[Ffmpeg-devel-irc] ffmpeg-devel.log.20120417

[burek]

[04:51] <CIA-17> ffmpeg: 03Anton Khirnov 07master * r2e21526790 10ffmpeg/avconv.c: 
[04:51] <CIA-17> ffmpeg: avconv: change {input,output}_{streams,files} into arrays of pointers.
[04:51] <CIA-17> ffmpeg: Right now, they are arrays of structs, reallocated when new
[04:51] <CIA-17> ffmpeg: streams/files are added. This makes storing pointers to those structs
[04:51] <CIA-17> ffmpeg: harder than necessary.
[04:52] <CIA-17> ffmpeg: 03Anton Khirnov 07master * rea9367e921 10ffmpeg/avconv.c: 
[04:52] <CIA-17> ffmpeg: avconv: don't pass input/output streams to some functions.
[04:52] <CIA-17> ffmpeg: They are globals, no point in pretending they are not.
[04:52] <CIA-17> ffmpeg: 03Anton Khirnov 07master * rb7327887ea 10ffmpeg/ (avconv.c avplay.c cmdutils.c cmdutils.h tests/ref/lavf/xwd): 
[04:52] <CIA-17> ffmpeg: avconv: get output pixel format from lavfi.
[04:52] <CIA-17> ffmpeg: This way we don't require a clearly defined corresponding input stream.
[04:52] <CIA-17> ffmpeg: The result for the xwd test changes because rgb24 is now chosen instead
[04:52] <CIA-17> ffmpeg: of bgra.
[04:52] <CIA-17> ffmpeg: 03Anton Khirnov 07master * r9dced85426 10ffmpeg/avconv.c: avconv: remove a useless variable from OutputStream.
[04:52] <CIA-17> ffmpeg: 03Anton Khirnov 07master * re77c86629f 10ffmpeg/avconv.c: 
[04:52] <CIA-17> ffmpeg: avconv: move resample_{width,height,pix_fmt} to InputStream.
[04:52] <CIA-17> ffmpeg: This is a more proper place for them, since they store parameters of the
[04:52] <CIA-17> ffmpeg: input, not output, stream.
[04:52] <CIA-17> ffmpeg: 03Anton Khirnov 07master * r6c9eac1998 10ffmpeg/avconv.c: 
[04:52] <CIA-17> ffmpeg: avconv: don't set output width/height directly from input value.
[04:52] <CIA-17> ffmpeg: Always take it from lavfi. This way we don't need a clearly defined
[04:52] <CIA-17> ffmpeg: corresponding input stream.
[04:52] <CIA-17> ffmpeg: 03Anton Khirnov 07master * r836ce90566 10ffmpeg/avconv.c: 
[04:52] <CIA-17> ffmpeg: avconv: move filtered_frame from InputStream to OutputStream.
[04:52] <CIA-17> ffmpeg: than one input and/or output.
[04:52] <CIA-17> ffmpeg: 03Anton Khirnov 07master * r3b266da3d3 10ffmpeg/ (avconv.c doc/avconv.texi doc/filters.texi): avconv: add support for complex filtergraphs.
[04:52] <CIA-17> ffmpeg: 03Michael Niedermayer 07master * r4778783160 10ffmpeg/: (log message trimmed)
[04:52] <CIA-17> (40 lines omitted)
[09:38] <ubitux> "It will never be in libav master, which also happens to be the true
[09:38] <ubitux> master.  The ffmpeg fork is only kept on life support by a few holdouts.
[09:38] <ubitux> "
[09:38] <ubitux> seriously& :/
[09:38] <ubitux> (Måns, about libswresample, on vlc-devel)
[09:39] <ubitux> i'd say it's better to avoid to raise again a flame, but well&
[10:32] <TimNich> don't let them grind you down...
[10:36] <ubitux> i'm more angry about this disinformation actually
[10:37] <ubitux> but whatever.
[10:37] <ubitux> let's continue coding, i guess we are better at coding than political crap
[10:37] <av500> it's usally being used by both sides during "warfare" :)
[10:38] <ubitux> well it's too bad a new battleground was just introduced here
[10:44] <ohsix> as if his ego is foreign to you :]
[10:46] <ohsix> not to mention the "troll" driven development style that wasn't even tounge in cheek any time i've heard someone mention it, other people signed on of course; but it was obviously parts of some peoples personality when they arrived
[12:11] <CIA-17> ffmpeg: 03Nicolas George 07master * rb0e7321cf0 10ffmpeg/libavutil/parseutils.c: parseutils: include errors in test output.
[12:11] <CIA-17> ffmpeg: 03Nicolas George 07master * r68ca49dc72 10ffmpeg/tests/ (Makefile fate/libavutil.mak): fate: libavutil tests do not require SAMPLES.
[12:27] <CIA-17> ffmpeg: 03Anton Khirnov 07master * r3e901cbc75 10ffmpeg/avconv.c: avconv: reindent.
[12:27] <CIA-17> ffmpeg: 03Anton Khirnov 07master * r89605e4aa0 10ffmpeg/avconv.c: avconv: send EOF to vsrc_buffer.
[12:27] <CIA-17> ffmpeg: 03Anton Khirnov 07master * r9e8aae443b 10ffmpeg/avconv.c: 
[12:27] <CIA-17> ffmpeg: avconv: flush decoders immediately after an EOF.
[12:27] <CIA-17> ffmpeg: Prevents extensive buffering when overlaying a single picture.
[12:27] <CIA-17> ffmpeg: 03Anton Khirnov 07master * r7ce118bae5 10ffmpeg/doc/filters.texi: doc/filters: update documentation.
[12:27] <CIA-17> ffmpeg: 03Diego Biurrun 07master * r05afc5f57d 10ffmpeg/libavcodec/ (dv.c dvdec.c dvquant.h): dv: Move static tables only used in one place to where they are used.
[12:27] <CIA-17> ffmpeg: 03Diego Biurrun 07master * ra3dbd459ff 10ffmpeg/libavcodec/ (dv.c dvdata.c dvdata.h dvdec.c dvquant.h): dv: Merge dvquant.h into dvdata.c where all other DV tables reside.
[12:27] <CIA-17> ffmpeg: 03Diego Biurrun 07master * rb553a425fb 10ffmpeg/doc/developer.texi: doc: Update sample Vim config with suitable (function) indentation settings.
[12:27] <CIA-17> ffmpeg: 03Ronald S. Bultje 07master * r87a246341b 10ffmpeg/libavcodec/x86/h264_weight_10bit.asm: 
[12:27] <CIA-17> ffmpeg: h264: use proper PROLOGUE statement for a function using 8 registers.
[12:27] <CIA-17> ffmpeg: Fixes crashes when using biweight on win64.
[12:27] <CIA-17> ffmpeg: 03Mans Rullgard 07master * r8b84e082ed 10ffmpeg/libavcodec/indeo3data.h: 
[12:27] <CIA-17> ffmpeg: indeo3: add parens around some macro arguments
[12:27] <CIA-17> ffmpeg: Without these, the expansion contains things like --1 with
[12:27] <CIA-17> ffmpeg: some compilers resulting in build errors.
[12:27] <CIA-17> ffmpeg: Signed-off-by: Mans Rullgard <mans at mansr.com>
[12:27] <CIA-17> ffmpeg: 03Michael Niedermayer 07master * r15141f939d 10ffmpeg/: (log message trimmed)
[12:27] <CIA-17> ffmpeg: Merge remote-tracking branch 'qatar/master'
[12:27] <CIA-17> ffmpeg:  dv: Merge dvquant.h into dvdata.c where all other DV tables reside.
[12:27] <CIA-17> ffmpeg: 03Anton Khirnov 07master * raff01de641 10ffmpeg/libavfilter/graphparser.c: 
[12:27] <CIA-17> ffmpeg: graphparser: set next to NULL on an entry extracted from inputs list
[12:27] <CIA-17> ffmpeg: Prevents it from referring to the rest of the list.
[12:50] <merbanan> ubitux: anyway the libav resample lib work will be a step forward in handling audio
[12:50] <merbanan> but as you say we should code more and troll less
[12:57] <ubitux> merbanan: forward in what sense?
[12:57] <ubitux> i'd say i mean ffmpeg already has the features, and the audio filtering&
[12:57] <ubitux> libav ppl are just rewriting everything differently
[12:58] <ubitux> the only thing they will achieve is to piss off users
[12:59] <merbanan> simd optimization and delay compensation for example
[13:00] <ubitux> that couldn't be done in libswresample?
[13:00] <merbanan> sure it could
[13:01] <merbanan> but I'm not doing the work
[13:01] <ubitux> i really don't think it's a step forward
[13:06] <merbanan> it's a step forward in the sense that someone works on improving audio handling
[13:06] <merbanan> we'll see how good it is after the work is done
[13:07] <michaelni> merbanan, we maintain libswresample
[13:07] <michaelni> SIMD optims will be added
[13:07] <michaelni> and in case libav ever adds optimizations to their rewriten version ill merge them very very quickly
[13:08] <michaelni> also just in case its unclear libswresample supports everything libavresample does
[13:09] <michaelni> its a big mystery to me why that rewrite was started
[13:09] <Compn> they wanted it, but they didnt want to merge things from ffmpeg, apparently
[13:09] <Compn> :P
[13:09] <ubitux> merbanan: at least michaelni, nicolas and i worked on audio handling this last year
[13:10] <ubitux> libav just completly ignored this, and now they realise they are far behind; but instead of reusing the work done, they're just rewritting that stuff completely
[13:10] <ubitux> so no, sorry, but it's not a step forward
[13:11] <michaelni> also merbanan i volunteered to implement all features libav or ffmtech want for free in libswresample
[13:11] <michaelni> why is ffmtech wasting money on this rewrite ?
[13:11] <michaelni> this is not in the interrest of the community not ffmpegs not libavs or anyone elses ..
[13:16] <Compn> it is if libav doesnt want any "Copyright (C) 2011 Michael Niedermayer" in thier tree :P
[13:19] <Compn> but i'm just trolling
[13:20] <ubitux> well, afaik the libavresample code is based on the old audio api, from which michael is the author (and i remember seeing credits to him actually)
[13:21] <ubitux> but i guess code is "different", to say "hey we are doing things better [/differently]"
[13:21] <ubitux> but i'm just trolling too ;)
[13:22] <av500> cool, let me troll as well
[13:24] <ubitux> no time is up, it's time to code now
[13:25] <av500> void troll( void );
[13:30] <iive> I thought the problem is with git. Aka it cannot merge things from the downstream into the upstream.
[13:33] <Compn> :D
[13:35] <iive> talking about ffmtech, have there been new election? It was about this time last year.
[13:36] <Compn> libav removing lowres support eh?
[13:37] <iive> Compn: well, i'm not fan of lowres either. I've talked against it even before the takeover&fork. But I would have liked a few benchmarks, before and after.
[13:37] <av500> benchmark of what?
[13:37] <iive> of decoding speed.
[13:38] <av500> of the lowres decode?
[13:38] <av500> or of highres decode?
[13:38] <iive> of decoding with lowres=0 and decoding with lowres removed.
[13:38] <av500> prolly no difference
[13:39] <ubitux> lowres could be used for visual analysis i guess
[13:39] <iive> there is always a difference. it may be less than 1%, but it could be measured.
[13:39] <av500> I dont think removing lowres made the code slower
[13:40] <iive> never implied that. I just would have liked to see the benchmarks.
[13:41] <av500> I dont see the point
[13:41] <iive> if the difference is 1%, then it is quite huge improvement, for a feature that is never used.
[13:42] <iive> And it would have been good PR for libav folks. "We made libav faster"
[13:44] <iive> instead of "we didn't make libav slower" ;)
[14:00] <angel> ciao
[14:00] <angel> !list
[14:27] <CIA-17> ffmpeg: 03Michael Niedermayer 07master * r9bb3b7d028 10ffmpeg/libavformat/oggdec.c: 
[14:27] <CIA-17> ffmpeg: oggdec: Recreate streams only in the 1 stream case.
[14:27] <CIA-17> ffmpeg: Other cases are not supported and lead to inconsistencies which
[14:27] <CIA-17> ffmpeg: can lead to out of array writes.
[14:27] <CIA-17> ffmpeg: Reported-by: Dale Curtis <dalecurtis at chromium.org>
[14:27] <CIA-17> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[15:49] <Compn> does ffac3 downmix automatically ?
[15:49] <Compn> this was one of the features that mplayer used in the liba52 codec 
[15:49] <Compn> now that ffmpeg has audio filters, auto downmix would be nice :)
[15:51] <Compn> ubitux ? 
[15:57] <ubitux> Compn: yes, i guess& but audio filtering is in a too bad state ATM
[16:26] <CIA-17> ffmpeg: 03Michael Niedermayer 07master * rc7c976c6d7 10ffmpeg/libavformat/oggdec.c: 
[16:26] <CIA-17> ffmpeg: oggdec: Safety check against stream counts being inconsistent in seek()
[16:26] <CIA-17> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[16:26] <CIA-17> ffmpeg: 03Robert Nagy 07master * ra80217a5ee 10ffmpeg/libavfilter/avcodec.c: 
[16:26] <CIA-17> ffmpeg: copy pts and format props between lavfi buffer and frame.
[16:26] <CIA-17> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[16:33] <nevcairiel> ffac3 supports request_channels, you can do a stereo downmix with it inside the decoder
[16:33] <nevcairiel> who knows if mplayer leverages that though =)
[16:33] <Daemon404> this sounds like a mess lol
[16:34] <ubitux> mplayer could use libswr ;)
[16:34] <nevcairiel> how so? AC3 even has downmix coeffs in the bitstream, it just uses those
[16:34] <Daemon404> i mean having some stuff in decoders
[16:34] <Daemon404> soem stuff in libswr
[16:34] <Daemon404> etc
[16:35] <nevcairiel> well exporting the bitstream coeffs to some exteral mixer sounds even more like a mess. :D
[16:35] <Daemon404> true.
[16:35] <Daemon404> No Good Solution (TM)
[17:51] <michaelni> Tjoppen, theres a ping in "[FFmpeg-devel] [PATCH]: mxf: Fix frame layout values"
[18:17] <CIA-17> ffmpeg: 03Michael Niedermayer 07master * r2677697ac9 10ffmpeg/libavformat/matroskadec.c: 
[18:17] <CIA-17> ffmpeg: matroskadec: check headerstrip data availability.
[18:17] <CIA-17> ffmpeg: Fixes null ptr dereference
[18:17] <CIA-17> ffmpeg: Found-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
[18:17] <CIA-17> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[18:17] <CIA-17> ffmpeg: 03Michael Niedermayer 07master * r9789b6ca48 10ffmpeg/libavformat/avidec.c: 
[18:17] <CIA-17> ffmpeg: avidec: document early_exit argument
[18:17] <CIA-17> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[18:17] <CIA-17> ffmpeg: 03Michael Niedermayer 07master * r096231d497 10ffmpeg/libavformat/avidec.c: 
[18:17] <CIA-17> ffmpeg: avidec: Dont crash on avi packets that belong to dv streams in dv in avi
[18:17] <CIA-17> ffmpeg: Fixes null pointer dereference
[18:17] <CIA-17> ffmpeg: Found-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
[18:17] <CIA-17> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[19:08] <elinenbe> is anyone here?
[19:11] <Daemon404> perhaps
[19:14] <elinenbe> QQ that I can't find an answer to.  Using 10.2 -- is is possible to pipe something to amovie (instead of a filename)?
[19:14] <elinenbe> eg, this works: ffmpeg -y -f lavfi -i amovie=elvisradio.flv,silencedetect=n=-23dB:d=1 -f null -
[19:14] <elinenbe> but cat elvisradio.flv | ffmpeg -y -f lavfi -i amovie=pipe,silencedetect=n=-23dB:d=1 -f null -
[19:14] <elinenbe> does not work
[19:15] <Daemon404> er
[19:15] <Daemon404> im not sure what the heck youre trying to do?
[19:15] <Daemon404> oh
[19:15] <Daemon404> pipe to the filter?
[19:15] <Daemon404> and then back to output?
[19:21] <elinenbe> Yes, I'm trying to pipe to a filter.
[19:21] <elinenbe> But it seems to only work with a filename.
[19:22] <elinenbe> This is the silencedetect filter that outputs text to standard out when silences occur
[19:24] <Daemon404> im not sure amovie has pipe support
[19:24] <Daemon404> saste would know, but he isnt here right now
[19:25] <ubitux> elinenbe: maybe you can simulate the behaviour with a mkpipe
[19:25] <Daemon404> named pipes yo
[19:25] <ubitux> mkfifo sorry
[19:26] <elinenbe> mkfifo?  not famalir with it... let me take a look.
[19:28] <elinenbe> looks cool... trying it out now
[19:29] <elinenbe> very ineresting.  Lemme see if I can get everything working properly
[19:43] <elinenbe> eh... no dice.
[20:26] <Compn> elinenbe : you have to do -i - i think
[20:26] <Compn> cat file.flv | ffmpeg -i -
[20:26] <Compn> but it will also input ffmpeg text to stdout which isnt good
[20:26] <Compn> need to silence ffmpeg output
[20:29] <CIA-17> ffmpeg: 03Michael Niedermayer 07master * rebf6d1d295 10ffmpeg/libavcodec/nuv.c: 
[20:29] <CIA-17> ffmpeg: nuv: check buffer size before checking content.
[20:29] <CIA-17> ffmpeg: Found-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
[20:29] <CIA-17> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[20:29] <CIA-17> ffmpeg: 03sebist 07master * re128182afd 10ffmpeg/libavcodec/crystalhd.c: 
[20:29] <CIA-17> ffmpeg: CrystalHD: fix pStride value.
[20:29] <CIA-17> ffmpeg: Signed-off-by: sebist <sebok.istvan at gmail.com>
[20:29] <CIA-17> ffmpeg: Reviewed-by: Philip Langdale <philipl at overt.org>
[20:29] <CIA-17> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[20:29] <CIA-17> ffmpeg: 03sebist 07master * r7a3d2258a6 10ffmpeg/libavcodec/crystalhd.c: 
[20:29] <CIA-17> ffmpeg: CrystalHD: Set aspect ratio.
[20:29] <CIA-17> ffmpeg: Signed-off-by: sebist <sebok.istvan at gmail.com>
[20:29] <CIA-17> ffmpeg: Reviewed-by: Philip Langdale <philipl at overt.org>
[20:29] <CIA-17> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[20:29] <CIA-17> ffmpeg: 03Robert Nagy 07master * re90dfa6881 10ffmpeg/libavfilter/vf_yadif.c: 
[20:29] <CIA-17> ffmpeg: yadif: Improve pts calculation for is_second.
[20:29] <CIA-17> ffmpeg: Tested-by: Robert Nagy <ronag89 at gmail.com>
[20:29] <CIA-17> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[20:29] <CIA-17> ffmpeg: 03Michael Niedermayer 07master * ra66675268f 10ffmpeg/libavcodec/indeo5.c: 
[20:29] <CIA-17> ffmpeg: indeo5: dont run the wavelet transform over partially decoded bands.
[20:29] <CIA-17> ffmpeg: This fixes a null pointer dereference.
[20:29] <CIA-17> ffmpeg: Found-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
[20:29] <CIA-17> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[21:12] <Daemon404> michaelni, im struggling to understand what you said about abe0b8e9f378a0f8781c1a3da6714d20cfd19594
[21:12] Action: Daemon404 isnt sure what michaelni means by color bias
[21:15] <elinenbe> Hi there... anyone here?  trying to push a stream coming from librtmp through amovie, but I can't seem to make this work... any suggestions?
[21:37] <michaelni> Daemon404, its simple
[21:38] <michaelni> in 8bit 128 is gray
[21:38] <michaelni> neutral gray, not redish gray not bluish
[21:38] <michaelni> in 12bit 2048 is gray
[21:38] <michaelni> 128<<4 == 2048
[21:39] <Daemon404> i guess that makes sense for PC range
[21:39] <Daemon404> btw michaelni.. afaik nothing even uses that function for PC rage
[21:39] <Daemon404> range*
[21:39] <Daemon404> everything full/pc range is said to be "YUVJ"
[21:39] <Daemon404> e.g. the h264 decoder
[21:40] <michaelni> 128 is gray in both variants
[21:40] <Daemon404> i know
[21:40] <Daemon404> but im saying i dont think anythign uses c->srcRange
[21:40] <nevcairiel> swscale internally renames YUVJ to YUV and sets the range flag
[21:40] <Daemon404> i couldnt even get it to trigger manually
[21:42] <Daemon404> YUVJ certainly doesn't,
[21:43] <Daemon404> (i put debug messages in and they were never tripped)
[21:44] <michaelni> c->srcRange  = handle_jpeg(&srcFormat); <---- this should
[21:46] <Daemon404> michaelni, well that wrapper seems to never be called
[21:46] <Daemon404> :|
[21:46] <Daemon404> (the one changed in abe0b8e9f378a0f8781c1a3da6714d20cfd19594)
[23:44] <ubitux> elinenbe: atm amovie is a limited workaround to use the audio filtering; it's meant to evolve, but in the current state i'm afraid the mkfifo is one of the only workaround (if you're able to make it work)
[23:44] <ubitux> also, no point in asking libav, they don't have audio filtering atm
[23:44] <ubitux> (refering to your question on #libav)
[23:45] <ubitux> i wonder if hacking the complex filter system for audio will require much work&
[00:00] --- Wed Apr 18 2012