[Ffmpeg-devel-irc] ffmpeg-devel.log.20120211

burek burek021 at gmail.com
Sun Feb 12 02:05:03 CET 2012


[00:54] Action: Compn grumbles something about bugzilla not returning full bug list
[00:54] <Compn> http://bugzilla.libav.org/buglist.cgi?query_format=advanced&short_desc=%20&bug_status=NEW&bug_status=ASSIGNED&bug_status=REOPENED&short_desc_type=allwordssubstr
[00:57] <gnafu> Compn: http://bugzilla.libav.org/buglist.cgi?order=bug_id
[01:35] <CIA-67> ffmpeg: 03Martin Storsjö 07master * re7d4a2a721 10ffmpeg/libavcodec/bitstream_filter.c: 
[01:35] <CIA-67> ffmpeg: libavcodec: Don't do av_free(av_malloc(0)) for bitstream filters
[01:35] <CIA-67> ffmpeg: This fixes crashes on exit when closing a bitstream filter that
[01:35] <CIA-67> ffmpeg: hasn't allocated any private data, on OS X.
[01:35] <CIA-67> ffmpeg: Signed-off-by: Martin Storsjö <martin at martin.st>
[01:35] <CIA-67> ffmpeg: 03Alex Converse 07master * r37bed6ff3f 10ffmpeg/ (4 files in 2 dirs): 
[01:35] <CIA-67> ffmpeg: aacdec: Try to sniff a reasonable channel layout for PCE based configurations.
[01:35] <CIA-67> ffmpeg: This changes the output order of multichannel PCE based streams.
[01:35] <CIA-67> ffmpeg: 03Alex Converse 07master * rf1ecd0802d 10ffmpeg/libavcodec/ (aacdec.c aacdectab.h): aacdec: Support native channel layout when requested.
[01:35] <CIA-67> ffmpeg: 03Alex Converse 07master * r7b05025856 10ffmpeg/libavcodec/aacdec.c: aacdec: Unify preconfigured layout and PCE layout.
[01:36] <CIA-67> ffmpeg: 03Michael Niedermayer 07master * re6d9fa66f1 10ffmpeg/libavcodec/ (ac3_parser.c ac3dec.c): (log message trimmed)
[01:36] <CIA-67> ffmpeg: ac3dec: Move center and surround mix level tables to the parser.
[01:36] <CIA-67> ffmpeg: That way all mix levels as exported by avpriv_ac3_parse_header()
[01:36] <CIA-67> ffmpeg: will have the same meaning.
[01:36] <CIA-67> ffmpeg: Previously the 3-bit center mix level for E-AC-3 was used to index in a
[01:36] <CIA-67> ffmpeg: 4-entry table, leading to out-of-array reads.
[01:36] <CIA-67> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[01:36] <CIA-67> ffmpeg: 03Justin Ruggles 07master * rb758cf7343 10ffmpeg/libavcodec/utils.c: avcodec: set avpkt->size to 0 if encode2() did not output a packet
[01:36] <CIA-67> ffmpeg: 03Alex Converse 07master * r0ab3687924 10ffmpeg/libavformat/dv.c: 
[01:36] <CIA-67> ffmpeg: dv: Fix small overread in audio frequency table.
[01:36] <CIA-67> ffmpeg: Found-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
[01:36] <CIA-67> ffmpeg: 03Justin Ruggles 07master * rd483bb58c3 10ffmpeg/libavcodec/x86/ac3dsp.asm: 
[01:36] <CIA-67> ffmpeg: ac3dsp: do not use pshufb in ac3_extract_exponents_ssse3()
[01:36] <CIA-67> ffmpeg: We need to do unsigned saturation in order to cover the corner case when the
[01:36] <CIA-67> ffmpeg: absolute coefficient value is 16777215 (the maximum value).
[01:36] <CIA-67> ffmpeg: Fixes Bug #216
[01:36] <CIA-67> ffmpeg: 03Justin Ruggles 07master * ra75bc764ec 10ffmpeg/ (avconv.c libavcodec/utils.c): 
[01:36] <CIA-67> ffmpeg: avcodec: for audio encoding, set packet dts to packet pts.
[01:36] <CIA-67> ffmpeg: There are no audio encoders which do frame reordering.
[01:36] <CIA-67> ffmpeg: 03Ronald S. Bultje 07master * r45b7bd7c53 10ffmpeg/libavcodec/ (h264.c h264.h h264_cabac.c h264_cavlc.c svq3.c): (log message trimmed)
[01:36] <CIA-67> ffmpeg: h264: disallow constrained intra prediction modes for luma.
[01:36] <CIA-67> ffmpeg: Conversion of the luma intra prediction mode to one of the constrained
[01:36] <CIA-67> ffmpeg: ("alzheimer") ones can happen by crafting special bitstreams, causing
[01:36] <CIA-67> ffmpeg: a crash because we'll call a NULL function pointer for 16x16 block intra
[01:36] <CIA-67> ffmpeg: prediction, since constrained intra prediction functions are only
[01:36] <CIA-67> ffmpeg: implemented for chroma (8x8 blocks).
[01:36] <CIA-67> ffmpeg: 03Alex Converse 07master * r81749f30cd 10ffmpeg/libavcodec/ (dvdata.c dvdata.h): dv: Move tables from dvdata.h to dvdata.c
[01:36] <CIA-67> ffmpeg: 03Paul B Mahol 07master * rdcd2b55e1a 10ffmpeg/libavcodec/utils.c: 
[01:36] <CIA-67> ffmpeg: Check AVCodec.pix_fmts in avcodec_open2()
[01:36] <CIA-67> ffmpeg: Signed-off-by: Paul B Mahol <onemda at gmail.com>
[01:36] <CIA-67> ffmpeg: Signed-off-by: Justin Ruggles <justin.ruggles at gmail.com>
[01:36] <CIA-67> ffmpeg: 03Alex Converse 07master * r9e1db721c4 10ffmpeg/libavcodec/svq3.c: 
[01:36] <CIA-67> ffmpeg: svq3: Prevent illegal reads while parsing extradata.
[01:36] <CIA-67> ffmpeg: Found-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
[01:36] <CIA-67> ffmpeg: 03Joakim Plate 07master * r3371e89640 10ffmpeg/libavformat/bink.c: 
[01:36] <CIA-67> ffmpeg: bink: Check return value of avio_seek and avoid modifying state if it fails
[01:36] <CIA-67> ffmpeg: Signed-off-by: Paul B Mahol <onemda at gmail.com>
[01:36] <CIA-67> ffmpeg: Signed-off-by: Justin Ruggles <justin.ruggles at gmail.com>
[01:36] <CIA-67> ffmpeg: 03Joakim Plate 07master * r066ad0926a 10ffmpeg/libavformat/tmv.c: 
[01:36] <CIA-67> ffmpeg: tmv: Check return value of avio_seek and avoid modifying state if it fails
[01:36] <CIA-67> ffmpeg: Signed-off-by: Paul B Mahol <onemda at gmail.com>
[01:36] <CIA-67> ffmpeg: Signed-off-by: Justin Ruggles <justin.ruggles at gmail.com>
[01:36] <CIA-67> ffmpeg: 03Joakim Plate 07master * r8dcd00eabd 10ffmpeg/libavformat/nsvdec.c: 
[01:37] <CIA-67> ffmpeg: nsvdec: Check return value of avio_seek and avoid modifying state if it fails
[01:37] <CIA-67> ffmpeg: Signed-off-by: Paul B Mahol <onemda at gmail.com>
[01:37] <CIA-67> ffmpeg: Signed-off-by: Justin Ruggles <justin.ruggles at gmail.com>
[01:37] <CIA-67> ffmpeg: 03Paul B Mahol 07master * rc388558d31 10ffmpeg/libavcodec/wavpack.c: 
[01:37] <CIA-67> ffmpeg: wavpack: allow user to disable CRC checking
[01:37] <CIA-67> ffmpeg: Signed-off-by: Paul B Mahol <onemda at gmail.com>
[01:37] <CIA-67> ffmpeg: Signed-off-by: Justin Ruggles <justin.ruggles at gmail.com>
[01:37] <CIA-67> ffmpeg: 03Joakim Plate 07master * rd7c11b114b 10ffmpeg/libavformat/ffmdec.c: 
[01:37] <CIA-67> ffmpeg: ffmdec: Check return value of avio_seek and avoid modifying state if it fails
[01:37] <CIA-67> ffmpeg: Signed-off-by: Paul B Mahol <onemda at gmail.com>
[01:37] <CIA-67> (9 lines omitted)
[02:40] <CIA-67> ffmpeg: 03Ingo Brückl 07master * ra697a185e4 10ffmpeg/libavformat/libavformat.v: 
[02:40] <CIA-67> ffmpeg: buildsys: Fix shared lib build of MPlayer.
[02:40] <CIA-67> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[02:40] <CIA-67> ffmpeg: 03Michael Niedermayer 07master * red83f97235 10ffmpeg/libavcodec/tta.c: 
[02:40] <CIA-67> ffmpeg: tta: Add some safety precautions to avoid freeing things that have not been allocated.
[02:40] <CIA-67> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[02:40] <CIA-67> ffmpeg: 03Paul B Mahol 07master * rc6abf21492 10ffmpeg/libavcodec/tta.c: 
[02:40] <CIA-67> ffmpeg: tta: do not leak memory if bps == 3
[02:40] <CIA-67> ffmpeg: In bps == 3 case output is decoded directly.
[02:40] <CIA-67> ffmpeg: Signed-off-by: Paul B Mahol <onemda at gmail.com>
[02:40] <CIA-67> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[05:54] <lionaneesh> Hey
[06:06] <michaelni> hey
[06:10] <lionaneesh> michaelni, Whats your TZ
[06:10] <lionaneesh> TZ = Time Zone
[08:34] <CIA-67> ffmpeg: 03Michael Niedermayer 07master * rfd0b8d5986 10ffmpeg/libavcodec/ansi.c: 
[08:34] <CIA-67> ffmpeg: ansi: Fix use of uninitalized width/height warning.
[08:34] <CIA-67> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[16:24] <^prelude^> hello
[16:24] <^prelude^> anyone here have experience with GPU encoding
[16:24] <^prelude^> i am a bit confused by that and I need some pointing in the right direction
[16:40] <Compn> ^prelude^ : i heard videolan had some exp with that
[16:40] <Compn> might want to ask in their chan
[17:04] <JEEB> ^prelude^, most GPU-using encoders suck, as you really can't do most already set formats with good efficiency with them (GPUs tend to need o9k threads for the thing to be fully utilized). Of course, that doesn't say that there are zero things where the GPU couldn't be used. There's the lookahead and motion estimation in H.264 encoding for example. The problem of course is that a good implementation will need time and money, it's not exactly easy t
[17:04] <JEEB> o make an efficient algorithm.
[17:05] <JEEB> There's currently some folk doing an opencl lookahead for x264, which is the first really usable'ish thing that's popped up in years.
[17:44] <Compn> JEEB : would you say the problem was the gpu manf arent releasing specs/api/drivers for people to utilize ?
[17:47] <nevcairiel> the GPUs have their own full decoder stacks, which are very fast, but usually dont make people happy quality wise
[17:48] <nevcairiel> partial acceleration has to be implemented manually, and stuff like OpenCL simply only works very fast with extreme parallel processing, which is not easy to find
[17:48] <nevcairiel> decoder=encoder
[17:48] <JEEB> Compn, that's usually a problem with decoding, with encoding you have all the CUDA/OpenCL crap which in theory is "free for you to use".
[17:48] <Compn> lol
[17:49] <JEEB> And yes, nev now said pretty much what I said
[17:49] <JEEB> needs lots o' parallel stuff (threads)
[17:49] <JEEB> to keep the GPU busy
[17:49] <Compn> i see
[17:49] <JEEB> the driver pretty much needs to have data already in line to be fed the thing
[17:49] <nevcairiel> Sure could use QuickSync etc to encode things, but the quality is below x264, even though the speed may be much higher
[17:49] <Compn> probably a hw h264 capture card would work better :P
[17:50] <JEEB> the last I had someone test quicksync
[17:50] <JEEB> it was ~ultrafast on QP encoding
[17:50] <JEEB> on a i5
[17:50] <JEEB> http://phoronix.com/forums/showthread.php?50697-Test-tool-for-snb-h264-encoder-in-libva-git&p=205580#post205580
[17:50] <JEEB> s/ultrafast/superfast/
[17:50] <iive> I would think that an encoder allows a lot of places for parallel processing.
[17:50] <JEEB> ultrafast was 2x the speed
[17:51] <JEEB> but yes, intel's hw encoder ASIC seems to be the most useful thing if you absolutely positively want something like that
[17:52] <nevcairiel> did that ever say how quality compares quicksync vs x264 superfast?
[20:39] <^sandro^-> hello
[20:39] <^sandro^-> anyone around?
[21:17] <iive> no
[22:51] <burek> can someone please help with static build of ffmpeg..? http://pastebin.com/zzhqVhaS
[22:51] <burek> the main problem is this
[22:51] <burek> /usr/lib/libasound.a(dlmisc.o): In function `snd_dlopen': dlmisc.c:(.text+0xe4): undefined reference to `dlopen'
[22:52] <burek> but I compiled libasound statically and even glibc statically
[22:52] <burek> why doesn't it recognize it :/
[22:52] <^sandro^-> hello
[22:52] <^sandro^-> anyone around that can help with UDP capture
[22:52] <^sandro^-> raw vide input from Sat. 
[22:53] <^sandro^-> im having a hard time transcoding it to H264 > . TS
[22:53] <burek> try in #ffmpeg ?
[22:53] <^sandro^-> i am trying to create a live HLS stream
[22:55] <burek> I can't wait for the forum to get online, because this way I'm never gonna make it to finally build ffmpeg statically :s
[22:56] <durandal_1707> burek: funny, dlopen is call for dynamic lib
[22:57] <burek> well i guess
[22:58] <burek> so, what do you suggest durandal_1707 :/
[22:59] <durandal_1707> first find out if alsa can be statically compiled at all
[22:59] <funman> add -ldl
[22:59] <burek> funman, there is -ldl :)
[22:59] <durandal_1707> and that heavily depends on kernel so i see no point to build that statically
[23:00] <burek> you mean, I should drop alsa support?
[23:00] <durandal_1707> no it should really be shared
[23:00] <funman> add -ldl after -lasound?
[23:00] <durandal_1707> because it will not magically work better if it is staticaly in on incompatible kernels anyway
[23:01] <burek> durandal_1707, I need it to be static :(
[23:01] <burek> funman, how can i do it using ./configure script
[23:01] <nevcairiel> --extra-ldflags="-ldl" probably
[23:01] <burek> durandal_1707, well, ok, I understand that, but if it works for latest kernels thats good enough
[23:02] <burek> ./configure --enable-static --disable-shared --enable-gpl --enable-libx264 --enable-nonfree --enable-libaacplus --enable-libmp3lame --enable-libvorbis --enable-libtheora --enable-gray --enable-vdpau --enable-bzlib --enable-libass --enable-libfreetype --enable-libopenjpeg --enable-libspeex --enable-libxvid --enable-x11grab --enable-version3 --enable-libopencore-amrnb --enable-libopencore-amrwb --enable-runtime-cpudetect --disable-doc --disable-ffser
[23:02] <burek> ver --disable-ffprobe --extra-cflags='-I/usr/local/glibc-static/include/ -I/usr/local/include' --extra-ldflags='-static -L/usr/local/glibc-static/lib -L/usr/local/lib -L/usr/lib -ldl'
[23:02] <burek> that's the line for configure
[23:03] <burek> I'd like to make it static to be able to just "download and run"
[23:03] <burek> without much pain
[23:03] <nevcairiel> i was wondering something else the other day, can you compile a dynamic version of the ffmpeg libraries, but include statically linked versions of some of the 3rd party libraries in them? :d
[23:03] <nevcairiel> without manually hackery
[23:04] <burek> well, I guess you should tell ffmpeg explicitly which lib you'd like to link statically
[23:04] <burek> so I guess it won't work with the standard ./configure
[23:05] <iive> burek: can you try `pkg-config --static --libs alsa`
[23:05] <burek> yes just as ec
[23:05] <burek> -lasound -lm -ldl -lpthread -lrt
[23:06] <iive> if it fails again, you may want to try adding some of these libraries, or completely disable alsa support (audio recording i guess)
[23:06] <burek> as extra0kdfkags>
[23:06] <burek> omg
[23:06] <burek> extra-ldflags*
[23:07] <burek> right?
[23:09] <burek> iive, i tried with --extra-ldflags='-s
[23:09] <burek> tatic -L/usr/local/glibc-static/lib -L/usr/local/lib -L/usr/lib -lasound -lm -ldl -lpthread -lrt'
[23:09] <burek> same error :/
[23:09] <burek> this is really frustrating :S
[23:10] <^sandro^-> anyone know how to select the propert map input
[23:10] <^sandro^-> Input #0, mpegts, from 'udp://@:5000':
[23:10] <^sandro^->   Duration: N/A, start: 31060.480611, bitrate: 41672 kb/s
[23:10] <^sandro^->   Program 54 
[23:10] <^sandro^->     Metadata:
[23:10] <^sandro^->       service_name    : BN BOSNIA
[23:11] <^sandro^-> oops
[23:11] <^sandro^-> sorry.. 
[23:11] <iive> use pastebin site to paste long texts
[23:11] <^sandro^-> wasn't supposed to do that.. in any case.. any way to capture it without -map 0:0 .. problem is that it keeps changing and what captues one time doesn't the next
[23:11] <^sandro^-> yes that was copied and i hit the paste button as i was typing :(
[23:12] <iive> also, this is development channel, not usage one
[23:12] <^sandro^-> nobody seems to know anything about the mpeg2video stuff :(
[23:12] <^sandro^-> ive tried both channels .. nada
[23:13] <durandal_1707> aren'y you need to use ffprobe and write some kind of script to parse output...
[23:14] <iive> you won't get more help by been impudent. sometimes you need more patience.
[23:46] <CIA-67> ffmpeg: 03Carl Eugen Hoyos 07master * rcadd9ab6aa 10ffmpeg/libavcodec/ffv1.c: 
[23:46] <CIA-67> ffmpeg: Cosmetics: Move a block in ffv1's encode_init().
[23:46] <CIA-67> ffmpeg: Makes the next patch smaller.
[23:46] <CIA-67> ffmpeg: 03Carl Eugen Hoyos 07master * r2396206fb4 10ffmpeg/libavcodec/ffv1.c: ffv1: Assume encoding all bits is wanted if bits_per_raw_sample == 0.
[23:57] <burek> durandal_1707, I have libasound.a in /usr/lib/
[23:57] <burek> i compiled it, so alsa can be built statically
[23:58] <burek> when the forum is up, I'll create a thread for compiling ffmpeg statically and I really hope we can make it
[23:59] <burek> after that I'll just setup a cron job to build it each day
[23:59] <durandal_1707> what forum
[23:59] <durandal_1707> ?
[23:59] <burek> so that people, who want to try ffmpeg quickly, can just download and run it
[23:59] <burek> ffmpeg's forum
[00:00] --- Sun Feb 12 2012


More information about the Ffmpeg-devel-irc mailing list