[Ffmpeg-devel-irc] ffmpeg-devel.log.20141120

[burek]

[00:22] <cone-71> ffmpeg.git 03Michael Niedermayer 07master:0dba982bb4f7: avformat/dtsdec: dts_probe: check reserved bit, check lfe, check sr_code similarity
[01:23] <cone-71> ffmpeg.git 03Michael Niedermayer 07master:4388e78a0f02: swscale/x86/rgb2rgb_template: handle the first 2 lines with C in rgb24toyv12_*()
[01:30] <cehoyos> michaelni: I would like to backport three patches to 2.4 that do not fix regressions:
[01:31] <cehoyos> The two patches from Peter Ross that export Phantom Cine Metadata about the colours and your swscale patch that fixes the crash when scaling bayer.
[01:31] <cehoyos> Any objections?
[01:36] <cone-71> ffmpeg.git 03Michael Niedermayer 07release/2.4:e5f5df37c8e1: avformat/dtsdec: dts_probe: check reserved bit, check lfe, check sr_code similarity
[01:54] <michaelni> cehoyos, should be ok if they apply without conflicts and fate passes
[01:54] <cehoyos> The Bayer patch also requires to backport fba89461 which contains a minor version bump.
[01:54] <michaelni> version bumps are not ok in releases
[01:54] <cehoyos> If it is ok to backport this patch (it works, fate is running atm), what should I do with the version bump?
[01:55] <cehoyos> Is simply not bumping ok?
[01:55] <cehoyos> fate passes
[01:56] <michaelni> yes not bumping is an option 
[01:56] <iive> why is there a version bump?
[01:56] <cehoyos> Sio is adding fba89461 to 2.4 ok in your opinion?
[01:56] <cehoyos> The other three patches look safe to me but I cannot comment on this one.
[01:56] <cehoyos> An internal symbol was added.
[01:57] <iive> so, nothing that breaks abi/api.
[01:58] <cehoyos> Sorry, it was a micro bump, not a minor bump
[01:58] <cehoyos> I misread the conflict.
[02:00] <michaelni> fba89461 is ok if it applies cleanly (except the omited version change) and it works
[02:01] <cehoyos> It applies cleanly, I will test now.
[02:04] <cone-71> ffmpeg.git 03Michael Niedermayer 07release/2.4:633a2a082fc1: swscale: support internal scaler cascades
[02:04] <cehoyos> Tested and pushed, thank you!
[02:04] <cone-71> ffmpeg.git 03Michael Niedermayer 07release/2.4:222236317bf4: swscale/utils: support bayer input + scaling, and bayer input + any supported output
[02:04] <cone-71> ffmpeg.git 03Peter Ross 07release/2.4:e386241d54a0: cinedec: report white balance gain coefficients using metadata
[02:04] <cone-71> ffmpeg.git 03Peter Ross 07release/2.4:057ee3592418: avfilter/vf_lut: gammaval709()
[02:06] <cehoyos> Good night!
[03:17] <jamrial> michaelni: you didn't add a changelog entry about the webp muxer
[03:17] <jamrial> considering that unlike image2 it supports creating animated files i'd say it's worth a mention
[04:32] <cone-71> ffmpeg.git 03Michael Niedermayer 07master:c661601f4583: Changelog: add WebP muxer
[04:32] <michaelni> jamrial, done
[07:45] <ubitux> haha, aac patch v9
[07:45] <ubitux> it will never end @_@
[07:51] <rcombs> ooh boy
[07:51] <rcombs> meanwhile: anyone around who knows a fair bit about movenc.c?
[07:51] <rcombs> I'm currently attempting to get dashenc to output segments starting from the _middle_ of a stream
[07:53] <rcombs> as in, start at the beginning, client plays a few segments, then the user seeks and jumps to, say, 2:30; restart the transcoder with `-copyts -start_at_zero -ss 150` and start writing new segments at the offset, using the same MPD
[07:54] <rcombs> (I'm using custom-generated MPDs instead of the ones dashenc.c writes, so the duration is always set to the full final duration of the output file)
[07:54] <rcombs> current problem is that the output segments get shifted so their timestamps start at zero, despite the -copyts
[07:54] <rcombs> so I'm playing around to try to add an option to not do that
[07:55] <rcombs> (Fun? Fun.)
[08:09] <rcombs> I ended up hitting the "Track %d starts with a nonzero dts" warning after a bit of screwing around
[09:09] <lookatme> hello
[11:34] <cone-976> ffmpeg.git 03jessejiang 07master:29d208d5d47b: avutil/arm/float_dsp_init_vfp: replace restrict by av_restrict
[12:10] <cone-976> ffmpeg.git 03Carl Eugen Hoyos 07master:25ccf5df723c: lavf/mux: Always call write_trailer() from av_write_trailer() to avoid a leak.
[12:10] <cone-976> ffmpeg.git 03Michael Niedermayer 07master:9266eb0c620d: Merge remote-tracking branch 'cehoyos/master'
[12:22] <cone-976> ffmpeg.git 03Matthew Oliver 07master:e39f8fad321f: configure: Prevent icl being incorrectly detected as msvc.
[14:01] <j-b> ANyone knows about vdowave3 codec?
[14:14] <lookatmeyou> hello, I've got a problem
[14:14] <lookatmeyou> me at ubuntu:~/ffmpeg-2.4.3/doc/examples$ ./decoding_encoding h264
[14:14] <lookatmeyou> Encode video file test.h264
[14:14] <lookatmeyou> Codec not found
[14:14] <lookatmeyou> I have already install x264, but still get this error
[14:15] <lookatmeyou> help!
[14:15] <lookatmeyou> anyone here?
[14:40] <J_Darnley> lookatmeyou: what does ffmpeg say when you try to use libx264?
[14:43] <lookatmeyou> later, I find I havn't build ffmpeg with --enable-x264
[14:43] <lookatmeyou> thank you
[15:17] <Daemon404> something make ffmpeg mighty unhappy on windows/msvc....
[15:17] <Daemon404> it crashes instandtly
[15:17] <Daemon404> as of today
[15:17] <cone-976> ffmpeg.git 03Michael Niedermayer 07master:88b4c1a7316c: avcodec/mpeg12dec: Print error/warning messages on issues in mpeg1_decode_sequence()
[15:17] <cone-976> ffmpeg.git 03Michael Niedermayer 07master:0f8908aa1b66: avcodec/mpeg12dec: do not fail on invalid frame_rate_index
[15:20] <Daemon404> loooks like matt broke icl
[15:23] <benoit-> Daemon404: do you have more input than that?
[15:25] <Daemon404> benoit-, i sent a link to fate on the ML
[15:25] <benoit-> Daemon404: k
[16:06] <Daemon404> running 64bit icl test
[16:13] <Daemon404> has ffmpeg moved to tim's nodejs fate btw?
[17:13] <cone-976> ffmpeg.git 03Michael Niedermayer 07master:1852b2a0f497: avcodec/mpeg12dec: Use more specific error codes
[17:13] <cone-976> ffmpeg.git 03Michael Niedermayer 07master:bab11fe7bf4b: avcodec/mpeg12dec: forward error codes
[17:30] <Daemon404> urg
[17:30] <Daemon404> icl instances is actually hanging my fate box
[17:31] <Daemon404> https://www.dropbox.com/s/o7omv923gsiztez/cil.png?dl=0
[17:31] <Daemon404> lots of that
[17:31] <Daemon404> you cant disable that dialogue apparently
[17:32] <nevcairiel> Good job fixing icl in that one commit huh
[18:09] <kierank> j-b: [FFmpeg-devel] [PATCH 3/6] Add AV_PIX_FMT_NV12T.
[18:09] <kierank> what
[18:09] <kierank> the
[18:09] <kierank> fuck
[18:14] <wm4> sounds like it should be a codec
[18:29] <Compn> wm4 : can be both...
[18:29] <Compn> raw codec and colorspace
[18:30] <Daemon404> i personally prefer not to call different methods of packing bits/bytes a "Colorspace"
[18:36] <j-b> WHY?
[18:36] <av500> its not a colorspace
[18:36] <av500> colorspace is CIElab etc..
[18:36] <av500> or RGB
[18:36] <av500> and SRGB
[18:38] <wm4> j-b: to make something work with v4l I suppose
[18:38] <wm4> but I thought v4l has a transparent conversion lib for such things
[18:38] <wm4> which ffmpeg can use
[18:39] <wm4> vf_v4l2_m2m.c
[18:39] <wm4> maybe because of this
[18:40] <j-b> What's the use of the filter without the decoder
[18:45] <kierank> I remember they talked about memory to memory devices at kernel summit
[18:45] <kierank> can't remember what it does
[19:07] <iive> kierank: itn't that a virtual device? aka you have a device that gives you video frames generated by another program?
[19:07] <kierank> dunno
[19:08] <Compn> Daemon404 : sounds like you want to bikeshed over a name :P
[19:09] <kierank> erm no
[20:26] <cone-976> ffmpeg.git 03Matthew Oliver 07master:17b7f99d810d: configure: disable strip when using icl.
[22:39] <cone-976> ffmpeg.git 03Michael Niedermayer 07master:8bce5c8e74b5: avcodec/ac3dec: Use avpriv_float_dsp_alloc()
[22:39] <cone-976> ffmpeg.git 03Michael Niedermayer 07master:9018bd11979f: avcodec/atrac1: Use avpriv_float_dsp_alloc()
[23:01] <cone-976> ffmpeg.git 03Michael Niedermayer 07master:b054054c9b8f: avcodec/dcadec: Use avpriv_float_dsp_alloc()
[23:01] <cone-976> ffmpeg.git 03Michael Niedermayer 07master:21ded9ce67d8: avcodec/imc: Use avpriv_float_dsp_alloc()
[23:34] <ubitux> is ffplay broken or what?
[23:35] <ubitux> can't play mp3s anymore, is it on my side only?
[23:35] <ubitux> oh well, just this pulse crap again probably
[23:36] <ubitux> (only audio looks broken)
[23:54] <pross> works here
[00:00] --- Fri Nov 21 2014