[Ffmpeg-devel-irc] ffmpeg-devel.log.20111202

[burek]

[00:03] <michaelni> Daemon404, :)
[00:27] <CIA-41> ffmpeg: 03Carl Eugen Hoyos 07master * r8b08f81949 10ffmpeg/libavcodec/bmp.c: Support transparency in 32bit bmp files.
[01:00] <CIA-41> ffmpeg: 03Michael Niedermayer 07master * r0dd704dbd4 10ffmpeg/libavcodec/vble.c: 
[01:00] <CIA-41> ffmpeg: vble: remove unused variable len.
[01:00] <CIA-41> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[01:00] <CIA-41> ffmpeg: Signed-off-by: Ronald S. Bultje <rsbultje at gmail.com>
[01:00] <CIA-41> ffmpeg: 03Derek Buitenhuis 07master * re46abbcf77 10ffmpeg/libavcodec/vble.c: 
[01:00] <CIA-41> ffmpeg: vble: merge len and val arrays
[01:00] <CIA-41> ffmpeg: There's no reason to use two arrays for this.
[01:00] <CIA-41> ffmpeg: Based off commit 2fea60c60084c4e70d7cef128ea3bca5690ce465
[01:00] <CIA-41> ffmpeg: to FFmpeg by Michael Niedermayer.
[01:00] <CIA-41> ffmpeg: Signed-off-by: Derek Buitenhuis <derek.buitenhuis at gmail.com>
[01:00] <CIA-41> ffmpeg: Signed-off-by: Ronald S. Bultje <rsbultje at gmail.com>
[01:00] <CIA-41> ffmpeg: 03Michael Niedermayer 07master * r317ea97bff 10ffmpeg/libavcodec/vble.c: 
[01:00] <CIA-41> ffmpeg: vble: remove flags copy, its not used in any speed relevant code.
[01:00] <CIA-41> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[01:00] <CIA-41> ffmpeg: Signed-off-by: Ronald S. Bultje <rsbultje at gmail.com>
[01:01] <CIA-41> ffmpeg: Signed-off-by: Martin Storsjö <martin at martin.st>
[01:01] <CIA-41> ffmpeg: 03Martin Storsjö 07master * r6af354436c 10ffmpeg/libavformat/md5proto.c: 
[01:01] <CIA-41> ffmpeg: md5proto: Remove the get_file_handle function
[01:01] <CIA-41> ffmpeg: The private data pointer isn't a file handle, this protocol
[01:01] <CIA-41> ffmpeg: doesn't have any file handle to return.
[01:01] <CIA-41> ffmpeg: Signed-off-by: Martin Storsjö <martin at martin.st>
[01:01] <CIA-41> ffmpeg: 03Martin Storsjö 07master * r02490bf358 10ffmpeg/libavformat/rtmpproto.c: 
[01:01] <CIA-41> ffmpeg: rtmp: Clean up properly if the handshake failed
[01:01] <CIA-41> ffmpeg: This prevents memory leaks if this function returns an error.
[01:01] <CIA-41> ffmpeg: Signed-off-by: Martin Storsjö <martin at martin.st>
[01:01] <CIA-41> ffmpeg: 03Reimar Döffinger 07master * rf28070a123 10ffmpeg/ (5 files in 3 dirs): 
[01:01] <CIA-41> ffmpeg: Add PlayStation Portable PMP format demuxer
[01:01] <CIA-41> ffmpeg: Not yet complete, for demuxing AAC the AAC header must be generated
[01:01] <CIA-41> ffmpeg: manually.
[01:01] <CIA-41> ffmpeg: Possibly the decoder could accept the header as extradata to simplify
[01:01] <CIA-41> (78 lines omitted)
[01:09] <Daemon404> michaelni, the Libav folks asked me to add a FATE test for VBLE, so I did. I'm assuming FFmpeg wants it too.
[01:10] <Daemon404> Who handle's FFmpeg's FATE rsync server?
[01:17] <michaelni> Daemon404, flowers-partial-2MB.avi uploaded to our rsync server, thanks
[01:17] <Daemon404> ok, cool.
[01:18] <ubitux> AVFilterBufferRef has data/linesize/format which are common with AVFilterBuffer
[01:18] <ubitux> and AVFilterBuffer has w and h which are common with AVFilterBufferRefVideoProps
[01:19] <ubitux> but AVFilterBufferRefVideoProps also has data/linesize/format
[01:19] <ubitux> ._.
[01:19] <michaelni> a buffer has its width & height and allocated space
[01:20] <michaelni> a reference in that buffer may have different width/height/linesize, it may be just a part of the whole
[01:21] <ubitux> this what the "crop" comment is about i guess
[01:27] <michaelni> and the ref is common stuff with video/audio in seperate structs
[03:06] <ubitux> can we assume av_read_frame() will return a packet containing exactly one subtitle?
[03:07] <michaelni> normally the demuxer/parser should split things in sane units (frames)
[03:09] <ubitux> according to the doxy we have 1 pkt = 1 frame for video, the audio case is a bit special, but for subtitle nothing is specified
[03:09] <michaelni> Zeranoe, btw, i totally forgot, about the pthread/w32thread default thing, if you received no reply from wbs then iam happy to apply a patch with whatever change you want
[03:09] <ubitux> since a pkt is associated with a pts, i guess we will have 1 pkt = 1 subtitle
[03:11] <michaelni> I think that with the current lack of full subs support the read_frame should be adapted to what makes most sense for a better and more complete implementation
[03:12] <ubitux> ok
[03:12] <ubitux> i wonder how we will deal with karaoke stuff&
[03:15] <ubitux> i also wonder if packetizing the subtitles will make sense with libass
[03:16] <ubitux> ah, ass_process_chunk, one event in matroska format
[03:17] <ubitux> so if we demux an external ass file, we will have to "mux" it before feeding it to the subtitles chain
[03:37] <CIA-41> ffmpeg: 03Carl Eugen Hoyos 07master * rd7fc100694 10ffmpeg/libavcodec/tiffenc.c: Support tiff RGBA encoding.
[03:37] <CIA-41> ffmpeg: 03Carl Eugen Hoyos 07master * r40c36196d5 10ffmpeg/libavcodec/libopenjpegenc.c: Support yuva420p encoding via libopenjpeg.
[03:37] <CIA-41> ffmpeg: 03Carl Eugen Hoyos 07master * rbba8975a9c 10ffmpeg/libavcodec/h264_ps.c: 
[03:37] <CIA-41> ffmpeg: Be less verbose when skipping RBSP data in H264 PPS.
[03:37] <CIA-41> ffmpeg: Fixes ticket #693.
[03:37] <CIA-41> ffmpeg: 03Carl Eugen Hoyos 07master * rfab5973fa5 10ffmpeg/libavcodec/mpeg4videodec.c: 
[03:37] <CIA-41> ffmpeg: Be less verbose when skipping MPEG-4 ASP frames.
[03:37] <CIA-41> ffmpeg: Fixes ticket #560.
[04:08] <Zeranoe> michaelni: Thank you, I haven't had much time as of late so I won't be able to get a patch to you for a while but I will work on it in my free time. Was there any objections to my suggested method?
[04:12] <michaelni> i got no reply sadly
[04:16] <Daemon404> interesting that vble is actually faster when build with clang/llvm
[04:16] <Daemon404> built*
[04:17] <Daemon404> too bad debian's version of GNU gold is busted as shit
[04:17] <Daemon404> so i cant try out LTO
[11:43] <vega13>  If I make a custom avcodeccontext and my mpeg4 packets that I want to decode have extradata, how do I go about setting my avcodeccontext?
[11:44] <vega13>  if I decode it with my avcodeccontext I get the warning: first frame is no keyframe for every frame
[11:44] <vega13>  when I load a non custom avcodecctx with pCodecCtx=pFormatCtx->streams[videoStream]->codec;
[11:44] <vega13> I don't get this error and frames are complete and don't look messed up
[11:44] <vega13> on inspection of difference in data between the two avcodeccontexts it seems the one that works contains extradata..
[11:47] <vega13> What am i doing wrong, it's not possible to use a custom avcodeccontext on a stream like this?
[14:42] <CIA-41> ffmpeg: 03Peter Ross 07master * r1fd6924320 10ffmpeg/libavcodec/pictordec.c: 
[14:42] <CIA-41> ffmpeg: pictordec: support uncompressed images
[14:42] <CIA-41> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[19:02] <Node_164> mjpeg: unsupported coding type (ce)? thx..
[19:03] <gnafu> Node_164: You're welcome.
[19:31] <Compn> Node_164 : is it all different unsupported coding types ?
[19:31] <Compn> like each frame reports a diff coding type ?
[19:31] <Compn> if so, its probably not mjpeg
[22:27] <ubitux> would it make sense to add a AVSubtitle in the AVFilterBuffer* struct?
[22:27] <ubitux> or should i duplicate the "interesting" content, just like it's done with the a/v data?
[22:31] <ubitux> it would be equivalent to put a AVFrame in the filterchain (i wonder why it's not the case actually, maybe because it has too much unrelated information in it?)
[22:55] <CIA-41> ffmpeg: 03Michael Niedermayer 07master * r096701d9b2 10ffmpeg/doc/examples/muxing.c: 
[22:55] <CIA-41> ffmpeg: muxing example: set encoder defaults
[22:55] <CIA-41> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[22:59] <michaelni> ubitux, i think stefano wanted to keep the lavc/lavfi APIs not too entangled
[23:01] <ubitux> ok
[23:01] <ubitux> you need various things of lavc anyway
[23:01] <ubitux> in the source filter at least
[23:01] <ubitux> but well ok
[23:02] <ubitux> it sounds overkill to duplicate the AVSubtitles though (even if it's a subset of the fields)
[23:03] <michaelni> ive no oppinion on this either way, i just remember sastes opposition to passing a AVFrame as param to a lavfi func 
[23:06] <ubitux> may it be related to ABI issues too?
[23:06] <michaelni> could be
[23:07] <michaelni> also AVFRame is alot bigger tham AVSubtitles
[23:33] <CIA-41> ffmpeg: 03Michael Niedermayer 07master * rcf257f30b9 10ffmpeg/doc/examples/muxing.c: 
[23:33] <CIA-41> ffmpeg: doc/example/muxing: fix video timestamps
[23:33] <CIA-41> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[00:00] --- Sat Dec  3 2011