| Branch: | Revision:

ffmpeg / tests / ffmpeg.regression.ref @ 40b67a56

History | View | Annotate | Download (9.22 KB)

# Date Author Comment
40b67a56 11/05/2007 09:24 AM Benoit Fouet

Fix mov checksums following r10852.

Originally committed as revision 10928 to svn://

f6128014 09/11/2007 10:46 PM Michael Niedermayer

fix timestamps used for ratecontrol
these were wrong (in pts vs dts sense) when b frames were in use
they were also wrong if the average framerate was smaller than 1/timebase
resulting in totally wrong final bitrate

Originally committed as revision 10477 to svn://

e559c30a 09/09/2007 03:06 PM Michael Niedermayer

allow changing of the spatial_decomposition_count and quantization tables
per frame (this is untested as the encoder does not use it yet)

Originally committed as revision 10458 to svn://

7d7f57d9 09/08/2007 02:51 PM Michael Niedermayer

store halfpel filter coefficients in the header as well as the
flag for diagonal interpolation

the primary reason for this change is that previously MC up to 1/4 pel
matched H.264 exactly and that increases the risk of stumbling over

secondly this allows 0.10 db or more quality gain by choosing a longer...

a68ca08e 09/08/2007 03:14 AM Michael Niedermayer

cleanup mc_block()
perform interpolation steps in such an order that halfpel interpolation
could be done per picture
this also makes mc_block() match h.264 for the 1/4 pel cases so that the
use of the h264 functions for some cases does not introduce a fantastic mess...

d593e329 08/25/2007 03:00 AM Michael Niedermayer

use 16bit IDWT (a SIMD implementation of it should be >2x faster then with
the old 32bit code)
disable mmx/sse2 optimizations as they need a rewrite now

Originally committed as revision 10218 to svn://

b538791b 08/23/2007 10:49 AM Michael Niedermayer

use more bits on the encoder side
negligible increase in quality

Originally committed as revision 10195 to svn://

ce611a27 08/21/2007 04:29 PM Michael Niedermayer

Change rounding of the horizontal DWT to match the vertical one.
This allows some simplifications and optimizations and should
not have any effect on quality.

Originally committed as revision 10172 to svn://

d7826b58 08/05/2007 08:03 PM Michael Niedermayer

make wmv1 and wmv2 playable with M$ DMO decoder

Originally committed as revision 9950 to svn://

d2f8f410 08/04/2007 08:35 PM Michael Niedermayer

flashsv regression test
closes issue5

Originally committed as revision 9912 to svn://

d31c98bf 07/03/2007 03:41 PM Benoit Fouet

update regression checksums after r9447

Originally committed as revision 9468 to svn://

4fbd52f0 06/07/2007 10:36 PM Diego Biurrun

Merge tests Makefile into the top-level Makefile.
This gets rid of some recursion and simplifies the code.

Originally committed as revision 9255 to svn://

1c87ed3a 03/24/2007 01:48 AM Michael Niedermayer

1000000l for me ive misstakely reverted the regression checksum change from r8444 in r8486
how did i succeed doing such a ridiculously silly thing? well i think it happened like:
1. verifying that the regression tests pass with old resample2.c
2. updating the regressions to the new resample2.c ... failed svn complained...

8ec04d34 03/22/2007 02:27 PM Michael Niedermayer

set more reasonable cutoff for audio rate conversion

Originally committed as revision 8486 to svn://

4ba67eee 03/22/2007 01:52 PM Michael Niedermayer

update resgression tests after the resampler changes

Originally committed as revision 8485 to svn://

4140a523 03/18/2007 11:18 PM Bill O'Shaughnessy

AC-3 encoded volume is too high. Revert revision 7160:
bring AC3 encoder output up to input volume level
patch by Bill O'Shaughnessy % bill P oshaughnessy A %
+ reg tests update gruntwork by me
Original thread:
date: Nov 21, 2006 11:36PM
subject: [Ffmpeg-devel] Simpler Patch to bring AC3 encoder output up to input level...

75b213ed 03/05/2007 05:54 PM Michael Niedermayer

prevent audio packets from being fragmented, wmp mac doesnt seem to like that and microshitty also doesnt seem to ever fragment them

Originally committed as revision 8265 to svn://

ff0e0b62 03/05/2007 02:54 PM Michael Niedermayer

update forgotten regression test chcksums for ====
Author: michael
Date: Mon Mar 5 03:41:49 2007
New Revision: 8240


create codec_comment_header which looks more like what M$ creates, sane or not ...

Originally committed as revision 8260 to svn://

1d7d9935 02/12/2007 11:32 PM Michael Niedermayer

always use audio spread (instead of "video no spread" for audio ...)

Originally committed as revision 7960 to svn://

a98d3e82 02/12/2007 10:55 PM Michael Niedermayer

increaseing preroll time to the minimum of the asf files i have

Originally committed as revision 7955 to svn://

d9076fe7 02/12/2007 07:59 PM Michael Niedermayer

remove md5 check for wma as there are rounding differences between cpus

Originally committed as revision 7954 to svn://

0786943a 02/12/2007 07:50 PM Michael Niedermayer

disabling vorbis regression test for now, both encoder and decoder behave strange and have to be fixed

Originally committed as revision 7953 to svn://

0d9f8633 02/12/2007 07:24 PM Michael Niedermayer

dont set the keyframe flag for audio as microshitty doesnt do it and some things have problems with our asfs, maybe that has a positive effect?

Originally committed as revision 7952 to svn://

5600b6e1 02/12/2007 06:44 PM Michel Bardiaux

Regression changes after addition of fact chunk to non-PCM wav

Originally committed as revision 7951 to svn://

2d241e66 02/09/2007 08:49 PM Michael Niedermayer

fix broken memset(padding, 0) code
add a few asserts

Originally committed as revision 7903 to svn://

190972a0 02/07/2007 11:06 PM Michael Niedermayer

iam pretty sure send time is dts and not pts ...

Originally committed as revision 7879 to svn://

3160138f 02/07/2007 05:04 PM Michael Niedermayer

vorbis regression test
(using .asf as our .ogg muxer depends on libogg, nut muxer depends on libnut and vorbis in avi/mpeg is not really a good idea)

Originally committed as revision 7874 to svn://

6d3d8dae 02/06/2007 11:59 PM Michael Niedermayer

wma regression test

Originally committed as revision 7865 to svn://

45ddd8b0 01/14/2007 11:55 PM Måns Rullgård

merge fragmented mpeg4adv section in

Originally committed as revision 7519 to svn://

396a5e68 01/02/2007 12:24 PM Michael Niedermayer

store a few values in the header as difference to the last
this makes frames a few bytes smaller (0.1% for high bitrate but >1% for low bitrates)

Originally committed as revision 7401 to svn://

148c9bdb 12/28/2006 12:35 PM Allan Hsu

Patch for the FLV muxer to supply more complete metadata
in the onMetaData header. Patch by Allan Hsu <allan at counterpop dot net>.

Originally committed as revision 7382 to svn://

d951bb9a 12/22/2006 11:30 PM Michael Niedermayer

test DIRECT{0,0} too if flag mv0 is used
slight PSNR/bitrate increase

Originally committed as revision 7361 to svn://

49f1ac5a 12/21/2006 05:55 PM Baptiste Coudurier

change mpeg2 intra vlc test to advanced one, add qprd trell and b frames

Originally committed as revision 7349 to svn://

d6975eb8 12/21/2006 03:45 PM Michael Niedermayer

use qscale_table in adaptive quant
some PSNR/bitrate improvement for adaptive quant

Originally committed as revision 7346 to svn://

59743d16 12/21/2006 02:24 PM Michael Niedermayer

dont randomly disallow intr4v in adaptive quant
some PSNR/bitrate gain if adaptive quant is used
initalize qscale_table correctly (it was pretty much random since the qp->lambda change)
this probably has not much effect as the table isnt used currently IIRC...

51b1a6c9 12/21/2006 01:23 PM Michael Niedermayer

dont randomly dissallow direct MBs
some PSNR/bitrate gain for b frames with adaptive quant or QPRD

Originally committed as revision 7341 to svn://

2d1e5071 12/21/2006 01:03 PM Michael Niedermayer

QPRD & adaptive quantization regression tests

Originally committed as revision 7340 to svn://

e4d0e2ed 12/20/2006 05:11 PM Michael Niedermayer

fix special casing of the 0,0 MV for b frames (slight PSNR/bitrate gain)

Originally committed as revision 7334 to svn://

c8aee695 12/11/2006 08:18 PM Reimar Döffinger

Fix JPEG-LS encoder 0xff-escaping writing too much or uninitialized data.

Originally committed as revision 7284 to svn://

ed1dfc49 12/01/2006 08:38 PM Michael Niedermayer

remove pretty much useless skiping of some predictors
this has pretty much no quality or speed effect except very small random changes

Originally committed as revision 7202 to svn://

f7f8af46 11/30/2006 10:44 PM Michael Niedermayer

1 step diamond search around the median MV predictor
+0.01 PSNR
idea from x264

Originally committed as revision 7189 to svn://

3e954443 11/23/2006 10:21 PM Bill O'Shaughnessy

bring AC3 encoder output up to input volume level
patch by Bill O'Shaughnessy % bill P oshaughnessy A %
+ reg tests update gruntwork by me
Original thread:
date: Nov 21, 2006 11:36 PM
subject: [Ffmpeg-devel] Simpler Patch to bring AC3 encoder output up to input level...

3c141817 11/22/2006 01:53 PM Baptiste Coudurier

fix bitstream to be correctly decoded by quicktime

Originally committed as revision 7157 to svn://

8cd15e47 11/22/2006 12:30 AM Michael Niedermayer

fix b frames + -mbd rd without -psnr (yes that was seriously buggy and probably hasnt been noticed as everyone who tested stuff used -psnr)

Originally committed as revision 7151 to svn://

3c8f30a7 09/04/2006 03:33 AM Roman Shaposhnik
  • Restructuring the division of labor between DV codec and DV format
    [ Based on a patch by Brian Brice (bbrice at newtek dot com) ]

Originally committed as revision 6161 to svn://

5340a0bc 08/24/2006 10:32 AM Baptiste Coudurier

wide atom is only defined in mov, use free for mp4/3gp

Originally committed as revision 6072 to svn://

0bd2483a 08/22/2006 06:24 AM Michael Niedermayer

replace MULL by MULH + shift in dct32() (code is very slightly faster, and the compiler should be able to optimize the shift away on 64bit archs)

Originally committed as revision 6045 to svn://

57060f89 08/18/2006 04:48 PM David Conrad

AVI tag reading and writing patch by David Conrad.
umovimus at gmail dot com

Originally committed as revision 6022 to svn://

efbe19fd 08/07/2006 02:36 PM Baptiste Coudurier

update regression tests for patches r5945-r5950 based on patch by Francois Draperi <>

Originally committed as revision 5951 to svn://

634b8cfa 08/06/2006 03:29 PM Baptiste Coudurier

always write duration and file size, fix seeking, progress bar is now active

Originally committed as revision 5939 to svn://

fd0fb306 07/17/2006 09:51 PM Michael Niedermayer

add header info to flv format based on a patch by (Philipp Klaus >ffmpegdevel ad pylonsoft mot ch )

Originally committed as revision 5779 to svn://

a403fc03 07/02/2006 10:22 AM Justin Ruggles

flac-lpc patch by (Justin Ruggles jruggle earthlink net)
tabs removed and fixed (it was missing in the patch) by me

Originally committed as revision 5572 to svn://

e71bcc37 06/27/2006 09:01 PM Justin Ruggles

flacenc - rice param search patch by (Justin Ruggles jruggle earthlink net

1) search for optimal rice parameters and partition order. i also
modified the stereo method estimation to use this to calculate estimated
bit count instead of using just the pure sums....

f33aa120 06/26/2006 06:00 AM Michael Niedermayer

stereo decorrelation support by (Justin Ruggles jruggle earthlink net>)

Originally committed as revision 5528 to svn://

aeb2904e 06/24/2006 11:40 AM Michael Niedermayer

flac regression test

Originally committed as revision 5515 to svn://

4a387d7d 06/14/2006 07:10 PM Loren Merritt

update regression tests for r5449 (adpcm)

Originally committed as revision 5476 to svn://

871371a7 05/30/2006 05:44 AM Loren Merritt

Snow: fix subband weighting in wavelet cmp functions. use 32x32 cmp in iterative motion estimation.

Originally committed as revision 5427 to svn://

8c36eaaa 05/28/2006 09:44 PM Loren Merritt

Snow multiple reference frames
(bitstream is not backwards compatible, even if refs aren't used)

Originally committed as revision 5417 to svn://

ed90d370 05/18/2006 11:32 PM Michael Niedermayer

dont store mov style meta tags in mp4 (fixes ffmpeg -title + gtkpod)

Originally committed as revision 5405 to svn://

307eb248 05/13/2006 05:16 PM Baptiste Coudurier

add MPEG-2 intra vlc support

Originally committed as revision 5372 to svn://

e27b6e62 04/28/2006 05:03 PM Michael Niedermayer

missmatch control for mpeg2 intra dequantization if bitexact=1

Originally committed as revision 5328 to svn://

ddbddce5 04/24/2006 01:04 PM Benjamin Larsson

Fix samples to bytes relation.

Originally committed as revision 5314 to svn://

f1f7807a 04/10/2006 12:12 PM Benjamin Larsson

AC3 encoding volume fix, by Justin Ruggles jruggle earthlink net.

Originally committed as revision 5281 to svn://

52dc09a5 04/07/2006 09:30 PM Corey Hickey

Fix regression tests to match changed output of AC3 encoder
(ac3enc.c r1.30)

Originally committed as revision 5275 to svn://

0816152a 03/13/2006 05:23 AM Loren Merritt

seed iterative_me with mvs from conventional search.
+15% speed, +.1 dB

Originally committed as revision 5159 to svn://

0b297700 03/06/2006 08:54 AM Daniel Maas

DVCPRO50 support.

Patch by Daniel Maas dmaas at maasdigital dot com

Originally committed as revision 5113 to svn://

26f86eaf 03/02/2006 07:28 PM Baptiste Coudurier

ftyp patch by (Baptiste COUDURIER <baptiste.coudurier smartjog com)

Originally committed as revision 5098 to svn://

1522767d 03/02/2006 01:37 AM Baptiste Coudurier

compute output format before setting language patch by (Baptiste COUDURIER <baptiste.coudurier smartjog com)

Originally committed as revision 5093 to svn://

40116978 02/28/2006 02:54 AM Dan Maas

Assign class 2 to most macroblocks by default, instead of a more conservative
method defined by SMPTE

Patch by Dan Maas

Originally committed as revision 5074 to svn://

2e52b44e 02/27/2006 04:50 AM Dan Maas

AC coefficient weighing (per SMPTE 314M) for the DV encoder and decoder. The
performance impact is less than 1%.

Patch by Dan Maas (dmaas at maasdigital dot com)

Originally committed as revision 5070 to svn://

6df5f6ae 02/25/2006 10:58 PM Dan Maas

size[0-3] are not initialized (and can get random negative trash
values), so the comparison with vs_total_ac_bits is messed up on the
first couple loop iterations, leading to AC underflows.

the b->prev[] pointers were not being maintained correctly. We
potentially have to update b->prev[] both before and after the area...

6661b213 02/23/2006 01:05 PM Michael Niedermayer

dont lowwer qnos of more blocks then needed

Originally committed as revision 5055 to svn://

d2d230a7 02/23/2006 08:56 AM Michael Niedermayer

fixing dv_guess_qnos()
roman, dont hesitate to reverse this and solve it differntly if you want ...

Originally committed as revision 5053 to svn://

b4e021e8 02/02/2006 03:34 PM Michael Niedermayer

jpeg-ls regression test

Originally committed as revision 4925 to svn://

82e415b1 01/25/2006 05:14 PM Michael Niedermayer

avoid using non constant fields of AVCodecContext in avi muxer

Originally committed as revision 4897 to svn://

12943a53 01/24/2006 08:15 AM Benjamin Larsson

Fix the wrongly commited regressions stuff for psp compatible mp4 files.

Originally committed as revision 4890 to svn://

dcfdb046 01/24/2006 08:03 AM Benjamin Larsson

Fix for title field in PSP compatible mp4 files. Patch by
Olivier, oandrieu <at>

Originally committed as revision 4889 to svn://

29f7ae79 01/23/2006 06:49 PM Michael Niedermayer

scale video for snow iterative me down so the egression test is faster

Originally committed as revision 4888 to svn://

b29af723 01/23/2006 02:12 PM Michael Niedermayer

4gb fixes by (Dirk Musfeldt | d.musfeldt meilenstein de)

Originally committed as revision 4886 to svn://

48d1b9a1 01/21/2006 02:35 AM Loren Merritt

fix convergence of snow iterative_me

Originally committed as revision 4876 to svn://

d3ac6ed6 01/19/2006 09:21 AM Loren Merritt

detect when iterative_me fails to converge, and stop searching

Originally committed as revision 4868 to svn://

c8a596d2 01/12/2006 09:37 PM Michael Niedermayer

bugs found by valgrind

Originally committed as revision 4841 to svn://

47ad9aba 01/08/2006 05:06 PM Michael Niedermayer

fixing second last time François Revol applies a patch without running the regression tests, next time his account is disabled
Note, if you think thats too harsh, look at the cvs history he has broken the regression tests many times and has not once
updated the checksums ......

80e44bc3 01/08/2006 01:50 AM Michael Niedermayer

use h264 MC functions for 2xX Xx2 blocks in snow too

Originally committed as revision 4824 to svn://

2692ceab 01/07/2006 06:34 PM Michael Niedermayer

use h264 MC functions for rectangular blocks too

Originally committed as revision 4821 to svn://

a588a2c6 01/07/2006 03:53 PM Michael Niedermayer

regression test for iterative ME & wavelet cmp functions

Originally committed as revision 4820 to svn://

88e51e1f 12/29/2005 03:46 PM Michael Niedermayer

second try of improved lambda/qp guessing for motion estimation RD in first pass

Originally committed as revision 4785 to svn://

aaaa6f15 12/25/2005 04:34 PM Michael Niedermayer

dont discard least significant bits

Originally committed as revision 4773 to svn://

b7c51ebc 12/24/2005 03:12 AM Michael Niedermayer

slightly better mv_penalty table

Originally committed as revision 4770 to svn://

377798d6 12/23/2005 12:40 AM Michael Niedermayer

minor typo

Originally committed as revision 4767 to svn://

50c6f6cd 12/22/2005 02:17 PM Michael Niedermayer

improved lambda/qp guessing for motion estimation RD in first pass
this only affects encoding with bframes and rate control

Originally committed as revision 4765 to svn://

f79432a2 09/05/2005 10:26 AM Michael Niedermayer

set a few variables correctly

Originally committed as revision 4560 to svn://

c619ff6d 08/30/2005 08:03 PM Michael Niedermayer

dc coeff rounding fix
class=3 num of bits fix
do interlaced check & idct only if CODEC_FLAG_INTERLACED_DCT

Originally committed as revision 4542 to svn://

8fce2707 08/30/2005 07:44 PM Michael Niedermayer

regression tests for dv

Originally committed as revision 4541 to svn://

d8cbeba1 07/25/2005 02:36 PM Michael Niedermayer

yamaha adpcm regression test

Originally committed as revision 4473 to svn://

3b5ffe7a 05/18/2005 07:42 PM Michael Niedermayer

fix mpeg2 chroma_420_type

Originally committed as revision 4274 to svn://

132041f0 05/05/2005 02:10 PM Justin Ruggles

While adding stereo rematrixing, I came across something that needs to
be fixed even without adding the feature. The output correctly uses 4
dummy values for the rematrixing flags in block-0, but the bit
allocation routine does not take these bits into account. From what I...

8c2515bb 04/17/2005 09:27 AM Yartrebo

fix overflow in mc_block() patch by (Yartrebo )yartrebo earthlink net

Originally committed as revision 4132 to svn://

b44985ba 04/15/2005 01:24 PM Michael Niedermayer

store the number of runs to avoid storing the last run value
about 10% lower bitrate for -qscale 32 (forman & some music video)
worst case bitrate increase <0.1% (lossless or low qscale)
and now the bad news, even though this just adds a single subtraction and an if() into the medium sized unpack_coeffs() loop and the if() will only be false once per unpac_coeff() call, gcc produces 50% slower code, i didnt look at the generated asm yet, not sure if i want to ......

f5a71928 04/10/2005 02:21 PM Michael Niedermayer

rescale coefficients during IDWT, that way the lifting steps are much simpler and faster

Originally committed as revision 4119 to svn://