[Ffmpeg-devel-irc] ffmpeg-devel.log.20141126

[burek]

[01:11] <cone-477> ffmpeg.git 03Carl Eugen Hoyos 07master:5badcdf20d98: Rename sync() functions in libavformat.
[01:11] <cone-477> ffmpeg.git 03Benoit Fouet 07master:4acefd25215a: avformat/apngdec: account for blend and dispose operations.
[01:55] <llogan> michaelni: we could always ask users on twitter to come up with a name. they'll probably like that, and we'll get a free name such as "Ur mom".
[02:18] <cone-477> ffmpeg.git 03Michael Niedermayer 07master:2ad38c6e02cc: avformat/mxfdec: dont ask for samples with field dominance 0 anymore
[02:25] <kierank> what's the difference between yuv2yuvX and yuv2planeX
[02:32] <kierank> ah the latter has accurate rounding
[04:30] <cone-477> ffmpeg.git 03Michael Niedermayer 07master:e5c01ccdf5a9: avcodec/flacdec: Call ff_flacdsp_init() unconditionally
[04:30] <cone-477> ffmpeg.git 03Michael Niedermayer 07master:5f30522894d1: avcodec/flacdec: fix off by 1 error
[06:05] <Zeranoe> Perhaps another DeckLink issue: http://ffmpeg.zeranoe.com/forum/viewtopic.php?f=10&t=1823#p7554
[12:00] <cone-393> ffmpeg.git 03Martin Storsjö 07master:ee37620b6ae4: movenc: Add a flag for indicating a discontinuous fragment
[12:00] <cone-393> ffmpeg.git 03Michael Niedermayer 07master:b78074fd13ad: Merge commit 'ee37620b6ae4783cda637408422044b2d14a688c'
[13:34] <cone-393> ffmpeg.git 03Martin Storsjö 07master:234fb81e3145: movenc: Expose the fragment index as an avoption
[13:34] <cone-393> ffmpeg.git 03Michael Niedermayer 07master:0df95fa327c8: Merge commit '234fb81e3145e9c9aec4ec16266676fab7dc21fa'
[13:34] <cone-393> ffmpeg.git 03Benoit Fouet 07master:8b8cb30d1188: avformat/apngdec: use packet pts and duration instead of altering stream framerate.
[14:59] <kierank> michaelni: can you apply my v210 stuff
[15:18] <michaelni> kierank, is there a patch that took care of James comments ?
[15:19] <michaelni> 1122 23:10 James Almer     (5.9K) 
[15:30] <kierank> michaelni: ok I will squash the two patches and resubmit 
[16:14] <cone-393> ffmpeg.git 03Michael Niedermayer 07master:79ceaf827be0: avcodec/pngdec: Check IHDR/IDAT order
[17:00] <kierank> michaelni: patch sent with changes
[17:04] <kierank> hah and now libav has a patch
[17:39] <cone-393> ffmpeg.git 03Michael Niedermayer 07master:9a53707e86eb: avcodec/pngdec: Fix paeth prediction with small images
[17:46] <michaelni> kierank, reply sent, seems it breaks fate
[17:47] <kierank> how do I run that single fate test?
[17:47] <Daemon404> make fate-vsynth1-v210
[17:48] <kierank> wtf is that test doing
[17:49] <ubitux> use V=1
[17:49] <kierank> it compares a 422 file by converting it to 420
[17:49] <kierank> i did use V=1
[18:08] <jamrial> it fails even without running the asm
[18:11] <saste> arwa: http://trac.ffmpeg.org/wiki/SponsoringPrograms/OPW/SponsoringPrograms/OPW/2014-12/MiscLibavfilterExtension
[18:11] <saste> fast&furious edit, I have to dig it more myself
[18:12] <saste> about eq/eq2, James Darnley was working on it, you should contact him and ask for more
[18:12] <saste> arwa: ^
[18:13] <kierank> is an upsampling to 422 and a downsampling back to 420 meant to be bitexact?
[18:14] <nevcairiel> i would be surprised
[18:15] <nevcairiel> maybe if you use a point scaler for downsampling or something :p
[18:25] <kierank> a lot of pixels are off by one
[18:26] <Compn> who said we had bit exact colorspace conversion? :P
[18:27] <Compn> it would be interesting to see if the c version was bitexact vs the asm 
[18:27] <wm4> basically libswscale is being liquid shit _again_
[18:28] <Compn> or rather , no one is making a proper bugreport so it can be fixed
[18:28] <Compn> instead wasting time on complaining :)
[18:55] <kierank> i'm pretty sure there's an off by one somewhere
[18:55] <kierank> or a clipping ug
[18:59] <michaelni> kierank, in your code or swscale ?
[19:00] <kierank> don't know
[19:02] <kierank> ./ffmpeg -pix_fmt yuv422p -s 1280x720 -i in.yuv -vcodec v210 -y out.mov && ./ffmpeg -i out.mov -pix_fmt yuv422p10 -y out.yuv && ./ffmpeg -pix_fmt yuv422p -s 1280x720 -i in.yuv -f rawvideo -pix_fmt yuv422p10 -y in10.yuv && md5sum in10.yuv out.yuv
[19:02] <kierank> in10.yuv and out.yuv match
[19:02] <kierank> but in.yuv and out.yuv don't
[19:02] <kierank> so there's something going on somewhere
[19:03] <kierank> some broken colourspace conversion
[19:05] <kierank> michaelni: ^
[19:07] <kierank> the c seems to be ok
[19:07] <kierank> the asm doesn't
[19:24] <cone-393> ffmpeg.git 03Michael Niedermayer 07master:1b5d11240692: avformat/mov: Fix memleaks for duplicate STCO/CO64/STSC atoms
[19:24] <cone-393> ffmpeg.git 03Michael Niedermayer 07master:1d3a3b9f8907: avcodec/rawdec: Check the return code of avpicture_get_size()
[20:12] <kierank> Gramner: ping
[20:12] <Gramner> pong
[20:12] <kierank> Gramner: do you mind being a second pair of eyes and looking at my v210 8-bit chroma shuffles?
[20:12] <kierank> there's something subtly wrong with the code 
[20:13] <Gramner> yeah sure, in a few minutes
[20:13] <kierank> thanks
[20:22] <kierank> ah I think it's because I don't merge the luma and chroma correctly
[20:23] <kierank> i mean the u and v planes actually
[20:24] <michaelni> kierank, i found the issue
[20:25] <kierank> the shuffle is wrong in my asm for one thing
[20:26] <michaelni> yes, posted correction to ml a sec ago
[21:14] <cone-393> ffmpeg.git 03Kieran Kunhya 07master:36091742d182: v210enc: Add SIMD optimised 8-bit and 10-bit encoders
[21:57] <ramiro> a user got in touch with me asking to add strftime to the segment muxer
[21:57] <ramiro> the segment muxer uses av_get_frame_filename() to get the filename
[21:58] <ramiro> would it be A Good Thing to make av_get_frame_filename() use strftime, or better to just add strftime to the segment muxer?
[21:59] <cone-393> ffmpeg.git 03Clément BSsch 07master:b424e67abf0d: avfilter/signalstats: fix different buffers for out frame if burn is enabled
[21:59] <cone-393> ffmpeg.git 03Clément BSsch 07master:c7e8f610f281: avfilter/signalstats: remove pointless sub filter init system
[21:59] <cone-393> ffmpeg.git 03Clément BSsch 07master:56b98dfc4f09: avfilter/signalstats: integrate height loop into subfilters
[21:59] <cone-393> ffmpeg.git 03Clément BSsch 07master:cc5c667eb13b: avfilter/signalstats: reindent after previous commit
[21:59] <cone-393> ffmpeg.git 03Clément BSsch 07master:fad6865748c7: avfilter/signalstats: fix repitition/repetition typo
[21:59] <cone-393> ffmpeg.git 03Clément BSsch 07master:9db78a296c37: avfilter/signalstats: add slice threading for subfilters
[21:59] <cone-393> ffmpeg.git 03Clément BSsch 07master:7acbd56a8a99: avfilter/signalstats: isolate sat hue computation metrics in a function
[21:59] <cone-393> ffmpeg.git 03Clément BSsch 07master:82dda8e4eaa6: avfilter/signalstats: add threading in compute_sat_hue_metrics
[21:59] <cone-393> ffmpeg.git 03Clément BSsch 07master:9cb1d81a60bd: avfilter/signalstats: localize a few variables
[21:59] <cone-393> ffmpeg.git 03Clément BSsch 07master:c2ea7069c489: avfilter/signalstats: re-use yuv/yuvu/yuvv vars in diff
[22:35] <Zeranoe> Perhaps another DeckLink issue: http://ffmpeg.zeranoe.com/forum/viewtopic.php?f=10&t=1823#p7554
[22:42] <nevcairiel> the free of the name in that line should use SysFreeString, not free()
[22:42] <nevcairiel> on win32 anyway
[22:42] <nevcairiel> no idea about the other operating systems
[22:43] <nevcairiel> (obviously not SysFreeString on other operating systems)
[22:43] <ramiro> nevcairiel: thanks
[22:44] <ramiro> I wonder why I chose free...
[22:44] <nevcairiel> on osx its a CFString, whatever is required to free that, i don't know!
[22:45] <ramiro> Zeranoe: care to make a patch?
[22:47] <Zeranoe> ramiro: Pretty busy right now
[22:49] <ramiro> nevcairiel: how about you? =)
[22:50] <nevcairiel> i dont write patches i cant even compile 
[23:44] <cone-393> ffmpeg.git 03Lukasz Marek 07master:e98aced69955: ffserver_config: cosmetic: simplify functions calls.
[23:44] <cone-393> ffmpeg.git 03Lukasz Marek 07master:d57a6d20875f: ffserver_config: cosmetic: move line_num into FFServerConfig
[23:44] <cone-393> ffmpeg.git 03Lukasz Marek 07master:f61cb6453d22: ffserver_config: map ffserver options to AVOptions
[23:45] <cone-393> ffmpeg.git 03Lukasz Marek 07master:6c2ed67c2f6f: ffserver_config: remove ffserver_apply_stream_config function
[23:45] <cone-393> ffmpeg.git 03Lukasz Marek 07master:aaf6cc925f7f: ffserver: allow skip setting defaults
[23:45] <cone-393> ffmpeg.git 03Lukasz Marek 07master:ec6e035b8b1f: ffserver: export recommented encoder configuration
[23:45] <cone-393> ffmpeg.git 03Lukasz Marek 07master:3cb0bec6870c: ffserver: dont leak child arguments
[23:45] <cone-393> ffmpeg.git 03Lukasz Marek 07master:3d0867917faa: ffserver: dont leak pb_buffer
[23:45] <cone-393> ffmpeg.git 03Lukasz Marek 07master:345cfd04d093: lavc/options: fix leaks in avcodec_free_context
[23:54] <rcombs> >HEVC >interlaced
[23:54] <rcombs> weren't there plans to not even support that
[23:57] <gnafu> >broadcast
[23:57] <gnafu> >industry
[23:57] <gnafu> :-P
[23:57] <cone-393> ffmpeg.git 03Benoit Fouet 07master:e2b8b4caf6c0: avformat/apngdec: validate frame dimensions.
[23:58] Action: michaelni has the suspicion that the plans to kill interlaced are as old as interlaced formats
[00:00] --- Thu Nov 27 2014