[Ffmpeg-devel-irc] ffmpeg-devel.log.20120721

burek burek021 at gmail.com
Sun Jul 22 02:05:02 CEST 2012


[00:06] <juanmabc> mmm, i think i need some data[0] = ; data[1] = ; data[2] = ; linesize[0] = ; linesize[1] = ; linesize[2] = ; love
[00:06] <juanmabc> the example seems to say the they use a specific YUV420P conversion, and i'm using RGB24
[00:07] <iive> juanmabc: you need to align the data[x] pointers, and also have linesize that is multiple of 16 or 32 (sse or avx)
[00:10] <juanmabc> so i have to allocate a RGB24 of w*h buffer, and pass pointers aligned to data[] and linesize[], isn't it?
[00:10] <juanmabc> testing with random values, i get a green screen, but no warning of speedloss ;P
[00:11] <iive> juanmabc: you first have to calculate the linesize as round up of the width
[00:11] <iive> then use h*linesize to allocate the buffer.
[00:12] <juanmabc> example does size * 3 / 2, for a YUV420
[00:12] <juanmabc> is that my srcfrm too? i mean, default ffmpeg?
[00:13] <juanmabc> to what then i convert to RGB24 power of two for opengl
[00:13] <iive> ffmpeg allocation routines would allocate aligned buffers.
[00:13] <juanmabc> if you check my paste i only use av_mallocs
[00:14] <juanmabc> so then, data[0] = ?
[00:15] <juanmabc> i could guess buffer as it is, the trouble is data[1] and data[2]
[00:16] <iive> i think av_malloc always returns aligned buffer.
[00:17] <juanmabc> using that: http://fpaste.org/MMwU/ i get a green screen, but i'm just happy to scape for first time speedloss warning
[00:17] <iive> rgb24 is packed, so there is only one plane. only data[0]
[00:17] <juanmabc> good, that's the stuff i need to know
[00:17] <juanmabc> (i think)
[00:20] <juanmabc> then data[1] = ? or how can i say just one plane?
[00:27] <kierank> iive: av_image_alloc should do all this
[00:27] <kierank> so it's probably a bug if it doesn't
[00:27] <iive> is it he one in avutil?
[00:28] <iive> some of the routines are moved there.
[00:28] <iive> doesn't sws sample show how to allocate a frame?
[00:29] <kierank> yet it's in avutil
[00:31] <kierank> yes*
[00:31] <juanmabc> again, i can't use av_image_alloc since i'm in 0.10.4
[00:32] <juanmabc> (which for a distro is pretty decent)
[00:32] <juanmabc> stop moving api! ;P
[00:32] <juanmabc> gonna check what does internally
[00:33] <juanmabc> imgutils
[00:34] <juanmabc> buf too dense, gonna go old style
[00:36] <juanmabc> i got data[0]  = buffer; i have to set data1 and data2 or it will warn, and also linesize which would be deduced from data
[00:38] <iive> juanmabc: unfortunately ffmpeg follow the liav fork, and that fork is dedicated on rewriting everything in slightly different way.
[00:41] <Compn> they should just change the git history if they want to get rid of certain people from git blame list :P
[01:10] <juanmabc> so where do i use av_image_alloc before decode_video2 ? i still don't have linesize then
[01:10] <juanmabc> after it, then i'll have to alloc each decode_video2 ?
[01:11] <juanmabc> mmm, i guess, it *writes* to linesize
[01:19] <juanmabc> ok so using av_image_alloc at init, does not solve speedloss, any hints on how to use it if it does all that things
[01:19] <juanmabc> frm->data[1] and 2 % 1 are 8 not 0
[01:19] <juanmabc> s/%16/
[01:20] <juanmabc> kierank: well, you suggested it, are you still around? (sorry for highlighting)
[03:18] <Compn> where is j-b :P
[03:21] <juanmabc> it just ended seeming a bug, decode_video2 does not return an aligned frm for my sws_scale, feel free to keep the link as a reference http://code.google.com/p/openmedialibrary
[04:31] <Hawk-TAO> hi
[04:43] <Hawk-TAO> the function av_opt_set_defaults2(),what dose the "2" stand for?
[04:44] <_Nike_> Hi! With RTSP connection what function or var can I detect the codec? I told if the stream is H264, MPEG4, JPG...
[04:45] <durandal_1707> Hawk-TAO: 2 stands for version
[04:46] <Hawk-TAO> version?
[04:46] <Hawk-TAO> thx
[04:46] <durandal_1707> it just means that other function without 2 is deprecated and going to be replaced with 2 variant on next major API bump
[04:47] <Hawk-TAO> ok,thank you so much
[04:47] <Hawk-TAO> it is hot in your city?
[04:53] <CIA-41> ffmpeg: 03Justin Ruggles 07master * r4cd22b7738 10ffmpeg/libavcodec/alac.c: 
[04:53] <CIA-41> ffmpeg: alac: fix channel pointer assignment for 24 and 32-bit
[04:53] <CIA-41> ffmpeg: Needs to be done separately for each element.
[04:53] <CIA-41> ffmpeg: 03Justin Ruggles 07master * r1749e12f45 10ffmpeg/libavformat/cafdec.c: 
[04:53] <CIA-41> ffmpeg: cafdec: allow larger ALAC magic cookie
[04:53] <CIA-41> ffmpeg: It already skips any extra bytes at the end, and apparently there are some
[04:53] <CIA-41> ffmpeg: samples in the wild with larger 'kuki' chunks.
[04:53] <CIA-41> ffmpeg: 03Anton Khirnov 07master * r5eb9c495af 10ffmpeg/tests/fate/cover-art.mak: FATE: add a test for flac cover art.
[04:53] <CIA-41> ffmpeg: 03Michael Niedermayer 07master * r3c033d00f5 10ffmpeg/: (log message trimmed)
[04:53] <CIA-41> ffmpeg: Merge remote-tracking branch 'qatar/master'
[04:53] <CIA-41> ffmpeg: * qatar/master:
[04:53] <CIA-41> ffmpeg:  libopenjpeg: introduce lowres and lowqual private options
[04:53] <CIA-41> ffmpeg:  FATE: add a test for flac cover art.
[04:53] <CIA-41> ffmpeg:  cafdec: allow larger ALAC magic cookie
[04:53] <CIA-41> ffmpeg:  alac: fix channel pointer assignment for 24 and 32-bit
[04:53] <CIA-41> ffmpeg: 03Luca Barbato 07master * rce64e5bfd1 10ffmpeg/libavcodec/libopenjpegdec.c: 
[04:53] <CIA-41> ffmpeg: libopenjpeg: introduce lowres and lowqual private options
[04:54] <CIA-41> ffmpeg: OpenJPEG can decode in lower resolution or decode only a number
[04:54] <CIA-41> ffmpeg: of enhancement layers.
[04:56] <durandal_1707> michaelni: why you set reduce twice to same value in libopenjpegdec, there was no reason to merge that libav crap
[04:59] <durandal_1707> and setting cp_layer have only sense with full decoding and not with LIMIT_TO_MAIN_HEADER
[05:46] <CIA-41> ffmpeg: 03Paul B Mahol 07master * r296c1b6bd1 10ffmpeg/ (6 files in 3 dirs): libcaca output device
[05:49] <CIA-41> ffmpeg: 03Marton Balint 07master * r016a472009 10ffmpeg/ (4 files in 2 dirs): 
[05:49] <CIA-41> ffmpeg: avcodec: add decode_error_flags field to AVFrame
[05:49] <CIA-41> ffmpeg: Signed-off-by: Marton Balint <cus at passwd.hu>
[05:49] <CIA-41> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[05:49] <CIA-41> ffmpeg: 03Marton Balint 07master * rb99e9ee938 10ffmpeg/libavcodec/ac3dec.c: 
[05:49] <CIA-41> ffmpeg: ac3dec: set decode_error_flags field in ac3 codec
[05:49] <CIA-41> ffmpeg: Signed-off-by: Marton Balint <cus at passwd.hu>
[05:49] <CIA-41> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[05:49] <CIA-41> ffmpeg: 03Piotr Bandurski 07master * r3800facac6 10ffmpeg/libavformat/riff.c: 
[05:49] <CIA-41> ffmpeg: riff: add G723_1 wav tag
[05:49] <CIA-41> ffmpeg: Reveiwed-by: compn <tempn at twmi.rr.com>
[05:49] <CIA-41> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[06:12] <durandal_1707> michaelni: your alac fix does not work for 5.1 sample
[06:14] <durandal_1707> look here https://ffmpeg.org/trac/ffmpeg/ticket/859
[06:43] <CIA-41> ffmpeg: 03Michael Niedermayer 07master * r17352ad315 10ffmpeg/libavcodec/alac.c: 
[06:43] <CIA-41> ffmpeg: alacdec: fix packed sample output with 5.1
[06:43] <CIA-41> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[08:22] <Hawk-TAO> hi
[08:23] <Hawk-TAO> someone is online?
[08:27] <Hawk-TAO> i have a question
[08:28] <Hawk-TAO> the function   if ((ret = s->iformat->read_header(s)) < 0) in utils.c line 642
[08:29] <Hawk-TAO> which function does the read_header() point to?
[08:31] <nevcairiel> that depends on your input format
[08:31] <Hawk-TAO> if avi,where does it point to?
[08:32] <nevcairiel> avi_read_header
[08:34] <Hawk-TAO> ok,thx
[08:37] <Hawk-TAO> if some other format,like mp4 or rmvb,where dose it point to?
[08:40] <nevcairiel> just look at the files in libavformat for those formats, there is a struct at the bottom that defines it
[08:43] <Hawk-TAO> bottom?
[10:18] <CIA-41> ffmpeg: 03Stefano Sabatini 07master * rbeef7a6328 10ffmpeg/doc/outdevs.texi: doc/outdevs: fix @code{} syntax in caca device docs
[10:19] <CIA-41> ffmpeg: 03Stefano Sabatini 07master * rf1a7124d30 10ffmpeg/Changelog: Changelog: add entry for the caca device
[12:31] <CIA-41> ffmpeg: 03Paul B Mahol 07master * rad69e778b7 10ffmpeg/libavformat/iff.c: 
[12:31] <CIA-41> ffmpeg: iff: remove dupe ID_ANNO definition
[12:31] <CIA-41> ffmpeg: Signed-off-by: Paul B Mahol <onemda at gmail.com>
[15:14] <Hawk-TAO> hi
[15:14] <Hawk-TAO> morning
[15:15] <Hawk-TAO>      len = s->read_packet(s->opaque, dst, len); in aviobuf.c,line 413
[15:15] <Hawk-TAO> where does the function pointer read_packet() point to?
[15:17] <durandal_1707> to the relavant demuxer
[15:17] <nevcairiel> if its in avio, to the relevant protocol
[15:18] <Hawk-TAO> which file?
[15:18] <durandal_1707> there are many files
[15:20] <Hawk-TAO> hmm,if reading the avi video header,where does it point to?
[15:28] <Hawk-TAO> is this question so simple?
[15:30] <durandal_1707> Hawk-TAO: you are on wrong channel
[15:32] <Hawk-TAO> i just wanna know how it works
[18:07] <CIA-41> ffmpeg: 03Paul B Mahol 07master * r881e0de640 10ffmpeg/libavformat/iff.c: 
[18:07] <CIA-41> ffmpeg: iff: pass context to av_log_ask_for_sample
[18:07] <CIA-41> ffmpeg: Signed-off-by: Paul B Mahol <onemda at gmail.com>
[18:14] <CIA-41> ffmpeg: 03Paul B Mahol 07master * ra36f19e942 10ffmpeg/libavcodec/libopenjpegdec.c: 
[18:14] <CIA-41> ffmpeg: libopenjpegdec: set cp_reduce value only once
[18:14] <CIA-41> ffmpeg: Signed-off-by: Paul B Mahol <onemda at gmail.com>
[21:12] <CIA-41> ffmpeg: 03Reimar Döffinger 07master * r313d1981ad 10ffmpeg/libavcodec/zerocodec.c: (log message trimmed)
[21:12] <CIA-41> ffmpeg: zerocodec: fix direct rendering.
[21:12] <CIA-41> ffmpeg: Set picture type before calling get_buffer.
[21:12] <CIA-41> ffmpeg: This allows the DR application to make better decisions.
[21:12] <CIA-41> ffmpeg: It also fixes a resource leak in case of missing reference frames
[21:12] <CIA-41> ffmpeg: since it would call get_buffer but never release_buffer.
[21:12] <CIA-41> ffmpeg: Also use FFSWAP to ensure that the AVFrame is properly initialized
[22:32] <CIA-41> ffmpeg: 03Robert Nagy 07master * rb70d89a033 10ffmpeg/ (doc/APIchanges libavfilter/avfilter.h libavfilter/buffer.c): 
[22:32] <CIA-41> ffmpeg: lavfi: add avfilter_unref_bufferp()
[22:32] <CIA-41> ffmpeg: Signed-off-by: Anton Khirnov <anton at khirnov.net>
[22:32] <CIA-41> ffmpeg: 03Anton Khirnov 07master * r51a8049be8 10ffmpeg/libavfilter/vf_aspect.c: 
[22:32] <CIA-41> ffmpeg: vf_aspect: clear AVFilterLink.cur_buf in start_frame().
[22:32] <CIA-41> ffmpeg: The buffer is passed on to the next filter, so we shouldn't keep any
[22:32] <CIA-41> ffmpeg: pointers to it.
[22:32] <CIA-41> ffmpeg: 03Anton Khirnov 07master * rd19499c796 10ffmpeg/libavfilter/vf_gradfun.c: vf_gradfun: don't store two pointers to one AVFilterBufferRef.
[22:32] <CIA-41> ffmpeg: 03Anton Khirnov 07master * rad8b6326aa 10ffmpeg/libavfilter/vf_overlay.c: vf_overlay: don't access a buffer reference that's been given away.
[22:32] <CIA-41> ffmpeg: 03Anton Khirnov 07master * rf493c644cd 10ffmpeg/libavfilter/vf_pad.c: vf_pad: don't give up its own reference to the output buffer.
[22:32] <CIA-41> ffmpeg: 03Anton Khirnov 07master * r9203fa0b32 10ffmpeg/libavfilter/vf_fieldorder.c: vf_fieldorder: don't give up its own reference to the output buffer.
[22:32] <CIA-41> ffmpeg: 03Anton Khirnov 07master * r0393af4f0b 10ffmpeg/libavfilter/vf_settb.c: 
[22:32] <CIA-41> ffmpeg: vf_settb: simplify start_frame().
[22:32] <CIA-41> ffmpeg: Avoid pointless filter ref/unref. Just use the reference passed to
[22:32] <CIA-41> ffmpeg: start_frame().
[22:32] <CIA-41> ffmpeg: 03Anton Khirnov 07master * r4c9080a7ef 10ffmpeg/libavfilter/ (10 files): 
[22:32] <CIA-41> ffmpeg: 03Michael Niedermayer 07master * rf7e0c18549 10ffmpeg/libavfilter/vf_overlay.c: 
[22:32] <CIA-41> ffmpeg: vf_overlay: prevent premature freeing of cur_buf
[22:32] <CIA-41> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[22:32] <CIA-41> ffmpeg: 03Michael Niedermayer 07master * r9e1508f1e7 10ffmpeg/: (log message trimmed)
[22:32] <CIA-41> ffmpeg: Merge remote-tracking branch 'qatar/master'
[22:32] <CIA-41> ffmpeg: * qatar/master:
[22:32] <CIA-41> ffmpeg:  lavfi: unref AVFilterLink.out_buf in ff_end_frame().
[22:32] <CIA-41> ffmpeg:  lavfi: unref AVFilterLink.cur_buf in ff_end_frame().
[22:32] <CIA-41> ffmpeg:  vsrc_testsrc: avoid an unnecessary avfilter_ref_buffer().
[22:32] <CIA-41> ffmpeg:  vf_slicify: clear AVFilterLink.cur_buf in start_frame().
[22:32] <CIA-41> ffmpeg: 03Anton Khirnov 07master * r0fe010a36f 10ffmpeg/libavfilter/vf_delogo.c: vf_delogo: don't store two pointers to one AVFilterBufferRef.
[22:32] <CIA-41> ffmpeg: 03Anton Khirnov 07master * r7155763b78 10ffmpeg/libavfilter/vf_drawtext.c: 
[22:32] <CIA-41> ffmpeg: vf_drawtext: don't give up its own reference to the input buffer.
[22:32] <CIA-41> ffmpeg: Directly forwarding the input buffer to the next filter means that
[22:32] <CIA-41> ffmpeg: drawtext no longer owns any references to it and thus shouldn't refer to
[22:32] <CIA-41> (12 lines omitted)
[23:13] <_Nike_> When I use (avformat_open_input(&context, "rtsp://..." works! when I use: void myfunc (const char *URL)  { avformat_open_input(&context, URL,... } don't work!
[00:00] --- Sun Jul 22 2012


More information about the Ffmpeg-devel-irc mailing list