[Ffmpeg-devel-irc] ffmpeg-devel.log.20130430

burek burek021 at gmail.com
Wed May 1 02:05:02 CEST 2013


[00:36] <ubitux> HEADERS-$(CONFIG_OPENCL)                += opencl.h
[00:36] <ubitux> erm
[00:37] <ubitux> this is not supposed to be installed, right?
[00:37] <ubitux> or do we?
[00:37] <ubitux> we export some opencl api?
[00:38] <cone-334> ffmpeg.git 03Clément BSsch 07master:d558e46469e0: lavfi: add two opencl headers to SKIPHEADERS.
[01:21] <cone-334> ffmpeg.git 03Clément BSsch 07master:a466c2c470f7: fate: fix filtfmts-test dependency after 9a5924f.
[02:16] <cone-334> ffmpeg.git 03Clément BSsch 07master:e881057c982e: fate: add a rule for pixfmts tests.
[02:16] <cone-334> ffmpeg.git 03Clément BSsch 07master:1e0a0aafba62: fate: port our pixfmts change to fate-run.sh.
[02:16] <cone-334> ffmpeg.git 03Clément BSsch 07master:93aede41ce05: fate: move kerndeint test to the new system.
[02:16] <cone-334> ffmpeg.git 03Clément BSsch 07master:45734dea9e03: fate: port remaining filters to new system.
[02:16] <cone-334> ffmpeg.git 03Clément BSsch 07master:2609434b45ab: fate: remove forgotten transpose filter results from old system.
[02:16] <ubitux> ok now we are good
[02:16] <ubitux> ...and i need to fix what i broke :p
[02:17] <ubitux> i hope filter=${filter%_*} will work properly (sh/dash looks ok)
[02:17] <ubitux> michaelni: was there anything else while i'm still in fate?
[02:19] <michaelni> hmm dunno, tests for every decoder, encoder, muxer, demuxer and ffplay
[02:20] <ubitux> michaelni: i meant from the fork merge PoV :p
[02:20] <ubitux> sorry i'm not motivated to work for days on adding tests :p
[02:20] <michaelni> i know and noone is
[02:21] <ubitux> :)
[02:21] <ubitux> socis!
[02:21] <ubitux> :)
[02:21] <michaelni> hmm, yes that sounds like a good  idea
[03:00] <burek> is there any work planned on intel's quick sync (in ffmpeg of course): http://en.wikipedia.org/wiki/Intel_Quick_Sync_Video
[03:07] <Compn> do you sample ?
[03:07] <Compn> oh nm
[03:07] <Compn> newer than atom stuff then ?
[03:07] <Compn> or is it the same api as the atom chips ?
[08:54] <cone-480> ffmpeg.git 03Diego Biurrun 07master:97c56ad7961e: x86: dsputil: Remove a set of pointless #ifs around function declarations
[08:54] <cone-480> ffmpeg.git 03Michael Niedermayer 07master:01a5a3a2e8ac: Merge remote-tracking branch 'qatar/master'
[11:14] <durandal_1707> hmm 3 test are disabled or?
[11:18] <durandal_1707> ubitux: gifenc is failing on big-endian?
[11:18] <michaelni> gif fails due to designb flaw in sidedata, its endian dependant IIRC
[11:19] <durandal_1707> ah
[11:28] <cehoyos> michaelni: The bug ("blue") that you fixed in xyz is not in Matthias Buercher's original filter: http://www.belle-nuit.com/xyz2rgb-video-filter
[11:40] <cone-480> ffmpeg.git 03Paul B Mahol 07master:2bdd434931e1: lavfi/stereo3d: add support for alternating frames input & output format
[12:25] <cone-480> ffmpeg.git 03Paul B Mahol 07master:e85f02e3f5e2: fate: add test for telecine filter
[12:25] <cone-480> ffmpeg.git 03Paul B Mahol 07master:468ec1714c18: fate: add test for swapuv filter
[12:25] <cone-480> ffmpeg.git 03Paul B Mahol 07master:a0adeb1f9f9b: fate: add test for separatefields filter
[12:54] <durandal_1707> beastd: could you make art contest to get something like this:
[12:54] <beastd> hi durandal_1707
[12:54] <durandal_1707> https://lh6.googleusercontent.com/-UVCROnHgAVw/UWa7bDpj2lI/AAAAAAAAAMk/jtiDl4WAMdQ/s1550-fcrop64=1,00000000ffffff6e/gplus.png
[12:55] <durandal_1707> perhaps even new logo, (zigzag is boring)
[12:56] <beastd> durandal_1707: ok. will do. but please give me a few days. you can nag me here daily on IRC though
[12:57] <beastd> durandal_1707: You want new art for g+ ?
[12:58] <durandal_1707> it does not g+ mentioned in contest
[12:58] <durandal_1707> its just different size of logo
[12:59] <durandal_1707> but not done wiht -vf scale=
[12:59] <beastd> of course not
[13:00] <beastd> durandal_1707: ok will make the contest description generic.
[13:01] <beastd> see you
[13:01] Action: beastd goes AFK
[13:42] <cone-480> ffmpeg.git 03Michael Niedermayer 07master:147a45fca4d8: framecrcenc: workaround design flaw in sidedata palette
[13:42] <cone-480> ffmpeg.git 03Michael Niedermayer 07master:53efb2fa09c0: gif: Fix little endian specific code
[14:20] <durandal_1707> is it possible to set outlink format depending on inlink format?
[14:21] <durandal_1707> say input format is 8bit, i want output format to be 8bit to and not 16bit
[14:37] <durandal_1707> so the only place one can insert outpads is in filter .init ?
[15:10] <durandal_1707> the limitation that its only possible to insert outpads in .init is really bad
[15:38] <durandal_1707> is it even possible to fix this without writing another api?
[15:44] <durandal_1707> michaelni: i hope you applied with "useless" removed
[15:47] <michaelni> durandal_1707, why should i have remvoed it ?
[15:48] <michaelni> after all you choose not to remove it when sending the patch :)
[15:51] <durandal_1707> if that issue about dynamic pads is resolved it will certainly become more useful
[15:57] <durandal_1707> omg, this is ridiculous, renaming filter contexts ...
[16:01] <BuxiNess> Well, have a nice Worker day, and don't forget the lily of the valley
[16:09] <durandal_1707> instead of doing one filter that extract planes, add several filter that extract single plane (alphaextract is already preset), that gives 6 extra filters, (or even more for > 8 bit)
[16:28] <Bor0> what variable in ffplay is used to specify size of buffer for reading videos through http?
[16:30] <durandal_1707> -inbuf?
[16:31] <Bor0> can't find such variable in ffplay
[16:31] <durandal_1707> infbuf
[16:32] <Bor0> I'll look into that, thanks
[16:51] <Bor0> what's the easiest way to check for "buffering" indicator?
[16:52] <durandal_1707> i don't think there is such thing
[16:52] <Bor0> could you tell me around which part of ffplay would I look at to implement it?
[16:54] <durandal_1707> i can not tell you, because i do not know anything about ffplay
[16:55] <durandal_1707> but anyway, why you want that in ffplay?
[16:55] <Bor0> I see, okay. I'll look around av_read_frame perhaps.
[16:55] <Bor0> I have converted it to libffplay, which then I am embedding it to my browser plugin
[16:56] <durandal_1707> ther buffering thing is part of protocol, so its internal in libavformat, and afaik user can not control it
[16:56] <Bor0> I only need indicator when the video is buffering, instead of getting frozen frame
[16:57] <durandal_1707> i don't think there is such thing, if you want it open feature request on our trac
[16:58] <ubitux> durandal_1707: i need to check something for gce, i think i missed something when testing with browsers
[16:58] <ubitux> (like a cache issue or something, making me think it worked)
[16:58] <durandal_1707> aha...,
[17:42] <durandal_1707> ubitux: looks like anto is making changes to allow config_props stuff to be called multiple times between single init/uninit
[17:42] <ubitux> possibly live reconfiguration or something
[17:45] <durandal_1707> i wonder how i could resolve this dynamic output thing
[17:59] <durandal_1707> is nullsrc supposed to return uninitialized data?
[18:20] <durandal_1707> is it possible to save each filter output to separate file?
[18:36] <michaelni> no but it sounds like such feature could be usefull
[18:38] <durandal_1707> i mean filter that have several outputs
[18:40] <durandal_1707> ffmpeg -f lavfi -i testsrc -filter_comples split=4 ... 
[18:40] <durandal_1707> i want that each 4 "streams" are saved in separate file and not "magically" interleaved in single one
[18:49] <durandal_1707> got it: ffmpeg -f lavfi -i smptebars,format=yuva444p -filter_complex "extractplanes[y][u][v][a]" -frames 5 -map '[y]' /tmp/y%d.png -frames 5 -map '[u]' /tmp/u%d.png -frames 5 -map '[v]' /tmp/v%d.png -frames 5 -map '[a]' /tmp/a%d.png
[22:32] <ubitux> couldn't we make use of openmp in most filters?
[22:32] <ubitux> also, how could we that would work with our threading?
[23:10] <seanDuncan> I'm trying to use x264's qpfile option through ffmpeg's 'x264opts' option but qpfile doesn't seem to be working from there. is it called something else in ffmpeg?  
[23:11] <durandal11707> seanDuncan: wrong channel
[23:11] <seanDuncan> I oh sorry
[23:11] <seanDuncan> #ffmpeg more appropriate?
[00:00] --- Wed May  1 2013


More information about the Ffmpeg-devel-irc mailing list