[Ffmpeg-devel-irc] ffmpeg-devel.log.20111201

[burek]

[00:29] <CIA-41> ffmpeg: 03Carl Eugen Hoyos 07master * r9dcd8b86f9 10ffmpeg/libavcodec/pictordec.c: 
[00:29] <CIA-41> ffmpeg: Use correct context for av_log.
[00:29] <CIA-41> ffmpeg: Fixes ticket #695.
[01:14] <CIA-41> ffmpeg: 03Michael Bradshaw 07master * r0275b75a7e 10ffmpeg/libavcodec/libopenjpegenc.c: BPP fix, YUVP9, 10, 16 support, make RGB24 default
[03:02] <CIA-41> ffmpeg: 03Anton Khirnov 07master * r91b412e786 10ffmpeg/avconv.c: 
[03:02] <CIA-41> ffmpeg: avconv: update InputStream.pts in the streamcopy case.
[03:02] <CIA-41> ffmpeg: This was broken in 2a651b719c309c5e2fc663a5a9d6ca36153ab98f.
[03:02] <CIA-41> ffmpeg: 03Anton Khirnov 07master * r2092232581 10ffmpeg/libavformat/utils.c: 
[03:02] <CIA-41> ffmpeg: lavf: estimate frame duration from r_frame_rate.
[03:02] <CIA-41> ffmpeg: If r_frame_rate is set, it should be more reliable for this than either
[03:02] <CIA-41> ffmpeg: codec or stream timebase.
[03:02] <CIA-41> ffmpeg: 03Anton Khirnov 07master * r741a05a293 10ffmpeg/avconv.c: avconv: compute next_pts from input packet duration when possible.
[03:02] <CIA-41> ffmpeg: 03Anton Khirnov 07master * rd3e3ffa501 10ffmpeg/libavformat/rmdec.c: 
[03:02] <CIA-41> ffmpeg: rmdec: don't set codec timebase.
[03:02] <CIA-41> ffmpeg: It's not supposed to be set outside of lavc.
[03:02] <CIA-41> ffmpeg: Set r_frame_rate instead.
[03:02] <CIA-41> ffmpeg: 03Anton Khirnov 07master * rc98c1f434e 10ffmpeg/libavformat/matroskadec.c: 
[03:02] <CIA-41> ffmpeg: matroskadec: don't set codec timebase.
[03:02] <CIA-41> ffmpeg: It's not supposed to be set outside of lavc.
[03:02] <CIA-41> ffmpeg: 03Anton Khirnov 07master * r7bb3e62594 10ffmpeg/avconv.c: 
[03:02] <CIA-41> ffmpeg: avconv: make copy_tb on by default.
[03:02] <CIA-41> ffmpeg: I.e. on streamcopy set output codec timebase from input stream timebase
[03:02] <CIA-41> ffmpeg: (as opposed to input codec timebase). This should be more sane, because
[03:02] <CIA-41> ffmpeg: since the stream is not decoded, the input codec tb has no relation to
[03:02] <CIA-41> ffmpeg: the timestamps of the copied packets.
[03:02] <CIA-41> ffmpeg: 03Martin Storsjö 07master * r2583660664 10ffmpeg/libavformat/ (rtpdec.h rtsp.c): 
[03:02] <CIA-41> ffmpeg: rtpdec: Add an init function that can do custom codec context initialization
[03:02] <CIA-41> ffmpeg: Signed-off-by: Martin Storsjö <martin at martin.st>
[03:02] <CIA-41> ffmpeg: 03Miroslav SlugeH 07master * r06d7325ab1 10ffmpeg/libavformat/ (Makefile rtpdec.c rtpdec_formats.h rtpdec_g726.c): (log message trimmed)
[03:03] <CIA-41> (18 lines omitted)
[03:54] <CIA-41> ffmpeg: 03Michael Niedermayer 07master * r54e40b9c14 10ffmpeg/ffmpeg.c: 
[03:54] <CIA-41> ffmpeg: ffmpeg: factorize duration calculation in transcode_video()
[03:54] <CIA-41> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[03:54] <CIA-41> ffmpeg: 03Michael Niedermayer 07master * r6e96ad8215 10ffmpeg/ffmpeg.c: 
[03:54] <CIA-41> ffmpeg: ffmpeg: make copy_tb tri state so the user can select either timebase or
[03:54] <CIA-41> ffmpeg: leave it at auto
[03:54] <CIA-41> ffmpeg: Inspired-by a change from Anton Khirnov
[03:54] <CIA-41> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[04:26] <CIA-41> ffmpeg: 03Michael Niedermayer 07master * rec20fc1581 10ffmpeg/libavformat/ (avformat.h options.c utils.c version.h): 
[04:26] <CIA-41> ffmpeg: lavf: allow grouping packets in chunks of a user specified size and duration.
[04:26] <CIA-41> ffmpeg: This is similar to MP4Boxs -inter
[04:26] <CIA-41> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[04:26] <CIA-41> ffmpeg: 03Michael Niedermayer 07master * r31f9032b78 10ffmpeg/libavformat/ (avformat.h options.c utils.c version.h): 
[04:26] <CIA-41> ffmpeg: lavf: add audio_preload option, this allows interleaving audio earlier
[04:26] <CIA-41> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[04:26] <CIA-41> ffmpeg: 03Michael Niedermayer 07master * r03b60509cb 10ffmpeg/ffmpeg.c: 
[04:26] <CIA-41> ffmpeg: ffmpeg: fix changing sample rate handling
[04:26] <CIA-41> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[04:52] <CIA-41> ffmpeg: 03Michael Niedermayer 07master * r67eb95ce15 10ffmpeg/ffmpeg.c: 
[04:52] <CIA-41> ffmpeg: ffmpeg: fix segfault with ffmpeg -h
[04:52] <CIA-41> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[09:59] <Bankin> I have a question about your task in Google code in for FFmpeg:create a logo, an icon or slogan - for icon or logo what shoul be the size of the logo/icon and for the slogan should it be just text or it shoulb be an image again ?
[10:44] <Bankin> I have a question about your task in Google code in for FFmpeg:create a logo, an icon or a slogan - for icon or logo what shoul be the size of the logo/icon and for the slogan should it be just text or it should be an image again ?
[11:44] Last message repeated 1 time(s).
[12:30] <Compn> Bankin : make it a vector image, so it can be scaled small or large ;)
[12:31] <Compn> it can be either, but i think a logo is what we are looking for, not the ffmpeg text 
[12:45] <Bankin> Thank you : )
[15:12] <ubitux> we really need to add various cmd line checks like ffmpeg/ffprobe/... -h
[15:12] <ubitux> it seems i'm not the only one to make the OptionDef mistakes :p
[15:36] <CIA-41> ffmpeg: 03Peter Ross 07master * rf3f488423a 10ffmpeg/libavcodec/pictordec.c: 
[15:36] <CIA-41> ffmpeg: pictordec: prevent segfault when reading corrupted files
[15:36] <CIA-41> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[15:37] <CIA-41> ffmpeg: 03Peter Ross 07master * r43a2a8b4a3 10ffmpeg/libavcodec/pictordec.c: 
[15:37] <CIA-41> ffmpeg: pictordec: use correct context for another av_log
[15:37] <CIA-41> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[16:35] <Tjoppen> hm
[16:35] <Tjoppen> georg's single_eubc patch changes the seek test's result. I'll have to investigate why
[17:37] <CIA-41> ffmpeg: 03Michael Niedermayer 07master * r3e1951e8f4 10ffmpeg/libavcodec/h264_ps.c: 
[17:37] <CIA-41> ffmpeg: h264: More restrictive more_rbsp_data_in_pps()
[17:37] <CIA-41> ffmpeg: Fixes Ticket697
[17:37] <CIA-41> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[18:02] <CIA-41> ffmpeg: 03Carl Eugen Hoyos 07master * rd9cc6bb432 10ffmpeg/ffmpeg.c: 
[18:02] <CIA-41> ffmpeg: Support ffmpeg -dn to disable copying data streams.
[18:02] <CIA-41> ffmpeg: Fixes ticket #647.
[18:47] <CIA-41> ffmpeg: 03Michael Niedermayer 07master * rb50be4e38d 10ffmpeg/libavcodec/vc1.c: 
[18:47] <CIA-41> ffmpeg: vc1_parser: fix NULL pointer dereference
[18:47] <CIA-41> ffmpeg: Fixes Ticket456
[18:47] <CIA-41> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[18:57] <cbsrobot> weird - converting qtrle to h264 gives me wrong colors and a unplayable file in quicktime, while qtrle to prores seems ok
[18:58] <Daemon404> that could merely be because quicktime's h264 decoder is a piel of crap
[18:58] <Daemon404> and can't handle a lot of stuff
[18:58] <Daemon404> it needs a specific subset of h264 iirc
[18:59] <cbsrobot> its just the prores to h264 is fine afterwards - i do not think its quicktime - at least not this time
[19:00] <cbsrobot> well I need to investigate later
[19:00] <Daemon404> oh
[19:00] <Daemon404> that sounds strange.
[19:00] <Compn> sounds like
[19:01] <Compn> wrong colorspace conversion
[19:01] <Compn> qtrle is paletted isnt it ?
[19:01] <cbsrobot> it's rgb24 
[19:02] <cbsrobot> It autoinserts following scale:
[19:02] <cbsrobot>  w:720 h:576 fmt:rgb24 -> w:720 h:576 fmt:yuv420p flags:0x4
[19:02] <cbsrobot> w:720 h:576 fmt:yuv420p -> w:720 h:576 fmt:rgb24 flags:0x4
[19:02] <cbsrobot> fo the qtrle to h264 conversion
[19:02] <cbsrobot> *for
[19:03] <cbsrobot> while qtrle to prores uses:
[19:03] <cbsrobot> w:720 h:576 fmt:rgb24 -> w:720 h:576 fmt:yuv420p flags:0x4
[19:03] <cbsrobot> w:720 h:576 fmt:yuv420p -> w:720 h:576 fmt:yuv422p10le flags:0x4
[19:03] <cbsrobot> hmm - I need to run
[19:04] <cbsrobot> I try to add a bugreport
[19:13] <cbsrobot> hmm - ffplay can play it, mplayer says: unsupportet pixel fmt: 83, quicktime shows a black frame, vlc a purple one ... nice 
[20:47] <CIA-41> ffmpeg: 03Michael Niedermayer 07master * r4a8e3324fb 10ffmpeg/libavcodec/libx264.c: 
[20:47] <CIA-41> ffmpeg: libx264: fix generic boolean support
[20:47] <CIA-41> ffmpeg: Fixes Ticket660
[20:47] <CIA-41> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[20:57] <cbsrobot> ^ nice
[20:58] <cbsrobot> a nice gci task would be to add a "-target pal-bluray" support
[20:58] <cbsrobot> and ntsc-bluray aswell
[20:58] <nevcairiel> oh please no, people should just forget about pal and ntsc with bluray
[20:58] <nevcairiel> make it all the same
[20:59] <cbsrobot> well sometime sourcematerial does not allow you to - i guess
[20:59] <nevcairiel> besides, bluray doesnt really have ntsc or pal modes
[20:59] <Compn> nevcairiel : you think converting framerates (from pal to ntsc) is a better idea ?
[20:59] <Compn> or from pal to (whatever bluray default you want) is a good idea ?
[21:00] <nevcairiel> Blu-ray doesnt support 25fps progressive, so you have to do something badly with it. .)
[21:00] <cbsrobot> see http://www.x264bluray.com/home/1080p
[21:00] <Compn> lol
[21:01] Action: Compn has no clue about bray
[21:01] Action: Compn calling it bray from now on
[21:01] <Daemon404> [15:00] < nevcairiel> Blu-ray doesnt support 25fps progressive, so you have to do something badly with it. .)
[21:01] <Daemon404> you can cheat teh system
[21:01] <Daemon404> you can flag it as 50i
[21:01] <Daemon404> whe nit is actually 25p
[21:01] <Compn> ugh
[21:01] <Daemon404> i have a bunch of blurays that do this
[21:01] <Daemon404> liek sherlock
[21:01] Action: Compn puts his fingers in his ears and runs away
[21:01] <nevcairiel> its a ugly hack
[21:01] <nevcairiel> if a bad player trys to deinterlace it, quality degrades =(
[21:02] <Daemon404> nevcairiel, id rather have that ugly hack than no 25p
[21:02] <Daemon404> :P
[21:02] <Daemon404> (i only use a computer to play)
[21:02] <Daemon404> and i dont really understand why they didnt include a 25p spec...
[21:02] <cbsrobot> isnt 25pfs using the same hack ?
[21:02] <Daemon404> no
[21:03] <cbsrobot> eh 25psf
[21:03] Action: pasteeater still waiting on x264 --device
[21:04] <cbsrobot> Progressive segmented Frame (PsF, sF, SF) is a scheme designed to acquire, store, modify, and distribute progressive-scan video using interlaced equipment and media.
[21:04] <cbsrobot> ^ wiki
[21:04] <nevcairiel> its really the same hack, yea
[21:04] <nevcairiel> assuming the processing software understands 2:2 pulldown
[21:06] <brontosaurusrex> hi, is there something special to do at compile time to enable multithreaded prores decoder/encoder?
[21:07] <cbsrobot> brontosaurusrex: no
[21:08] <brontosaurusrex> hrm, i'am getting zero speed boost
[21:09] <CIA-41> ffmpeg: 03Michael Niedermayer 07master * rdd974c1bc1 10ffmpeg/libavcodec/ (allcodecs.c libx264.c): 
[21:09] <CIA-41> ffmpeg: libx264: Implement rgb24 support through a seperate AVCodec.
[21:09] <CIA-41> ffmpeg: This avoids people mistakely encoding in a way that many players dont support.
[21:09] <CIA-41> ffmpeg: Fixes Ticket658
[21:09] <CIA-41> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[21:10] <Daemon404> cbsrobot, in ffmpeg you might
[21:11] <Daemon404> since there are 2 prores decoders
[21:11] <Daemon404> it might matter which is used.
[21:11] <cbsrobot> ^ michaelni : thanks
[21:13] <cbsrobot> Daemon404: both have CODEC_CAP_SLICE_THREADS
[21:14] <Daemon404> i do seem to recall one being faster though
[21:14] <Daemon404> but i see your point
[21:18] <brontosaurusrex> ffmpeg -threads 0 and the rest of cli - sounds good or...?
[21:18] <brontosaurusrex> getting really small speed boost
[21:20] <brontosaurusrex> for example ffmpeg -threads 8 -i "$files" -vcodec prores -profile 2 -acodec pcm_s16le "$out.mov"
[21:21] <cbsrobot> brontosaurusrex: what format is your infile ?
[21:21] <brontosaurusrex> prores
[21:21] <cbsrobot> ah prores -> prores ?
[21:22] <brontosaurusrex> cbsrobot, yes, just testing
[21:22] <Daemon404> is teh encoder multithreaded?
[21:22] Action: Daemon404 checks
[21:23] <Daemon404> (that may be your bottleneck)
[21:23] <brontosaurusrex> Daemon404, yeah, iam not sure
[21:23] <Daemon404> try outputting to /dev/null 
[21:23] <Daemon404> ?
[21:23] <brontosaurusrex> or getting some other input than prores to test?
[21:24] <brontosaurusrex> or output
[21:24] <Daemon404> i thought your goal was to test prores decoding speed multithreaded
[21:24] <brontosaurusrex> Daemon404, both
[21:37] <CIA-41> ffmpeg: 03Michael Niedermayer 07master * r59a58884b9 10ffmpeg/ffplay.c: 
[21:37] <CIA-41> ffmpeg: ffplay: Copy audio side data too. This fixes handling of some rare nellymoser files that change the sample rate mid stream (sample file at: http://trac.videolan.org/vlc/ticket/5586)
[21:37] <CIA-41> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[21:37] <CIA-41> ffmpeg: Signed-off-by: Marton Balint <cus at passwd.hu>
[21:37] <CIA-41> ffmpeg: 03Michael Niedermayer 07master * r8c15f9d79c 10ffmpeg/: 
[21:37] <CIA-41> ffmpeg: Merge remote-tracking branch 'cus/stable'
[21:37] <CIA-41> ffmpeg: * cus/stable:
[21:37] <CIA-41> ffmpeg:  ffplay: Copy audio side data too. This fixes handling of some rare nellymoser files that change the sample rate mid stream (sample file at: http://trac.videolan.org/vlc/ticket/5586)
[21:37] <CIA-41> ffmpeg: Merged-by: Michael Niedermayer <michaelni at gmx.at>
[21:39] <ubitux> michaelni: 5a15602a4e99c730036c33b467f60248889219e1  any reason this wasn't merge?
[21:46] <ubitux> i was thinking about merging fate{,2}.mak
[21:57] <michaelni> ubitux, i dont remember, feel free to remove it
[21:58] <ubitux> ok
[21:59] <cbsrobot> this aint my day
[21:59] <cbsrobot> now with latest x264 and ffmpeg all h264 files look dark ...
[22:00] <michaelni> do they look better with old ffmpeg ?
[22:00] <michaelni> old x264 ?
[22:00] <cbsrobot> i'm testing
[22:02] <cbsrobot> michaelni: a nice gci task would be to add a "-target pal-bluray" support
[22:03] <michaelni> AFAIK tasks added now wil only show up in 2 weeks
[22:07] <cbsrobot> seems older ffmpeg
[22:10] <michaelni> older works ?
[22:10] <michaelni> can you bisect ?
[22:13] <cbsrobot> I wont have time but my last revision was quite new ...
[22:16] <michaelni> which revission and how can i reproduce it ?
[22:20] <michaelni> cbsrobot, yuv or rgb x264 ?
[22:20] <cbsrobot> well ok
[22:21] <cbsrobot> so it seems quicktime player shows the file really dark
[22:21] <cbsrobot> vlc - no problem
[22:22] <michaelni> is it >8bit or rgb and does qt support that ?
[22:22] <michaelni> some change in the default encoding options maybe ...
[22:23] <cbsrobot> I need to check
[22:24] <brontosaurusrex> cbsrobot, lion?
[22:24] <cbsrobot> no snow leo
[22:25] <brontosaurusrex> same here, if you have a small sample i can test vlc vs qt on this system
[22:25] <cbsrobot> wait i'll get back to master
[22:26] <michaelni> we dont have a OSX fate client yet, anyone of you want to change that ?
[22:26] <brontosaurusrex> what would be a fate client?
[22:27] <cbsrobot> it's also way too dark with ffmpeg version N-34970-g6f326b5
[22:27] <cbsrobot> which i used before
[22:29] <cbsrobot> hmmm - how to show in git what revision i'm using ?
[22:31] <michaelni> git show
[22:31] <cbsrobot> ok
[22:31] <michaelni> brontosaurusrex, a fate client is something that tests ffmpeg regularly so we know early when we messed up
[22:32] <brontosaurusrex> michaelni, the git getting, compiling, test running platform?
[22:32] <michaelni> yes
[22:32] <brontosaurusrex> michaelni, is that something automagic or mainly manual thing?
[22:32] <michaelni> automatic
[22:33] <michaelni> it just needs a kick once a month when it got mysteriously stuck
[22:34] <CIA-41> ffmpeg: 03Michael Niedermayer 07master * r18abf46b6f 10ffmpeg/configure: 
[22:34] <CIA-41> ffmpeg: configure: add forgotten libx264rgb_encoder_deps
[22:34] <CIA-41> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[22:34] <brontosaurusrex> so the software/script is allready written?
[22:37] <michaelni> yes, you just need a few lines that calls it in regular intervals can be a cronjob, manual or a loop with sleep
[22:37] <michaelni> brontosaurusrex, i can forward you a mail that describes precissely how to set it up
[22:38] <michaelni> and theres doc/fate.txt in git 
[22:38] <brontosaurusrex> michaelni, right, so fate.sh, is it written in a way to autoupdate itself?
[22:38] <brontosaurusrex> ok, ill check the git, unless you have more info in that mail?
[22:38] <michaelni> i think the mail says the same
[22:41] <michaelni> mail forwarded
[22:42] <brontosaurusrex> michaelni, thanks
[22:43] <cbsrobot> michaelni: all fine
[22:43] <cbsrobot> its quicktime that sucks
[22:44] <cbsrobot> it has problems with high profile - as usual
[22:45] <michaelni> that should be documented better, maybe a warning when one encodes with high profile
[22:48] <cbsrobot> gci ...
[23:34] <CIA-41> ffmpeg: 03Adrian Drzewiecki 07master * rdd7453a24e 10ffmpeg/libavformat/id3v2.c: 
[23:34] <CIA-41> ffmpeg: Fix id3v2 extended header handling.
[23:34] <CIA-41> ffmpeg: When skipping over the extended header, take into account
[23:34] <CIA-41> ffmpeg: that the size field has already been read. The extended header
[23:34] <CIA-41> ffmpeg: also takes up space, so adjust total header length accordingly.
[23:34] <CIA-41> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[23:42] <CIA-41> ffmpeg: 03Derek Buitenhuis 07master * r69035f3ec7 10ffmpeg/libavcodec/libutvideo.cpp: 
[23:42] <CIA-41> ffmpeg: Revert "libutvideodec: Make buf_size local"
[23:42] <CIA-41> ffmpeg: This change was dependent on a different patch that
[23:42] <CIA-41> ffmpeg: never actually made it into FFmpeg, and it actually
[23:42] <CIA-41> ffmpeg: ended up breaking builds.
[23:42] <CIA-41> ffmpeg: This reverts commit 70cf7bb958899c369f778e29b8bd4203723bed04.
[23:42] <CIA-41> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[23:54] Action: Daemon404 sees michaelni got in a remark about git trees ;)
[23:56] <cbsrobot> bcoudurier: will you submit vf_sub to ffmpeg ?
[00:00] --- Fri Dec  2 2011