1Ppmtompeg User Manual(0) Ppmtompeg User Manual(0)
2
6 ppmtompeg - encode an MPEG-1 bitstream
7
10 ppmtompeg [options] parameter-file
11
14 This program is part of Netpbm(1).
15 ppmtompeg produces an MPEG-1 video stream. MPEG-1 is the first great
17 video compression method, and is what is used in Video CDs (VCD). ppm‐
18 tompeg originated in the year 1995. DVD uses a more advanced method,
19 MPEG-2. There is an even newer method called MPEG-4 which is also
20 called Divx. I don't know where one finds that used.
21 There's technically a difference between a compression method for video
23 and an actual file (stream) format for a movie, and I don't know if it
24 can be validly said that the format of the stream ppmtompeg produces is
25 MPEG-1.
26 Mencoder from the Mplayer package ⟨http://www.mplayerhq.hu⟩ is proba‐
28 bly superior for most video format generation needs, if for no other
29 reason than that it is more popular.
30 The programming library PM2V ⟨http://pm2v.free.fr⟩ generates MPEG-2
32 streams.
33 Use Mplayer ⟨http://www.mplayerhq.hu⟩ (not part of Netpbm) to do the
35 reverse conversion: to create a series of PNM files from an MPEG
36 stream.
37 param_file is a parameter file which includes a list of input files and
39 other parameters. The file is described in detail below.
40 To understand this program, you need to understand something about the
42 complex MPEG-1 format. One source of information about this standard
43 format is the section Introduction to MPEG in the Compression FAQ
44 ⟨http://www.faqs.org/faqs/compression-faq/⟩ .
45
48 The -gop, -combine_gops, -frames, and -combine_frames options are all
49 mutually exclusive.
50 -stat stat_file
54 This option causes ppmtompeg to append the statistics that it
55 write to Standard Output to the file stat_file as well. The
56 statistics use the following abbreviations: bits per block
57 (bpb), bits per frame (bpf), seconds per frame (spf), and bits
58 per second (bps).
59 These statistics include how many I, P, and B frames there were,
61 and information about compression and quality.
62 -quiet num_seconds
66 causes ppmtompeg not to report remaining time more often than
67 every num_seconds seconds (unless the time estimate rises, which
68 will happen near the beginning of the run). A negative value
69 tells ppmtompeg not to report at all. 0 is the default (reports
70 once after each frame). Note that the time remaining is an es‐
71 timate and does not take into account time to read in frames.
72 -realquiet
75 causes ppmtompeg to run silently, with the only screen output
76 being errors. Particularly useful when reading input from
77 stdin. The equivalent of the -quiet common option of most other
78 Netpbm programs.
79 -no_frame_summary
83 This option prevents ppmtompeg from printing a summary line for
84 each frame
85 -float_dct
88 forces ppmtompeg to use a more accurate, yet more computation‐
89 ally expensive version of the DCT.
90 -gop gop_num
93 causes ppmtompeg to encode only the numbered GOP (first GOP is
94 0). The parameter file is the same as for normal usage. The
95 output file will be the normal output file with the suffix
96 .gop.gop_num. ppmtompeg does not output any sequence informa‐
97 tion.
98 -combine_gops
101 causes ppmtompeg simply to combine some GOP files into a single
102 MPEG output stream. ppmtompeg inserts a sequence header and
103 trailer. In this case, the parameter file needs only to contain
104 the SIZE value, an output file, and perhaps a list of input GOP
105 files (see below).
106 If you don't supply a list of input GOP files is used, then ppm‐
108 tompeg assumes you're using the same parameter file you used
109 when you created the input (with the -gop option) and calculates
110 the corresponding gop filenames itself. If this is not the
111 case, you can specify input GOP files in the same manner as nor‐
112 mal input files -- except instead of using INPUT_DIR, INPUT, and
113 END_INPUT, use GOP_INPUT_DIR, GOP_INPUT, and GOP_END_INPUT. If
114 no input GOP files are specified, then the default is to use the
115 output file name with suffix .gop.gop_num, with gop_num starting
116 from 0, as the input files.
117 Thus, unless you're mixing and matching GOP files from different
119 sources, you can simply use the same parameter file for creating
120 the GOP files (-gop) and for later turning them into an MPEG
121 stream (-combine_gops).
122 -frames first_frame last_frame
126 This option causes ppmtompeg to encode only the frames numbered
127 first_frame to last_frame, inclusive. The parameter file is the
128 same as for normal usage. The output will be placed in separate
129 files, one per frame, with the file names being the normal out‐
130 put file name with the suffix .frame.frame_num. No GOP header
131 information is output. (Thus, the parameter file need not in‐
132 clude the GOP_SIZE value)
133 Use ppmtompeg -combine_frames to combine these frames later into
135 an MPEG stream.
136 -combine_frames
140 This option causes ppmtompeg simply to combine some individual
141 MPEG frames (such as you might have created with an earlier run
142 of ppmtompeg -frames) into a single MPEG stream. Sequence and
143 GOP headers are inserted appropriately. In this case, the pa‐
144 rameter file needs to contain only the SIZE value, the GOP_SIZE
145 value, an output file, and perhaps a list of frame files (see
146 below).
147 The parameter file may specify input frame files in the same
149 manner as normal input files -- except instead of using IN‐
150 PUT_DIR, INPUT, and END_INPUT, use FRAME_INPUT_DIR, FRAME_INPUT,
151 and FRAME_END_INPUT. If no input frame files are specified, then
152 the default is to use the output file name with suffix
153 .frame.frame_num, with frame_num starting from 0, as the input
154 files.
155 -nice This option causes ppmtompeg to run any remote processes
160 "nicely," i.e. at low priority. (This is relevant only if you
161 are running ppmtompeg in parallel mode. Otherwise, there are no
162 remote processes). See 'man nice.'
163 -max_machines num_machines
166 This option causes ppmtompeg to use no more than num_machines
167 machines as slaves for use in parallel encoding.
168 -snr This option causes ppmtompeg to include the signal-to-noise ra‐
171 tio in the reported statistics. Prints SNR (Y U V) and peak SNR
172 (Y U V) for each frame. In summary, prints averages of lumi‐
173 nance only (Y). SNR is defined as 10*log(variance of origi‐
174 nal/variance of error). Peak SNR is defined as
175 20*log(255/RMSE). Note that ppmtompeg runs a little slower when
176 you use this option.
177 -mse This option causes ppmtompeg to report the mean squared error
180 per block. It also automatically reports the quality of the im‐
181 ages, so there is no need to specify -snr then.
182 -bit_rate_info rate_file
185 This option makes ppmtompeg write bit rate information into the
186 file rate_file. Bit rate information is bits per frame, and
187 also bits per I-frame-to-I-frame.
188 -mv_histogram
191 This option causes ppmtompeg to print a histogram of the motion
192 vectors as part of statistics. There are three histograms --
193 one for P frame, one for forward B frame, and one for backward B
194 frame motion vectors.
195 The output is in the form of a matrix, each entry corresponding
197 to one motion vector in the search window. The center of the ma‐
198 trix represents (0,0) motion vectors.
199 -debug_sockets
202 This option causes ppmtompeg to print to Standard Output mes‐
203 sages that narrate the communication between the machines when
204 you run ppmtompeg in parallel mode ⟨#parallel⟩ .
205 -debug_machines
208 This option causes ppmtompeg to print to Standard Output mes‐
209 sages that narrate the progress of the conversion on the various
210 machines when you run ppmtompeg in parallel mode ⟨#parallel⟩ .
211
216 The parameter file must contain the following lines (except when using
217 the -combine_gops or -combine_frames options):
218 PATTERN pattern
223 This statement specifies the pattern (sequence) of I frames, P
224 frames, and B frames. pattern is just a sequence of the letters
225 I, P, and B with nothing between. Example:
226 PATTERN IBBPBBPBBPBBPBB
228 </pre>
229 See
231 I Frames, P Frames, B Frames
232 ⟨#ipb⟩
233 .
234 OUTPUT output file
237 This names the file where the output MPEG stream goes.
238 INPUT_DIR directory
241 This statement tells where the input images (frames) come from.
242 If each frame is in a separate file, directory is the directory
243 where they all are. You may use . to refer to the current
244 directory. A null directory refers to the root directory of the
245 system file tree.
246 To have ppmtompeg read all the frames serially from Standard
248 Input, specify
249 INPUT_DIR stdin
250 INPUT This line must be followed by a list of the input files (in dis‐
254 play order) and then the line END_INPUT.
255 There are three types of lines between INPUT and END_INPUT.
257 First, a line may simply be the name of an input file. Second,
258 the line may be of the form single_star_expr [x-y]. sin‐
259 gle_star_expr can have a single * in it. It is replaced by all
260 the numbers between x and y inclusive. So, for example, the
261 line tennis*.ppm [12-15] refers to the files tennis12.ppm, ten‐
262 nis13.ppm, tennis14.ppm, tennis15.ppm.
263 Uniform zero-padding occurs, as well. For example, the line
265 football.*.ppm [001-130] refers to the files football.001.ppm,
266 football.002.ppm, ..., football.009.ppm, football.010.ppm, ...,
267 football.130.ppm.
268 The third type of line is: single_star_expr [x-y+s], where the
270 line is treated exactly as above, except that we skip by s.
271 Thus, the line football.*.ppm [001-130+4] refers to the files
272 football.001.ppm, football.005.ppm, football.009.ppm, foot‐
273 ball.013.ppm, etc.
274 Furthermore, a line may specify a shell command to execute to
276 generate lines to be interpreted as described above, as if those
277 lines were in the parameter file instead. Use back ticks, like
278 in the Bourne Shell, like this:
279 `cat myfilelist`
281 If input is from Standard Input (per the INPUT_DIR statement),
284 ppmtompeg ignores the INPUT/END_INPUT block, but it still must
285 be present.
286 BASE_FILE_FORMAT {PPM | PNM | YUV |
289 JPEG | JMOVIE} ppmtompeg must convert all input files to
290 one of the following formats as a first step of processing: PNM,
291 YUV, JPEG(v4), or JMOVIE. (The conversion may be trivial if
292 your input files are already in one of these formats). This
293 line specifies which of the four formats. PPM is actually a
294 subset of PNM. The separate specification is allowed for back‐
295 ward compatibility. Use PNM instead of PPM in new applications.
296 INPUT_CONVERT conversion_command
299 You must specify how to convert a file to the base file format.
300 If no conversion is necessary, then you would just say:
301 INPUT_CONVERT *
303 Otherwise, conversion_command is a shell command that causes an
306 image in the format your specified with BASE_FILE_FORMAT to be
307 written to Standard Output. ppmtompeg executes the command once
308 for each line between INPUT and END_INPUT (which is normally,
309 but not necessarily, a file name). In the conversion command,
310 ppmtompeg replaces each '*' with the contents of that line.
311 If you had a bunch of gif files, you might say:
313 INPUT_CONVERT giftopnm *
314 If you have a bunch of separate a.Y, a.U, and a.V files
317 (where
318 the U and V have already been subsampled), then you might
319 say:
320 INPUT_CONVERT cat *.Y *.U *.V
322 Input conversion is not allowed with input from stdin, so use
325 INPUT_CONVERT *
327 as described above.
330 SIZE widthxheight
333 width and height are the width and height of each frame in pix‐
335 els.
336 When ppmtompeg can get this information from the input image
338 files, it ignores the SIZE parameter and you may omit it.
339 When the image files are in YUV format, the files don't contain
341 dimension information, so SIZE is required.
342 When ppmtompeg is running in parallel mode, not all of the pro‐
344 cesses in the network have access to the image files, so SIZE is
345 required and must give the same dimensions as the input image
346 files.
347 YUV_SIZE widthxheight
350 This is an obsolete synonym of SIZE.
351 YUV_FORMAT {ABEKAS | PHILLIPS | UCB |
354 EYUV | pattern} This is meaningful only
355 when BASE_FILE_FORMAT specifies YUV format, and then it is re‐
356 quired. It specifies the sub-format of the YUV class.
357 GOP_SIZE n
361 n is the number of frames in a Group of Pictures. Except that
362 because a GOP must start with an I frame, ppmtompeg makes a GOP
363 as much longer than n as it has to to make the next GOP start
364 with an I frame.
365 Normally, it makes sense to make your GOP size a multiple of
367 your pattern length (the latter is determined by the PATTERN pa‐
368 rameter file statement).
369 See Group Of Pictures ⟨#gop⟩ .
371 SLICES_PER_FRAME n
374 n is roughly the number of slices per frame. Note, at least one
375 MPEG player may complain if slices do not start at the left side
376 of an image. To ensure this does not happen, make sure the num‐
377 ber of rows is divisible by SLICES_PER_FRAME.
378 PIXEL {FULL | HALF}
381 use half-pixel motion vectors, or just full-pixel ones It is
382 usually important that you use half-pixel motion vectors, be‐
383 cause it results in both better quality and better compression.
384 RANGE n
388 Use a search range of n pixels in each of the four directions
389 from a subject pixel. (So the search window is a square n*2
390 pixels on a side).
391 PSEARCH_ALG {EXHAUSTIVE | TWOLEVEL |
394 SUBSAMPLE | LOGARITHMIC} This statement tells ppmtompeg
395 what kind of search
396 technique (algorithm) to use for P frames. You select the
397 desired
398 combination of speed and compression. EXHAUSTIVE gives the
399 best compression, but LOGARITHMIC is the fastest.
400 TWOLEVEL is an exhaustive full-pixel search, followed by a
401 local half- pixel search around the best full-pixel vector
402 (the
403 PIXEL option is ignored for this search technique).
404 BSEARCH_ALG {SIMPLE | CROSS2 | EXHAUSTIVE}
407 This statement tells ppmtompeg what kind of search
408 technique (algorithm) to use for B frames. SIMPLE means
409 find best forward and backward vectors, then interpolate.
410 CROSS2 means find those two vectors, then see what backward
411 vector best matches the best forward vector, and vice versa.
412 EXHAUSTIVE does an n-squared search and is
413 extremely slow in relation to the others (CROSS2
414 is about half as fast as SIMPLE).
415 IQSCALE n
418 Use n as the qscale for I frames.
419 See Qscale ⟨#qscale⟩ .
420 PQSCALE n
423 Use n as the qscale for P frames.
424 See Qscale ⟨#qscale⟩ .
425 BQSCALE n
428 Use n as the qscale for B frames.
429 See Qscale ⟨#qscale⟩ .
430 REFERENCE_FRAME {ORIGINAL | DECODED}
433 This statement determines whether ppmtompeg uses the original
434 images or the decoded images when computing motion vectors. Us‐
435 ing decoded images is more accurate and should increase the
436 playback quality of the output, but it makes the encoding take
437 longer and seems to give worse compression. It also causes some
438 complications with parallel encoding. (see the section on paral‐
439 lel encoding). One thing you can do as a trade-off is select
440 ORIGINAL here, and lower the qscale (see QSCALE if the quality
441 is not good enough.
442 Original or Decoded? (Normalized)
444 ────────────────────────────────────────────────────────────────────
446 Reference Compression Speed Quality I Quality P Quality B
447 Decoded 1000 1000 1000 969 919
448 Original 885 1373 1000 912 884
449 The following lines are optional:
456 FORCE_ENCODE_LAST_FRAME
461 This statement is obsolete. It does nothing.
462 Before Netpbm 10.26 (January 2005), ppmtompeg would drop trail‐
464 ing B frames from your movie, since a movie can't end with a B
465 frame. (See I Frames, P Frames, B Frames ⟨#ipb⟩ .) You would
466 have to specify FORCE_ENCODE_LAST_FRAME to stop that from hap‐
467 pening and get the same function that ppmtompeg has today.
468 NIQTABLE
472 This statement specifies a custom non-intra quantization table.
473 If you don't specify this statement, ppmtompeg uses a default
474 non-intra quantization table.
475 The 8 lines immediately following NIQTABLE specify the quantiza‐
477 tion table. Each line defines a table row and consists of 8 in‐
478 tegers, whitespace-delimited, which define the table columns.
479 IQTABLE
482 This is analogous to NIQTABLE, but for the intra quantization
483 table.
484 ASPECT_RATIO ratio
487 This statement specifies the aspect ratio for ppmtompeg to spec‐
488 ify in the MPEG output. I'm not sure what this is used for.
489 ratio must be 1.0, 0.6735, 0.7031, 0.7615, 0.8055, 0.8437,
491 0.8935, 0.9157, 0.9815, 1.0255, 1.0695, 1.0950, 1.1575, or
492 1.2015.
493 FRAME_RATE rate
496 This specifies the frame rate for ppmtompeg to specify in the
497 MPEG output. Some players use this value to determine the play‐
498 back rate.
499 rate must be 23.976, 24, 25, 29.97, 30, 50, 59.94, or 60.
501 BIT_RATE rate
504 This specifies the bit rate for Constant Bit Rate (CBR) encod‐
505 ing.
506 rate must be an integer.
508 BUFFER_SIZE size
511 This specifies the value ppmtompeg is to specify in the MPEG
512 output for the Video Buffering Verifier (VBV) buffer size needed
513 to decode the sequence.
514 A Video Verifying Buffer is a buffer in which a decoder keeps
516 the decoded bits in order to match the uneven speed of the de‐
517 coding with the required constant playback speed.
518 As ppmtompeg encodes the image, it simulates the decoding
520 process in terms of how many bits would be in the VBV as each
521 frame gets decoded, assuming a VBV of the size you indicate.
522 If you specify the WARN_VBV_UNDERFLOW statement, ppmtompeg is‐
524 sues a warning each time the simulation underflows the buffer,
525 which suggests that an underflow would occur on playback, which
526 suggests the buffer is too small.
527 If you specify the WARN_VBV_OVERFLOW statement, ppmtompeg issues
529 a warning each time the simulation overflows the buffer, which
530 suggests that an overflow would occur on playback, which sug‐
531 gests the buffer is too small.
532 WARN_VBV_UNDERFLOW
535 WARN_VBV_OVERFLOW
537 See BUFFER_SIZE.
538 These options were new in Netpbm 10.26 (January 2005). Before
540 that, ppmtompeg issued the warnings always.
541 The following statements apply only to parallel operation:
546 PARALLEL
551 This statement, paired with END PARALLEL, is what causes ppmtom‐
552 peg to operate in parallel mode. See Parallel Operation
553 ⟨#parallel⟩ .
554 END PARALLEL
557 This goes with PARALLEL.
558 PARALLEL_TEST_FRAMES n
561 The master starts off by measuring each slave's speed. It does
562 this by giving each slave n frames to encode and noting how long
563 the slave takes to finish. These are not just test frames,
564 though -- they're real frames and the results become part of the
565 output. ppmtompeg is old and measures time in undivided sec‐
566 onds, so to get useful timings, specify enough frames that it
567 will take at least 5 seconds to process them. The default is
568 10.
569 If you specify FORCE_I_ALIGN, ppmtompeg will increase the test
571 frames value enough to maintain the alignment.
572 If there aren't enough frames for every slave to have the indi‐
574 cated number of test frames, ppmtompeg will give some slaves
575 fewer.
576 PARALLEL_TIME_CHUNKS t
580 When you specify this statement, the master attempts to feed
581 work to the slaves in chunks that take t seconds to process. It
582 uses the speed measurement it made when it started up (see PAR‐
583 ALLEL_TEST_FRAMES) to decide how many frames to put in the
584 chunk. This statement obviously doesn't affect the first batch
585 of work sent to each slave, which is the one used to measure the
586 slave's speed.
587 Smaller values of t increase communication, but improve load
589 balancing. The default is 30 seconds.
590 You may specify only one of PARALLEL_TIME_CHUNKS, PARAL‐
592 LEL_CHUNK_TAPER, and PARALLEL_PERFECT. PARALLEL_CHUNK_TAPER is
593 usually best.
594 PARALLEL_CHUNK_TAPER
597 When you specify this statement, the master distributes work
598 like with PARALLEL_TIME_CHUNKS, except that the master chooses
599 the number of seconds for the chunks. It starts with a large
600 number and, as it gets closer to finishing the job, reduces it.
601 That way, it reduces scheduling overhead when precise scheduling
602 isn't helpful, but still prevents a slave from finishing early
603 after all the work has already been handed out to the other
604 slaves, and then sitting idle while there's still work to do.
605 You may specify only one of PARALLEL_TIME_CHUNKS, PARAL‐
607 LEL_CHUNK_TAPER, and PARALLEL_PERFECT. PARALLEL_CHUNK_TAPER is
608 usually best.
609 PARALLEL_PERFECT
613 If this statement is present, ppmtompeg schedules on the assump‐
614 tion that each machine is about the same speed. The master will
615 simply divide up the frames evenly between the slaves -- each
616 slave gets the same number of frames. If some slaves are faster
617 than others, they will finish first and remain idle while the
618 slower slaves continue.
619 This has the advantage of minimal scheduling overhead. Where
621 slaves have different speeds, though, it makes inefficient use
622 of the fast ones. Where slaves are the same speed, it also has
623 the disadvantage that they all finish at the same time and feed
624 their output to the single Combine Server in a burst, which
625 makes less efficient use of the Combine Server and thus can in‐
626 crease the total elapsed time.
627 You may specify only one of PARALLEL_TIME_CHUNKS, PARAL‐
629 LEL_CHUNK_TAPER, and PARALLEL_PERFECT. PARALLEL_CHUNK_TAPER is
630 usually best.
631 RSH remote_shell_command
634 ppmtompeg executes the shell command remote_shell_command to
635 start a process on another machine. The default command is rsh,
636 and whatever command you specify must have compatible semantics.
637 ssh is usually compatible. The command ppmtompeg uses is one
638 like this: ssh remote.host.com -l username shellcommand.
639 Be sure to set up .rhosts files or SSH key authorizations where
641 needed. Otherwise, you'll have to type in passwords.
642 On some HP machines, rsh is the restricted shell, and you want
644 to specify remsh.
645 FORCE_I_ALIGN
648 This statement forces each slave to encode a chunk of frames
649 which is a multiple of the pattern length (see PATTERN). Since
650 the first frame in any pattern is an I frame, this forces each
651 chunk encoded by a slave to begin with an I frame.
652 This document used to say there was an argument to FORCE_I_ALIGN
654 which was the number of frames ppmtompeg would use (and was re‐
655 quired to be a multiple of the pattern length). But ppmtompeg
656 has apparently always ignored that argument, and it does now.
657 KEEP_TEMP_FILES
660 This statement causes ppmtompeg not to delete the temporary
661 files it uses to transmit encoded frames to the combine server.
662 This means you will be left with a file for each frame, the same
663 as you would get with the -frames option.
664 This is mostly useful for debugging.
666 This works only if you're using a shared filesystem to communi‐
668 cate between the servers.
669 This option was new in Netpbm 10.26 (January 2005).
671 Parameter File Notes
677 If you use the -combine_gops option, then you need to specify only the
678 SIZE and OUTPUT values in the parameter file. In addition, the parame‐
679 ter file may specify input GOP files in the same manner as normal input
680 files -- except instead of using INPUT_DIR, INPUT, and END_INPUT, use
681 GOP_INPUT_DIR, GOP_INPUT, and GOP_END_INPUT. If you specify no input
682 GOP files, then ppmtompeg uses by default the output file name with
683 suffix .gop.gop_num, with gop_num starting from 0, as the input files.
684 If you use the -combine_frames option, then you need to specify only
686 the SIZE, GOP_SIZE, and OUTPUT values in the parameter file. In addi‐
687 tion, the parameter file may specify input frame files in the same man‐
688 ner as normal input files -- except instead of using INPUT_DIR, INPUT,
689 and END_INPUT, use FRAME_INPUT_DIR, FRAME_INPUT, and FRAME_END_INPUT.
690 If no input frame files are specified, then the default is to use the
691 output file name with suffix .frame.frame_num, with frame_num starting
692 from 0, as the input files.
693 Any number of spaces and tabs may come between each option and value.
695 Lines beginning with # are ignored. Any other lines are ignored except
696 for those between INPUT and END_INPUT. This allows you to use the same
697 parameter file for normal usage and for -combine_gops and -com‐
698 bine_frames.
699 The file format is case-sensitive so all keywords should be in upper
701 case.
702 The statements may appear in any order, except that the order within a
704 block statement (such as INPUT ... END INPUT) is significant.
705 ppmtompeg is prepared to handle up to 16 B frames between reference
707 frames when encoding with input from stdin. (To build a modified ppm‐
708 tompeg with a higher limit, change the constant B_FRAME_RUN in frame.c
709 and recompile).
710
713 Qscale
714 The quantization scale values (qscale) give a trade-off between quality
715 and compression. Using different Qscale values has very little effect
716 on speed. The qscale values can be set separately for I, P, and B
717 frames.
718 You select the qscale values with the IQSCALE, PQSCALE, and BSCALE pa‐
720 rameter file statements.
721 A qscale value is an integer from 1 to 31. Larger numbers give better
723 compression, but worse quality. In the following, the quality numbers
724 are peak signal-to-noise ratio, defined as: signal-to-noise formula
725 where MSE is the mean squared error.
726 Flower garden tests:
729 Qscale vs Quality
731 ────────────────────────────────────────
733 Qscale I Frames P Frames B Frames
734 1 43.2 46.3 46.5
735 6 32.6 34.6 34.3
736 11 28.6 29.5 30.0
737 16 26.3 26.8 28.6
738 21 24.7 25.0 27.9
739 26 23.5 23.9 27.5
740 31 22.6 23.0 27.3
741 Qscale vs Compression
743 ────────────────────────────────────────
745 Qscale I Frames P Frames B Frames
746 1 2 2 2
747 6 7 10 15
748 11 11 18 43
749 16 15 29 97
750 21 19 41 173
751 26 24 56 256
752 31 28 73 330
753 Search Techniques
757 There are several different motion vector search techniques available.
758 There are different techniques available for P frame search and B frame
759 search. Using different search techniques present little difference in
760 quality, but a large difference in compression and speed.
761 There are 4 types of P frame search: Exhaustive, TwoLevel, SubSample,
764 and Logarithmic.
765 There are 3 types of B frame search: Exhaustive, Cross2, and Simple.
768 The recommended search techniques are TwoLevel and Logarithmic for P
770 frame search, and Cross2 and Simple for B frame search. Here are some
771 numbers comparing the different search methods:
772 P frame Motion Vector Search (Normalized)
774 ─────────────────────────────────────────────────────────────────────
776 Technique Compression 1 Speed 2 Quality 3
777 ⟨#smallbetter⟩ ⟨#largefaster⟩ ⟨#largebetter⟩
778 Exhaustive 1000 1000 1000
779 SubSample 1008 2456 1000
780 TwoLevel 1009 3237 1000
781 Logarithmic 1085 8229 998
782 B frame Motion Vector Search (Normalized)
784 ────────────────────────────────────────────────────────────────────
786 Technique Compression 1 Speed 2 Quality 3
787 ⟨#smallbetter⟩ ⟨#largefaster⟩ ⟨#largebetter⟩
788 Exhaustive 1000 1000 1000
789 Cross2 975 1000 996
790 Simple 938 1765 991
791 1Smaller numbers are better compression.
793 2Larger numbers mean faster execution.
795 3Larger numbers mean better quality.
797 For some reason, Simple seems to give better compression, but it de‐
799 pends on the image sequence.
800 Select the search techniques with the PSEARCH_ALG and BSEARCH_ALG pa‐
802 rameter file statements.
803 Group Of Pictures (GOP)
807 A Group of Pictures (GOP) is a roughly independently decodable sequence
808 of frames. An MPEG video stream is made of one or more GOP's. You may
809 specify how many frames should be in each GOP with the GOP_SIZE parame‐
810 ter file statement. A GOP always starts with an I frame.
811 Instead of encoding an entire sequence, you can encode a single GOP.
813 To do this, use the -gop command option. You can later join the re‐
814 sulting GOP files at any time by running ppmtompeg with the -com‐
815 bine_gops command option.
816 Slices
820 A slice is an independently decodable unit in a frame. It can be as
821 small as one macroblock, or it can be as big as the entire frame. Bar‐
822 ring transmission error, adding slices does not change quality or
823 speed; the only effect is slightly worse compression. More slices are
824 used for noisy transmission so that errors are more recoverable. Since
825 usually errors are not such a problem, we usually just use one slice
826 per frame.
827 Control the slice size with the SLICES_PER_FRAME parameter file state‐
830 ment.
831 Some MPEG playback systems require that each slice consist of whole
833 rows of macroblocks. If you are encoding for this kind of player, if
834 the height of the image is H pixels, then you should set the
835 SLICES_PER_FRAME to some number which divides H/16. For example, if
836 the image is 240 pixels (15 macroblocks) high, then you should use only
837 15, 5, 3, or 1 slices per frame.
838 Note: these MPEG playback systems are really wrong, since the MPEG
841 standard says this doesn't have to be so.
842 Search Window
847 The search window is the window in which ppmtompeg searches for motion
848 vectors. The window is a square. You can specify the size of the
849 square, and whether to allow half-pixel motion vectors or not, with the
850 RANGE and PIXEL parameter file statements.
851 I Frames, P Frames, B Frames
854 In MPEG-1, a movie is represented as a sequence of MPEG frames, each of
855 which is an I Frame, a P Frame, or a B Frame. Each represents an ac‐
856 tual frame of the movie (don't get confused by the dual use of the word
857 "frame." A movie frame is a graphical image. An MPEG frame is a set
858 of data that describes a movie frame).
859 An I frame ("intra" frame) describes a movie frame in isolation --
861 without respect to any other frame in the movie. A P frame ("predic‐
862 tive" frame) describes a movie frame by describing how it differs from
863 the movie frame described by the latest preceding I or P frame. A B
864 frame ("bidirectional" frame) describes a movie frame by describing how
865 it differs from the movie frames described by the nearest I or P frame
866 before and after it.
867 Note that the first frame of a movie must be described by an I frame
869 (because there is no previous movie frame) and the last movie frame
870 must be described by an I or P frame (because there is no subsequent
871 movie frame).
872 Beyond that, you can choose which frames are represented by which
874 types. You specify a pattern, such as IBPBP and ppmtompeg simply re‐
875 peats it over and over throughout the movie. The pattern affects
876 speed, quality, and stream size. Here is a chart which shows some of
877 the trade-offs:
878 Comparison of I/P/B Frames (Normalized)
880 ────────────────────────────────────
882 Frame Type Size Speed Quality
883 I frames 1000 1000 1000
884 P frames 409 609 969
885 B frames 72 260 919
886 (this is with constant qscale)
888 A standard sequence is IBBPBBPBBPBBPBB.
891 Select the sequence with the PATTERN parameter file statement.
894 Since the last MPEG frame cannot be a B frame (see above), if the pat‐
896 tern you specify indicates a B frame for the last movie frame of the
897 movie, ppmtompeg makes it an I frame instead.
898 Before Netpbm 10.26 (January 2005), ppmtompeg instead drops the trail‐
900 ing B frames by default, and you need the FORCE_ENCODE_LAST_FRAME pa‐
901 rameter file statement to make it do this.
902 The MPEG frames don't appear in the MPEG-1 stream in the same order
904 that the corresponding movie frames appear in the movie -- the B frames
905 come after the I and P frames on which they are based. For example, if
906 the movie is 4 frames that you will represent with the pattern IBBP,
907 the MPEG-1 stream will start with an I frame describing movie frame 0.
908 The next frame in the MPEG-1 stream is a P frame describing movie frame
909 3. The last two frames in the MPEG-1 stream are B frames describing
910 movie frames 1 and 2, respectively.
911 Specifying Input and Output Files
915 Specify the input frame images with the INPUT_DIR, INPUT, END_INPUT,
916 BASE_FILE_FORMAT, SIZE, YUV_FORMAT and INPUT_CONVERT parameter file
917 statements.
918 Specify the output file with the OUTPUT parameter file statement.
920 Statistics
924 ppmtompeg can generate a variety of statistics about the encoding. See
925 the -stat, -snr, -mv_histogram, -quiet, -no_frame_summary, and
926 -bit_rate_info options.
927
931 You can run ppmtompeg on multiple machines at once, encoding the same
932 MPEG stream. When you do, the machines are used as shown in the fol‐
933 lowing diagram. We call this "parallel mode."
934 ppmtompeg-par.gif
936 To do parallel processing, put the statement
938 PARALLEL
940 in the parameter file, followed by a listing of the machines, one ma‐
943 chine per line, then
944 END_PARALLEL
946 Each of the machine lines must be in one of two forms. If the machine
949 has filesystem access to the input files, then the line is:
950 machine user executable
952 The executable is normally ppmtompeg (you may need to give the complete
954 path if you've built for different architectures). If the machine does
955 not have filesystem access to the input files, the line is:
956 REMOTE machine user executable parameter file
958 The -max_machines command option limits the number of machines ppmtom‐
960 peg will use. If you specify more machines in the parameter file than
961 -max_machines allows, ppmtompeg uses only the machines listed first.
962 This is handy if you want to experiment with different amounts of par‐
963 allelism.
964 In general, you should use full path file names when describing exe‐
966 cutables and parameter files. This includes the parameter file argu‐
967 ment on the original invocation of ppmtompeg.
968 All file names must be the same on all systems (so if e.g. you're using
970 an NFS filesystem, you must make sure it is mounted at the same mount‐
971 point on all systems).
972 Because not all of the processes involved in parallel operation have
974 easy access to the input files, you must specify the SIZE parameter
975 file statement when you do parallel operation.
976 The machine on which you originally invoke ppmtompeg is the master ma‐
978 chine. It hosts a "combine server,", a "decode server," and a number
979 of "i/o servers," all as separate processes. The other machines in the
980 network (listed in the parameter file) are slave machines. Each hosts
981 a single process that continuously requests work from the master and
982 does it. The slave process does the computation to encode MPEG frames.
983 It processes frames in batches identified by the master.
984 The master uses a remote shell command to start a process on a slave
986 machine. By default, it uses an rsh shell command to do this. But use
987 the RSH parameter file statement to control this. The shell command
988 the master executes remotely is ppmtompeg, but with options to indicate
989 that it is to perform slave functions.
990 The various machines talk to each other over TCP connections. Each ma‐
992 chine finds and binds to a free TCP port number and tells its partners
993 the port number. These port numbers are at least 2048.
994 Use the PARALLEL_TEST_FRAMES, PARALLEL_TIME_CHUNKS, and PARALLEL_PER‐
996 FECT parameter file statements to control the way the master divides up
997 work among the slaves.
998 Use the -nice command option to cause all slave processes to run
1000 "nicely," i.e. as low priority processes. That way, this substantial
1001 and long-running CPU load will have minimal impact on other, possibly
1002 interactive, users of the systems.
1003
1006 Here is a look at ppmtompeg speed, in single-node (not parallel) opera‐
1007 tion:
1008 Compression Speed
1010 ───────────────────────────────────────
1012 Machine Type Macroblocks per second1
1013 HP 9000/755 280
1014 DEC 3000/400 247
1015 HP 9000/750 191
1016 Sparc 10 104
1017 DEC 5000 68
1018 1A macroblock is a 16x16 pixel square
1019 The measurements in the table are with inputs and outputs via a conven‐
1021 tional locally attached filesystem. If you are using a network
1022 filesystem over a single 10 MB/s Ethernet, that constrains your speed
1023 more than your CPU speed. In that case, don't expect to get better
1024 than 4 or 5 frames per second no matter how fast your CPUs are.
1025 Network speed is even more of a bottleneck when the slaves do not have
1027 filesystem access to the input files -- i.e. you declare them REMOTE.
1028 Where I/O is the bottleneck, size of the input frames can make a big
1030 difference. So YUV input is better than PPM, and JPEG is better than
1031 both.
1032 When you're first trying to get parallel mode working, be sure to use
1034 the -debug_machines option so you can see what's going on. Also, -de‐
1035 bug_sockets can help you diagnose communication problems.
1036
1040 • Kevin Gong - University of California, Berkeley, kev‐
1041 ing@cs.berkeley.edu
1042 • Ketan Patel - University of California, Berkeley, kpa‐
1045 tel@cs.berkeley.edu
1046 • Dan Wallach - University of California, Berkeley, dwal‐
1049 lach@cs.berkeley.edu
1050 • Darryl Brown - University of California, Berkeley, dar‐
1053 ryl@cs.berkeley.edu
1054 • Eugene Hung - University of California, Berkeley, ey‐
1057 hung@cs.berkeley.edu
1058 • Steve Smoot - University of California, Berkeley,
1061 smoot@cs.berkeley.edu
1062
1064 This manual page was generated by the Netpbm tool 'makeman' from HTML
1065 source. The master documentation is at
1066 http://netpbm.sourceforge.net/doc/ppmtompeg.html
1068netpbm documentation 23 July 2006 Ppmtompeg User Manual(0)