between 0 and UINT32_MAX. If not specified, or if
22068 explicitly set to -1, the filter will try to use a good random seed
22069 on a best effort basis.
22070 rule
22072 Set the cellular automaton rule, it is a number ranging from 0 to
22073 255. Default value is 110.
22074 size, s
22076 Set the size of the output video. For the syntax of this option,
22077 check the "Video size" section in the ffmpeg-utils manual.
22078 If filename or pattern is specified, the size is set by default to
22080 the width of the specified initial state row, and the height is set
22081 to width * PHI.
22082 If size is set, it must contain the width of the specified pattern
22084 string, and the specified pattern will be centered in the larger
22085 row.
22086 If a filename or a pattern string is not specified, the size value
22088 defaults to "320x518" (used for a randomly generated initial
22089 state).
22090 scroll
22092 If set to 1, scroll the output upward when all the rows in the
22093 output have been already filled. If set to 0, the new generated row
22094 will be written over the top row just after the bottom row is
22095 filled. Defaults to 1.
22096 start_full, full
22098 If set to 1, completely fill the output with generated rows before
22099 outputting the first frame. This is the default behavior, for
22100 disabling set the value to 0.
22101 stitch
22103 If set to 1, stitch the left and right row edges together. This is
22104 the default behavior, for disabling set the value to 0.
22105 Examples
22107 • Read the initial state from pattern, and specify an output of size
22109 200x400.
22110 cellauto=f=pattern:s=200x400
22112 • Generate a random initial row with a width of 200 cells, with a
22114 fill ratio of 2/3:
22115 cellauto=ratio=2/3:s=200x200
22117 • Create a pattern generated by rule 18 starting by a single alive
22119 cell centered on an initial row with width 100:
22120 cellauto=p=@s=100x400:full=0:rule=18
22122 • Specify a more elaborated initial pattern:
22124 cellauto=p='@@ @ @@':s=100x400:full=0:rule=18
22126 coreimagesrc
22128 Video source generated on GPU using Apple's CoreImage API on OSX.
22129 This video source is a specialized version of the coreimage video
22131 filter. Use a core image generator at the beginning of the applied
22132 filterchain to generate the content.
22133 The coreimagesrc video source accepts the following options:
22135 list_generators
22137 List all available generators along with all their respective
22138 options as well as possible minimum and maximum values along with
22139 the default values.
22140 list_generators=true
22142 size, s
22144 Specify the size of the sourced video. For the syntax of this
22145 option, check the "Video size" section in the ffmpeg-utils manual.
22146 The default value is "320x240".
22147 rate, r
22149 Specify the frame rate of the sourced video, as the number of
22150 frames generated per second. It has to be a string in the format
22151 frame_rate_num/frame_rate_den, an integer number, a floating point
22152 number or a valid video frame rate abbreviation. The default value
22153 is "25".
22154 sar Set the sample aspect ratio of the sourced video.
22156 duration, d
22158 Set the duration of the sourced video. See the Time duration
22159 section in the ffmpeg-utils(1) manual for the accepted syntax.
22160 If not specified, or the expressed duration is negative, the video
22162 is supposed to be generated forever.
22163 Additionally, all options of the coreimage video filter are accepted.
22165 A complete filterchain can be used for further processing of the
22166 generated input without CPU-HOST transfer. See coreimage documentation
22167 and examples for details.
22168 Examples
22170 • Use CIQRCodeGenerator to create a QR code for the FFmpeg homepage,
22172 given as complete and escaped command-line for Apple's standard
22173 bash shell:
22174 ffmpeg -f lavfi -i coreimagesrc=s=100x100:filter=CIQRCodeGenerator@inputMessage=https\\\\\://FFmpeg.org/@inputCorrectionLevel=H -frames:v 1 QRCode.png
22176 This example is equivalent to the QRCode example of coreimage
22178 without the need for a nullsrc video source.
22179 gradients
22181 Generate several gradients.
22182 size, s
22184 Set frame size. For the syntax of this option, check the "Video
22185 size" section in the ffmpeg-utils manual. Default value is
22186 "640x480".
22187 rate, r
22189 Set frame rate, expressed as number of frames per second. Default
22190 value is "25".
22191 c0, c1, c2, c3, c4, c5, c6, c7
22193 Set 8 colors. Default values for colors is to pick random one.
22194 x0, y0, y0, y1
22196 Set gradient line source and destination points. If negative or out
22197 of range, random ones are picked.
22198 nb_colors, n
22200 Set number of colors to use at once. Allowed range is from 2 to 8.
22201 Default value is 2.
22202 seed
22204 Set seed for picking gradient line points.
22205 duration, d
22207 Set the duration of the sourced video. See the Time duration
22208 section in the ffmpeg-utils(1) manual for the accepted syntax.
22209 If not specified, or the expressed duration is negative, the video
22211 is supposed to be generated forever.
22212 speed
22214 Set speed of gradients rotation.
22215 mandelbrot
22217 Generate a Mandelbrot set fractal, and progressively zoom towards the
22218 point specified with start_x and start_y.
22219 This source accepts the following options:
22221 end_pts
22223 Set the terminal pts value. Default value is 400.
22224 end_scale
22226 Set the terminal scale value. Must be a floating point value.
22227 Default value is 0.3.
22228 inner
22230 Set the inner coloring mode, that is the algorithm used to draw the
22231 Mandelbrot fractal internal region.
22232 It shall assume one of the following values:
22234 black
22236 Set black mode.
22237 convergence
22239 Show time until convergence.
22240 mincol
22242 Set color based on point closest to the origin of the
22243 iterations.
22244 period
22246 Set period mode.
22247 Default value is mincol.
22249 bailout
22251 Set the bailout value. Default value is 10.0.
22252 maxiter
22254 Set the maximum of iterations performed by the rendering algorithm.
22255 Default value is 7189.
22256 outer
22258 Set outer coloring mode. It shall assume one of following values:
22259 iteration_count
22261 Set iteration count mode.
22262 normalized_iteration_count
22264 set normalized iteration count mode.
22265 Default value is normalized_iteration_count.
22267 rate, r
22269 Set frame rate, expressed as number of frames per second. Default
22270 value is "25".
22271 size, s
22273 Set frame size. For the syntax of this option, check the "Video
22274 size" section in the ffmpeg-utils manual. Default value is
22275 "640x480".
22276 start_scale
22278 Set the initial scale value. Default value is 3.0.
22279 start_x
22281 Set the initial x position. Must be a floating point value between
22282 -100 and 100. Default value is
22283 -0.743643887037158704752191506114774.
22284 start_y
22286 Set the initial y position. Must be a floating point value between
22287 -100 and 100. Default value is
22288 -0.131825904205311970493132056385139.
22289 mptestsrc
22291 Generate various test patterns, as generated by the MPlayer test
22292 filter.
22293 The size of the generated video is fixed, and is 256x256. This source
22295 is useful in particular for testing encoding features.
22296 This source accepts the following options:
22298 rate, r
22300 Specify the frame rate of the sourced video, as the number of
22301 frames generated per second. It has to be a string in the format
22302 frame_rate_num/frame_rate_den, an integer number, a floating point
22303 number or a valid video frame rate abbreviation. The default value
22304 is "25".
22305 duration, d
22307 Set the duration of the sourced video. See the Time duration
22308 section in the ffmpeg-utils(1) manual for the accepted syntax.
22309 If not specified, or the expressed duration is negative, the video
22311 is supposed to be generated forever.
22312 test, t
22314 Set the number or the name of the test to perform. Supported tests
22315 are:
22316 dc_luma
22318 dc_chroma
22319 freq_luma
22320 freq_chroma
22321 amp_luma
22322 amp_chroma
22323 cbp
22324 mv
22325 ring1
22326 ring2
22327 all
22328 max_frames, m
22329 Set the maximum number of frames generated for each test,
22330 default value is 30.
22331 Default value is "all", which will cycle through the list of all
22333 tests.
22334 Some examples:
22336 mptestsrc=t=dc_luma
22338 will generate a "dc_luma" test pattern.
22340 frei0r_src
22342 Provide a frei0r source.
22343 To enable compilation of this filter you need to install the frei0r
22345 header and configure FFmpeg with "--enable-frei0r".
22346 This source accepts the following parameters:
22348 size
22350 The size of the video to generate. For the syntax of this option,
22351 check the "Video size" section in the ffmpeg-utils manual.
22352 framerate
22354 The framerate of the generated video. It may be a string of the
22355 form num/den or a frame rate abbreviation.
22356 filter_name
22358 The name to the frei0r source to load. For more information
22359 regarding frei0r and how to set the parameters, read the frei0r
22360 section in the video filters documentation.
22361 filter_params
22363 A '|'-separated list of parameters to pass to the frei0r source.
22364 For example, to generate a frei0r partik0l source with size 200x200 and
22366 frame rate 10 which is overlaid on the overlay filter main input:
22367 frei0r_src=size=200x200:framerate=10:filter_name=partik0l:filter_params=1234 [overlay]; [in][overlay] overlay
22369 life
22371 Generate a life pattern.
22372 This source is based on a generalization of John Conway's life game.
22374 The sourced input represents a life grid, each pixel represents a cell
22376 which can be in one of two possible states, alive or dead. Every cell
22377 interacts with its eight neighbours, which are the cells that are
22378 horizontally, vertically, or diagonally adjacent.
22379 At each interaction the grid evolves according to the adopted rule,
22381 which specifies the number of neighbor alive cells which will make a
22382 cell stay alive or born. The rule option allows one to specify the rule
22383 to adopt.
22384 This source accepts the following options:
22386 filename, f
22388 Set the file from which to read the initial grid state. In the
22389 file, each non-whitespace character is considered an alive cell,
22390 and newline is used to delimit the end of each row.
22391 If this option is not specified, the initial grid is generated
22393 randomly.
22394 rate, r
22396 Set the video rate, that is the number of frames generated per
22397 second. Default is 25.
22398 random_fill_ratio, ratio
22400 Set the random fill ratio for the initial random grid. It is a
22401 floating point number value ranging from 0 to 1, defaults to 1/PHI.
22402 It is ignored when a file is specified.
22403 random_seed, seed
22405 Set the seed for filling the initial random grid, must be an
22406 integer included between 0 and UINT32_MAX. If not specified, or if
22407 explicitly set to -1, the filter will try to use a good random seed
22408 on a best effort basis.
22409 rule
22411 Set the life rule.
22412 A rule can be specified with a code of the kind "SNS/BNB", where NS
22414 and NB are sequences of numbers in the range 0-8, NS specifies the
22415 number of alive neighbor cells which make a live cell stay alive,
22416 and NB the number of alive neighbor cells which make a dead cell to
22417 become alive (i.e. to "born"). "s" and "b" can be used in place of
22418 "S" and "B", respectively.
22419 Alternatively a rule can be specified by an 18-bits integer. The 9
22421 high order bits are used to encode the next cell state if it is
22422 alive for each number of neighbor alive cells, the low order bits
22423 specify the rule for "borning" new cells. Higher order bits encode
22424 for an higher number of neighbor cells. For example the number
22425 6153 = "(12<<9)+9" specifies a stay alive rule of 12 and a born
22426 rule of 9, which corresponds to "S23/B03".
22427 Default value is "S23/B3", which is the original Conway's game of
22429 life rule, and will keep a cell alive if it has 2 or 3 neighbor
22430 alive cells, and will born a new cell if there are three alive
22431 cells around a dead cell.
22432 size, s
22434 Set the size of the output video. For the syntax of this option,
22435 check the "Video size" section in the ffmpeg-utils manual.
22436 If filename is specified, the size is set by default to the same
22438 size of the input file. If size is set, it must contain the size
22439 specified in the input file, and the initial grid defined in that
22440 file is centered in the larger resulting area.
22441 If a filename is not specified, the size value defaults to
22443 "320x240" (used for a randomly generated initial grid).
22444 stitch
22446 If set to 1, stitch the left and right grid edges together, and the
22447 top and bottom edges also. Defaults to 1.
22448 mold
22450 Set cell mold speed. If set, a dead cell will go from death_color
22451 to mold_color with a step of mold. mold can have a value from 0 to
22452 255.
22453 life_color
22455 Set the color of living (or new born) cells.
22456 death_color
22458 Set the color of dead cells. If mold is set, this is the first
22459 color used to represent a dead cell.
22460 mold_color
22462 Set mold color, for definitely dead and moldy cells.
22463 For the syntax of these 3 color options, check the "Color" section
22465 in the ffmpeg-utils manual.
22466 Examples
22468 • Read a grid from pattern, and center it on a grid of size 300x300
22470 pixels:
22471 life=f=pattern:s=300x300
22473 • Generate a random grid of size 200x200, with a fill ratio of 2/3:
22475 life=ratio=2/3:s=200x200
22477 • Specify a custom rule for evolving a randomly generated grid:
22479 life=rule=S14/B34
22481 • Full example with slow death effect (mold) using ffplay:
22483 ffplay -f lavfi life=s=300x200:mold=10:r=60:ratio=0.1:death_color=#C83232:life_color=#00ff00,scale=1200:800:flags=16
22485 allrgb, allyuv, color, colorspectrum, haldclutsrc, nullsrc, pal75bars,
22487 pal100bars, rgbtestsrc, smptebars, smptehdbars, testsrc, testsrc2,
22488 yuvtestsrc
22489 The "allrgb" source returns frames of size 4096x4096 of all rgb colors.
22490 The "allyuv" source returns frames of size 4096x4096 of all yuv colors.
22492 The "color" source provides an uniformly colored input.
22494 The "colorspectrum" source provides a color spectrum input.
22496 The "haldclutsrc" source provides an identity Hald CLUT. See also
22498 haldclut filter.
22499 The "nullsrc" source returns unprocessed video frames. It is mainly
22501 useful to be employed in analysis / debugging tools, or as the source
22502 for filters which ignore the input data.
22503 The "pal75bars" source generates a color bars pattern, based on EBU PAL
22505 recommendations with 75% color levels.
22506 The "pal100bars" source generates a color bars pattern, based on EBU
22508 PAL recommendations with 100% color levels.
22509 The "rgbtestsrc" source generates an RGB test pattern useful for
22511 detecting RGB vs BGR issues. You should see a red, green and blue
22512 stripe from top to bottom.
22513 The "smptebars" source generates a color bars pattern, based on the
22515 SMPTE Engineering Guideline EG 1-1990.
22516 The "smptehdbars" source generates a color bars pattern, based on the
22518 SMPTE RP 219-2002.
22519 The "testsrc" source generates a test video pattern, showing a color
22521 pattern, a scrolling gradient and a timestamp. This is mainly intended
22522 for testing purposes.
22523 The "testsrc2" source is similar to testsrc, but supports more pixel
22525 formats instead of just "rgb24". This allows using it as an input for
22526 other tests without requiring a format conversion.
22527 The "yuvtestsrc" source generates an YUV test pattern. You should see a
22529 y, cb and cr stripe from top to bottom.
22530 The sources accept the following parameters:
22532 level
22534 Specify the level of the Hald CLUT, only available in the
22535 "haldclutsrc" source. A level of "N" generates a picture of "N*N*N"
22536 by "N*N*N" pixels to be used as identity matrix for 3D lookup
22537 tables. Each component is coded on a "1/(N*N)" scale.
22538 color, c
22540 Specify the color of the source, only available in the "color"
22541 source. For the syntax of this option, check the "Color" section in
22542 the ffmpeg-utils manual.
22543 size, s
22545 Specify the size of the sourced video. For the syntax of this
22546 option, check the "Video size" section in the ffmpeg-utils manual.
22547 The default value is "320x240".
22548 This option is not available with the "allrgb", "allyuv", and
22550 "haldclutsrc" filters.
22551 rate, r
22553 Specify the frame rate of the sourced video, as the number of
22554 frames generated per second. It has to be a string in the format
22555 frame_rate_num/frame_rate_den, an integer number, a floating point
22556 number or a valid video frame rate abbreviation. The default value
22557 is "25".
22558 duration, d
22560 Set the duration of the sourced video. See the Time duration
22561 section in the ffmpeg-utils(1) manual for the accepted syntax.
22562 If not specified, or the expressed duration is negative, the video
22564 is supposed to be generated forever.
22565 Since the frame rate is used as time base, all frames including the
22567 last one will have their full duration. If the specified duration
22568 is not a multiple of the frame duration, it will be rounded up.
22569 sar Set the sample aspect ratio of the sourced video.
22571 alpha
22573 Specify the alpha (opacity) of the background, only available in
22574 the "testsrc2" source. The value must be between 0 (fully
22575 transparent) and 255 (fully opaque, the default).
22576 decimals, n
22578 Set the number of decimals to show in the timestamp, only available
22579 in the "testsrc" source.
22580 The displayed timestamp value will correspond to the original
22582 timestamp value multiplied by the power of 10 of the specified
22583 value. Default value is 0.
22584 type
22586 Set the type of the color spectrum, only available in the
22587 "colorspectrum" source. Can be one of the following:
22588 black
22590 white
22591 all
22592 Examples
22594 • Generate a video with a duration of 5.3 seconds, with size 176x144
22596 and a frame rate of 10 frames per second:
22597 testsrc=duration=5.3:size=qcif:rate=10
22599 • The following graph description will generate a red source with an
22601 opacity of 0.2, with size "qcif" and a frame rate of 10 frames per
22602 second:
22603 color=c=red@0.2:s=qcif:r=10
22605 • If the input content is to be ignored, "nullsrc" can be used. The
22607 following command generates noise in the luminance plane by
22608 employing the "geq" filter:
22609 nullsrc=s=256x256, geq=random(1)*255:128:128
22611 Commands
22613 The "color" source supports the following commands:
22615 c, color
22617 Set the color of the created image. Accepts the same syntax of the
22618 corresponding color option.
22619 openclsrc
22621 Generate video using an OpenCL program.
22622 source
22624 OpenCL program source file.
22625 kernel
22627 Kernel name in program.
22628 size, s
22630 Size of frames to generate. This must be set.
22631 format
22633 Pixel format to use for the generated frames. This must be set.
22634 rate, r
22636 Number of frames generated every second. Default value is '25'.
22637 For details of how the program loading works, see the program_opencl
22639 filter.
22640 Example programs:
22642 • Generate a colour ramp by setting pixel values from the position of
22644 the pixel in the output image. (Note that this will work with all
22645 pixel formats, but the generated output will not be the same.)
22646 __kernel void ramp(__write_only image2d_t dst,
22648 unsigned int index)
22649 {
22650 int2 loc = (int2)(get_global_id(0), get_global_id(1));
22651 float4 val;
22653 val.xy = val.zw = convert_float2(loc) / convert_float2(get_image_dim(dst));
22654 write_imagef(dst, loc, val);
22656 }
22657 • Generate a Sierpinski carpet pattern, panning by a single pixel
22659 each frame.
22660 __kernel void sierpinski_carpet(__write_only image2d_t dst,
22662 unsigned int index)
22663 {
22664 int2 loc = (int2)(get_global_id(0), get_global_id(1));
22665 float4 value = 0.0f;
22667 int x = loc.x + index;
22668 int y = loc.y + index;
22669 while (x > 0 || y > 0) {
22670 if (x % 3 == 1 && y % 3 == 1) {
22671 value = 1.0f;
22672 break;
22673 }
22674 x /= 3;
22675 y /= 3;
22676 }
22677 write_imagef(dst, loc, value);
22679 }
22680 sierpinski
22682 Generate a Sierpinski carpet/triangle fractal, and randomly pan around.
22683 This source accepts the following options:
22685 size, s
22687 Set frame size. For the syntax of this option, check the "Video
22688 size" section in the ffmpeg-utils manual. Default value is
22689 "640x480".
22690 rate, r
22692 Set frame rate, expressed as number of frames per second. Default
22693 value is "25".
22694 seed
22696 Set seed which is used for random panning.
22697 jump
22699 Set max jump for single pan destination. Allowed range is from 1 to
22700 10000.
22701 type
22703 Set fractal type, can be default "carpet" or "triangle".
22704
22706 Below is a description of the currently available video sinks.
22707 buffersink
22709 Buffer video frames, and make them available to the end of the filter
22710 graph.
22711 This sink is mainly intended for programmatic use, in particular
22713 through the interface defined in libavfilter/buffersink.h or the
22714 options system.
22715 It accepts a pointer to an AVBufferSinkContext structure, which defines
22717 the incoming buffers' formats, to be passed as the opaque parameter to
22718 "avfilter_init_filter" for initialization.
22719 nullsink
22721 Null video sink: do absolutely nothing with the input video. It is
22722 mainly useful as a template and for use in analysis / debugging tools.
22723
22725 Below is a description of the currently available multimedia filters.
22726 abitscope
22728 Convert input audio to a video output, displaying the audio bit scope.
22729 The filter accepts the following options:
22731 rate, r
22733 Set frame rate, expressed as number of frames per second. Default
22734 value is "25".
22735 size, s
22737 Specify the video size for the output. For the syntax of this
22738 option, check the "Video size" section in the ffmpeg-utils manual.
22739 Default value is "1024x256".
22740 colors
22742 Specify list of colors separated by space or by '|' which will be
22743 used to draw channels. Unrecognized or missing colors will be
22744 replaced by white color.
22745 adrawgraph
22747 Draw a graph using input audio metadata.
22748 See drawgraph
22750 agraphmonitor
22752 See graphmonitor.
22753 ahistogram
22755 Convert input audio to a video output, displaying the volume histogram.
22756 The filter accepts the following options:
22758 dmode
22760 Specify how histogram is calculated.
22761 It accepts the following values:
22763 single
22765 Use single histogram for all channels.
22766 separate
22768 Use separate histogram for each channel.
22769 Default is "single".
22771 rate, r
22773 Set frame rate, expressed as number of frames per second. Default
22774 value is "25".
22775 size, s
22777 Specify the video size for the output. For the syntax of this
22778 option, check the "Video size" section in the ffmpeg-utils manual.
22779 Default value is "hd720".
22780 scale
22782 Set display scale.
22783 It accepts the following values:
22785 log logarithmic
22787 sqrt
22789 square root
22790 cbrt
22792 cubic root
22793 lin linear
22795 rlog
22797 reverse logarithmic
22798 Default is "log".
22800 ascale
22802 Set amplitude scale.
22803 It accepts the following values:
22805 log logarithmic
22807 lin linear
22809 Default is "log".
22811 acount
22813 Set how much frames to accumulate in histogram. Default is 1.
22814 Setting this to -1 accumulates all frames.
22815 rheight
22817 Set histogram ratio of window height.
22818 slide
22820 Set sonogram sliding.
22821 It accepts the following values:
22823 replace
22825 replace old rows with new ones.
22826 scroll
22828 scroll from top to bottom.
22829 Default is "replace".
22831 aphasemeter
22833 Measures phase of input audio, which is exported as metadata
22834 "lavfi.aphasemeter.phase", representing mean phase of current audio
22835 frame. A video output can also be produced and is enabled by default.
22836 The audio is passed through as first output.
22837 Audio will be rematrixed to stereo if it has a different channel
22839 layout. Phase value is in range "[-1, 1]" where "-1" means left and
22840 right channels are completely out of phase and 1 means channels are in
22841 phase.
22842 The filter accepts the following options, all related to its video
22844 output:
22845 rate, r
22847 Set the output frame rate. Default value is 25.
22848 size, s
22850 Set the video size for the output. For the syntax of this option,
22851 check the "Video size" section in the ffmpeg-utils manual. Default
22852 value is "800x400".
22853 rc
22855 gc
22856 bc Specify the red, green, blue contrast. Default values are 2, 7 and
22857 1. Allowed range is "[0, 255]".
22858 mpc Set color which will be used for drawing median phase. If color is
22860 "none" which is default, no median phase value will be drawn.
22861 video
22863 Enable video output. Default is enabled.
22864 phasing detection
22866 The filter also detects out of phase and mono sequences in stereo
22868 streams. It logs the sequence start, end and duration when it lasts
22869 longer or as long as the minimum set.
22870 The filter accepts the following options for this detection:
22872 phasing
22874 Enable mono and out of phase detection. Default is disabled.
22875 tolerance, t
22877 Set phase tolerance for mono detection, in amplitude ratio. Default
22878 is 0. Allowed range is "[0, 1]".
22879 angle, a
22881 Set angle threshold for out of phase detection, in degree. Default
22882 is 170. Allowed range is "[90, 180]".
22883 duration, d
22885 Set mono or out of phase duration until notification, expressed in
22886 seconds. Default is 2.
22887 Examples
22889 • Complete example with ffmpeg to detect 1 second of mono with 0.001
22891 phase tolerance:
22892 ffmpeg -i stereo.wav -af aphasemeter=video=0:phasing=1:duration=1:tolerance=0.001 -f null -
22894 avectorscope
22896 Convert input audio to a video output, representing the audio vector
22897 scope.
22898 The filter is used to measure the difference between channels of stereo
22900 audio stream. A monaural signal, consisting of identical left and right
22901 signal, results in straight vertical line. Any stereo separation is
22902 visible as a deviation from this line, creating a Lissajous figure. If
22903 the straight (or deviation from it) but horizontal line appears this
22904 indicates that the left and right channels are out of phase.
22905 The filter accepts the following options:
22907 mode, m
22909 Set the vectorscope mode.
22910 Available values are:
22912 lissajous
22914 Lissajous rotated by 45 degrees.
22915 lissajous_xy
22917 Same as above but not rotated.
22918 polar
22920 Shape resembling half of circle.
22921 Default value is lissajous.
22923 size, s
22925 Set the video size for the output. For the syntax of this option,
22926 check the "Video size" section in the ffmpeg-utils manual. Default
22927 value is "400x400".
22928 rate, r
22930 Set the output frame rate. Default value is 25.
22931 rc
22933 gc
22934 bc
22935 ac Specify the red, green, blue and alpha contrast. Default values are
22936 40, 160, 80 and 255. Allowed range is "[0, 255]".
22937 rf
22939 gf
22940 bf
22941 af Specify the red, green, blue and alpha fade. Default values are 15,
22942 10, 5 and 5. Allowed range is "[0, 255]".
22943 zoom
22945 Set the zoom factor. Default value is 1. Allowed range is "[0,
22946 10]". Values lower than 1 will auto adjust zoom factor to maximal
22947 possible value.
22948 draw
22950 Set the vectorscope drawing mode.
22951 Available values are:
22953 dot Draw dot for each sample.
22955 line
22957 Draw line between previous and current sample.
22958 Default value is dot.
22960 scale
22962 Specify amplitude scale of audio samples.
22963 Available values are:
22965 lin Linear.
22967 sqrt
22969 Square root.
22970 cbrt
22972 Cubic root.
22973 log Logarithmic.
22975 swap
22977 Swap left channel axis with right channel axis.
22978 mirror
22980 Mirror axis.
22981 none
22983 No mirror.
22984 x Mirror only x axis.
22986 y Mirror only y axis.
22988 xy Mirror both axis.
22990 Examples
22992 • Complete example using ffplay:
22994 ffplay -f lavfi 'amovie=input.mp3, asplit [a][out1];
22996 [a] avectorscope=zoom=1.3:rc=2:gc=200:bc=10:rf=1:gf=8:bf=7 [out0]'
22997 Commands
22999 This filter supports the all above options as commands except options
23001 "size" and "rate".
23002 bench, abench
23004 Benchmark part of a filtergraph.
23005 The filter accepts the following options:
23007 action
23009 Start or stop a timer.
23010 Available values are:
23012 start
23014 Get the current time, set it as frame metadata (using the key
23015 "lavfi.bench.start_time"), and forward the frame to the next
23016 filter.
23017 stop
23019 Get the current time and fetch the "lavfi.bench.start_time"
23020 metadata from the input frame metadata to get the time
23021 difference. Time difference, average, maximum and minimum time
23022 (respectively "t", "avg", "max" and "min") are then printed.
23023 The timestamps are expressed in seconds.
23024 Examples
23026 • Benchmark selectivecolor filter:
23028 bench=start,selectivecolor=reds=-.2 .12 -.49,bench=stop
23030 concat
23032 Concatenate audio and video streams, joining them together one after
23033 the other.
23034 The filter works on segments of synchronized video and audio streams.
23036 All segments must have the same number of streams of each type, and
23037 that will also be the number of streams at output.
23038 The filter accepts the following options:
23040 n Set the number of segments. Default is 2.
23042 v Set the number of output video streams, that is also the number of
23044 video streams in each segment. Default is 1.
23045 a Set the number of output audio streams, that is also the number of
23047 audio streams in each segment. Default is 0.
23048 unsafe
23050 Activate unsafe mode: do not fail if segments have a different
23051 format.
23052 The filter has v+a outputs: first v video outputs, then a audio
23054 outputs.
23055 There are nx(v+a) inputs: first the inputs for the first segment, in
23057 the same order as the outputs, then the inputs for the second segment,
23058 etc.
23059 Related streams do not always have exactly the same duration, for
23061 various reasons including codec frame size or sloppy authoring. For
23062 that reason, related synchronized streams (e.g. a video and its audio
23063 track) should be concatenated at once. The concat filter will use the
23064 duration of the longest stream in each segment (except the last one),
23065 and if necessary pad shorter audio streams with silence.
23066 For this filter to work correctly, all segments must start at timestamp
23068 0.
23069 All corresponding streams must have the same parameters in all
23071 segments; the filtering system will automatically select a common pixel
23072 format for video streams, and a common sample format, sample rate and
23073 channel layout for audio streams, but other settings, such as
23074 resolution, must be converted explicitly by the user.
23075 Different frame rates are acceptable but will result in variable frame
23077 rate at output; be sure to configure the output file to handle it.
23078 Examples
23080 • Concatenate an opening, an episode and an ending, all in bilingual
23082 version (video in stream 0, audio in streams 1 and 2):
23083 ffmpeg -i opening.mkv -i episode.mkv -i ending.mkv -filter_complex \
23085 '[0:0] [0:1] [0:2] [1:0] [1:1] [1:2] [2:0] [2:1] [2:2]
23086 concat=n=3:v=1:a=2 [v] [a1] [a2]' \
23087 -map '[v]' -map '[a1]' -map '[a2]' output.mkv
23088 • Concatenate two parts, handling audio and video separately, using
23090 the (a)movie sources, and adjusting the resolution:
23091 movie=part1.mp4, scale=512:288 [v1] ; amovie=part1.mp4 [a1] ;
23093 movie=part2.mp4, scale=512:288 [v2] ; amovie=part2.mp4 [a2] ;
23094 [v1] [v2] concat [outv] ; [a1] [a2] concat=v=0:a=1 [outa]
23095 Note that a desync will happen at the stitch if the audio and video
23097 streams do not have exactly the same duration in the first file.
23098 Commands
23100 This filter supports the following commands:
23102 next
23104 Close the current segment and step to the next one
23105 ebur128
23107 EBU R128 scanner filter. This filter takes an audio stream and analyzes
23108 its loudness level. By default, it logs a message at a frequency of
23109 10Hz with the Momentary loudness (identified by "M"), Short-term
23110 loudness ("S"), Integrated loudness ("I") and Loudness Range ("LRA").
23111 The filter can only analyze streams which have sample format is double-
23113 precision floating point. The input stream will be converted to this
23114 specification, if needed. Users may need to insert aformat and/or
23115 aresample filters after this filter to obtain the original parameters.
23116 The filter also has a video output (see the video option) with a real
23118 time graph to observe the loudness evolution. The graphic contains the
23119 logged message mentioned above, so it is not printed anymore when this
23120 option is set, unless the verbose logging is set. The main graphing
23121 area contains the short-term loudness (3 seconds of analysis), and the
23122 gauge on the right is for the momentary loudness (400 milliseconds),
23123 but can optionally be configured to instead display short-term loudness
23124 (see gauge).
23125 The green area marks a +/- 1LU target range around the target loudness
23127 (-23LUFS by default, unless modified through target).
23128 More information about the Loudness Recommendation EBU R128 on
23130 <http://tech.ebu.ch/loudness>.
23131 The filter accepts the following options:
23133 video
23135 Activate the video output. The audio stream is passed unchanged
23136 whether this option is set or no. The video stream will be the
23137 first output stream if activated. Default is 0.
23138 size
23140 Set the video size. This option is for video only. For the syntax
23141 of this option, check the "Video size" section in the ffmpeg-utils
23142 manual. Default and minimum resolution is "640x480".
23143 meter
23145 Set the EBU scale meter. Default is 9. Common values are 9 and 18,
23146 respectively for EBU scale meter +9 and EBU scale meter +18. Any
23147 other integer value between this range is allowed.
23148 metadata
23150 Set metadata injection. If set to 1, the audio input will be
23151 segmented into 100ms output frames, each of them containing various
23152 loudness information in metadata. All the metadata keys are
23153 prefixed with "lavfi.r128.".
23154 Default is 0.
23156 framelog
23158 Force the frame logging level.
23159 Available values are:
23161 info
23163 information logging level
23164 verbose
23166 verbose logging level
23167 By default, the logging level is set to info. If the video or the
23169 metadata options are set, it switches to verbose.
23170 peak
23172 Set peak mode(s).
23173 Available modes can be cumulated (the option is a "flag" type).
23175 Possible values are:
23176 none
23178 Disable any peak mode (default).
23179 sample
23181 Enable sample-peak mode.
23182 Simple peak mode looking for the higher sample value. It logs a
23184 message for sample-peak (identified by "SPK").
23185 true
23187 Enable true-peak mode.
23188 If enabled, the peak lookup is done on an over-sampled version
23190 of the input stream for better peak accuracy. It logs a message
23191 for true-peak. (identified by "TPK") and true-peak per frame
23192 (identified by "FTPK"). This mode requires a build with
23193 "libswresample".
23194 dualmono
23196 Treat mono input files as "dual mono". If a mono file is intended
23197 for playback on a stereo system, its EBU R128 measurement will be
23198 perceptually incorrect. If set to "true", this option will
23199 compensate for this effect. Multi-channel input files are not
23200 affected by this option.
23201 panlaw
23203 Set a specific pan law to be used for the measurement of dual mono
23204 files. This parameter is optional, and has a default value of
23205 -3.01dB.
23206 target
23208 Set a specific target level (in LUFS) used as relative zero in the
23209 visualization. This parameter is optional and has a default value
23210 of -23LUFS as specified by EBU R128. However, material published
23211 online may prefer a level of -16LUFS (e.g. for use with podcasts or
23212 video platforms).
23213 gauge
23215 Set the value displayed by the gauge. Valid values are "momentary"
23216 and s "shortterm". By default the momentary value will be used, but
23217 in certain scenarios it may be more useful to observe the short
23218 term value instead (e.g. live mixing).
23219 scale
23221 Sets the display scale for the loudness. Valid parameters are
23222 "absolute" (in LUFS) or "relative" (LU) relative to the target.
23223 This only affects the video output, not the summary or continuous
23224 log output.
23225 Examples
23227 • Real-time graph using ffplay, with a EBU scale meter +18:
23229 ffplay -f lavfi -i "amovie=input.mp3,ebur128=video=1:meter=18 [out0][out1]"
23231 • Run an analysis with ffmpeg:
23233 ffmpeg -nostats -i input.mp3 -filter_complex ebur128 -f null -
23235 interleave, ainterleave
23237 Temporally interleave frames from several inputs.
23238 "interleave" works with video inputs, "ainterleave" with audio.
23240 These filters read frames from several inputs and send the oldest
23242 queued frame to the output.
23243 Input streams must have well defined, monotonically increasing frame
23245 timestamp values.
23246 In order to submit one frame to output, these filters need to enqueue
23248 at least one frame for each input, so they cannot work in case one
23249 input is not yet terminated and will not receive incoming frames.
23250 For example consider the case when one input is a "select" filter which
23252 always drops input frames. The "interleave" filter will keep reading
23253 from that input, but it will never be able to send new frames to output
23254 until the input sends an end-of-stream signal.
23255 Also, depending on inputs synchronization, the filters will drop frames
23257 in case one input receives more frames than the other ones, and the
23258 queue is already filled.
23259 These filters accept the following options:
23261 nb_inputs, n
23263 Set the number of different inputs, it is 2 by default.
23264 duration
23266 How to determine the end-of-stream.
23267 longest
23269 The duration of the longest input. (default)
23270 shortest
23272 The duration of the shortest input.
23273 first
23275 The duration of the first input.
23276 Examples
23278 • Interleave frames belonging to different streams using ffmpeg:
23280 ffmpeg -i bambi.avi -i pr0n.mkv -filter_complex "[0:v][1:v] interleave" out.avi
23282 • Add flickering blur effect:
23284 select='if(gt(random(0), 0.2), 1, 2)':n=2 [tmp], boxblur=2:2, [tmp] interleave
23286 latency, alatency
23288 Measure filtering latency.
23289 Report previous filter filtering latency, delay in number of audio
23291 samples for audio filters or number of video frames for video filters.
23292 On end of input stream, filter will report min and max measured latency
23294 for previous running filter in filtergraph.
23295 metadata, ametadata
23297 Manipulate frame metadata.
23298 This filter accepts the following options:
23300 mode
23302 Set mode of operation of the filter.
23303 Can be one of the following:
23305 select
23307 If both "value" and "key" is set, select frames which have such
23308 metadata. If only "key" is set, select every frame that has
23309 such key in metadata.
23310 add Add new metadata "key" and "value". If key is already available
23312 do nothing.
23313 modify
23315 Modify value of already present key.
23316 delete
23318 If "value" is set, delete only keys that have such value.
23319 Otherwise, delete key. If "key" is not set, delete all metadata
23320 values in the frame.
23321 print
23323 Print key and its value if metadata was found. If "key" is not
23324 set print all metadata values available in frame.
23325 key Set key used with all modes. Must be set for all modes except
23327 "print" and "delete".
23328 value
23330 Set metadata value which will be used. This option is mandatory for
23331 "modify" and "add" mode.
23332 function
23334 Which function to use when comparing metadata value and "value".
23335 Can be one of following:
23337 same_str
23339 Values are interpreted as strings, returns true if metadata
23340 value is same as "value".
23341 starts_with
23343 Values are interpreted as strings, returns true if metadata
23344 value starts with the "value" option string.
23345 less
23347 Values are interpreted as floats, returns true if metadata
23348 value is less than "value".
23349 equal
23351 Values are interpreted as floats, returns true if "value" is
23352 equal with metadata value.
23353 greater
23355 Values are interpreted as floats, returns true if metadata
23356 value is greater than "value".
23357 expr
23359 Values are interpreted as floats, returns true if expression
23360 from option "expr" evaluates to true.
23361 ends_with
23363 Values are interpreted as strings, returns true if metadata
23364 value ends with the "value" option string.
23365 expr
23367 Set expression which is used when "function" is set to "expr". The
23368 expression is evaluated through the eval API and can contain the
23369 following constants:
23370 VALUE1, FRAMEVAL
23372 Float representation of "value" from metadata key.
23373 VALUE2, USERVAL
23375 Float representation of "value" as supplied by user in "value"
23376 option.
23377 file
23379 If specified in "print" mode, output is written to the named file.
23380 Instead of plain filename any writable url can be specified.
23381 Filename ``-'' is a shorthand for standard output. If "file" option
23382 is not set, output is written to the log with AV_LOG_INFO loglevel.
23383 direct
23385 Reduces buffering in print mode when output is written to a URL set
23386 using file.
23387 Examples
23389 • Print all metadata values for frames with key
23391 "lavfi.signalstats.YDIF" with values between 0 and 1.
23392 signalstats,metadata=print:key=lavfi.signalstats.YDIF:value=0:function=expr:expr='between(VALUE1,0,1)'
23394 • Print silencedetect output to file metadata.txt.
23396 silencedetect,ametadata=mode=print:file=metadata.txt
23398 • Direct all metadata to a pipe with file descriptor 4.
23400 metadata=mode=print:file='pipe\:4'
23402 perms, aperms
23404 Set read/write permissions for the output frames.
23405 These filters are mainly aimed at developers to test direct path in the
23407 following filter in the filtergraph.
23408 The filters accept the following options:
23410 mode
23412 Select the permissions mode.
23413 It accepts the following values:
23415 none
23417 Do nothing. This is the default.
23418 ro Set all the output frames read-only.
23420 rw Set all the output frames directly writable.
23422 toggle
23424 Make the frame read-only if writable, and writable if read-
23425 only.
23426 random
23428 Set each output frame read-only or writable randomly.
23429 seed
23431 Set the seed for the random mode, must be an integer included
23432 between 0 and "UINT32_MAX". If not specified, or if explicitly set
23433 to "-1", the filter will try to use a good random seed on a best
23434 effort basis.
23435 Note: in case of auto-inserted filter between the permission filter and
23437 the following one, the permission might not be received as expected in
23438 that following filter. Inserting a format or aformat filter before the
23439 perms/aperms filter can avoid this problem.
23440 realtime, arealtime
23442 Slow down filtering to match real time approximately.
23443 These filters will pause the filtering for a variable amount of time to
23445 match the output rate with the input timestamps. They are similar to
23446 the re option to "ffmpeg".
23447 They accept the following options:
23449 limit
23451 Time limit for the pauses. Any pause longer than that will be
23452 considered a timestamp discontinuity and reset the timer. Default
23453 is 2 seconds.
23454 speed
23456 Speed factor for processing. The value must be a float larger than
23457 zero. Values larger than 1.0 will result in faster than realtime
23458 processing, smaller will slow processing down. The limit is
23459 automatically adapted accordingly. Default is 1.0.
23460 A processing speed faster than what is possible without these
23462 filters cannot be achieved.
23463 segment, asegment
23465 Split single input stream into multiple streams.
23466 This filter does opposite of concat filters.
23468 "segment" works on video frames, "asegment" on audio samples.
23470 This filter accepts the following options:
23472 timestamps
23474 Timestamps of output segments separated by '|'. The first segment
23475 will run from the beginning of the input stream. The last segment
23476 will run until the end of the input stream
23477 frames, samples
23479 Exact frame/sample count to split the segments.
23480 In all cases, prefixing an each segment with '+' will make it relative
23482 to the previous segment.
23483 Examples
23485 • Split input audio stream into three output audio streams, starting
23487 at start of input audio stream and storing that in 1st output audio
23488 stream, then following at 60th second and storing than in 2nd
23489 output audio stream, and last after 150th second of input audio
23490 stream store in 3rd output audio stream:
23491 asegment=timestamps="60|150"
23493 select, aselect
23495 Select frames to pass in output.
23496 This filter accepts the following options:
23498 expr, e
23500 Set expression, which is evaluated for each input frame.
23501 If the expression is evaluated to zero, the frame is discarded.
23503 If the evaluation result is negative or NaN, the frame is sent to
23505 the first output; otherwise it is sent to the output with index
23506 "ceil(val)-1", assuming that the input index starts from 0.
23507 For example a value of 1.2 corresponds to the output with index
23509 "ceil(1.2)-1 = 2-1 = 1", that is the second output.
23510 outputs, n
23512 Set the number of outputs. The output to which to send the selected
23513 frame is based on the result of the evaluation. Default value is 1.
23514 The expression can contain the following constants:
23516 n The (sequential) number of the filtered frame, starting from 0.
23518 selected_n
23520 The (sequential) number of the selected frame, starting from 0.
23521 prev_selected_n
23523 The sequential number of the last selected frame. It's NAN if
23524 undefined.
23525 TB The timebase of the input timestamps.
23527 pts The PTS (Presentation TimeStamp) of the filtered frame, expressed
23529 in TB units. It's NAN if undefined.
23530 t The PTS of the filtered frame, expressed in seconds. It's NAN if
23532 undefined.
23533 prev_pts
23535 The PTS of the previously filtered frame. It's NAN if undefined.
23536 prev_selected_pts
23538 The PTS of the last previously filtered frame. It's NAN if
23539 undefined.
23540 prev_selected_t
23542 The PTS of the last previously selected frame, expressed in
23543 seconds. It's NAN if undefined.
23544 start_pts
23546 The first PTS in the stream which is not NAN. It remains NAN if not
23547 found.
23548 start_t
23550 The first PTS, in seconds, in the stream which is not NAN. It
23551 remains NAN if not found.
23552 pict_type (video only)
23554 The type of the filtered frame. It can assume one of the following
23555 values:
23556 I
23558 P
23559 B
23560 S
23561 SI
23562 SP
23563 BI
23564 interlace_type (video only)
23565 The frame interlace type. It can assume one of the following
23566 values:
23567 PROGRESSIVE
23569 The frame is progressive (not interlaced).
23570 TOPFIRST
23572 The frame is top-field-first.
23573 BOTTOMFIRST
23575 The frame is bottom-field-first.
23576 consumed_sample_n (audio only)
23578 the number of selected samples before the current frame
23579 samples_n (audio only)
23581 the number of samples in the current frame
23582 sample_rate (audio only)
23584 the input sample rate
23585 key This is 1 if the filtered frame is a key-frame, 0 otherwise.
23587 pos the position in the file of the filtered frame, -1 if the
23589 information is not available (e.g. for synthetic video)
23590 scene (video only)
23592 value between 0 and 1 to indicate a new scene; a low value reflects
23593 a low probability for the current frame to introduce a new scene,
23594 while a higher value means the current frame is more likely to be
23595 one (see the example below)
23596 concatdec_select
23598 The concat demuxer can select only part of a concat input file by
23599 setting an inpoint and an outpoint, but the output packets may not
23600 be entirely contained in the selected interval. By using this
23601 variable, it is possible to skip frames generated by the concat
23602 demuxer which are not exactly contained in the selected interval.
23603 This works by comparing the frame pts against the
23605 lavf.concat.start_time and the lavf.concat.duration packet metadata
23606 values which are also present in the decoded frames.
23607 The concatdec_select variable is -1 if the frame pts is at least
23609 start_time and either the duration metadata is missing or the frame
23610 pts is less than start_time + duration, 0 otherwise, and NaN if the
23611 start_time metadata is missing.
23612 That basically means that an input frame is selected if its pts is
23614 within the interval set by the concat demuxer.
23615 The default value of the select expression is "1".
23617 Examples
23619 • Select all frames in input:
23621 select
23623 The example above is the same as:
23625 select=1
23627 • Skip all frames:
23629 select=0
23631 • Select only I-frames:
23633 select='eq(pict_type\,I)'
23635 • Select one frame every 100:
23637 select='not(mod(n\,100))'
23639 • Select only frames contained in the 10-20 time interval:
23641 select=between(t\,10\,20)
23643 • Select only I-frames contained in the 10-20 time interval:
23645 select=between(t\,10\,20)*eq(pict_type\,I)
23647 • Select frames with a minimum distance of 10 seconds:
23649 select='isnan(prev_selected_t)+gte(t-prev_selected_t\,10)'
23651 • Use aselect to select only audio frames with samples number > 100:
23653 aselect='gt(samples_n\,100)'
23655 • Create a mosaic of the first scenes:
23657 ffmpeg -i video.avi -vf select='gt(scene\,0.4)',scale=160:120,tile -frames:v 1 preview.png
23659 Comparing scene against a value between 0.3 and 0.5 is generally a
23661 sane choice.
23662 • Send even and odd frames to separate outputs, and compose them:
23664 select=n=2:e='mod(n, 2)+1' [odd][even]; [odd] pad=h=2*ih [tmp]; [tmp][even] overlay=y=h
23666 • Select useful frames from an ffconcat file which is using inpoints
23668 and outpoints but where the source files are not intra frame only.
23669 ffmpeg -copyts -vsync 0 -segment_time_metadata 1 -i input.ffconcat -vf select=concatdec_select -af aselect=concatdec_select output.avi
23671 sendcmd, asendcmd
23673 Send commands to filters in the filtergraph.
23674 These filters read commands to be sent to other filters in the
23676 filtergraph.
23677 "sendcmd" must be inserted between two video filters, "asendcmd" must
23679 be inserted between two audio filters, but apart from that they act the
23680 same way.
23681 The specification of commands can be provided in the filter arguments
23683 with the commands option, or in a file specified by the filename
23684 option.
23685 These filters accept the following options:
23687 commands, c
23689 Set the commands to be read and sent to the other filters.
23690 filename, f
23692 Set the filename of the commands to be read and sent to the other
23693 filters.
23694 Commands syntax
23696 A commands description consists of a sequence of interval
23698 specifications, comprising a list of commands to be executed when a
23699 particular event related to that interval occurs. The occurring event
23700 is typically the current frame time entering or leaving a given time
23701 interval.
23702 An interval is specified by the following syntax:
23704 <START>[-<END>] <COMMANDS>;
23706 The time interval is specified by the START and END times. END is
23708 optional and defaults to the maximum time.
23709 The current frame time is considered within the specified interval if
23711 it is included in the interval [START, END), that is when the time is
23712 greater or equal to START and is lesser than END.
23713 COMMANDS consists of a sequence of one or more command specifications,
23715 separated by ",", relating to that interval. The syntax of a command
23716 specification is given by:
23717 [<FLAGS>] <TARGET> <COMMAND> <ARG>
23719 FLAGS is optional and specifies the type of events relating to the time
23721 interval which enable sending the specified command, and must be a non-
23722 null sequence of identifier flags separated by "+" or "|" and enclosed
23723 between "[" and "]".
23724 The following flags are recognized:
23726 enter
23728 The command is sent when the current frame timestamp enters the
23729 specified interval. In other words, the command is sent when the
23730 previous frame timestamp was not in the given interval, and the
23731 current is.
23732 leave
23734 The command is sent when the current frame timestamp leaves the
23735 specified interval. In other words, the command is sent when the
23736 previous frame timestamp was in the given interval, and the current
23737 is not.
23738 expr
23740 The command ARG is interpreted as expression and result of
23741 expression is passed as ARG.
23742 The expression is evaluated through the eval API and can contain
23744 the following constants:
23745 POS Original position in the file of the frame, or undefined if
23747 undefined for the current frame.
23748 PTS The presentation timestamp in input.
23750 N The count of the input frame for video or audio, starting from
23752 0.
23753 T The time in seconds of the current frame.
23755 TS The start time in seconds of the current command interval.
23757 TE The end time in seconds of the current command interval.
23759 TI The interpolated time of the current command interval, TI = (T
23761 - TS) / (TE - TS).
23762 If FLAGS is not specified, a default value of "[enter]" is assumed.
23764 TARGET specifies the target of the command, usually the name of the
23766 filter class or a specific filter instance name.
23767 COMMAND specifies the name of the command for the target filter.
23769 ARG is optional and specifies the optional list of argument for the
23771 given COMMAND.
23772 Between one interval specification and another, whitespaces, or
23774 sequences of characters starting with "#" until the end of line, are
23775 ignored and can be used to annotate comments.
23776 A simplified BNF description of the commands specification syntax
23778 follows:
23779 <COMMAND_FLAG> ::= "enter" | "leave"
23781 <COMMAND_FLAGS> ::= <COMMAND_FLAG> [(+|"|")<COMMAND_FLAG>]
23782 <COMMAND> ::= ["[" <COMMAND_FLAGS> "]"] <TARGET> <COMMAND> [<ARG>]
23783 <COMMANDS> ::= <COMMAND> [,<COMMANDS>]
23784 <INTERVAL> ::= <START>[-<END>] <COMMANDS>
23785 <INTERVALS> ::= <INTERVAL>[;<INTERVALS>]
23786 Examples
23788 • Specify audio tempo change at second 4:
23790 asendcmd=c='4.0 atempo tempo 1.5',atempo
23792 • Target a specific filter instance:
23794 asendcmd=c='4.0 atempo@my tempo 1.5',atempo@my
23796 • Specify a list of drawtext and hue commands in a file.
23798 # show text in the interval 5-10
23800 5.0-10.0 [enter] drawtext reinit 'fontfile=FreeSerif.ttf:text=hello world',
23801 [leave] drawtext reinit 'fontfile=FreeSerif.ttf:text=';
23802 # desaturate the image in the interval 15-20
23804 15.0-20.0 [enter] hue s 0,
23805 [enter] drawtext reinit 'fontfile=FreeSerif.ttf:text=nocolor',
23806 [leave] hue s 1,
23807 [leave] drawtext reinit 'fontfile=FreeSerif.ttf:text=color';
23808 # apply an exponential saturation fade-out effect, starting from time 25
23810 25 [enter] hue s exp(25-t)
23811 A filtergraph allowing to read and process the above command list
23813 stored in a file test.cmd, can be specified with:
23814 sendcmd=f=test.cmd,drawtext=fontfile=FreeSerif.ttf:text='',hue
23816 setpts, asetpts
23818 Change the PTS (presentation timestamp) of the input frames.
23819 "setpts" works on video frames, "asetpts" on audio frames.
23821 This filter accepts the following options:
23823 expr
23825 The expression which is evaluated for each frame to construct its
23826 timestamp.
23827 The expression is evaluated through the eval API and can contain the
23829 following constants:
23830 FRAME_RATE, FR
23832 frame rate, only defined for constant frame-rate video
23833 PTS The presentation timestamp in input
23835 N The count of the input frame for video or the number of consumed
23837 samples, not including the current frame for audio, starting from
23838 0.
23839 NB_CONSUMED_SAMPLES
23841 The number of consumed samples, not including the current frame
23842 (only audio)
23843 NB_SAMPLES, S
23845 The number of samples in the current frame (only audio)
23846 SAMPLE_RATE, SR
23848 The audio sample rate.
23849 STARTPTS
23851 The PTS of the first frame.
23852 STARTT
23854 the time in seconds of the first frame
23855 INTERLACED
23857 State whether the current frame is interlaced.
23858 T the time in seconds of the current frame
23860 POS original position in the file of the frame, or undefined if
23862 undefined for the current frame
23863 PREV_INPTS
23865 The previous input PTS.
23866 PREV_INT
23868 previous input time in seconds
23869 PREV_OUTPTS
23871 The previous output PTS.
23872 PREV_OUTT
23874 previous output time in seconds
23875 RTCTIME
23877 The wallclock (RTC) time in microseconds. This is deprecated, use
23878 time(0) instead.
23879 RTCSTART
23881 The wallclock (RTC) time at the start of the movie in microseconds.
23882 TB The timebase of the input timestamps.
23884 Examples
23886 • Start counting PTS from zero
23888 setpts=PTS-STARTPTS
23890 • Apply fast motion effect:
23892 setpts=0.5*PTS
23894 • Apply slow motion effect:
23896 setpts=2.0*PTS
23898 • Set fixed rate of 25 frames per second:
23900 setpts=N/(25*TB)
23902 • Set fixed rate 25 fps with some jitter:
23904 setpts='1/(25*TB) * (N + 0.05 * sin(N*2*PI/25))'
23906 • Apply an offset of 10 seconds to the input PTS:
23908 setpts=PTS+10/TB
23910 • Generate timestamps from a "live source" and rebase onto the
23912 current timebase:
23913 setpts='(RTCTIME - RTCSTART) / (TB * 1000000)'
23915 • Generate timestamps by counting samples:
23917 asetpts=N/SR/TB
23919 setrange
23921 Force color range for the output video frame.
23922 The "setrange" filter marks the color range property for the output
23924 frames. It does not change the input frame, but only sets the
23925 corresponding property, which affects how the frame is treated by
23926 following filters.
23927 The filter accepts the following options:
23929 range
23931 Available values are:
23932 auto
23934 Keep the same color range property.
23935 unspecified, unknown
23937 Set the color range as unspecified.
23938 limited, tv, mpeg
23940 Set the color range as limited.
23941 full, pc, jpeg
23943 Set the color range as full.
23944 settb, asettb
23946 Set the timebase to use for the output frames timestamps. It is mainly
23947 useful for testing timebase configuration.
23948 It accepts the following parameters:
23950 expr, tb
23952 The expression which is evaluated into the output timebase.
23953 The value for tb is an arithmetic expression representing a rational.
23955 The expression can contain the constants "AVTB" (the default timebase),
23956 "intb" (the input timebase) and "sr" (the sample rate, audio only).
23957 Default value is "intb".
23958 Examples
23960 • Set the timebase to 1/25:
23962 settb=expr=1/25
23964 • Set the timebase to 1/10:
23966 settb=expr=0.1
23968 • Set the timebase to 1001/1000:
23970 settb=1+0.001
23972 • Set the timebase to 2*intb:
23974 settb=2*intb
23976 • Set the default timebase value:
23978 settb=AVTB
23980 showcqt
23982 Convert input audio to a video output representing frequency spectrum
23983 logarithmically using Brown-Puckette constant Q transform algorithm
23984 with direct frequency domain coefficient calculation (but the transform
23985 itself is not really constant Q, instead the Q factor is actually
23986 variable/clamped), with musical tone scale, from E0 to D#10.
23987 The filter accepts the following options:
23989 size, s
23991 Specify the video size for the output. It must be even. For the
23992 syntax of this option, check the "Video size" section in the
23993 ffmpeg-utils manual. Default value is "1920x1080".
23994 fps, rate, r
23996 Set the output frame rate. Default value is 25.
23997 bar_h
23999 Set the bargraph height. It must be even. Default value is "-1"
24000 which computes the bargraph height automatically.
24001 axis_h
24003 Set the axis height. It must be even. Default value is "-1" which
24004 computes the axis height automatically.
24005 sono_h
24007 Set the sonogram height. It must be even. Default value is "-1"
24008 which computes the sonogram height automatically.
24009 fullhd
24011 Set the fullhd resolution. This option is deprecated, use size, s
24012 instead. Default value is 1.
24013 sono_v, volume
24015 Specify the sonogram volume expression. It can contain variables:
24016 bar_v
24018 the bar_v evaluated expression
24019 frequency, freq, f
24021 the frequency where it is evaluated
24022 timeclamp, tc
24024 the value of timeclamp option
24025 and functions:
24027 a_weighting(f)
24029 A-weighting of equal loudness
24030 b_weighting(f)
24032 B-weighting of equal loudness
24033 c_weighting(f)
24035 C-weighting of equal loudness.
24036 Default value is 16.
24038 bar_v, volume2
24040 Specify the bargraph volume expression. It can contain variables:
24041 sono_v
24043 the sono_v evaluated expression
24044 frequency, freq, f
24046 the frequency where it is evaluated
24047 timeclamp, tc
24049 the value of timeclamp option
24050 and functions:
24052 a_weighting(f)
24054 A-weighting of equal loudness
24055 b_weighting(f)
24057 B-weighting of equal loudness
24058 c_weighting(f)
24060 C-weighting of equal loudness.
24061 Default value is "sono_v".
24063 sono_g, gamma
24065 Specify the sonogram gamma. Lower gamma makes the spectrum more
24066 contrast, higher gamma makes the spectrum having more range.
24067 Default value is 3. Acceptable range is "[1, 7]".
24068 bar_g, gamma2
24070 Specify the bargraph gamma. Default value is 1. Acceptable range is
24071 "[1, 7]".
24072 bar_t
24074 Specify the bargraph transparency level. Lower value makes the
24075 bargraph sharper. Default value is 1. Acceptable range is "[0,
24076 1]".
24077 timeclamp, tc
24079 Specify the transform timeclamp. At low frequency, there is trade-
24080 off between accuracy in time domain and frequency domain. If
24081 timeclamp is lower, event in time domain is represented more
24082 accurately (such as fast bass drum), otherwise event in frequency
24083 domain is represented more accurately (such as bass guitar).
24084 Acceptable range is "[0.002, 1]". Default value is 0.17.
24085 attack
24087 Set attack time in seconds. The default is 0 (disabled). Otherwise,
24088 it limits future samples by applying asymmetric windowing in time
24089 domain, useful when low latency is required. Accepted range is "[0,
24090 1]".
24091 basefreq
24093 Specify the transform base frequency. Default value is
24094 20.01523126408007475, which is frequency 50 cents below E0.
24095 Acceptable range is "[10, 100000]".
24096 endfreq
24098 Specify the transform end frequency. Default value is
24099 20495.59681441799654, which is frequency 50 cents above D#10.
24100 Acceptable range is "[10, 100000]".
24101 coeffclamp
24103 This option is deprecated and ignored.
24104 tlength
24106 Specify the transform length in time domain. Use this option to
24107 control accuracy trade-off between time domain and frequency domain
24108 at every frequency sample. It can contain variables:
24109 frequency, freq, f
24111 the frequency where it is evaluated
24112 timeclamp, tc
24114 the value of timeclamp option.
24115 Default value is "384*tc/(384+tc*f)".
24117 count
24119 Specify the transform count for every video frame. Default value is
24120 6. Acceptable range is "[1, 30]".
24121 fcount
24123 Specify the transform count for every single pixel. Default value
24124 is 0, which makes it computed automatically. Acceptable range is
24125 "[0, 10]".
24126 fontfile
24128 Specify font file for use with freetype to draw the axis. If not
24129 specified, use embedded font. Note that drawing with font file or
24130 embedded font is not implemented with custom basefreq and endfreq,
24131 use axisfile option instead.
24132 font
24134 Specify fontconfig pattern. This has lower priority than fontfile.
24135 The ":" in the pattern may be replaced by "|" to avoid unnecessary
24136 escaping.
24137 fontcolor
24139 Specify font color expression. This is arithmetic expression that
24140 should return integer value 0xRRGGBB. It can contain variables:
24141 frequency, freq, f
24143 the frequency where it is evaluated
24144 timeclamp, tc
24146 the value of timeclamp option
24147 and functions:
24149 midi(f)
24151 midi number of frequency f, some midi numbers: E0(16), C1(24),
24152 C2(36), A4(69)
24153 r(x), g(x), b(x)
24155 red, green, and blue value of intensity x.
24156 Default value is "st(0, (midi(f)-59.5)/12); st(1,
24158 if(between(ld(0),0,1), 0.5-0.5*cos(2*PI*ld(0)), 0)); r(1-ld(1)) +
24159 b(ld(1))".
24160 axisfile
24162 Specify image file to draw the axis. This option override fontfile
24163 and fontcolor option.
24164 axis, text
24166 Enable/disable drawing text to the axis. If it is set to 0, drawing
24167 to the axis is disabled, ignoring fontfile and axisfile option.
24168 Default value is 1.
24169 csp Set colorspace. The accepted values are:
24171 unspecified
24173 Unspecified (default)
24174 bt709
24176 BT.709
24177 fcc FCC
24179 bt470bg
24181 BT.470BG or BT.601-6 625
24182 smpte170m
24184 SMPTE-170M or BT.601-6 525
24185 smpte240m
24187 SMPTE-240M
24188 bt2020ncl
24190 BT.2020 with non-constant luminance
24191 cscheme
24193 Set spectrogram color scheme. This is list of floating point values
24194 with format "left_r|left_g|left_b|right_r|right_g|right_b". The
24195 default is "1|0.5|0|0|0.5|1".
24196 Examples
24198 • Playing audio while showing the spectrum:
24200 ffplay -f lavfi 'amovie=a.mp3, asplit [a][out1]; [a] showcqt [out0]'
24202 • Same as above, but with frame rate 30 fps:
24204 ffplay -f lavfi 'amovie=a.mp3, asplit [a][out1]; [a] showcqt=fps=30:count=5 [out0]'
24206 • Playing at 1280x720:
24208 ffplay -f lavfi 'amovie=a.mp3, asplit [a][out1]; [a] showcqt=s=1280x720:count=4 [out0]'
24210 • Disable sonogram display:
24212 sono_h=0
24214 • A1 and its harmonics: A1, A2, (near)E3, A3:
24216 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),
24218 asplit[a][out1]; [a] showcqt [out0]'
24219 • Same as above, but with more accuracy in frequency domain:
24221 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),
24223 asplit[a][out1]; [a] showcqt=timeclamp=0.5 [out0]'
24224 • Custom volume:
24226 bar_v=10:sono_v=bar_v*a_weighting(f)
24228 • Custom gamma, now spectrum is linear to the amplitude.
24230 bar_g=2:sono_g=2
24232 • Custom tlength equation:
24234 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)))'
24236 • Custom fontcolor and fontfile, C-note is colored green, others are
24238 colored blue:
24239 fontcolor='if(mod(floor(midi(f)+0.5),12), 0x0000FF, g(1))':fontfile=myfont.ttf
24241 • Custom font using fontconfig:
24243 font='Courier New,Monospace,mono|bold'
24245 • Custom frequency range with custom axis using image file:
24247 axisfile=myaxis.png:basefreq=40:endfreq=10000
24249 showfreqs
24251 Convert input audio to video output representing the audio power
24252 spectrum. Audio amplitude is on Y-axis while frequency is on X-axis.
24253 The filter accepts the following options:
24255 size, s
24257 Specify size of video. For the syntax of this option, check the
24258 "Video size" section in the ffmpeg-utils manual. Default is
24259 "1024x512".
24260 mode
24262 Set display mode. This set how each frequency bin will be
24263 represented.
24264 It accepts the following values:
24266 line
24268 bar
24269 dot
24270 Default is "bar".
24272 ascale
24274 Set amplitude scale.
24275 It accepts the following values:
24277 lin Linear scale.
24279 sqrt
24281 Square root scale.
24282 cbrt
24284 Cubic root scale.
24285 log Logarithmic scale.
24287 Default is "log".
24289 fscale
24291 Set frequency scale.
24292 It accepts the following values:
24294 lin Linear scale.
24296 log Logarithmic scale.
24298 rlog
24300 Reverse logarithmic scale.
24301 Default is "lin".
24303 win_size
24305 Set window size. Allowed range is from 16 to 65536.
24306 Default is 2048
24308 win_func
24310 Set windowing function.
24311 It accepts the following values:
24313 rect
24315 bartlett
24316 hanning
24317 hamming
24318 blackman
24319 welch
24320 flattop
24321 bharris
24322 bnuttall
24323 bhann
24324 sine
24325 nuttall
24326 lanczos
24327 gauss
24328 tukey
24329 dolph
24330 cauchy
24331 parzen
24332 poisson
24333 bohman
24334 Default is "hanning".
24336 overlap
24338 Set window overlap. In range "[0, 1]". Default is 1, which means
24339 optimal overlap for selected window function will be picked.
24340 averaging
24342 Set time averaging. Setting this to 0 will display current maximal
24343 peaks. Default is 1, which means time averaging is disabled.
24344 colors
24346 Specify list of colors separated by space or by '|' which will be
24347 used to draw channel frequencies. Unrecognized or missing colors
24348 will be replaced by white color.
24349 cmode
24351 Set channel display mode.
24352 It accepts the following values:
24354 combined
24356 separate
24357 Default is "combined".
24359 minamp
24361 Set minimum amplitude used in "log" amplitude scaler.
24362 data
24364 Set data display mode.
24365 It accepts the following values:
24367 magnitude
24369 phase
24370 delay
24371 Default is "magnitude".
24373 showspatial
24375 Convert stereo input audio to a video output, representing the spatial
24376 relationship between two channels.
24377 The filter accepts the following options:
24379 size, s
24381 Specify the video size for the output. For the syntax of this
24382 option, check the "Video size" section in the ffmpeg-utils manual.
24383 Default value is "512x512".
24384 win_size
24386 Set window size. Allowed range is from 1024 to 65536. Default size
24387 is 4096.
24388 win_func
24390 Set window function.
24391 It accepts the following values:
24393 rect
24395 bartlett
24396 hann
24397 hanning
24398 hamming
24399 blackman
24400 welch
24401 flattop
24402 bharris
24403 bnuttall
24404 bhann
24405 sine
24406 nuttall
24407 lanczos
24408 gauss
24409 tukey
24410 dolph
24411 cauchy
24412 parzen
24413 poisson
24414 bohman
24415 Default value is "hann".
24417 overlap
24419 Set ratio of overlap window. Default value is 0.5. When value is 1
24420 overlap is set to recommended size for specific window function
24421 currently used.
24422 showspectrum
24424 Convert input audio to a video output, representing the audio frequency
24425 spectrum.
24426 The filter accepts the following options:
24428 size, s
24430 Specify the video size for the output. For the syntax of this
24431 option, check the "Video size" section in the ffmpeg-utils manual.
24432 Default value is "640x512".
24433 slide
24435 Specify how the spectrum should slide along the window.
24436 It accepts the following values:
24438 replace
24440 the samples start again on the left when they reach the right
24441 scroll
24443 the samples scroll from right to left
24444 fullframe
24446 frames are only produced when the samples reach the right
24447 rscroll
24449 the samples scroll from left to right
24450 lreplace
24452 the samples start again on the right when they reach the left
24453 Default value is "replace".
24455 mode
24457 Specify display mode.
24458 It accepts the following values:
24460 combined
24462 all channels are displayed in the same row
24463 separate
24465 all channels are displayed in separate rows
24466 Default value is combined.
24468 color
24470 Specify display color mode.
24471 It accepts the following values:
24473 channel
24475 each channel is displayed in a separate color
24476 intensity
24478 each channel is displayed using the same color scheme
24479 rainbow
24481 each channel is displayed using the rainbow color scheme
24482 moreland
24484 each channel is displayed using the moreland color scheme
24485 nebulae
24487 each channel is displayed using the nebulae color scheme
24488 fire
24490 each channel is displayed using the fire color scheme
24491 fiery
24493 each channel is displayed using the fiery color scheme
24494 fruit
24496 each channel is displayed using the fruit color scheme
24497 cool
24499 each channel is displayed using the cool color scheme
24500 magma
24502 each channel is displayed using the magma color scheme
24503 green
24505 each channel is displayed using the green color scheme
24506 viridis
24508 each channel is displayed using the viridis color scheme
24509 plasma
24511 each channel is displayed using the plasma color scheme
24512 cividis
24514 each channel is displayed using the cividis color scheme
24515 terrain
24517 each channel is displayed using the terrain color scheme
24518 Default value is channel.
24520 scale
24522 Specify scale used for calculating intensity color values.
24523 It accepts the following values:
24525 lin linear
24527 sqrt
24529 square root, default
24530 cbrt
24532 cubic root
24533 log logarithmic
24535 4thrt
24537 4th root
24538 5thrt
24540 5th root
24541 Default value is sqrt.
24543 fscale
24545 Specify frequency scale.
24546 It accepts the following values:
24548 lin linear
24550 log logarithmic
24552 Default value is lin.
24554 saturation
24556 Set saturation modifier for displayed colors. Negative values
24557 provide alternative color scheme. 0 is no saturation at all.
24558 Saturation must be in [-10.0, 10.0] range. Default value is 1.
24559 win_func
24561 Set window function.
24562 It accepts the following values:
24564 rect
24566 bartlett
24567 hann
24568 hanning
24569 hamming
24570 blackman
24571 welch
24572 flattop
24573 bharris
24574 bnuttall
24575 bhann
24576 sine
24577 nuttall
24578 lanczos
24579 gauss
24580 tukey
24581 dolph
24582 cauchy
24583 parzen
24584 poisson
24585 bohman
24586 Default value is "hann".
24588 orientation
24590 Set orientation of time vs frequency axis. Can be "vertical" or
24591 "horizontal". Default is "vertical".
24592 overlap
24594 Set ratio of overlap window. Default value is 0. When value is 1
24595 overlap is set to recommended size for specific window function
24596 currently used.
24597 gain
24599 Set scale gain for calculating intensity color values. Default
24600 value is 1.
24601 data
24603 Set which data to display. Can be "magnitude", default or "phase",
24604 or unwrapped phase: "uphase".
24605 rotation
24607 Set color rotation, must be in [-1.0, 1.0] range. Default value is
24608 0.
24609 start
24611 Set start frequency from which to display spectrogram. Default is
24612 0.
24613 stop
24615 Set stop frequency to which to display spectrogram. Default is 0.
24616 fps Set upper frame rate limit. Default is "auto", unlimited.
24618 legend
24620 Draw time and frequency axes and legends. Default is disabled.
24621 drange
24623 Set dynamic range used to calculate intensity color values. Default
24624 is 120 dBFS. Allowed range is from 10 to 200.
24625 limit
24627 Set upper limit of input audio samples volume in dBFS. Default is 0
24628 dBFS. Allowed range is from -100 to 100.
24629 The usage is very similar to the showwaves filter; see the examples in
24631 that section.
24632 Examples
24634 • Large window with logarithmic color scaling:
24636 showspectrum=s=1280x480:scale=log
24638 • Complete example for a colored and sliding spectrum per channel
24640 using ffplay:
24641 ffplay -f lavfi 'amovie=input.mp3, asplit [a][out1];
24643 [a] showspectrum=mode=separate:color=intensity:slide=1:scale=cbrt [out0]'
24644 showspectrumpic
24646 Convert input audio to a single video frame, representing the audio
24647 frequency spectrum.
24648 The filter accepts the following options:
24650 size, s
24652 Specify the video size for the output. For the syntax of this
24653 option, check the "Video size" section in the ffmpeg-utils manual.
24654 Default value is "4096x2048".
24655 mode
24657 Specify display mode.
24658 It accepts the following values:
24660 combined
24662 all channels are displayed in the same row
24663 separate
24665 all channels are displayed in separate rows
24666 Default value is combined.
24668 color
24670 Specify display color mode.
24671 It accepts the following values:
24673 channel
24675 each channel is displayed in a separate color
24676 intensity
24678 each channel is displayed using the same color scheme
24679 rainbow
24681 each channel is displayed using the rainbow color scheme
24682 moreland
24684 each channel is displayed using the moreland color scheme
24685 nebulae
24687 each channel is displayed using the nebulae color scheme
24688 fire
24690 each channel is displayed using the fire color scheme
24691 fiery
24693 each channel is displayed using the fiery color scheme
24694 fruit
24696 each channel is displayed using the fruit color scheme
24697 cool
24699 each channel is displayed using the cool color scheme
24700 magma
24702 each channel is displayed using the magma color scheme
24703 green
24705 each channel is displayed using the green color scheme
24706 viridis
24708 each channel is displayed using the viridis color scheme
24709 plasma
24711 each channel is displayed using the plasma color scheme
24712 cividis
24714 each channel is displayed using the cividis color scheme
24715 terrain
24717 each channel is displayed using the terrain color scheme
24718 Default value is intensity.
24720 scale
24722 Specify scale used for calculating intensity color values.
24723 It accepts the following values:
24725 lin linear
24727 sqrt
24729 square root, default
24730 cbrt
24732 cubic root
24733 log logarithmic
24735 4thrt
24737 4th root
24738 5thrt
24740 5th root
24741 Default value is log.
24743 fscale
24745 Specify frequency scale.
24746 It accepts the following values:
24748 lin linear
24750 log logarithmic
24752 Default value is lin.
24754 saturation
24756 Set saturation modifier for displayed colors. Negative values
24757 provide alternative color scheme. 0 is no saturation at all.
24758 Saturation must be in [-10.0, 10.0] range. Default value is 1.
24759 win_func
24761 Set window function.
24762 It accepts the following values:
24764 rect
24766 bartlett
24767 hann
24768 hanning
24769 hamming
24770 blackman
24771 welch
24772 flattop
24773 bharris
24774 bnuttall
24775 bhann
24776 sine
24777 nuttall
24778 lanczos
24779 gauss
24780 tukey
24781 dolph
24782 cauchy
24783 parzen
24784 poisson
24785 bohman
24786 Default value is "hann".
24788 orientation
24790 Set orientation of time vs frequency axis. Can be "vertical" or
24791 "horizontal". Default is "vertical".
24792 gain
24794 Set scale gain for calculating intensity color values. Default
24795 value is 1.
24796 legend
24798 Draw time and frequency axes and legends. Default is enabled.
24799 rotation
24801 Set color rotation, must be in [-1.0, 1.0] range. Default value is
24802 0.
24803 start
24805 Set start frequency from which to display spectrogram. Default is
24806 0.
24807 stop
24809 Set stop frequency to which to display spectrogram. Default is 0.
24810 drange
24812 Set dynamic range used to calculate intensity color values. Default
24813 is 120 dBFS. Allowed range is from 10 to 200.
24814 limit
24816 Set upper limit of input audio samples volume in dBFS. Default is 0
24817 dBFS. Allowed range is from -100 to 100.
24818 Examples
24820 • Extract an audio spectrogram of a whole audio track in a 1024x1024
24822 picture using ffmpeg:
24823 ffmpeg -i audio.flac -lavfi showspectrumpic=s=1024x1024 spectrogram.png
24825 showvolume
24827 Convert input audio volume to a video output.
24828 The filter accepts the following options:
24830 rate, r
24832 Set video rate.
24833 b Set border width, allowed range is [0, 5]. Default is 1.
24835 w Set channel width, allowed range is [80, 8192]. Default is 400.
24837 h Set channel height, allowed range is [1, 900]. Default is 20.
24839 f Set fade, allowed range is [0, 1]. Default is 0.95.
24841 c Set volume color expression.
24843 The expression can use the following variables:
24845 VOLUME
24847 Current max volume of channel in dB.
24848 PEAK
24850 Current peak.
24851 CHANNEL
24853 Current channel number, starting from 0.
24854 t If set, displays channel names. Default is enabled.
24856 v If set, displays volume values. Default is enabled.
24858 o Set orientation, can be horizontal: "h" or vertical: "v", default
24860 is "h".
24861 s Set step size, allowed range is [0, 5]. Default is 0, which means
24863 step is disabled.
24864 p Set background opacity, allowed range is [0, 1]. Default is 0.
24866 m Set metering mode, can be peak: "p" or rms: "r", default is "p".
24868 ds Set display scale, can be linear: "lin" or log: "log", default is
24870 "lin".
24871 dm In second. If set to > 0., display a line for the max level in the
24873 previous seconds. default is disabled: 0.
24874 dmc The color of the max line. Use when "dm" option is set to > 0.
24876 default is: "orange"
24877 showwaves
24879 Convert input audio to a video output, representing the samples waves.
24880 The filter accepts the following options:
24882 size, s
24884 Specify the video size for the output. For the syntax of this
24885 option, check the "Video size" section in the ffmpeg-utils manual.
24886 Default value is "600x240".
24887 mode
24889 Set display mode.
24890 Available values are:
24892 point
24894 Draw a point for each sample.
24895 line
24897 Draw a vertical line for each sample.
24898 p2p Draw a point for each sample and a line between them.
24900 cline
24902 Draw a centered vertical line for each sample.
24903 Default value is "point".
24905 n Set the number of samples which are printed on the same column. A
24907 larger value will decrease the frame rate. Must be a positive
24908 integer. This option can be set only if the value for rate is not
24909 explicitly specified.
24910 rate, r
24912 Set the (approximate) output frame rate. This is done by setting
24913 the option n. Default value is "25".
24914 split_channels
24916 Set if channels should be drawn separately or overlap. Default
24917 value is 0.
24918 colors
24920 Set colors separated by '|' which are going to be used for drawing
24921 of each channel.
24922 scale
24924 Set amplitude scale.
24925 Available values are:
24927 lin Linear.
24929 log Logarithmic.
24931 sqrt
24933 Square root.
24934 cbrt
24936 Cubic root.
24937 Default is linear.
24939 draw
24941 Set the draw mode. This is mostly useful to set for high n.
24942 Available values are:
24944 scale
24946 Scale pixel values for each drawn sample.
24947 full
24949 Draw every sample directly.
24950 Default value is "scale".
24952 Examples
24954 • Output the input file audio and the corresponding video
24956 representation at the same time:
24957 amovie=a.mp3,asplit[out0],showwaves[out1]
24959 • Create a synthetic signal and show it with showwaves, forcing a
24961 frame rate of 30 frames per second:
24962 aevalsrc=sin(1*2*PI*t)*sin(880*2*PI*t):cos(2*PI*200*t),asplit[out0],showwaves=r=30[out1]
24964 showwavespic
24966 Convert input audio to a single video frame, representing the samples
24967 waves.
24968 The filter accepts the following options:
24970 size, s
24972 Specify the video size for the output. For the syntax of this
24973 option, check the "Video size" section in the ffmpeg-utils manual.
24974 Default value is "600x240".
24975 split_channels
24977 Set if channels should be drawn separately or overlap. Default
24978 value is 0.
24979 colors
24981 Set colors separated by '|' which are going to be used for drawing
24982 of each channel.
24983 scale
24985 Set amplitude scale.
24986 Available values are:
24988 lin Linear.
24990 log Logarithmic.
24992 sqrt
24994 Square root.
24995 cbrt
24997 Cubic root.
24998 Default is linear.
25000 draw
25002 Set the draw mode.
25003 Available values are:
25005 scale
25007 Scale pixel values for each drawn sample.
25008 full
25010 Draw every sample directly.
25011 Default value is "scale".
25013 filter
25015 Set the filter mode.
25016 Available values are:
25018 average
25020 Use average samples values for each drawn sample.
25021 peak
25023 Use peak samples values for each drawn sample.
25024 Default value is "average".
25026 Examples
25028 • Extract a channel split representation of the wave form of a whole
25030 audio track in a 1024x800 picture using ffmpeg:
25031 ffmpeg -i audio.flac -lavfi showwavespic=split_channels=1:s=1024x800 waveform.png
25033 sidedata, asidedata
25035 Delete frame side data, or select frames based on it.
25036 This filter accepts the following options:
25038 mode
25040 Set mode of operation of the filter.
25041 Can be one of the following:
25043 select
25045 Select every frame with side data of "type".
25046 delete
25048 Delete side data of "type". If "type" is not set, delete all
25049 side data in the frame.
25050 type
25052 Set side data type used with all modes. Must be set for "select"
25053 mode. For the list of frame side data types, refer to the
25054 "AVFrameSideDataType" enum in libavutil/frame.h. For example, to
25055 choose "AV_FRAME_DATA_PANSCAN" side data, you must specify
25056 "PANSCAN".
25057 spectrumsynth
25059 Synthesize audio from 2 input video spectrums, first input stream
25060 represents magnitude across time and second represents phase across
25061 time. The filter will transform from frequency domain as displayed in
25062 videos back to time domain as presented in audio output.
25063 This filter is primarily created for reversing processed showspectrum
25065 filter outputs, but can synthesize sound from other spectrograms too.
25066 But in such case results are going to be poor if the phase data is not
25067 available, because in such cases phase data need to be recreated,
25068 usually it's just recreated from random noise. For best results use
25069 gray only output ("channel" color mode in showspectrum filter) and
25070 "log" scale for magnitude video and "lin" scale for phase video. To
25071 produce phase, for 2nd video, use "data" option. Inputs videos should
25072 generally use "fullframe" slide mode as that saves resources needed for
25073 decoding video.
25074 The filter accepts the following options:
25076 sample_rate
25078 Specify sample rate of output audio, the sample rate of audio from
25079 which spectrum was generated may differ.
25080 channels
25082 Set number of channels represented in input video spectrums.
25083 scale
25085 Set scale which was used when generating magnitude input spectrum.
25086 Can be "lin" or "log". Default is "log".
25087 slide
25089 Set slide which was used when generating inputs spectrums. Can be
25090 "replace", "scroll", "fullframe" or "rscroll". Default is
25091 "fullframe".
25092 win_func
25094 Set window function used for resynthesis.
25095 overlap
25097 Set window overlap. In range "[0, 1]". Default is 1, which means
25098 optimal overlap for selected window function will be picked.
25099 orientation
25101 Set orientation of input videos. Can be "vertical" or "horizontal".
25102 Default is "vertical".
25103 Examples
25105 • First create magnitude and phase videos from audio, assuming audio
25107 is stereo with 44100 sample rate, then resynthesize videos back to
25108 audio with spectrumsynth:
25109 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
25111 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
25112 ffmpeg -i magnitude.nut -i phase.nut -lavfi spectrumsynth=channels=2:sample_rate=44100:win_func=hann:overlap=0.875:slide=fullframe output.flac
25113 split, asplit
25115 Split input into several identical outputs.
25116 "asplit" works with audio input, "split" with video.
25118 The filter accepts a single parameter which specifies the number of
25120 outputs. If unspecified, it defaults to 2.
25121 Examples
25123 • Create two separate outputs from the same input:
25125 [in] split [out0][out1]
25127 • To create 3 or more outputs, you need to specify the number of
25129 outputs, like in:
25130 [in] asplit=3 [out0][out1][out2]
25132 • Create two separate outputs from the same input, one cropped and
25134 one padded:
25135 [in] split [splitout1][splitout2];
25137 [splitout1] crop=100:100:0:0 [cropout];
25138 [splitout2] pad=200:200:100:100 [padout];
25139 • Create 5 copies of the input audio with ffmpeg:
25141 ffmpeg -i INPUT -filter_complex asplit=5 OUTPUT
25143 zmq, azmq
25145 Receive commands sent through a libzmq client, and forward them to
25146 filters in the filtergraph.
25147 "zmq" and "azmq" work as a pass-through filters. "zmq" must be inserted
25149 between two video filters, "azmq" between two audio filters. Both are
25150 capable to send messages to any filter type.
25151 To enable these filters you need to install the libzmq library and
25153 headers and configure FFmpeg with "--enable-libzmq".
25154 For more information about libzmq see: <http://www.zeromq.org/>
25156 The "zmq" and "azmq" filters work as a libzmq server, which receives
25158 messages sent through a network interface defined by the bind_address
25159 (or the abbreviation "b") option. Default value of this option is
25160 tcp://localhost:5555. You may want to alter this value to your needs,
25161 but do not forget to escape any ':' signs (see filtergraph escaping).
25162 The received message must be in the form:
25164 <TARGET> <COMMAND> [<ARG>]
25166 TARGET specifies the target of the command, usually the name of the
25168 filter class or a specific filter instance name. The default filter
25169 instance name uses the pattern Parsed_<filter_name>_<index>, but you
25170 can override this by using the filter_name@id syntax (see Filtergraph
25171 syntax).
25172 COMMAND specifies the name of the command for the target filter.
25174 ARG is optional and specifies the optional argument list for the given
25176 COMMAND.
25177 Upon reception, the message is processed and the corresponding command
25179 is injected into the filtergraph. Depending on the result, the filter
25180 will send a reply to the client, adopting the format:
25181 <ERROR_CODE> <ERROR_REASON>
25183 <MESSAGE>
25184 MESSAGE is optional.
25186 Examples
25188 Look at tools/zmqsend for an example of a zmq client which can be used
25190 to send commands processed by these filters.
25191 Consider the following filtergraph generated by ffplay. In this
25193 example the last overlay filter has an instance name. All other filters
25194 will have default instance names.
25195 ffplay -dumpgraph 1 -f lavfi "
25197 color=s=100x100:c=red [l];
25198 color=s=100x100:c=blue [r];
25199 nullsrc=s=200x100, zmq [bg];
25200 [bg][l] overlay [bg+l];
25201 [bg+l][r] overlay@my=x=100 "
25202 To change the color of the left side of the video, the following
25204 command can be used:
25205 echo Parsed_color_0 c yellow | tools/zmqsend
25207 To change the right side:
25209 echo Parsed_color_1 c pink | tools/zmqsend
25211 To change the position of the right side:
25213 echo overlay@my x 150 | tools/zmqsend
25215
25217 Below is a description of the currently available multimedia sources.
25218 amovie
25220 This is the same as movie source, except it selects an audio stream by
25221 default.
25222 movie
25224 Read audio and/or video stream(s) from a movie container.
25225 It accepts the following parameters:
25227 filename
25229 The name of the resource to read (not necessarily a file; it can
25230 also be a device or a stream accessed through some protocol).
25231 format_name, f
25233 Specifies the format assumed for the movie to read, and can be
25234 either the name of a container or an input device. If not
25235 specified, the format is guessed from movie_name or by probing.
25236 seek_point, sp
25238 Specifies the seek point in seconds. The frames will be output
25239 starting from this seek point. The parameter is evaluated with
25240 "av_strtod", so the numerical value may be suffixed by an IS
25241 postfix. The default value is "0".
25242 streams, s
25244 Specifies the streams to read. Several streams can be specified,
25245 separated by "+". The source will then have as many outputs, in the
25246 same order. The syntax is explained in the "Stream specifiers"
25247 section in the ffmpeg manual. Two special names, "dv" and "da"
25248 specify respectively the default (best suited) video and audio
25249 stream. Default is "dv", or "da" if the filter is called as
25250 "amovie".
25251 stream_index, si
25253 Specifies the index of the video stream to read. If the value is
25254 -1, the most suitable video stream will be automatically selected.
25255 The default value is "-1". Deprecated. If the filter is called
25256 "amovie", it will select audio instead of video.
25257 loop
25259 Specifies how many times to read the stream in sequence. If the
25260 value is 0, the stream will be looped infinitely. Default value is
25261 "1".
25262 Note that when the movie is looped the source timestamps are not
25264 changed, so it will generate non monotonically increasing
25265 timestamps.
25266 discontinuity
25268 Specifies the time difference between frames above which the point
25269 is considered a timestamp discontinuity which is removed by
25270 adjusting the later timestamps.
25271 dec_threads
25273 Specifies the number of threads for decoding
25274 format_opts
25276 Specify format options for the opened file. Format options can be
25277 specified as a list of key=value pairs separated by ':'. The
25278 following example shows how to add protocol_whitelist and
25279 protocol_blacklist options:
25280 ffplay -f lavfi
25282 "movie=filename='1.sdp':format_opts='protocol_whitelist=file,rtp,udp\:protocol_blacklist=http'"
25283 It allows overlaying a second video on top of the main input of a
25285 filtergraph, as shown in this graph:
25286 input -----------> deltapts0 --> overlay --> output
25288 ^
25289 |
25290 movie --> scale--> deltapts1 -------+
25291 Examples
25293 • Skip 3.2 seconds from the start of the AVI file in.avi, and overlay
25295 it on top of the input labelled "in":
25296 movie=in.avi:seek_point=3.2, scale=180:-1, setpts=PTS-STARTPTS [over];
25298 [in] setpts=PTS-STARTPTS [main];
25299 [main][over] overlay=16:16 [out]
25300 • Read from a video4linux2 device, and overlay it on top of the input
25302 labelled "in":
25303 movie=/dev/video0:f=video4linux2, scale=180:-1, setpts=PTS-STARTPTS [over];
25305 [in] setpts=PTS-STARTPTS [main];
25306 [main][over] overlay=16:16 [out]
25307 • Read the first video stream and the audio stream with id 0x81 from
25309 dvd.vob; the video is connected to the pad named "video" and the
25310 audio is connected to the pad named "audio":
25311 movie=dvd.vob:s=v:0+#0x81 [video] [audio]
25313 Commands
25315 Both movie and amovie support the following commands:
25317 seek
25319 Perform seek using "av_seek_frame". The syntax is: seek
25320 stream_index|timestamp|flags
25321 • stream_index: If stream_index is -1, a default stream is
25323 selected, and timestamp is automatically converted from
25324 AV_TIME_BASE units to the stream specific time_base.
25325 • timestamp: Timestamp in AVStream.time_base units or, if no
25327 stream is specified, in AV_TIME_BASE units.
25328 • flags: Flags which select direction and seeking mode.
25330 get_duration
25332 Get movie duration in AV_TIME_BASE units.
25333
25335 ffmpeg(1), ffplay(1), ffprobe(1), libavfilter(3)
25336
25338 The FFmpeg developers.
25339 For details about the authorship, see the Git history of the project
25341 (git://source.ffmpeg.org/ffmpeg), e.g. by typing the command git log in
25342 the FFmpeg source directory, or browsing the online repository at
25343 <http://source.ffmpeg.org>.
25344 Maintainers for the specific components are listed in the file
25346 MAINTAINERS in the source code tree.
25347 FFMPEG-FILTERS(1)