[Ffmpeg-devel-irc] ffmpeg-devel.log.20120208

Thu Feb 9 02:05:03 CET 2012

[01:13] <Daemon404> Compn, just an fyi: BBB is being payed by google to write a perl wrapper for msvc for libav
[01:13] <Daemon404> re: that ML thread
[01:19] <Compn> Daemon404 : people compile ffmpeg with msvc all the time ... you mean for default compilation ?
[01:19] <Daemon404> um
[01:19] <Daemon404> thats kind of impossible.
[01:20] <Daemon404> every single file uses at the very least, named struct initialization
[01:20] <Daemon404> with msvc does not support.
[01:20] <Daemon404> which*
[01:20] <Daemon404> not to mention the inline gcc-style asm...
[01:20] <kierank> pffft just make them use icc
[01:20] <kierank> write a nice guide and people will use it
[01:20] <Daemon404> icl*
[01:20] <Daemon404> no icc on windows (Except for pengvado's special thingy)
[01:20] <Daemon404> :P
[01:21] <kierank> icl i mean
[01:21] <kierank> i always get those things confused
[01:21] <Daemon404> or do what x264 did
[01:21] <Daemon404> and integrate support
[01:21] <Compn> probably will intergrate support
[01:21] <Daemon404> but me significantly more annoying for libavcodec and friends
[01:21] <Daemon404> might*
[01:22] <Compn> lol
[01:22] <Compn> http://code.google.com/p/ffmpeg-msvc/
[01:23] <Compn> or theres this
[01:23] <Compn> https://launchpad.net/oah
[01:23] <Compn> A Build system and packaging tool for building open source projects with Visual Studio 2008 (Express) for the Microsoft Windows platform.
[01:24] <Compn> http://lists.mplayerhq.hu/pipermail/ffmpeg-devel/2008-March/052320.html
[01:25] <Compn> guess thats not going to well either
[01:25] <Compn> :P
[01:27] <funman> not sure what's the advantage of msvc
[01:28] <Compn> more idiotic apps by 'windows programmers' who cant figure out mingw...
[01:46] <Plorkyeran> vs's debugger is p superior to anything gdb-based that I've found
[01:46] <Plorkyeran> that's about it
[01:47] <Daemon404> ^ this
[01:47] <Daemon404> debugging on linux is balls
[01:59] <iive> Plorkyeran: vs-> visual studio?
[02:00] <Daemon404> yea
[02:00] <iive> unfortunately i never wrote program simple enough to not hang that debugger...
[02:00] <Daemon404> sounds liek doing it wrong
[02:00] <Daemon404> or 1998
[02:00] <Daemon404> :/
[02:01] Action: Daemon404 has never managed to hang it
[02:02] <iive> yes, about that time...
[02:02] <Plorkyeran> I've had issues when attaching it to an already running process, but even that's rare
[02:02] <iive> bad memories last long.
[02:03] <Daemon404> is that why im still traumtize by slackware?
[02:04] <iive> probably, It's still my favorite distro.
[02:04] <Daemon404> i see.
[02:18] <durandal_1707> michaelni: fine if you post patch to remove FF_API_OLD_AVOPTIONS stuff ?
[02:18] <durandal_1707> s/you/me :)))
[02:40] <sacrelege> hi all
[02:43] <sacrelege> I get a "-22" error code back from av_open_input_file(). Where can I lookup the error codes?
[02:48] <durandal_1707> -22 is EINVAL
[02:49] <sacrelege> durandal_1707: thanks for the answer, but how did you find out?
[02:50] <durandal_1707> sacrelege: /usr/include/errno.h
[02:56] <funman> strerror can help too
[03:37] <sacrelege> durandal_1707, funman thx
[06:04] <CIA-40> ffmpeg: 03Ronald S. Bultje 07master * r2dd7a1c030 10ffmpeg/libswscale/ (Makefile input.c swscale.c swscale_internal.h): 
[06:04] <CIA-40> ffmpeg: swscale: Split C input functions into separate file.
[06:04] <CIA-40> ffmpeg: Signed-off-by: Diego Biurrun <diego at biurrun.de>
[06:04] <CIA-40> ffmpeg: 03Aneesh Dogra 07master * rdb7d45237a 10ffmpeg/libavcodec/bytestream.h: 
[06:04] <CIA-40> ffmpeg: bytestream: Add bytestream2 writing API.
[06:04] <CIA-40> ffmpeg: Signed-off-by: Justin Ruggles <justin.ruggles at gmail.com>
[06:04] <CIA-40> ffmpeg: 03Ronald S. Bultje 07master * r2144941069 10ffmpeg/libswscale/ (Makefile output.c swscale.c swscale_internal.h): 
[06:04] <CIA-40> ffmpeg: swscale: split C output functions into separate file.
[06:04] <CIA-40> ffmpeg: Signed-off-by: Diego Biurrun <diego at biurrun.de>
[06:04] <CIA-40> ffmpeg: 03Justin Ruggles 07master * r32c61400c0 10ffmpeg/libavcodec/apedec.c: apedec: use sizeof(field) instead of sizeof(type)
[06:04] <CIA-40> ffmpeg: 03Justin Ruggles 07master * r37390d5cca 10ffmpeg/libavcodec/apedec.c: 
[06:04] <CIA-40> ffmpeg: apedec: allow the user to set the maximum number of output samples per call
[06:04] <CIA-40> ffmpeg: It makes sense in some cases to split up the output packet to save on memory
[06:04] <CIA-40> ffmpeg: usage (ape frames can be very large), but the current/default size is
[06:05] <CIA-40> ffmpeg: arbitrary. Allowing the user to configure this gives more flexibility and
[06:05] <CIA-40> ffmpeg: requires minimal additional code.
[06:05] <CIA-40> ffmpeg: 03Diego Biurrun 07master * r06b320ad78 10ffmpeg/libswscale/sparc/yuv2rgb_vis.c: swscale: K&R formatting cosmetics for SPARC code
[06:05] <CIA-40> ffmpeg: 03Anton Khirnov 07master * r4be1d7dc20 10ffmpeg/tests/ (fate/demux.mak ref/fate/pva-demux): 
[06:05] <CIA-40> ffmpeg: pva-demux test: add -vn
[06:05] <CIA-40> ffmpeg: The output is obviously not supposed to contain video (since only
[06:05] <CIA-40> ffmpeg: -acodec copy is specified), but that only happens because of the way -t
[06:05] <CIA-40> ffmpeg: handling is implemented currently.
[06:05] <CIA-40> ffmpeg: 03Anton Khirnov 07master * rc9af832623 10ffmpeg/avconv.c: 
[06:05] <CIA-40> ffmpeg: avconv: set encoder timebase for subtitles.
[06:05] <CIA-40> ffmpeg: The actual number (1/1000) will probably require some
[06:05] <CIA-40> ffmpeg: discussion/tweaking in the future, but should be good enough for now,
[06:05] <CIA-40> ffmpeg: since the timestamps in AVSubtitle are in this timebase by definition.
[06:05] <CIA-40> ffmpeg: 03Anton Khirnov 07master * r1270e12e49 10ffmpeg/ (16 files in 4 dirs): (log message trimmed)
[06:05] <CIA-40> ffmpeg: avconv: rework -t handling for encoding.
[06:05] <CIA-40> ffmpeg: Current code compares the desired recording time with InputStream.pts,
[06:05] <CIA-40> ffmpeg: which has a very unclear meaning. Change the code to use actual
[06:05] <CIA-40> ffmpeg: timestamps of the frames passed to the encoder.
[06:05] <CIA-40> ffmpeg: 03Anton Khirnov 07master * r6e98390258 10ffmpeg/avconv.c: 
[06:05] <CIA-40> (46 lines omitted)
[08:22] <CIA-40> ffmpeg: 03Michael Niedermayer 07master * rfe380adbfb 10ffmpeg/ffmpeg.c: 
[08:22] <CIA-40> ffmpeg: ffmpeg: add (next_dts/dts) variables.
[08:22] <CIA-40> ffmpeg: Strongly based on Anton Khirnovs code.
[08:22] <CIA-40> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[08:22] <CIA-40> ffmpeg: 03Michael Niedermayer 07master * rf72eaf69ab 10ffmpeg/ffmpeg.c: 
[08:22] <CIA-40> ffmpeg: ffmpeg: Init dts variables in output_packet()
[08:22] <CIA-40> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[08:23] <CIA-40> ffmpeg: 03Michael Niedermayer 07master * r4eea0cfc22 10ffmpeg/ (ffmpeg.c tests/ref/fate/vc1-ism): 
[08:23] <CIA-40> ffmpeg: ffmpeg: use ist->dts instead of passing an argument into transcode_video().
[08:23] <CIA-40> ffmpeg: This makes the code more similar to qatar
[08:23] <CIA-40> ffmpeg: And fixes decoding of the last frame of fate/vc1-ism
[08:23] <CIA-40> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[08:23] <CIA-40> ffmpeg: 03Michael Niedermayer 07master * rf070bf0894 10ffmpeg/ffmpeg.c: 
[08:23] <CIA-40> ffmpeg: ffmpeg: Fix (next)_dts/pts init.
[08:23] <CIA-40> ffmpeg: Based on Anton Khirnovs code.
[08:23] <CIA-40> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[08:23] <CIA-40> ffmpeg: 03Michael Niedermayer 07master * re806ceaa7b 10ffmpeg/ffmpeg.c: 
[08:23] <CIA-40> ffmpeg: ffmpeg: Use the correct (dts) variables for stream copy.
[08:23] <CIA-40> ffmpeg: Strongly based on Anton Khirnovs code.
[08:23] <CIA-40> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[10:09] <Tjoppen> herp. how do I encode multiple video streams now that -newvideo doesn't exist?
[10:10] <Tjoppen> with different options for the streams
[10:11] <Tjoppen> I'm capturing my desktop and webcam and want the former lossless but not the latter
[11:12] <ubitux> Tjoppen: with -map?
[11:13] <Tjoppen> I figured it out - append stream index to options
[11:13] <Tjoppen> manual should probably not mention -new*
[11:14] <Tjoppen> now I'm getting shitty sync though, but I probably have to set VFR mode somehow
[13:22] <Tjoppen> fun:
[13:23] <Tjoppen> "ffmpeg -f alsa -i default -f v4l2 -s 640x480 -i /dev/video0" maintains A/V sync
[13:23] <Tjoppen> "ffmpeg -f v4l2 -s 640x480 -i /dev/video0 -f alsa -i default" does not
[16:13] <salsaman> hi, can anybody help me
[16:13] <salsaman> i wrote a mpegts demuxer for LiVES
[16:13] <salsaman> its kind of working ok, the top part of images are showing up ok
[16:13] <salsaman> but the bottom part is messed up
[16:13] <salsaman> and i see errors like
[16:14] <salsaman> [mpeg2video @ 0x3148420] ac-tex damaged at 80 52
[16:14] <salsaman> and less commonly
[16:14] <salsaman> [mpeg2video @ 0x3148420] invalid mb type in I Frame at 77 57
[16:14] <salsaman> its odd because the got_picture flag is being set 
[16:15] <salsaman> i would love to get this working to have mpegts support directly in LiVES
[16:28] <CIA-40> ffmpeg: 03Michael Niedermayer 07master * rc0be67cac1 10ffmpeg/ffmpeg.c: 
[16:28] <CIA-40> ffmpeg: ffmpeg: use (next_)dts in timestamp discontinuity detection code.
[16:28] <CIA-40> ffmpeg: Strongly based on Anton Khirnovs code.
[16:28] <CIA-40> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[16:28] <CIA-40> ffmpeg: 03Michael Niedermayer 07master * r8ea81805d4 10ffmpeg/ffmpeg.c: 
[16:28] <CIA-40> ffmpeg: ffmpeg: print more details and rescaled timestamps in debug code.
[16:28] <CIA-40> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[16:28] <CIA-40> ffmpeg: 03Michael Niedermayer 07master * r7665da46ce 10ffmpeg/ffmpeg.c: 
[16:28] <CIA-40> ffmpeg: ffmpeg: update ist->pts in transcode_audio with the audio decoders output.
[16:28] <CIA-40> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[16:28] <CIA-40> ffmpeg: 03Michael Niedermayer 07master * r6028bd0dc6 10ffmpeg/ffmpeg.c: 
[16:28] <CIA-40> ffmpeg: ffmpeg: move next_dts/pts update out of transcode_video()
[16:28] <CIA-40> ffmpeg: this makes it placed similar to qatar
[16:28] <CIA-40> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[16:55] <CIA-40> ffmpeg: 03Michael Niedermayer 07master * r67700e877d 10ffmpeg/ffmpeg.c: 
[16:55] <CIA-40> ffmpeg: ffmpeg: update dts variables in transcode_audio()
[16:55] <CIA-40> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[16:55] <CIA-40> ffmpeg: 03Anton Khirnov 07master * rd97fa416d6 10ffmpeg/ffmpeg.c: (log message trimmed)
[16:55] <CIA-40> ffmpeg: ffmpeg: reduce overloading for InputStream.pts.
[16:55] <CIA-40> ffmpeg: It currently has different meanings at different times (dts of the last
[16:55] <CIA-40> ffmpeg: read packet/pts of the last decoded frame). Reduce obfuscation by
[16:55] <CIA-40> ffmpeg: storing pts of the decoded frame in the frame itself.
[16:55] <CIA-40> ffmpeg: Conflicts:
[16:55] <CIA-40> ffmpeg:  ffmpeg.c
[17:41] <CIA-40> ffmpeg: 03Anton Khirnov 07master * r67c63176ad 10ffmpeg/ffmpeg.c: 
[17:41] <CIA-40> ffmpeg: ffmpeg: remove setting fb->pts.
[17:41] <CIA-40> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[17:41] <CIA-40> ffmpeg: 03Michael Niedermayer 07master * re7f4de34ad 10ffmpeg/ffmpeg.c: 
[17:41] <CIA-40> ffmpeg: ffmpeg: use ist->dts in do_streamcopy()
[17:41] <CIA-40> ffmpeg: Based on change by Anton Khirnov.
[17:41] <CIA-40> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[17:41] <CIA-40> ffmpeg: 03Michael Niedermayer 07master * r160008a60f 10ffmpeg/ffmpeg.c: 
[17:41] <CIA-40> ffmpeg: ffmpeg: use ist->dts in rate_emu_sleep()
[17:41] <CIA-40> ffmpeg: Based on change by Anton Khirnov.
[17:41] <CIA-40> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[17:41] <CIA-40> ffmpeg: 03Michael Niedermayer 07master * rfba318a4b0 10ffmpeg/ffmpeg.c: 
[17:41] <CIA-40> ffmpeg: ffmpeg: update (next_)pts in the stream copy case
[17:41] <CIA-40> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[17:55] <ubitux> michaelni: do you plan to do more stuff on ffmpeg.c?
[17:55] <ubitux> i'm playing more with the vf and do_video_out()
[17:55] <ubitux> should i rebase now or wait more?
[18:00] <michaelni> ubitux, i dont plan major things but i see some bugs that id like to fix and until i fix them i dont know how much changes that will need
[18:00] <ubitux> ok
[18:48] <michaelni> ubitux, which of the patches you posted need a review and which have already been reviewed ?
[19:09] <ubitux> > 4 need review
[19:10] <ubitux> the first 4 are just the old ones rebased
[19:10] <ubitux> though, you can still review the unreviewed patch2 :)
[19:48] <CIA-40> ffmpeg: 03Clément BSsch 07master * r1aa8f5fcfb 10ffmpeg/libavfilter/libmpcodecs/vf_fspp.c: 
[19:48] <CIA-40> ffmpeg: vf_fspp: use LOCAL_MANGLE instead of MANGLE.
[19:48] <CIA-40> ffmpeg: The symbols are locals, thus the EXTERN_PREFIX should be omitted. This
[19:48] <CIA-40> ffmpeg: might fix the llvm-gcc fate instance.
[20:51] <burek> can people burn subs directly on video stream with ffmpeg?
[20:51] <burek> i know i did it in past month, but i dont remember did i use ffmpeg or not
[20:53] <durandal_1707> burek: you mean burn bitmap - hardcoded subs?
[20:53] <burek> yes, i figured libass :)
[20:53] <burek> but thanks anyway :)
[20:53] <burek> :beer:
[20:53] <durandal_1707> libass is for rendering ass subtitles
[20:54] <burek> and drawing it onto video
[20:54] <burek> its a video filter
[20:54] <burek> I'm waiting for the forum to finally get online
[20:54] <nevcairiel> no, libass is only the renderer
[20:54] <burek> so I can write those tutorials
[20:54] <burek> http://ffmpeg.org/ffmpeg.html#ass
[20:54] <burek> Draw ASS (Advanced Substation Alpha) subtitles on top of input video using the libass library.
[20:54] <durandal_1707> but didn't you ask for burned ones - in which case you don't need libass at all
[20:55] <nevcairiel> the drawing is not done with libass
[20:55] <nevcairiel> thats all ffmpeg
[20:55] <burek> i dont understand the question
[20:55] <burek> burek> can people burn subs directly on video stream with ffmpeg?
[20:55] <burek> i found out they can
[20:55] <durandal_1707> how?
[20:56] <burek> using -vf "ass=title.ass" ...
[20:56] <durandal_1707> ahh that is really nice
[20:56] <burek> i did it a month ago
[20:57] <CIA-40> ffmpeg: 03Paul B Mahol 07master * r3c939ad6e0 10ffmpeg/libavcodec/lagarith.c: 
[20:57] <CIA-40> ffmpeg: lagarith: refactor code
[20:57] <CIA-40> ffmpeg: Signed-off-by: Paul B Mahol <onemda at gmail.com>
[20:57] <CIA-40> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[21:26] <Compn> burek : ass works now? :)
[21:27] Action: Compn wonders what ubitux is up to
[21:27] <burek> works :)
[21:29] <gnafu> dat ass
[21:30] Action: Compn busy begging people for patches
[21:49] <burek> "how can i show the supported indevs in ffmpeg?"
[21:49] <burek> :)
[21:49] <burek> is there any switch, like -codecs, -formats ?
[21:50] <Tjoppen> -devices maybe?
[21:50] <Tjoppen> hm, no
[21:54] <durandal_1707> -protocols - beware output is ugly (there where times when it was better)
[21:54] <burek> i mean like alsa, oss
[21:54] <burek> and stuff
[21:55] <burek> just like when you do configure
[21:55] <durandal_1707> burek: it is in -formats with bunch of other stuff
[21:58] <burek> durandal_1707, thanks :)
[22:04] <burek> how to set pix_fmt MJPG
[22:04] <burek> it always reports something like failed to set value 'MJPEG' for option 'pix_fmt'
[22:04] <burek> is pix_fmt the correct option for setting input pixel format (From webcam for ex.)
[22:08] <durandal_1707> burek: you dont need to set pix_fmt for MJPEG
[22:10] <burek> well there is a webcam
[22:10] <ubitux> Compn: yes i haven't done much for the subtitles stuff recently, except the jacosub support :)
[22:10] <burek> that can be accessed using yuyv or mjpeg
[22:10] <burek> so i try to explicitly set mjpg
[22:11] <burek> v4l2-ctl --list-formats-ext reports pixel format MJPEG which can capture video in higher resolutions with higher fps
[22:11] <durandal_1707> than set -input_format
[22:11] <nevcairiel> mjpeg is not a pixel format, its a codec, you would have to go differently about telling ffmpeg that, but don't ask me if it even supports  capturing mjpeg from a v4l device
[22:12] <nevcairiel> hm
[22:12] <burek> durandal_1707, thank
[22:12] <burek> s
[22:12] <nevcairiel> apparently it does
[22:12] <ubitux> some webcam have a mjpeg output
[22:12] <durandal_1707> burek: actually pixel_format would work too, but pix_fmt will not
[22:12] <nevcairiel> i just peeked into v4l2.c, and it lists some input formats with mjpeg
[22:12] <ohsix> tons of them do, maybe all that you can buy at the moment new :]
[22:13] <burek> oh :) pixel_format and pix_fmt are not the same? :)
[22:13] <ohsix> (and logitech just released their first 1080p camera that does h264 a little while ago)
[22:13] <nevcairiel> pixel_format is internally just an alias for input_format, funny enough
[22:13] <ohsix> here's hoping everyone follows
[22:13] <burek> :)
[22:13] <burek> cool :)
[22:13] <burek> thanks :) :beer: :)
[23:00] <durandal_1707> heh that patch actually broke all darwins
[23:01] <ubitux> oh the fspp one?
[23:01] <ubitux> mmh why&
[23:01] <durandal_1707> not all, at least not yet ;)
[23:01] <ubitux> i'm going to revert, but any idea what could go wrong?
[23:02] <ubitux> btw, can mplayer be build out of the box on those plateform?
[23:02] <durandal_1707> dunno
[23:03] <durandal_1707> looks like ld could not find some symbols..
[23:03] <burek> one stupid question, how can i checkout this exact version of vlc from git: VLC media player 2.1.0-git Rincewind (revision 1.3.0-git-832-gd6a7a4d)
[23:04] <durandal_1707> burek: why it have something special?
[23:04] Action: durandal_1707 is very curious
[23:06] <burek> well, it doesn't crash :D
[23:07] <durandal_1707> ubitux: as I can see problem is in libmpcodecs, so do not waste your time on that
[23:07] <durandal_1707> better waste it in porting remaining mp filters to lavfi
[23:08] <ubitux> i'm curious about the issue
[23:08] <durandal_1707> well than you would need darwin box
[23:09] <ubitux> yes, that would be nice.
[23:11] <ubitux> i guess the "extern" mangle is needed because there is some sym resolution in ffmpeg
[23:12] <ubitux> michaelni: any idea before i'm reverting?
[23:20] <ubitux> well, can be figured it out later
[23:20] <CIA-40> ffmpeg: 03Clément BSsch 07master * rc84acc31b2 10ffmpeg/libavfilter/libmpcodecs/vf_fspp.c: 
[23:20] <CIA-40> ffmpeg: Revert "vf_fspp: use LOCAL_MANGLE instead of MANGLE."
[23:20] <CIA-40> ffmpeg: This reverts commit 1aa8f5fcfba6701eb0181ea07c070d09ab0f4efd. It broke
[23:20] <CIA-40> ffmpeg: the other x86_32 fate instances.
[23:24] <ubitux> :(
[23:53] <CIA-40> ffmpeg: 03Michael Niedermayer 07master * rd8710228ea 10ffmpeg/ffmpeg.c: 
[23:53] <CIA-40> ffmpeg: ffmpeg: change sync_opts code to be more similar to qatar.
[23:53] <CIA-40> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[00:00] --- Thu Feb  9 2012