[Ffmpeg-devel-irc] ffmpeg-devel.log.20111123

[burek]

[00:42] <Compn> i've hit that problem when git pushing
[00:42] <Compn> that i merge but forget to pull
[00:42] <Compn> stupid fast forward gets that extra merge message
[00:42] Action: Compn barely has any idea how git works :D
[00:42] <Compn> michaelni does a great job merging so many trees
[00:42] <Compn> :)
[00:45] <ubitux> Compn: git pull --rebase 
[00:46] <ubitux> but it's better to work in your dev branch anyway :P
[02:06] <CIA-36> ffmpeg: 03Lou Logan 07master * ra8e6d4d403 10ffmpeg/tools/patcheck: 
[02:06] <CIA-36> ffmpeg: patcheck: spelling and grammar nits
[02:06] <CIA-36> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[02:16] <bcoudurier> j-b t'es dans le coin encore ?
[02:26] <j-b> bcoudurier: ouais, je sais, je devrais être au lit
[02:51] <sam1> hi guys ,is ffmpeg  could hard burn  subtitles  into  movie?
[02:51] <relaxed> sam1: no
[02:52] <sam1> mencoder  could  do it ?
[02:52] <relaxed> yes
[02:55] <sam1> does ffmpeg  support this  features   in the future?
[02:55] <relaxed> sam1: ask in #ffmpeg or #mplayer, not here.
[02:55] <sam1> thanks  relaxed  
[03:55] <Compn> relaxed : ffmpeg has subtitle support now
[03:55] <Compn> sam1 : ffmpeg -newsubtitle file.srt
[03:56] <Compn> i thought ?
[03:56] <Compn> maybe i am wrong
[04:14] <CIA-36> ffmpeg: 03Anton Khirnov 07master * r7204ec1a88 10ffmpeg/avconv.c: 
[04:14] <CIA-36> ffmpeg: avconv: split off streamcopy handling into a separate loop.
[04:14] <CIA-36> ffmpeg: This is easier to understand and is less likely to break horribly when a
[04:14] <CIA-36> ffmpeg: stream is to be both decoded and copied.
[04:14] <CIA-36> ffmpeg: 03Anton Khirnov 07master * r2a651b719c 10ffmpeg/avconv.c: avconv: move streamcopy-only code out of decoding loop.
[04:14] <CIA-36> ffmpeg: 03Anton Khirnov 07master * r78162b4ea2 10ffmpeg/avconv.c: avconv: reindent.
[04:14] <CIA-36> ffmpeg: 03Anton Khirnov 07master * rded28ba35b 10ffmpeg/avconv.c: avconv: split audio transcoding out of output_packet().
[04:14] <CIA-36> ffmpeg: 03Anton Khirnov 07master * r45d4b66f6f 10ffmpeg/avconv.c: avconv: split video transcoding out of output_packet().
[04:14] <CIA-36> ffmpeg: 03Anton Khirnov 07master * r9595234c94 10ffmpeg/avconv.c: avconv: split subtitle transcoding out of output_packet().
[04:14] <CIA-36> ffmpeg: 03Mans Rullgard 07master * r8ee2b4672f 10ffmpeg/libavcodec/arm/asm.S: (log message trimmed)
[04:14] <CIA-36> ffmpeg: ARM: add explicit .arch and .fpu directives to asm.S
[04:14] <CIA-36> ffmpeg: This prevents build errors when compiler and assembler default
[04:14] <CIA-36> ffmpeg: targets differ. Ideally each file would declare the highest
[04:14] <CIA-36> ffmpeg: level it requires. This is however not easily possible as it
[04:14] <CIA-36> ffmpeg: complicates assembling pre-armv6t2 code in Thumb-2 mode.
[04:14] <CIA-36> ffmpeg: HAVE_NEON is used as indicator for ARMv7-A since no other
[04:14] <CIA-36> ffmpeg: 03Mans Rullgard 07master * r3fe5fc9325 10ffmpeg/ (29 files in 5 dirs): 
[04:14] <CIA-36> ffmpeg: regtest: split video encode/decode tests into individual targets
[04:14] <CIA-36> ffmpeg: Signed-off-by: Mans Rullgard <mans at mansr.com>
[04:14] <CIA-36> ffmpeg: 03Janne Grunau 07master * r0945eddec0 10ffmpeg/ (3 files in 2 dirs): (log message trimmed)
[04:14] <CIA-36> ffmpeg: pthread: do not touch has_b_frames
[04:14] <CIA-36> ffmpeg: Adding the thread count in frame level multithreading to has_b_frames
[04:14] <CIA-36> ffmpeg: as an additional delay causes more problems than it solves.
[04:14] <CIA-36> ffmpeg: For example inconsistent behaviour during timestamp calculation in
[04:14] <CIA-36> ffmpeg: libavformat.
[04:14] <CIA-36> ffmpeg: Thread count and frame level multithreading are both set by the user.
[04:14] <CIA-36> ffmpeg: 03Luca Barbato 07master * r384bdaceeb 10ffmpeg/libavutil/pixfmt.h: 
[04:14] <CIA-36> ffmpeg: doxy: cleanup pixfmt.h
[04:15] <CIA-36> ffmpeg: Remove the dubious warning about the header being private and
[04:15] <CIA-36> (40 lines omitted)
[05:59] <relaxed> Compn: I think he meant hard subbing
[06:07] <Compn> yeah i'm probably wrong
[06:07] <Compn> ubitux just started a thread about the subject ;)
[06:18] <relaxed> link?
[07:20] <ubitux> relaxed: http://lists.mplayerhq.hu/pipermail/ffmpeg-devel/2011-November/117059.html
[07:35] <relaxed> ubitux: The ability to hard sub srt/ass would certainly make ffmpeg more popular. I hope something comes of this.
[07:35] <ubitux> the ass should be supported soon
[07:37] <relaxed> I assume it will be a filter? Will it support custom ass options like mplayer's -ass-force-style?
[07:40] <ubitux> it's a filter yes: http://lists.mplayerhq.hu/pipermail/ffmpeg-devel/2011-September/115216.html
[11:37] <vivienschilis> hi guys
[11:38] <vivienschilis> I have an ffprobe issue since a recent commit to print the video resolution
[13:09] <vivienschilis> any devs here?
[13:11] <av500> all gone for the weekend
[13:23] <michaelni> vivienschilis, which commit ?
[13:24] <michaelni> and what issue exactly
[13:25] <vivienschilis> http://git.videolan.org/?p=ffmpeg.git;a=commit;h=29a29226bb01d8e46b731e73f637563e5c4cccfe
[13:25] <vivienschilis> ffprobe
[13:25] <vivienschilis> return a different width 
[13:25] <vivienschilis> if you do
[13:25] <vivienschilis> from ffmpeg
[13:25] <vivienschilis> if you do ffmpeg -i myvideo
[13:26] <vivienschilis> than if you do ffprobe show_streams
[13:26] <vivienschilis> example video
[13:26] <vivienschilis> http://panda-test-harness-videos.s3.amazonaws.com/panda.mp4
[13:26] <vivienschilis> ffprobe will return 304
[13:27] <vivienschilis> but the ffmpeg "classic" output will return 300
[13:31] <CIA-36> ffmpeg: 03Michael Niedermayer 07master * r8cda755c2f 10ffmpeg/libavfilter/formats.c: 
[13:31] <CIA-36> ffmpeg: avfilter: dont write out of array for duplicate formats in avfilter_merge_formats()
[13:31] <CIA-36> ffmpeg: Note, this is a theoretical fix, iam not aware of a case where this function receives
[13:31] <CIA-36> ffmpeg: duplicates in its input.
[13:31] <CIA-36> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[13:31] <CIA-36> ffmpeg: 03Michael Niedermayer 07master * r4d874fc2b1 10ffmpeg/libavfilter/vf_mp.c: 
[13:31] <CIA-36> ffmpeg: vf_mp: rewrite vf_mp filter warning message
[13:31] <CIA-36> ffmpeg: Warning text by: Clément BSsch <ubitux at gmail.com>
[13:31] <CIA-36> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[13:35] <vivienschilis> michaelni did I hit a bug here or I am just expecting the wrong value :)?
[13:42] <michaelni> vivienschilis, probably the width values at demuxer and decoder level differ
[13:45] <vivienschilis> michaelni, ok it's just that if I use the previous commit then I get the right value
[13:46] <vivienschilis> by previous I mean one commit before 29a29226bb01d8e46b731e73f637563e5c4cccfe
[14:15] <michaelni> vivienschilis, interresting issue ideed, ive a fix locally but iam not sure if it might have side effects
[14:28] <michaelni> posted an explanation of this issue in the thread from the patch that caused it "Re: [FFmpeg-devel] video corruption using h264 hw decoders"
[14:41] <CIA-36> ffmpeg: 03Michael Niedermayer 07master * r248bb81ebc 10ffmpeg/libavfilter/formats.c: 
[14:41] <CIA-36> ffmpeg: avfilter: Error out from avfilter_merge_formats() when there are duplicates
[14:41] <CIA-36> ffmpeg: instead of trying to continue.
[14:41] <CIA-36> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[14:42] <vivienschilis> thanks michaelni, I spent a few hours debugging avcodec_open2 but I don't have the knowledge of the ffmpeg's internal code to know what's going on ;)
[20:04] <ubitux> mmh
[20:04] <ubitux> we have a srt encoder
[20:04] <ubitux> but no muxer?
[20:04] <ubitux> so we can't mux as srt in mkv for instance?
[20:04] <ubitux> (but can output a srt?)
[20:05] <ubitux> seems supported thought&
[20:06] <Daemon404> ok 
[20:06] <Daemon404> question
[20:06] <Daemon404> why is it called an srt "encoder"
[20:06] <Daemon404> that just seems, well, wrong.
[20:06] <ubitux> because it encodes a AVSubtitle to a SubRip buffer
[20:06] <ubitux> ?
[20:07] <Daemon404> i dunno, i dotn call something like that encoding, per se
[20:07] <Daemon404> thats like me calling a regex that does ass -> srt a transcoder
[20:07] <ubitux> muxing is just a concat of those buffers i guess
[20:07] <ubitux> no it's not Daemon404 
[20:07] <Daemon404> and isnt AVSubtitle like ass + timestamp
[20:07] <ubitux> the srt encoder receives a AVSubtitle: which is the decoded form
[20:08] <Daemon404> er, pts
[20:08] <Daemon404> i dunno, i dont classify parsing text and putting it i na buffer as decodingh
[20:08] Action: Daemon404 shrugs
[20:08] <ubitux> there is no parsing in the encoder
[20:08] <ubitux>         if (st->font_name && strcmp(st->font_name, ASS_DEFAULT_FONT) ||
[20:08] <ubitux>             st->font_size != ASS_DEFAULT_FONT_SIZE ||
[20:08] <ubitux>             c != ASS_DEFAULT_COLOR) {
[20:08] <ubitux>             srt_print(s, "<font");
[20:08] <ubitux> this is the kind of code you have
[20:09] <ubitux> it is already parsed in a AVSubtitle
[20:09] Action: Daemon404 doesnt recognize either as decoding or encoding
[20:09] <ubitux> of course the internal struct match a ASS struct naming
[20:09] <Daemon404> anyway, no point in arguing over semantics
[20:10] <Daemon404> i do have a question about libavsubtitle or w/e
[20:10] <Daemon404> you arent planning and duplicating libass's functionality are you?
[20:11] <ubitux> i'm planning to use libass the more i can
[20:11] <Daemon404> ok
[20:11] <Daemon404> cause one thing libass has going for it that i find very important is
[20:11] <Daemon404> on windows, the possibility of no fontconfig dep
[20:11] <Daemon404> cause fontconfig on windows is a pile of crap and pain and waiting 30 mins fo rit to parse fonts
[20:11] <Daemon404> every time you install a new one
[20:12] <ubitux> libass will take care of the text rendering
[20:12] <Daemon404> but not font handling?
[20:12] Action: Daemon404 rules out ever usign libavsubtitle on windows
[20:13] <ubitux> i don't know how the fond will be handled, i forgot to mention it in the initial mail
[20:13] <ubitux> fonts*
[20:13] <Daemon404> its -very- important
[20:13] <ubitux> yes
[20:13] <ubitux> i mentionned it in the libass filter thought
[20:13] <ubitux> i'll raise the topic
[20:13] <Daemon404> the sole reason vsfilter is still prominent on windows for that is because nobody wanted to use libass with fontconfig on windows :P
[20:14] <ubitux> i need to look better at the code, collect my thought, and reply to everyone
[20:14] <Daemon404> and now it has part of a gdi backend, so a dshow filt is beign written almost instantly
[20:14] <Daemon404> ok. just dont read vsfilter's code. you'll go insane.
[20:22] <Compn> on windows one can use ffdshow vfw codec to burn subtitles :P
[20:23] <Compn> ffdshow is amazing, combining mplayer's filters and ffmpeg codecs
[20:24] <Daemon404> ffdshow
[20:24] <Daemon404> is a fucking pile of crap
[20:24] <Daemon404> at this point
[20:24] <Compn> lol
[20:24] <Daemon404> a giant mess of hacks 
[20:25] <Compn> any less hacky than windows directshow or video for windows ?
[20:25] <Daemon404> it's entire base is a fucked up fork of libav and ffmpeg mixed
[20:25] <Daemon404> with lots of /* ffdshow custom code */
[20:25] <Daemon404> inside of said fork, when theres no reason for it
[20:25] <Daemon404> and its own code is terrible
[20:26] <Daemon404> also on windows, why the heck would you use it to burn subtitles? avisynth + vsfilter or avisynth + libass
[20:27] <nevcairiel> too bad the libass no-fontconfig-on-windows branch is not finished yet
[20:27] <nevcairiel> it seems like he stopped working on it after GSoC ended =P
[20:28] <Daemon404> nevcairiel, a friend at MS of mine has interest in finishing it
[20:28] <Daemon404> and a dshow filt for libass is already being done
[20:28] <Daemon404> and semi-working
[20:28] <Daemon404> hope exists!
[20:28] <JEEB> <Daemon404> also on windows, why the heck would you use it to burn subtitles? avisynth + vsfilter or avisynth + libass <- I think of this as rather herp derp for a beginner, mplayer2's encoding branch would IMHO be much more easy-to-use and supports VFR/libass and all that herp derp
[20:29] <nevcairiel> burning subs onto a image is easy, especially with libass which has a nice API
[20:29] <nevcairiel> the hard part is getting it right
[20:29] <Daemon404> JEEB, no
[20:29] <Daemon404> avisynth is easier to use for beginners
[20:29] <Daemon404> simpyl cause no cmd line
[20:29] <Daemon404> see: o9k newb encoders
[20:30] <Daemon404> btw nevcairiel i found a strange bug in either lav splitter or ffdshow with rmvb... did i already tell you about this?
[20:30] Action: Daemon404 cannot remember
[20:30] <Daemon404> hmm yes i did
[20:30] <nevcairiel> If the libass font branch is finished some day, i actually have it on my roadmap to add it to lav video
[20:30] Action: Daemon404 stabs brain
[20:30] <nevcairiel> you did
[20:31] <JEEB> not to mention that people are being idiots, I just had someone on aegisub who wanted to break conventions and make the subtitles rendered so that the X/Y are on the rendered surface and not on the original picture
[20:31] <nevcairiel> lol
[20:32] <JEEB> thus pretty much breaking most libass/vsfilter using combos, except for MPC-HC's internal renderer and mplayer2 that has an option now
[20:32] <JEEB> Why can't people just understand that they're just making this mess worse and worse -.-
[20:32] <Daemon404> JEEB, to be perfectly fair
[20:32] <Daemon404> that shit is nto exactly well documented anywhere
[20:32] <Daemon404> and is pretty obfuscated
[20:33] Action: Daemon404 only knows all this shit after 6 years of fanshrubban
[20:33] <JEEB> he perfectly knew about fsx/fsy
[20:33] <Daemon404> he clearly didnt understand the problem tho
[20:33] <Daemon404> if he wanted that
[20:33] <JEEB> I don't know
[20:34] <JEEB> I think it was more of a case "I wanna do it like this and that's it >:| "
[20:34] Action: Daemon404 shrugs
[20:54] <Compn> yeah, well i never said i used ffdshow to burn subs
[20:55] <Compn> in fact, i've only used mencoder to do it once or twice
[20:55] <Compn> and that was years ago
[20:55] Action: Compn uses softsubs as much as possible
[20:55] <Daemon404> well it's basically only used to flashy kara now
[20:55] <Daemon404> s/to/for/
[20:56] <Daemon404> rest is softsubs or RGBA overlays
[20:59] <Compn> i wonder which anime company will be the first to release subtitled episodes on the same week as the japanese versions 
[21:00] <Compn> i guess funimation took over naruto
[21:00] <Compn> thats pretty close
[21:01] <JEEB> crunchyroll has done "on the same week/day", niconico's fate/zero is "released after the Tokyo MX airing ends"
[21:01] <JEEB> so speed-wise some official outlets have gotten quite quick
[21:01] <Compn> why would ffdshow have libav stuff? no, i'm better off not knowing
[21:01] <JEEB> -tryouts = a fork of its own
[21:01] <JEEB> although xhmikosr tells me it's supposedly libav
[21:02] <JEEB> just that clsid can't properly switch over files so it's effing mess
[21:02] <Daemon404> it just pulls stuff from random sides
[21:02] <JEEB> (grepping for libav and ffmpeg headers results in a lulzy end result)
[21:02] <Daemon404> there is no coherence
[21:02] <Daemon404> and then svn commit with a "update stuff" message
[21:03] <JEEB> yeah
[21:03] <JEEB> -.-
[21:04] <gnafu> Wow.
[22:18] <CIA-36> ffmpeg: 03Michael Niedermayer 07master * r6a3a54f5e3 10ffmpeg/libavformat/utils.c: 
[22:18] <CIA-36> ffmpeg: lavf: minor cleanup of av_probe_input_format3()
[22:18] <CIA-36> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[22:18] <CIA-36> ffmpeg: 03Michael Niedermayer 07master * r98d1e9f915 10ffmpeg/libavformat/utils.c: 
[22:18] <CIA-36> ffmpeg: lavf: give more weight to the extension during probing if there is a id3
[22:18] <CIA-36> ffmpeg: tag that is larger than the probe buffer.
[22:18] <CIA-36> ffmpeg: Fixes Ticket336
[22:18] <CIA-36> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[22:35] <Orphis> Thanks michaelni :)
[22:36] <michaelni> :)
[22:36] <Orphis> But I don't work for the company that needed the fix, so I can't really test the fix
[22:57] <pasteeater> i read the FFmpeg git howto and some of the Git book, but I'm not totally sure how to push some simple ffmpeg-web changes to remote.
[22:59] <Daemon404> git is great, but it isn't exactly user-friendly or straightforward or intuitive
[22:59] <Daemon404> in some respects
[23:00] <Daemon404> pasteeater, git clone -> make changes -> git commit -> patch review -> git push?
[23:02] <pasteeater> that's pretty much what i do usually (excluding git push).
[23:02] <pasteeater> videolan.org/Git shows that the url has to be changed in .git/config
[23:04] Action: pasteeater stops asking dumb questions and goes back to reading Git again
[23:04] <cbsrobot> pasteeater: ask dumb questions
[23:06] <Daemon404> youre better off asking
[23:06] <Daemon404> git docs are honestly, a pile of crap
[23:07] <michaelni> also git push --dry-run and looking at the result with git log -p before the actual push is a good idea
[23:26] <pasteeater> what's the proper url for [remote "origin"] in ffmpeg-web/.git/config to push?
[23:27] <cbsrobot> i'd guess git://ffmpeg.org/ffmpeg-web
[23:36] <Reuben> Anyone here know much about the AAC decoder? Got some issues with it...
[23:37] <CIA-36> ffmpeg: 03Michael Niedermayer 07master * r183e832c53 10ffmpeg/doc/git-howto.txt: 
[23:37] <CIA-36> ffmpeg: githowto: git is generally available as package in most distros.
[23:37] <CIA-36> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[23:37] <CIA-36> ffmpeg: 03Michael Niedermayer 07master * rb329fc346e 10ffmpeg/ffmpeg.c: 
[23:37] <CIA-36> ffmpeg: ffmpeg: fix indention
[23:37] <CIA-36> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[23:37] <CIA-36> ffmpeg: 03Michael Niedermayer 07master * rfec2025936 10ffmpeg/ffmpeg.c: 
[23:37] <CIA-36> ffmpeg: ffmpeg: Fix ffmpeg.c:1874:21: warning: unused variable of [-Wunused-variable]
[23:37] <CIA-36> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[00:00] --- Thu Nov 24 2011