[Ffmpeg-devel-irc] ffmpeg-devel.log.20120710

[burek]

[00:54] <CIA-41> ffmpeg: 03Stefano Sabatini 07master * r2058b52cf8 10ffmpeg/ (doc/muxers.texi libavformat/segment.c libavformat/version.h): 
[00:54] <CIA-41> ffmpeg: lavf/segment: add -segment_times option
[00:54] <CIA-41> ffmpeg: Address trac ticket #1504.
[00:54] <CIA-41> ffmpeg: 03Stefano Sabatini 07master * raef9e94f73 10ffmpeg/ (doc/muxers.texi libavformat/segment.c libavformat/version.h): lavf/segment: add -segment_time_delta option
[01:31] <CIA-41> ffmpeg: 03Stefano Sabatini 07master * r21d5609826 10ffmpeg/ (4 files in 2 dirs): (log message trimmed)
[01:31] <CIA-41> ffmpeg: lavfi: add init_opaque callback
[01:31] <CIA-41> ffmpeg: This will be used by filters which require an opaque field to be passed
[01:31] <CIA-41> ffmpeg: in input. Should be required only for filters which imply a programmatic
[01:31] <CIA-41> ffmpeg: use.
[01:31] <CIA-41> ffmpeg: This is possibly a temporary solution, to be removed when we'll have a
[01:31] <CIA-41> ffmpeg: clearer and better agreememnt of how/if to pass binary data for
[01:31] <CIA-41> ffmpeg: 03Stefano Sabatini 07master * ra5382b50f2 10ffmpeg/libavfilter/sink_buffer.c: 
[01:31] <CIA-41> ffmpeg: lavfi/buffersink: restore possibility to pass opaque data
[01:31] <CIA-41> ffmpeg: Use the init_opaque callback for the purpose.
[01:31] <CIA-41> ffmpeg: Fix regression introduced in a5e8c41c28f. In particular, fix lavfi
[01:31] <CIA-41> ffmpeg: device in case a conversion to the supported (packed) formats is needed.
[01:40] <CIA-41> ffmpeg: 03Steven Robertson 07master * rac726a4f0c 10ffmpeg/libavfilter/src_movie.c: (log message trimmed)
[01:40] <CIA-41> ffmpeg: lavfi/movie: flush decoder frames on EOF
[01:40] <CIA-41> ffmpeg: movie currently forwards EOF (or begins looping) immediately upon
[01:40] <CIA-41> ffmpeg: reaching the last packet in the input stream, which can leave frames in
[01:40] <CIA-41> ffmpeg: the decoder. This change first tries to read any remaining packets from
[01:40] <CIA-41> ffmpeg: the decoder before forwarding EOF.
[01:40] <CIA-41> ffmpeg: Signed-off-by: Steven Robertson <steven at strobe.cc>
[01:41] <CIA-41> ffmpeg: 03Steven Robertson 07master * rc88d98fe4a 10ffmpeg/libavfilter/src_movie.c: 
[01:41] <CIA-41> ffmpeg: lavfi/movie: reindent after the previous patch
[01:41] <CIA-41> ffmpeg: Signed-off-by: Steven Robertson <steven at strobe.cc>
[01:41] <CIA-41> ffmpeg: Signed-off-by: Stefano Sabatini <stefasab at gmail.com>
[02:05] <CIA-41> ffmpeg: 03Paul B Mahol 07master * r409cc45883 10ffmpeg/tests/ (fate/audio.mak ref/fate/paf-audio): 
[02:05] <CIA-41> ffmpeg: fate: add test for paf audio decoder
[02:05] <CIA-41> ffmpeg: Signed-off-by: Paul B Mahol <onemda at gmail.com>
[02:06] <CIA-41> ffmpeg: 03Paul B Mahol 07master * r101615c7ff 10ffmpeg/tests/ (fate/video.mak ref/fate/paf-video): 
[02:06] <CIA-41> ffmpeg: fate: add test for paf video decoder
[02:06] <CIA-41> ffmpeg: Signed-off-by: Paul B Mahol <onemda at gmail.com>
[02:06] <CIA-41> ffmpeg: 03Paul B Mahol 07master * r1a8b0025f1 10ffmpeg/tests/ (fate/demux.mak ref/fate/paf-demux): 
[02:06] <CIA-41> ffmpeg: fate: add test for paf demuxer
[02:06] <CIA-41> ffmpeg: Signed-off-by: Paul B Mahol <onemda at gmail.com>
[02:19] <CIA-41> ffmpeg: 03Michael Niedermayer 07master * re8b9e1e294 10ffmpeg/libavcodec/ (dsputil.c dsputil.h dsputil_template.c): 
[02:19] <CIA-41> ffmpeg: dsputil: extend things to 12 and 14 bits
[02:19] <CIA-41> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[02:19] <CIA-41> ffmpeg: 03Michael Niedermayer 07master * r6ff6a51b86 10ffmpeg/libavcodec/dsputil_template.c: 
[02:19] <CIA-41> ffmpeg: h264: limit 10bpp MC pading trick to 10bpp
[02:19] <CIA-41> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[02:19] <CIA-41> ffmpeg: 03Michael Niedermayer 07master * r69d5e40e5a 10ffmpeg/libavcodec/h264idct.c: 
[02:19] <CIA-41> ffmpeg: h264idct: 12 and 14 bit support
[02:19] <CIA-41> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[02:20] <CIA-41> ffmpeg: 03Michael Niedermayer 07master * r5d05acba0e 10ffmpeg/libavcodec/h264pred.c: 
[02:20] <CIA-41> ffmpeg: h264pred: 12 and 14 bit support
[02:20] <CIA-41> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[02:20] <CIA-41> ffmpeg: 03Michael Niedermayer 07master * r1465ecf059 10ffmpeg/libavcodec/h264dsp.c: 
[02:20] <CIA-41> ffmpeg: h264dsp: 12 and 14 bit support
[02:20] <CIA-41> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[02:20] <CIA-41> ffmpeg: 03Michael Niedermayer 07master * re162733dcd 10ffmpeg/libavcodec/ (h264.c h264.h h264_ps.c): 
[02:20] <CIA-41> ffmpeg: h264: 12 and 14 bit support
[02:20] <CIA-41> ffmpeg: This works with I, P and B frames
[04:24] <CIA-41> ffmpeg: 03Paul B Mahol 07master * rb4d0c3d9d2 10ffmpeg/libavcodec/exr.c: 
[04:24] <CIA-41> ffmpeg: exr: return proper error code instead of -1
[04:24] <CIA-41> ffmpeg: Signed-off-by: Paul B Mahol <onemda at gmail.com>
[04:24] <CIA-41> ffmpeg: 03Paul B Mahol 07master * rdc0d551b52 10ffmpeg/libavcodec/exr.c: 
[04:24] <CIA-41> ffmpeg: exr: use buf_size instead of recalculating same value
[04:24] <CIA-41> ffmpeg: Signed-off-by: Paul B Mahol <onemda at gmail.com>
[04:29] <highgod> Hi,All.I add dxva2 to ffmpeg.exe to decode wmv3 video.In order to increase the speed,I init 4 sruface to hwaccel and delay 3 frame.but when the transfer operation finished.The last 3 frames can't be encoded,they are still in the surfaces.
[04:29] <highgod> I tried to modify code,but it so complex.
[04:33] <durandal_1707> michaelni: tiff encoder does not have init and close and use just av_malloc in enocde_frame, that must be really slow
[05:27] <durandal_1707> michaelni: hmm, tiffenc code needs allocated stuff to be zeroed, but doing just memset instead of malloc+memset should be still faster
[08:09] <rohankasat> hi , m getting an error "Exactly one scaler algorithm must be chosen" . can somebody help me out ?
[10:33] <TimNich> :rohankasat1 you have swscale flags set and they are defining two different scaling algorithms...
[11:18] <shroomM> vlc
[11:18] <shroomM> bah, sorry, wrong windows
[11:21] <rohankasat1> hey , guys 
[11:22] <rohankasat1> i got my video running , but its too fast , is there any frame rate set or delay api ?
[13:01] <michaelni> Tjoppen, there are mxfenc patches on ffmpeg-dev, if you have time to review
[13:23] <CIA-41> ffmpeg: 03Carl Eugen Hoyos 07master * r5bcabc56c0 10ffmpeg/libavcodec/snowenc.c: Define snow .pix_fmts to allow colour space auto-selection.
[13:28] <CIA-41> ffmpeg: 03Carl Eugen Hoyos 07master * r7307746444 10ffmpeg/libavformat/img2.c: 
[13:28] <CIA-41> ffmpeg: Recognize yuv10 image files as v210x.
[13:28] <CIA-41> ffmpeg: Reviewed-by: Kieran Kunhya
[13:34] <Tjoppen> michaelni: done
[13:47] <michaelni> Tjoppen, thx
[14:24] <CIA-41> ffmpeg: 03Michael Niedermayer 07master * r71c689e83e 10ffmpeg/libavcodec/dsputil.c: 
[14:24] <CIA-41> ffmpeg: dsputil: turn assert into if() as its possible to trigger with ffv1
[14:24] <CIA-41> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[14:33] <CIA-41> ffmpeg: 03Carl Eugen Hoyos 07master * r08ed3e8bf4 10ffmpeg/libavcodec/ffv1.c: Cosmetics: Move ffv1 encoder initialisation code.
[14:33] <CIA-41> ffmpeg: 03Carl Eugen Hoyos 07master * r4c4e125695 10ffmpeg/libavcodec/ffv1.c: Cosmetics: Return AVERROR_INVALIDDATA from ffv1 encoder initialisation if invalid data was provided.
[14:33] <CIA-41> ffmpeg: 03Carl Eugen Hoyos 07master * racf0283925 10ffmpeg/libavcodec/ffv1.c: 
[14:33] <CIA-41> ffmpeg: Set default ffv1 coder to -1.
[14:33] <CIA-41> ffmpeg: Autoselect coder 1 instead of default coder if bits_per_raw_sample > 8.
[14:33] <CIA-41> ffmpeg: Fixes ticket #1519.
[14:44] <CIA-41> ffmpeg: 03Nicolas George 07master * r0abab00303 10ffmpeg/libavcodec/dpx.c: dpx: invalid aspect ratio is 0/1, not 0/0.
[17:34] <durandal_1707> michaelni: why exr decoder instead of aborting decoding just add black pixels for missing scan lines?
[17:42] <michaelni> durandal_1707, in case of a damaged file we want to extract as much as possible
[17:43] <durandal_1707> michaelni: all other decoders (not just image ones) don't to that
[17:44] <durandal_1707> and qatar "devs" did not liked such behaviour (with bytestream2 it is trivial to show partial frames...)
[17:45] <durandal_1707> when i used such b2 feature when porting dfa to b2
[17:59] <michaelni> durandal_1707, maybe i misunderstood you but most decoders try hard to return as much data as possible and use error concealment (mpeg* / h26*)
[18:00] <michaelni> IMHO exr should not abort but return as much intact data as it has
[18:03] <michaelni> imagine someone has a exr image with an important text, he wants to read. getting no frame and an error messgae is a bad idea for that use case
[18:05] <durandal_1707> michaelni: look at other image decoders (game codecs are less important) none of them display partial frames
[18:07] <durandal_1707> michaelni: so decide which policy should be applied to FFmpeg code so that decoders can be improved
[18:11] <michaelni> the only reason i can see not to display a partial frame is when its the last frame in a movie so after playback one would end on a bad instead of a near identical good one
[18:11] <durandal_1707> michaelni: also why exr support is not done via OpenEXR library? Are there some reasons I'm not aware of?
[18:13] <michaelni> a openexr based decoder is welcome, so are improvments to exr.c
[18:20] <michaelni> about partial frames for codecs that dont support error concealment, we could add a flag that specifies if they should be shown or not
[18:20] <michaelni> so the user could specify her preferance
[18:35] <michaelni> durandal_1707, btw out jpeg decoder doesnt discard images that are damaged or truncated though its error concealment capabilities could be improved alot
[19:41] <CIA-41> ffmpeg: 03Michael Niedermayer 07master * r43c609374a 10ffmpeg/libavcodec/mpeg4videodec.c: 
[19:41] <CIA-41> ffmpeg: mpeg4videodec: use av_assert
[19:41] <CIA-41> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[19:42] <CIA-41> ffmpeg: 03Michael Niedermayer 07master * r84d902fb92 10ffmpeg/libavcodec/mpeg4videoenc.c: 
[19:42] <CIA-41> ffmpeg: mpeg4videoenc: use av_assert
[19:42] <CIA-41> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[20:34] <durandal_1707> michaelni: got almost working zip1 decompression in exr
[20:35] <durandal_1707> it would be nice if slice threading could be implemented in decoder
[21:35] <CIA-41> ffmpeg: 03Ronald S. Bultje 07master * rf80ddd5bf7 10ffmpeg/ (configure libavutil/internal.h): 
[21:35] <CIA-41> ffmpeg: lavu: use intrinsics for emms on systems lacking inline asm support
[21:35] <CIA-41> ffmpeg: Signed-off-by: Diego Biurrun <diego at biurrun.de>
[21:35] <CIA-41> ffmpeg: Signed-off-by: Martin Storsjö <martin at martin.st>
[21:35] <CIA-41> ffmpeg: 03Ronald S. Bultje 07master * r3f150ffba3 10ffmpeg/ (configure libavutil/x86/cpu.c): 
[21:35] <CIA-41> ffmpeg: x86/cpu: implement support for xgetbv through intrinsics
[21:35] <CIA-41> ffmpeg: Signed-off-by: Martin Storsjö <martin at martin.st>
[21:35] <CIA-41> ffmpeg: 03Martin Storsjö 07master * r620b1e7e98 10ffmpeg/libavutil/mem.c: 
[21:35] <CIA-41> ffmpeg: mem: Don't abort on av_malloc(0) in debug mode
[21:35] <CIA-41> ffmpeg: This makes the behaviour consistent between debug and release mode.
[21:35] <CIA-41> ffmpeg: Signed-off-by: Martin Storsjö <martin at martin.st>
[21:35] <CIA-41> ffmpeg: 03Ronald S. Bultje 07master * rc0ee695bd7 10ffmpeg/ (configure libavutil/x86/cpu.c): 
[21:35] <CIA-41> ffmpeg: x86/cpu: implement support for cpuid through intrinsics
[21:35] <CIA-41> ffmpeg: Signed-off-by: Martin Storsjö <martin at martin.st>
[21:36] <CIA-41> ffmpeg: 03Philip de Nier 07master * r5255ca24cd 10ffmpeg/libavformat/mxfdec.c: 
[21:36] <CIA-41> ffmpeg: mxfdec: set track edit rate num/den in expected order
[21:36] <CIA-41> ffmpeg: This matches the order used for the index table edit rate.
[21:36] <CIA-41> ffmpeg: Reviewed-by: Baptiste Coudurier <baptiste.coudurier at gmail.com>
[21:36] <CIA-41> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[21:36] <CIA-41> ffmpeg: Signed-off-by: Luca Barbato <lu_zero at gentoo.org>
[21:36] <CIA-41> ffmpeg: 03Philip de Nier 07master * rdfefede00e 10ffmpeg/libavformat/mxfdec.c: (log message trimmed)
[21:36] <CIA-41> ffmpeg: mxfdec: detect uncomp pictures using essence container ul
[21:36] <CIA-41> ffmpeg: This supports detection of uncompressed picture in files that
[21:36] <CIA-41> ffmpeg: didn't include a Picture Coding Label. The lables weren't
[21:36] <CIA-41> ffmpeg: available until SMPTE 377-1 and RP224v10
[21:36] <CIA-41> ffmpeg: Reviewed-by: Baptiste Coudurier <baptiste.coudurier at gmail.com>
[21:36] <CIA-41> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[21:36] <CIA-41> ffmpeg: 03Philip de Nier 07master * rfda891e108 10ffmpeg/libavformat/ (mxf.c mxf.h mxfdec.c): (log message trimmed)
[21:36] <CIA-41> ffmpeg: mxfdec: set pixel format for cdci picture formats
[21:36] <CIA-41> ffmpeg: The properties of the CDCI Descriptor are insufficient to specify
[21:36] <CIA-41> ffmpeg: the pixel format for uncompressed picture data. SMPTE 377-1 and
[21:36] <CIA-41> ffmpeg: RP224v10 have defined a set of picture coding labels to indicate what
[21:36] <CIA-41> ffmpeg: formatting was used.
[21:36] <CIA-41> ffmpeg: This patch uses 2 labels to detect UYVY422 or YUYV422 pixel formats.
[21:36] <CIA-41> ffmpeg: 03Tomas Härdin 07master * r1fadf53e00 10ffmpeg/libavformat/mxfdec.c: 
[21:36] <CIA-41> ffmpeg: mxfdec: truncate packets that extend past the next edit unit
[22:44] <CIA-41> ffmpeg: 03Nicolas George 07master * r5a9d6993cb 10ffmpeg/libavutil/parseutils.c: 
[22:44] <CIA-41> ffmpeg: parseutils: ignore digits below the microsecond.
[22:44] <CIA-41> ffmpeg: Accept 1.1234567 as simply 1.123456 instead of rejecting it.
[22:44] <CIA-41> ffmpeg: The rounding is towards 0, which is acceptable and much simpler.
[23:57] <teratorn> hey guys, i hope this isn't too off-topic, but..  we have a webm-based pseudo-live streaming server, which currenting starts to stream packets to new clients on the NEXT keframe after they connect, then the same packets are streamed to each client
[23:58] <teratorn> How would I go about starting the stream immediately (fabricating a new unique keyframe from the existing stream... sending a few frames until the GOP is exhausted, then resuming copying the original packet stream?
[23:59] <ohsix> you dont', you add a lot more keyframes since you're streaming :D
[23:59] <teratorn> well
[23:59] <teratorn> hrm :/
[23:59] <teratorn> that is the standard practice ? :/
[23:59] <teratorn> what do you set gop to?
[00:00] --- Wed Jul 11 2012