[Ffmpeg-devel-irc] ffmpeg-devel.log.20120810

burek burek021 at gmail.com
Sat Aug 11 02:05:02 CEST 2012


[00:03] <ubitux> showspectrum fixed \o/
[00:17] <CIA-41> ffmpeg: 03Michael Niedermayer 07master * r2f7e8dcf45 10ffmpeg/ (4 files in 2 dirs): (log message trimmed)
[00:17] <CIA-41> ffmpeg: yadif: redesign first and last 2 lines handling.
[00:17] <CIA-41> ffmpeg: The previous code dependent on the input buffer matching the
[00:17] <CIA-41> ffmpeg: buffer that has been provided by yadifs get_buffer.
[00:17] <CIA-41> ffmpeg: The API does in now way gurantee this though its often true.
[00:17] <CIA-41> ffmpeg: This fixes some out of array reads.
[00:17] <CIA-41> ffmpeg: The regression test checksums change due to "out of picture" values
[00:17] <CIA-41> ffmpeg: 03Michael Niedermayer 07master * r8f29901b54 10ffmpeg/libavfilter/vf_yadif.c: 
[00:17] <CIA-41> ffmpeg: yadif: remove get_video_buffer, its redundant now
[00:17] <CIA-41> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[00:17] <CIA-41> ffmpeg: 03Michael Niedermayer 07master * ra0d15027e3 10ffmpeg/libavfilter/vf_yadif.c: 
[00:17] <CIA-41> ffmpeg: yadif: check input video size.
[00:17] <CIA-41> ffmpeg: yadif needs at least 3x3 or it will likely crash with out of array reads
[00:17] <CIA-41> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[00:18] <michaelni> kierank, yadif should be fixed, if you still have crashes please ping me
[00:18] <kierank> michaelni: thank you
[00:56] <Mista_D> Possible to ensure ffprobe values are in the same sequesnce when outputing to CSV ?? Extracting video PID with `ffprobe...-show_streams -print_format csv|grep video|cut -d, -f19`. Sometimes PID is -f16 on MPEG2-TS files.
[02:41] <Compn> http://www.youtube.com/watch?v=pMas0tWc0sg
[04:44] <CIA-41> ffmpeg: 03Michael Niedermayer 07master * r969267482d 10ffmpeg/ffmpeg.c: 
[04:44] <CIA-41> ffmpeg: ffmpeg: Allocate new buffer for bitstream filter when buffer shifted
[04:44] <CIA-41> ffmpeg: fix crash with aac_adtstoasc bitstream filter
[04:44] <CIA-41> ffmpeg: Fixes Ticket1441
[04:44] <CIA-41> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[04:52] <CIA-41> ffmpeg: 03Diego Biurrun 07master * r24d3796d21 10ffmpeg/Makefile: 
[04:52] <CIA-41> ffmpeg: build: add HOSTOBJS to SUBDIR_VARS list
[04:52] <CIA-41> ffmpeg: Even though HOSTOBJS are not referenced directly in subdirectory Makefile
[04:52] <CIA-41> ffmpeg: snippets right now, robustness requires resetting the variable contents.
[04:52] <CIA-41> ffmpeg: 03Diego Biurrun 07master * r36a8c43073 10ffmpeg/libavcodec/qdm2.c: doxygen: qdm2: Drop documentation for non-existing function parameters
[04:52] <CIA-41> ffmpeg: 03Mans Rullgard 07master * r0db9eba48c 10ffmpeg/libavcodec/motion_est.c: 
[04:52] <CIA-41> ffmpeg: motion_est: drop inline from sad_hpel_motion_search()
[04:52] <CIA-41> ffmpeg: This function is only ever called through a function pointer,
[04:52] <CIA-41> ffmpeg: so marking it inline makes no sense.
[04:52] <CIA-41> ffmpeg: Signed-off-by: Mans Rullgard <mans at mansr.com>
[04:52] <CIA-41> ffmpeg: 03Mans Rullgard 07master * r74f82f92a4 10ffmpeg/libavcodec/motion_est.c: 
[04:52] <CIA-41> ffmpeg: motion_est: remove useless no_motion_search() function
[04:52] <CIA-41> ffmpeg: At both places this function is called, mb_[xy] == s->mb_[xy]
[04:52] <CIA-41> ffmpeg: making the call together with following code equivalent to
[04:52] <CIA-41> ffmpeg: simply assigning zeros.
[04:52] <CIA-41> ffmpeg: Signed-off-by: Mans Rullgard <mans at mansr.com>
[04:52] <CIA-41> ffmpeg: 03Mans Rullgard 07master * r5bf7bc625b 10ffmpeg/libavcodec/motion_est.c: 
[04:52] <CIA-41> ffmpeg: motion_est: remove unused macros
[04:52] <CIA-41> ffmpeg: Signed-off-by: Mans Rullgard <mans at mansr.com>
[04:52] <CIA-41> ffmpeg: 03Hendrik Leppkes 07master * rf9150c8ac0 10ffmpeg/libavcodec/lagarith.c: 
[04:52] <CIA-41> ffmpeg: lagarith: frame multithreading
[04:52] <CIA-41> ffmpeg: About 2x speedup going from 1 to 2 threads.
[04:52] <CIA-41> ffmpeg: 1.7s to 0.85s on foreman CIF.
[04:52] <CIA-41> ffmpeg: Signed-off-by: Janne Grunau <janne-libav at jannau.net>
[04:52] <CIA-41> ffmpeg: 03Michael Niedermayer 07master * r633b90ca8d 10ffmpeg/: (log message trimmed)
[04:52] <CIA-41> ffmpeg: Merge remote-tracking branch 'qatar/master'
[04:52] <CIA-41> ffmpeg: * qatar/master:
[04:52] <CIA-41> ffmpeg:  lagarith: frame multithreading
[04:53] <Compn> intermediate codecs making a resurgance
[11:58] <ubitux> saste: thanks, looks much better that way
[11:58] <ubitux> (the doc deps stuff)
[13:31] <Compn> j-b : 
[13:31] <Compn> The vlc ticket was a duplicate of FFmpeg ticket #780 (at least concerning 
[13:31] <Compn> the patch fixing it). The crash was fixed within 24 hours by Michael in 
[13:31] <Compn> December, it is fixed in all release branches since 0.5.
[13:31] <Compn> Carl Eugen
[13:32] <Compn> vlc ticket 6000 that is
[13:32] <Compn> hi carl reading this later on the mailing list ;)
[14:06] <ubitux> wtf haiku
[14:07] <ubitux> perl in /boot/common/bin/perl, but no /usr/bin/env to find it?
[14:07] <ubitux> how are you supposed to write perl scripts?
[14:18] <michaelni> haiku pod2man uses ---> #!/boot/common/bin/perl
[14:19] <ubitux> you mean the packaged pod2man?
[14:19] <ubitux> they patch each app? :/
[14:19] <michaelni> yes, the pad2man included with haiku or packaged or whatevef
[14:20] <michaelni> but #!perl seems working too on haiku
[14:25] <j-b> http://wiki.videolan.org/Bounties
[15:56] <CIA-41> ffmpeg: 03Michael Niedermayer 07master * rd9d0c1ccc3 10ffmpeg/libavcodec/mjpegdec.c: 
[15:56] <CIA-41> ffmpeg: mjpegdec: support AVRn interlaced
[15:56] <CIA-41> ffmpeg: Fixes Ticket1527
[15:56] <CIA-41> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[16:31] <CIA-41> ffmpeg: 03Kostya Shishkov 07master * r84f9d78a96 10ffmpeg/tests/ (7 files in 2 dirs): fate: add G.723.1 decoder tests
[16:31] <CIA-41> ffmpeg: 03Diego Biurrun 07master * r2804ba6a3b 10ffmpeg/libavformat/ (swf.h swfdec.c): swf: Move swf_audio_codec_tags table to the only place it is used
[16:31] <CIA-41> ffmpeg: 03Diego Biurrun 07master * r804d7a1aa6 10ffmpeg/ (5 files in 4 dirs): doxygen: Fix function parameter names to match the code
[16:31] <CIA-41> ffmpeg: 03Diego Biurrun 07master * rc72f12cbae 10ffmpeg/configure: configure: suncc: Add -xc99 to dependency flags, required on Solaris
[16:31] <CIA-41> ffmpeg: 03Michael Niedermayer 07master * r83c93fe74a 10ffmpeg/configure: (log message trimmed)
[16:31] <CIA-41> ffmpeg: configure: Check for a sctp struct instead of just the header
[16:31] <CIA-41> ffmpeg: This fixes build failures on debian/kfreebsd, which has the
[16:32] <CIA-41> ffmpeg: sctp.h header, but it is currently broken (a cpp test succeeds,
[16:32] <CIA-41> ffmpeg: but a compile test fails), see http://bugs.debian.org/684330 for
[16:32] <CIA-41> ffmpeg: details.
[16:32] <CIA-41> ffmpeg: Also remove the checked item from HAVE_LIST, since the corresponding
[16:32] <CIA-41> ffmpeg: 03Mans Rullgard 07master * rd913fd1f00 10ffmpeg/libavutil/internal.h: 
[16:32] <CIA-41> ffmpeg: libavutil: drop fallback definitions of INTxx_MIN/MAX
[16:32] <CIA-41> ffmpeg: This list is incomplete (we also use UINT16_MAX), so there does
[16:32] <CIA-41> ffmpeg: not appear to be any system we care about that needs these.
[16:32] <CIA-41> ffmpeg: Signed-off-by: Mans Rullgard <mans at mansr.com>
[16:32] <CIA-41> ffmpeg: 03Samuel Pitoiset 07master * r5e6001db8f 10ffmpeg/libavformat/rtmpproto.c: 
[16:32] <CIA-41> ffmpeg: rtmp: Factorize the code by adding handle_invoke_result
[16:32] <CIA-41> ffmpeg: Signed-off-by: Luca Barbato <lu_zero at gentoo.org>
[16:32] <CIA-41> ffmpeg: 03Mans Rullgard 07master * r6d3783d160 10ffmpeg/libavformat/ffmenc.c: 
[16:32] <CIA-41> ffmpeg: ffmenc: replace if/abort with assert()
[16:32] <CIA-41> ffmpeg: The condition is trivially true, but keeping the assert() is
[16:32] <CIA-41> ffmpeg: sensible to avoid FFM_HEADER_SIZE ever getting out of sync with
[16:32] <CIA-41> ffmpeg: the actual code.
[16:32] <CIA-41> ffmpeg: Signed-off-by: Mans Rullgard <mans at mansr.com>
[16:46] <j-b> http://wiki.videolan.org/Bounties btw
[16:55] <michaelni> j-b, the page doesnt say how much the bounties are ?
[16:56] <j-b> yet
[16:56] <j-b> michaelni: opinions on that?
[16:56] <michaelni> also frame size changing with mt isnt all that hard the hard part is to have it not exploitable
[16:57] <j-b> From a user point of view, hard or not-hard is meaningless: working or not is what matters :)
[16:57] <michaelni> if you just want it working and dont care if its exploitable, i can do that i think
[16:58] <j-b> I think expoitable is not good :)
[16:58] <j-b> but anyway, we are at the 'put money on tasks' part
[16:58] <Compn> you can make another bounty to make it not exploitable
[16:58] <Compn> :P
[16:58] <j-b> and 'gather important taks'
[16:59] <j-b> the -mt parts are probably between 500 and 1000 ?
[16:59] <nevcairiel> I wonder if its worth to try to figure out how to finish my vc-1i patch
[16:59] <Compn> voxware? people still use it ?
[17:00] <j-b> I hope not
[17:00] <Compn> curious why bounty for it :P
[17:01] <j-b> because it was popular in the past
[17:01] <j-b> but true, this will not be a lucrative one :)
[17:01] <Compn> :)
[17:01] <j-b> compared to DTS MA HD...
[17:02] <nevcairiel> technically its DTS-HD MA
[17:02] <nevcairiel> :p
[17:02] <iive> michaelni: about -mt resize. isn't the right way to hold new decoding until all threads with the old size finish? then reinit the codec and start anew?
[17:03] <j-b> nevcairiel: technically, it is crap :)
[17:03] <nevcairiel> crap or not, its t he goto lossless formats for Blu-rays now
[17:03] <nevcairiel> hm typing is off today
[17:04] <j-b> nevcairiel: indeed, hence the presence on the list
[17:04] <Compn> j-b : as for screengrabbing on windows, what about doing it thru ffmpeg ala this patch > http://ffmpeg.org/pipermail/ffmpeg-devel/2011-January/107042.html
[17:04] <j-b> nevcairiel: fixed.
[17:05] <j-b> screen grabbing is done and easy
[17:05] <j-b> the difficulty is audio dekstop grabbing on Win32 and Mac and UI
[17:05] <Compn> oh
[17:05] <Compn> j-b : does loopback mean you can use vlc as a fake webcam ?
[17:05] <Compn> e.g. to play video and upload it using flash client 
[17:05] <j-b> it means you can capture all the audio from your machine
[17:05] <Compn> like the source is a webcam 
[17:06] <Compn> ok i'm thinking of a different thing then :)
[17:06] <Compn> you want 'screen capture'
[17:07] <j-b> audio output capture, if you prefer
[17:07] <j-b> michaelni: if you have any idea on prices and on other missing ideas, please share
[17:08] <nevcairiel> how about java menus for libbluray? or is that coverd in some gsoc somewhere?
[17:08] <Compn> nevcairiel : you mean a wrapper for a jre ?
[17:08] <j-b> BD-J
[17:08] <j-b> hmm, ok
[17:08] <nevcairiel> Compn: its more then that, the support classes in libbluray also lack a lot of implementation
[17:09] <Compn> ok
[17:09] <michaelni> j-b, aac encoder that beats faac , h265 decoder and encoder
[17:10] <nevcairiel> although this might be somewhat self-serving, i'm hoping to work on that part some time in the future unless someone beats me to it. Not that i have a legal entity to ever receive anything. :p
[17:10] <j-b> hevc is not ok, until GSoC is closed.
[17:11] <Compn> libav has h265 porposal
[17:11] <Compn> in gsoc
[17:11] <michaelni> who is student & mentor ?
[17:11] <Compn> i dunno if it was accepted cuz they never updated the wiki
[17:11] <JEEB> the guy has gotten intra decoding to work AFAIK
[17:11] <JEEB> smarter is the student
[17:12] <JEEB> https://github.com/smarter/libav/commits/hevc
[17:12] <j-b> so bounty on top of GSoC is not a good idea :)
[17:29] <CIA-41> ffmpeg: 03Michael Niedermayer 07master * rf9f11c1000 10ffmpeg/libavformat/ (avformat.h version.h): 
[17:29] <CIA-41> ffmpeg: lavf: return APIC fourcc from query_format() if the codec is supported as an attachment pic
[17:29] <CIA-41> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[17:29] <CIA-41> ffmpeg: 03Michael Niedermayer 07master * r22d804b7f0 10ffmpeg/ffmpeg_opt.c: 
[17:29] <CIA-41> ffmpeg: ffmpeg: dont auto match videostreams with attachment pics.
[17:29] <CIA-41> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[17:29] <CIA-41> ffmpeg: 03Michael Niedermayer 07master * r7fd467f708 10ffmpeg/libavformat/mp3enc.c: 
[17:29] <CIA-41> ffmpeg: mp3/mp2: identify the video codecs as APIC through query_codec.
[17:29] <CIA-41> ffmpeg: Fixes Ticket1408
[17:29] <CIA-41> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[17:33] <Compn> j-b : what about a button on vlc to start saving whatever is playing to a new file ?
[17:33] <Compn> maybe is wish , to make vlc more like vcr :P
[17:34] <j-b> Compn: View->Advanced toolbar
[17:34] <j-b> an then click on the red button
[17:34] <Compn> ah
[17:34] <Compn> neat
[17:36] Action: Compn is a bad user
[18:50] <ubitux> saste: i'd like to continue the work on libdvdcss
[18:50] <ubitux> do you have a local branch you could push on your public git or something?
[18:53] <ubitux> s/libdvdcss/dvd device/
[21:09] <saste> ubitux: i'll push a public repo soon and ping you
[21:11] <ubitux> thank you very much :)
[21:30] <saste> ubitux: http://gitorious.org/~saste/ffmpeg/sastes-ffmpeg/commits/add-dvd-protocol-20120810
[21:31] <ubitux> perfect, thanks!
[23:22] <CIA-41> ffmpeg: 03Stefano Sabatini 07master * ra6d6b8a200 10ffmpeg/libavutil/opt.c: lavu/opt: fix crash in av_opt_next() in case the class has no options
[23:22] <CIA-41> ffmpeg: 03Stefano Sabatini 07master * rad5043e285 10ffmpeg/cmdutils.c: 
[23:22] <CIA-41> ffmpeg: cmdutils: show options in show_help_children only if the context has options
[23:22] <CIA-41> ffmpeg: Avoid to print a pointless name of the context followed by an empty list.
[23:22] <CIA-41> ffmpeg: 03Stefano Sabatini 07master * re874b29296 10ffmpeg/doc/Makefile: 
[23:22] <CIA-41> ffmpeg: doc/Makefile: invoke perl to interpret texi2pod.pl
[23:22] <CIA-41> ffmpeg: Do not rely on shebang directive in the script, whose exact behavior is
[23:22] <CIA-41> ffmpeg: system dependent.
[00:00] --- Sat Aug 11 2012


More information about the Ffmpeg-devel-irc mailing list