[Ffmpeg-devel-irc] ffmpeg-devel.log.20131203

[burek]

[00:14] <cone-672> ffmpeg.git 03Michael Niedermayer 07master:210a78aeb509: doc/examples/filtering_audio: init packet0.data
[01:32] <cone-672> ffmpeg.git 03Michael Niedermayer 07master:4d0ac9aee920: Makefile: add dependancies for ffversion.h
[04:08] <ian_m> I am suspicious of the pts set in x11grab.c. It is set to av_gettime but it should be in time_base units, which is set to 1/framerate
[04:17] <ian_m> Ah, that's a private time_base
[04:31] <cone-672> ffmpeg.git 03Michael Niedermayer 07master:0cc5011f9a1b: avfilter/vf_pad: fix req_end
[04:46] <michaelni> BBB, ubitux it seems there are some asan failures in vp9 fate tests (http://fate.ffmpeg.org/report.cgi?time=20131203002327&slot=x86_64-debian-asan-144800)
[04:54] <BBB> michaelni: file track issue, will look over weekend
[04:54] <BBB> probably missing size check in vp9 parser
[04:54] <BBB> (on flush)
[04:54] <michaelni> ok, thansk
[04:55] <BBB> np
[09:58] <cone-604> ffmpeg.git 03Michael Niedermayer 07release/2.1:cbcc18bd9f94: avfilter/vf_pad: fix req_end
[12:42] <cone-604> ffmpeg.git 03Anton Khirnov 07master:089fac77a6bf: alsa-audio-dec: explicitly cast the delay to a signed int64
[12:42] <cone-604> ffmpeg.git 03Michael Niedermayer 07master:0e15a344c1ce: Merge remote-tracking branch 'qatar/master'
[13:27] <BBB> michaelni: see email on list
[13:27] <BBB> michaelni: missing zero-size check afaics
[13:29] <BBB> or maybe this one is better
[13:29] <BBB> not sure
[14:37] <cone-604> ffmpeg.git 03Ronald S. Bultje 07master:9695fb262219: vp9_parser: handle zero-sized packets.
[15:31] <wm4> is there any reason the strength parameter in gradfun is limited to 64 or so?
[15:31] <wm4> the limit was 255 in mplayer
[15:33] <ubitux> yes
[15:33] <ubitux> while fixing the asm to match the C
[15:33] <ubitux> we needed to avoid overflows, which was reachable with such a large strength
[15:34] <ubitux> loren (iirc) said it made no sense to have such a huge stregnth anyway
[15:34] <ubitux> so we lowered it and fixed the asm
[15:34] <wm4> oh ok
[15:34] <wm4> huge strength is nice to get more blurrrrrrr
[15:34] <ubitux> :)
[15:38] <ubitux> wm4: see 2d66fc543b01995d6146fc132a778d3e722ca665
[15:39] <wm4> thanks
[15:39] <ubitux> and the other commits around
[15:39] <ubitux> for the asm fixes etc
[15:41] <ubitux> http://ffmpeg.org/pipermail/ffmpeg-devel/2012-December/135921.html
[15:42] <ubitux> wm4: are you looking at dropping some of your filters to use the ff ones?
[15:46] <wm4> I'm adding a weird redirection
[15:46] <wm4> that uses lavfi when available, and otherwise the builtin ones
[15:46] <ubitux> ah, fun
[15:47] <ubitux> with both built?
[15:47] <wm4> yes
[15:48] <wm4> the goal is not bothering users with the difference, as well as preventing users from using the old crappy implementations
[15:49] <ubitux> so you're adding options mapping for every common filter?
[15:49] <wm4> yes
[19:25] <ubitux> BBB: i'd like to add a local named register in addition to the named parameters on the cglobal prototype
[19:26] <ubitux> so i have cglobal foo, 3,3,0, a,b,c
[19:26] <ubitux> should i do sth like cglobal foo, 3,4,0,a,b,c,d?
[19:26] <nevcairiel> yes
[19:27] <ubitux> assuming foo(a,b,c) and d local
[19:27] <ubitux> ok
[19:27] <ubitux> thx
[19:27] <ubitux> :)
[19:33] <llogan> ffmpeg.org being slow (for me at least)
[20:35] <llogan> wm4: yeah... phase could use a zample
[20:35] <llogan> blame paul.
[20:35] <wm4> anyway, I figured out that phase doesn't use flags
[20:36] <wm4> I was just confused
[20:36] <wm4> but still, fuck the avoption parser
[21:13] <cone-661> ffmpeg.git 03Michael Niedermayer 07master:04e06cdf7d35: avcodec: split mp2 encoder into float and fixed
[21:13] <cone-661> ffmpeg.git 03Michael Niedermayer 07master:babb611d3541: libavcodec/mpegaudio: change CONFIG_FLOAT to USE_FLOAT
[21:18] <re-G> JEEBsv: hi. you asked information about my flv metadata problem
[21:20] <re-G> i'm using pipe to output flv to flash player and it seems that ffmpeg puts duration correct but for some reason flash player (via mediaelement.js) cannot read the duration unless i put duration manually with -metadata option
[21:20] <ubitux> http://www.csr.com/products/60/aptx does anyone has sample/doc/whateverinfo about this codec?
[21:21] <ubitux> i'd eventually like to look at this codec if no one is working RE etc
[21:22] <wm4> what was done to vf_noise is pretty stupid
[21:23] <wm4> it now supports specifying noise params per component
[21:23] <wm4> which is not helpful at all, because you can't predictably set noise for e.g. luma only
[21:23] <wm4> because if the format is RGB, the first component is R and not luma
[21:23] <wm4> so you'll get noise in R only
[21:38] <cone-661> ffmpeg.git 03Stefano Sabatini 07master:704cc5e75df3: ffprobe: free shared writer context on exit
[21:40] <llogan> saste: http://funding.openinitiative.com/ might be worth looking into
[21:44] <saste> llogan, interesting
[00:00] --- Wed Dec  4 2013