[Ffmpeg-devel-irc] ffmpeg-devel.log.20130708

burek burek021 at gmail.com
Tue Jul 9 02:05:02 CEST 2013


[00:01] <Daemon404> guess ill tryy ffbisect
[00:01] <Daemon404> inb4 it fails too
[00:53] <cone-757> ffmpeg.git 03Paul B Mahol 07master:19436308b48b: dxa: make code independent of sizeof(AVFrame)
[01:08] <cone-757> ffmpeg.git 03Paul B Mahol 07master:bd03f99d05d0: gif: do not set avctx->coded_frame
[01:08] <cone-757> ffmpeg.git 03Paul B Mahol 07master:ec4e042f33f9: gif: remove unused AVFrame from encoder private context
[01:17] <durandal_1707> michaelni: what release critical bugs remain?
[01:29] <michaelni> durandal_1707, stuff marked as regression (minus unimportant ones)
[01:29] <durandal_1707> all remaining are unimportant IMO
[02:03] <cone-757> ffmpeg.git 03Paul B Mahol 07master:3bec6a44eab4: zmbvenc: do not set avctx->coded_frame
[02:03] <cone-757> ffmpeg.git 03Paul B Mahol 07master:099e57bc38d7: zmbvenc: remove unused AVFrame from encoder private context
[02:58] <cone-757> ffmpeg.git 03Michael Niedermayer 07master:5312c319be98: avformat_find_stream_info: fallback to timestamps in analyzeduration check in more cases
[03:51] <cone-757> ffmpeg.git 03Michael Niedermayer 07master:5d876be87a11: avio: Handle AVERROR_EOF in the same way as the return value 0
[03:51] <cone-757> ffmpeg.git 03Michael Niedermayer 07master:d35b6cd37754: rmdec: Use the AVIOContext given as parameter in rm_read_metadata()
[03:51] <cone-757> ffmpeg.git 03Ronald S. Bultje 07master:6c86a63bad77: yuv4mpeg: Correctly round chroma up for odd luma sizes
[03:51] <cone-757> ffmpeg.git 03Ronald S. Bultje 07master:46d208e1e097: vp8: Wait for prev_frame to parse segment_map before reading it
[03:51] <cone-757> ffmpeg.git 03Michael Niedermayer 07master:38e9585de993: Makefile: Remove stray tabs
[03:51] <cone-757> ffmpeg.git 03Michael Niedermayer 07master:7023715e8880: Merge commit '38e9585de993c32899588ab037180f2c930ce74c'
[04:13] <cone-757> ffmpeg.git 03Martin Storsjö 07master:21732063a346: movenc: K&R formatting cosmetics
[04:13] <cone-757> ffmpeg.git 03Michael Niedermayer 07master:4cd066b69c08: Merge commit '21732063a346475eb22c332b27e8216b79f9ad4a'
[04:47] <cone-757> ffmpeg.git 03Luca Barbato 07master:636ced8e1dc8: cmdutils: wrap exit explicitly
[04:47] <cone-757> ffmpeg.git 03Michael Niedermayer 07master:f982d006bb9e: Merge commit '636ced8e1dc8248a1353b416240b93d70ad03edb'
[04:59] <cone-757> ffmpeg.git 03Luca Barbato 07master:c4abc9098cac: vqavideo: check the version
[04:59] <cone-757> ffmpeg.git 03Michael Niedermayer 07master:b5a87dc193d6: Merge commit 'c4abc9098cacb227dba39bac6aea16b2bceba0d0'
[05:08] <cone-757> ffmpeg.git 03Luca Barbato 07master:76f5dfbfd902: westwood_vqa: do not free extradata on error in read_header
[05:08] <cone-757> ffmpeg.git 03Luca Barbato 07master:f054e309c588: qdm2: use init_static_data
[05:08] <cone-757> ffmpeg.git 03Michael Niedermayer 07master:d28a8ad8f33d: Merge commit 'f054e309c58894450a5d18cce9799ef58aab9f14'
[05:10] <highgod> Hi, I want to ask a quesion that I found a option in deshow of ffmpeg { "pixel_format", "set video pixel format", OFFSET(pixel_format), AV_OPT_TYPE_PIXEL_FMT, {.i64 = AV_PIX_FMT_NONE}, -1, AV_PIX_FMT_NB-1, DEC },
[05:10] <highgod> how can I input it?
[05:10] <highgod> using command?
[05:10] <highgod> thanks
[05:11] <BBB> highgod: ffmpeg -pixel_format ... -i something [output options] outfile
[05:16] <cone-757> ffmpeg.git 03Luca Barbato 07master:76efedeadb1f: qdm2: formatting cosmetics
[05:16] <cone-757> ffmpeg.git 03Michael Niedermayer 07master:e1f98f22313b: Merge commit '76efedeadb1f6bf79020c44a71dd0cee13d932ad'
[05:27] <cone-757> ffmpeg.git 03Luca Barbato 07master:744a11c99664: qdm2: check and reset dithering index per channel
[05:27] <cone-757> ffmpeg.git 03Michael Niedermayer 07master:44d0e520f85a: Merge remote-tracking branch 'qatar/master'
[07:08] <highgod> thanks
[07:34] <highgod> Hi, I want to ask a question, I use 64bit ffmpeg libs and dlls, it cursh on my code, but I use the 32bit dlls and libs and using the same code ,the code run correct, can anyone give some help? thanks, my code is compiled by vs2010
[10:50] <michaelni> highgod, if something crashes, use gdb or another debugger to find out why/where
[10:58] <highgod> Hi, michaelni, my systems problem, the same code run on other win7 64 is correct
[11:22] <cone-757> ffmpeg.git 03Michael Niedermayer 07master:efce76a80d26: ffmpeg_filter: replace exit() by exit_program()
[11:22] <cone-757> ffmpeg.git 03Michael Niedermayer 07master:eeddeb6579d8: ffmpeg_opt: replace exit() by exit_program()
[11:23] <cone-757> ffmpeg.git 03Michael Niedermayer 07master:5d3c3035b798: cmdutils: replace exit() by exit_program()
[11:35] <xlinkz0> i have an AVCodecContext->extradata_size set to 475691104
[11:36] <xlinkz0> this is on ios, might this be because of the platform or maybe an older bug?
[12:26] <cone-757> ffmpeg.git 03Carl Eugen Hoyos 07master:b86651a208ee: Check the return value of ff_rv34_decode_init() in rv30.c and rv40.c
[13:16] <xlinkz0> how would i go about compiling the ffmpeg tool into a library?
[13:17] <cptspiff> why on earth would you like to compile the tool as a library?
[13:20] <xlinkz0> so i can use it on ios and android
[13:21] <xlinkz0> sadly i can't understand complex makefiles :(
[13:21] <xlinkz0> don't*
[13:21] <cptspiff> well then you know where to start..
[13:23] <cptspiff> it will be a matter of writing a wrapper for main()
[13:23] <cptspiff> plus probably renaming main..
[13:24] <saste> xlinkz0, don't do, unless you're a wizard
[13:24] <saste> use system(..) instead
[13:24] <cptspiff> no such thing on ios/droid.
[13:24] <xlinkz0> indeed, that still relies on processes
[13:25] <xlinkz0> i can't spawn processes 
[13:25] <xlinkz0> also wrapping it is only a matter of adding the thread local storage keyword in front of every global
[13:25] <xlinkz0> shouldn't be hard
[13:26] <xlinkz0> and compiling without threading support
[13:26] <xlinkz0> cptspiff: but if i don't know how to compile it into an .so it is useless , any tips on that?
[13:26] <cptspiff> uhm, -shared ?
[13:27] <xlinkz0> what?
[13:27] <cptspiff> http://gcc.gnu.org/onlinedocs/gcc/Link-Options.html
[13:27] <xlinkz0> i need the full command ffmpegs makefile uses to compile ffmpeg.c
[13:27] <cptspiff> make V=1
[13:28] <xlinkz0> sweet
[13:28] <xlinkz0> thank you
[14:24] <j-b> Good morning
[14:24] <durandal_1707> good night
[14:33] <durandal_1707> saste: are you porting any of mp filters?
[14:33] <saste> durandal_1707, no
[14:34] <durandal_1707> why?
[14:35] <saste> why should i?
[14:36] <durandal_1707> why should you not?
[14:41] <durandal_1707> ubitux: fieldmatch is missing 440
[16:07] <ubitux> durandal_1707: please hold until august, or feel free to take over the filter until then
[16:08] <durandal_1707> hold for what?
[16:09] <durandal_1707> michaelni: diracdec is still using deprecated stuff
[16:11] <durandal_1707> what's point of flatten attribute? it does not work for clang here and just spam
[16:12] <durandal_1707> we will never get stable api
[16:14] <durandal_1707> ubitux: stop watching anime
[16:14] <ubitux> :(
[16:15] <ubitux> honestly i'm not really available, and starting saturday i'll be even less
[16:15] <ubitux> i come back in august for the vpx project, so back on ffmpeg
[16:15] <ubitux> but until then that's not easy for me
[16:18] <durandal_1707> saste: well then explain it to me why you using old drawutils in rotate filter?
[16:19] <durandal_1707> there are 2 freaking drawutils and they are both suboptimal
[16:26] <saste> durandal_1707, because I ported the code, and I didn't bother to port it to the new drawutils
[16:29] <durandal_1707> saste: your AVERRROR(err); causes segv
[16:30] <cone-757> ffmpeg.git 03Michael Niedermayer 07master:1ba01d3d46e8: swscale: treat gray as a fullrage (0-255) format
[16:30] <saste> that's because you have too many "R"s
[16:32] <durandal_1707> :) no, that's typo above
[16:33] <durandal_1707> why calling AVERROR() twice?
[16:34] <saste> durandal_1707, err = AVERROR(errno);
[16:34] <saste> now you call log, which could change errno
[16:34] <saste> return err;
[16:45] <durandal11707> for metadata i use lavfi.psnr.mse.r/g/b , lavfi.psnr.s.r/g/b , lavfi.psnr.mse & lavfi.psnr.s
[17:03] <cone-757> ffmpeg.git 03Paul B Mahol 07master:6150bec3f873: lavfi: add psnr filter
[17:14] <cone-757> ffmpeg.git 03Michael Niedermayer 07master:e096283ea55b: mp3dec: detect CBR and use CBR axiom to seek
[18:14] <cone-757> ffmpeg.git 03Alexandre Sicard 07master:70b922f371e2: avformat/mov: read stream bitrates from isml manifest
[18:36] <durandal11707> saste: i move rotate to new drawutils
[18:47] <saste> durandal11707, nice
[18:48] <cone-757> ffmpeg.git 03Paul B Mahol 07master:d2dd1e075cf9: lavfi/pad: use FFSIGN
[18:48] <cone-757> ffmpeg.git 03Paul B Mahol 07master:55c94f48ce5a: lavfi/drawutils: fix planar rgb
[18:48] <cone-757> ffmpeg.git 03Paul B Mahol 07master:19c8f2271423: lavfi/rotate: switch to new drawutils
[19:25] <durandal11707> saste: how should metadata be displayed with drawtext?
[19:25] <durandal11707> some kind of expansion in text?
[19:25] <saste> durandal11707, %{metadata:key}
[19:26] <wm4> durandal11707: add lua scripting and let the scripts decide
[19:28] <durandal11707> easier said than done
[19:29] <wm4> true
[19:29] <wm4> %{metadata:key} is a decent solution too... it's yet another kind of special syntax though
[19:47] <cone-757> ffmpeg.git 03Michael Niedermayer 07master:7dee9c22686f: tests/fate/filter-audio: update asyncts reference
[19:56] <michaelni> did someone test if the msvc10 fate-filter-metadata-ebur128 is related to mmx/emms ?
[20:00] <nevcairiel> its a bug in the compiler
[20:00] <nevcairiel> the loop vectorizer does some wrong things when optimizing double variables in loops when sse2 is enabled
[20:00] <nevcairiel> ...which you cant disable on x64 :)
[20:02] <nevcairiel> MS ack'ed the bug, will be fixed sometime in the future
[20:02] <nevcairiel> possibly VS2013
[20:06] <durandal11707> i can't get expr in drawtext expansion to use?
[20:07] <durandal11707> and there are no such examples in doc
[20:36] <durandal11707> text=%{localtime:%a %b %d %Y} 
[20:36] <durandal11707> doesn't work here
[20:41] <mark4o> durandal11707: probably quoting issue
[20:41] <mark4o> works: -vf "drawtext=fontfile=/Library/Fonts/DejaVuSans-Bold.ttf:fontcolor=red:text='%{localtime\:%a %b %d %Y}'"
[20:42] <durandal11707> i use filter_script
[20:42] <mark4o> then you don't need the shell quoting but still probably need backslash
[20:47] <Daemon404> it's cause we use : in our parser
[20:47] <Daemon404> it drives me nuts too (on windows)
[20:47] <Daemon404> paths turn into a shitshoot
[21:00] <wm4> quoting hell
[21:02] <durandal11707> thing is metadata is added to frame in one filter, and in another filter its gone
[21:04] <wm4> eh, AVFrame.metadata exists?
[21:04] <wm4> weird
[21:04] <durandal11707> why would that be weird?
[21:04] <durandal11707> this worked when i did it manually in drawbox
[21:14] <durandal11707> wtf AV_DICT_IGNORE_SUFFIX means?
[21:31] <Daemon404> lets start that bisect again...
[21:31] <Daemon404> maybe this time a transformer wont blow up near my house
[21:44] Action: Daemon404 retries yet again
[21:44] <Daemon404> finally figured out how to ue ffbisect
[21:48] <wm4> huh, so ffbisect is a script created by another script?
[21:48] <nevcairiel> its rather confusing
[21:48] <Daemon404> wm4, it copies itself
[21:49] <Daemon404> so you always have the proper version
[21:49] <Daemon404> because it's also in git
[21:49] <Daemon404> so a previosu commit may not have it
[21:49] <Daemon404> you need to use it untracked
[21:49] <Daemon404> hence the copy.
[21:49] <wm4> that's... interesting
[22:04] <cone-757> ffmpeg.git 03Michael Niedermayer 07master:970c8df73528: mpeg12dec: avoid reinitializtation on PS changes when possible.
[22:15] <cone-757> ffmpeg.git 03Paul B Mahol 07master:a03e79ed5a71: lavfi: fix broken logic in metadata handling
[22:29] <durandal11707> i get this drawtext expansion working from script but from console there is no way
[22:31] <durandal11707> Daemon404: didn't thing asyncts have be done in swr ?
[22:32] <Daemon404> .. i cant parse that sentence
[22:33] <Daemon404> seriously tho, you guys are a bunch of children.
[22:34] <Daemon404> wm4, tl;dr to my email rpely to you ^
[22:34] <durandal11707> Daemon404: what asynts have?
[22:34] <wm4> Daemon404: huh?
[22:35] <Daemon404> wm4, re: why not enable avresample
[22:35] <wm4> ah
[22:35] <durandal11707> wm4 did replied?
[22:35] <Daemon404> durandal11707, see my email
[22:35] <Daemon404> i have explained what asyncts like 6 times
[22:35] <Daemon404> and im fucking tired of it
[22:35] <Daemon404> grep your goddamn logs
[22:35] <Daemon404> ive explaiend it to YOU before
[22:36] <durandal11707> i don't have logs, and you just act as giant douche
[22:36] <Daemon404> because i can only explain somethin so many times
[22:36] <Daemon404> before it gets on my nerves
[22:36] <Daemon404> becaue everyone ha the memory of a goldfish
[22:37] <Daemon404> what's that? i mention avresample?
[22:37] <Daemon404> TIME TO INTERROGATE DEREK AGAIN
[22:37] <Daemon404> every. single. time.
[22:37] <durandal11707> i wanted to port it to swresample...
[22:38] <Daemon404> if you recall, we collectively decided that was a bad idea
[22:38] <Daemon404> and that aresample should improve instead
[22:38] <Daemon404> fwiw, i see FAR more usage of avresample in the wild
[22:38] <Daemon404> i actually know of no project using swresample
[22:38] <Daemon404> even ffmpeg-only ones
[22:39] <durandal11707> and michaelni added something to swresample, i think delay in the beginning of audio...
[22:39] <durandal11707> i see no usage of avresample at all
[22:39] <Daemon404> then you are blind
[22:39] <Daemon404> lav filters, vlc, ffms2, mplayer2, mpv
[22:40] <durandal11707> hilarious
[22:40] <durandal11707> 3 of 5 don't use ffmpeg at all
[22:40] <Daemon404> because clearly we shouldnt count those.
[22:40] <Daemon404> cherry pick your stats etc
[22:41] <wm4> I have ~35 lines of ifdeffery to make it work with both, but that's fugly
[22:41] <durandal11707> since when we pick stuff becaues of "popularity"
[22:41] <Daemon404> i am so tired of this fork sit
[22:41] <Daemon404> shit*
[22:41] <wm4> only now?
[22:41] <Daemon404> wm4, if youre not on one side or the other
[22:42] <Daemon404> the only option is to go insane
[22:42] <Daemon404> and laugh at how childish everyone is
[22:42] <wm4> heh
[22:42] <Daemon404> also
[22:42] <Daemon404> [16:40] <@durandal11707> 3 of 5 don't use ffmpeg at all <-- i think you mean 1 of 5
[22:42] <Daemon404> only vlc is libav only.
[22:43] <Daemon404> at least get your facts straight.
[22:43] <durandal11707> i see no facts
[22:43] <wm4> lav filters is ffmpeg only, and I think mplayer2 is probably broken with recent git versions of libav
[22:43] <Daemon404> ffms, mpv, mplayer2, and lav all can use ffmpeg.
[22:44] <Daemon404> and in fact ffmpeg is the more supported of the two projects
[22:44] <durandal11707> i don't see facts, i see only bad choices
[22:44] <wm4> what does xbmc do?
[22:44] <Daemon404> you fucking people -_-
[22:44] <Daemon404> i am so fucking tired of thsi tribalistic bullshit
[22:44] <wm4> lol
[22:45] <iive> what is asyncts about?
[22:46] Action: beastd could have predicted a quit
[22:46] <durandal11707> stream timestamp syncronization
[22:47] <durandal11707> it inserts silence in gaps iirc
[22:48] <durandal11707> and it looks to me there is nothing similar when avresample is not enabled
[22:49] <durandal11707> which is like bad, as this should not be lib limitation, even if it may appear as such
[22:50] <durandal11707> and i don't see reason to have 2 libraries doing same thing enabled at same time
[23:07] <michaelni> durandal11707, swresample should supports all that silence insert and time sync and stretching samples to compensate timestamp drift and ...
[23:19] <durandal11707> should ... but does not ...
[23:20] <michaelni> that would be a bug
[23:28] <michaelni> also see "ffmpeg -help filter=aresample" for options to tune the async stuff with swresample
[00:00] --- Tue Jul  9 2013


More information about the Ffmpeg-devel-irc mailing list