[Ffmpeg-devel-irc] ffmpeg-devel.log.20140222

[burek]

[00:02] <cone-449> ffmpeg.git 03Michael Niedermayer 07master:8c55ff393340: avcodec/h264: use subsample factors of the used pixel format
[04:23] <cone-746> ffmpeg.git 03Michael Niedermayer 07master:44b22bba422b: avcodec/h264_ps: fix printed num_reorder_frames value
[09:56] <ubitux> michaelni: any comment on the png fix?
[10:37] <michaelni> ubitux, replied
[10:38] <ubitux> mmh ok
[11:56] <cone-550> ffmpeg.git 03Carl Eugen Hoyos 07master:b89596a43250: Fix FSF address in colormatrix and libzvbi license headers.
[13:14] <cone-550> ffmpeg.git 03Paul B Mahol 07master:8bcacd9f4262: SDR2 demuxer
[13:25] <aca> I have now working FFmpeg Debian packages, build with --enable-raise-major to avoid conflicts with libav packages. I named the development packages lib*-ffmpeg-dev, and only these are not coinstallabel with the libav versions, but having both dev packages wouldn't be useful anyway.
[13:27] <aca> I rebuilt the 108 reverse dependencies of src:libav and 50 failed to build (but even more fail to build with libav10). The most common failures were due to missing AVCODEC_MAX_AUDIO_FRAME_SIZE and CodecID. Is there something like a migration guide for these common errors?
[13:34] <JEEB> there's some migration documentation written over at libav at least, and since these changes generally are the same with ffmpeg, I would guess it's useful as well? https://wiki.libav.org/Migration/10
[13:34] <JEEB> CodecIDs became AV_CODEC_ID and AVCodecID
[13:35] <aca> Thanks, I'll look at that.
[14:37] <cone-550> ffmpeg.git 03Michael Niedermayer 07master:f288e1b67cba: avformat/utils: factorize h264/hevc checks out in compute_pkt_fields()
[14:37] <cone-550> ffmpeg.git 03Michael Niedermayer 07master:d4dfa97ae3ad: avformat/utils: reset pts_buffer in estimate_timings_from_pts()
[14:37] <cone-550> ffmpeg.git 03Michael Niedermayer 07master:3c096751ffe8: avformat/utils: compute_pkt_fields: Fix DTS for the case where has_b_frames is too large but the correct one is constant
[16:03] <cone-550> ffmpeg.git 03Luca Barbato 07master:d922c5a5fbaf: h264: Fix a typo from the previous commit
[16:03] <cone-550> ffmpeg.git 03Michael Niedermayer 07master:f98821dddb4a: Merge remote-tracking branch 'qatar/master'
[16:34] <BBB2> ubitux: ty!
[19:18] <BBB2> ubitux: question - is atom 64bit or 32bit only?
[19:20] <ubitux> it's 64
[19:20] <BBB2> thnx
[21:05] <cone-550> ffmpeg.git 03Peter Ross 07master:6236debe1a28: avcodec/raw: add bayer formats
[21:05] <cone-550> ffmpeg.git 03Peter Ross 07master:55479f42ce60: avformat/nut: add bayer colorspaces
[21:11] <cone-550> ffmpeg.git 03Anton Khirnov 07master:7e86c27b4ee9: lavr: add a function for checking whether AVAudioResampleContext is open
[21:11] <cone-550> ffmpeg.git 03Michael Niedermayer 07master:5367c0be6238: Merge commit '7e86c27b4ee9e5a3fbe6cf5249b9d918b2a5e731'
[21:13] <ubitux> it's really too bad hqx seems to work in the yuv domain
[21:15] <wm4> did you mean "not"?
[21:16] <cone-550> ffmpeg.git 03Anton Khirnov 07master:1db03a686411: lavr: return an error if a avresample_open() is called on an open context
[21:16] <cone-550> ffmpeg.git 03Michael Niedermayer 07master:246eae98cfe4: Merge remote-tracking branch 'qatar/master'
[21:18] <ubitux> wm4: no
[21:19] <ubitux> i mean it's for 8-bit rgb games
[21:19] <ubitux> it's too bad to require a convert for analysis
[21:26] <ubitux> just ran http://pastie.org/8759318 on a ffmpeg objdump
[21:26] <ubitux> not sure if it's accurate... but http://pastie.org/pastes/8759317/text
[21:26] Action: ubitux go back doing interesting stuff
[22:17] <BBB2> this will probably get lost in the netsplits, but ffvp9 blogpost: http://blogs.gnome.org/rbultje/2014/02/22/the-worlds-fastest-vp9-decoder-ffvp9/
[22:19] <beastd> BBB2: thanks for highlighting and quite elaborate blog post. will read it tomorrow or so.
[22:25] <ubitux> https://news.ycombinator.com/item?id=7283668
[22:41] <Daemon404> BBB2, that's an extremely useful link, thanks
[22:41] <Daemon404> your former coworker reached out to vimeo yesterday with some bogu claims
[22:41] <Daemon404> that link is helpful to me
[22:45] <ubitux> where is llogan when we need him
[22:46] <Daemon404> pff
[22:46] <Daemon404> i try and upvote it
[22:46] <Daemon404> but nothing happens
[22:46] <Daemon404> hn is broken for me \o/
[22:47] <ubitux> is it a newly created accoutn/
[22:47] <ubitux> ?
[22:47] <Daemon404> no
[22:47] <Daemon404> but i do only have one point
[22:47] <BBB2> Daemon404: which former coworker?
[22:48] <Daemon404> it was one of the main guys... i cant remember his name, sec
[22:48] Action: Daemon404 checks git
[22:48] <Daemon404> iirc he gave a talk about it before you once
[22:49] <Daemon404> crap i cant remember his name now
[22:50] <Daemon404> but he gave some ridiculous claims
[22:50] <Daemon404> like only 5x slower than x264 with 1/2 the size
[22:50] <Daemon404> for same quality
[22:50] <BBB2> marketing or engineer?
[22:50] <Daemon404> not sure... i swear it was an engineer, but i could be wrong
[22:51] <BBB2> lou, matt, jim, paul?
[22:51] <Daemon404> might be matt or jim... i'd have to ask my coworker who was on the call
[22:52] <BBB2> lou quillio is webmaster
[22:52] <BBB2> matt frost is product manager (marketing)
[22:52] <BBB2> jim bankoski is engineering manager
[22:52] <Daemon404> it was someone from the chrome team who reached out to us to try and sell us on using vp9
[22:52] <BBB2> paul wilkins is lead engineer
[23:11] <BBB2> kierank: oh you'd find that interesting also
[23:14] <cone-550> ffmpeg.git 03Michael Niedermayer 07master:f284e2a58a13: swresample: factorize clear_context() out
[23:14] <cone-550> ffmpeg.git 03Michael Niedermayer 07master:a89c01253190: avformat/mux: support shifting timestamps so they start at 0
[23:14] <cone-550> ffmpeg.git 03Michael Niedermayer 07master:717ec57c7e22: avformat/movenc: shift positive timestamps to 0 if edit lists cannot be used
[00:00] --- Sun Feb 23 2014