[Ffmpeg-devel-irc] ffmpeg-devel.log.20140109

[burek]

[00:35] <saste> michaelni, ping on muxing.c PTS handling patches
[00:40] <cone-780> ffmpeg.git 03Lou Logan 07master:00bef6431ed7: doc/filters: add rotate example using degrees
[00:42] <michaelni> saste, pong
[00:56] <llogan> ubitux: got a twitter tip? how about "the rotate filter can handle degrees too: ffmpeg -i input.avi -vf rotate=45*PI/180 output.mkv"
[00:58] <llogan> or "if you've been using qt-faststart you can alternatively use '-movflags +faststart' in ffmpeg"
[01:03] <cone-780> ffmpeg.git 03Michael Niedermayer 07master:c0a33c477094: avfilter: avfilter_register() that works in O(1) time
[04:37] <cone-206> ffmpeg.git 03Michael Niedermayer 07master:fb8f5d051061: avcodec/msvideo1enc: fix SKIPS_MAX
[07:27] <ubitux> llogan: dunno, maybe "fast 180 degree rotation, -vf vflip,hflip"?
[07:29] <clever> ubitux: do you know much about b frames in h264?
[07:29] <ubitux> no
[07:29] <clever> ah
[07:30] <clever> was just wondering if there was a simple way to fix my display order issues
[10:55] <cone-206> ffmpeg.git 03Stefano Sabatini 07master:eadc42125905: examples/muxing: set sample formats from list of codec supported sample formats
[10:55] <cone-206> ffmpeg.git 03Stefano Sabatini 07master:80bca6eabe8a: examples/muxing: honour distinction between encoder PTS timebase and stream timebase
[10:55] <cone-206> ffmpeg.git 03Stefano Sabatini 07master:9abe4a10fd58: examples/muxing: reuse global audio frame
[10:55] <cone-206> ffmpeg.git 03Stefano Sabatini 07master:8d22d37e548e: examples/muxing: use av_frame_free() in place of av_free()
[10:55] <cone-206> ffmpeg.git 03Stefano Sabatini 07master:094c500caf0c: examples/muxing: set timestamps in output audio packet
[11:28] <ubitux> saste: drawing borders is really complex
[11:28] <ubitux> even libass still fails at it with some fonts
[11:32] <saste> ubitux, i believe it
[11:32] <saste> but still the rendering result is pretty poor
[11:32] <saste> the feature should be probably implemented in libfreetype
[11:34] <ubitux> or in harfbuzz probably, but dunno
[11:38] <durandal_1707> saste: why is irc meeting again on dedicated channel?
[11:38] <wm4> why not use libass for drawing text
[11:39] <wm4> instead of NIH using freetype manually?
[11:39] <saste> btw we don't support right-to-left languages, nor japanese/chinese up to down
[11:39] <wm4> libass does
[11:39] <saste> wm4, you can already do it with the subtitles filters I suppose
[11:39] <saste> my idea is that drawtext should be only a thin wrapper around libfreetype, no more no less
[11:40] <wm4> how is that useful
[11:40] <saste> wm4, it is, if you want to stay thin and don't want to rely on libass and friends
[11:40] <wm4> libass is pretty thin
[11:41] <saste> is it fully multiplatform?
[11:41] <wm4> yes
[11:41] <ubitux> does it supports the same compilers and plateform ffmpeg has?
[11:41] <wm4> yes
[11:42] <ubitux> why isn't it merged with ffmpeg yet ? @_@
[11:42] <wm4> to troll ffmpeg devs
[11:42] <ubitux> !
[11:43] <durandal_1707> is there .ffmpegrc?
[11:43] <saste> libass? why should it be merged in ffmpeg?
[11:43] <ubitux> :)
[11:43] <saste> also rendering text is not the proper domain of libfreetype?
[11:43] <durandal_1707> because mplayer
[11:44] <ubitux> ft just does the rasterization, it won't ever support all the font effects you can imagine
[11:44] <saste> mplayer => the root of all the evil
[11:44] <wm4> saste: freetype alone is pretty low level, you can't even do international text
[11:45] <saste> wm4, yes so what's the next level?
[11:45] <saste> libfreetype -> libass
[11:45] <saste> what's the missing intermediate?
[11:45] <wm4> libass uses freetype, fontconfig and harfbuzz directly
[11:46] <saste> so i suppose harfbuzz is the answer
[11:46] <wm4> no
[11:47] <blastergod> Hello, I am trying to concat 2 mp4 files using a simple command "ffmpeg.exe -i "C:\vid1.mp4"  -i "C:\vid2.mp4"   -filter_complex "concat" -y "c:\zzzz.mp4""   . It works with all files but 1 . When I try to concat this 2 files, ffmpeg shows an error (after encoding about 300 frames) : "deprecated pixel format used, make sure you did set range correctly" and than the frame stops changing, 
[11:47] <blastergod> the timetsamp stops changing and the fps goes down from 200,199 etc... but it never reaches 0. It simply drops fps and that's it, it stays there without doing anything. If I press Q in command line (windows) the process stops and the video cannot be played. Now if I take each of the above videos and re-encode them first, than the concat works. What can cause this problem?
[11:47] <ubitux> blastergod: ’ #ffmpeg
[11:48] <blastergod> thanks, I am pretty in need of help, been searching and trying to fix this for over 20 hours :(
[11:53] <cone-206> ffmpeg.git 03Michael Niedermayer 07release/1.1:4de4eb60a1ba: avformat/oggdec: dont read timestamps from EOS pages of ogm videos
[11:53] <cone-206> ffmpeg.git 03Michael Niedermayer 07release/1.1:adb784ad8610: avformat/mxfdec: detect loops during header parsing
[11:53] <cone-206> ffmpeg.git 03Carl Eugen Hoyos 07release/1.1:3ada932202ae: Use the h264 parser when decoding VSSH in avi.
[11:53] <cone-206> ffmpeg.git 03Michael Niedermayer 07release/1.1:5ea2a8d43e69: avcodec/msvideo1enc: fix SKIPS_MAX
[11:53] <cone-206> ffmpeg.git 03Michael Niedermayer 07release/1.2:e24b33cd6835: avcodec/msvideo1enc: fix SKIPS_MAX
[11:53] <cone-206> ffmpeg.git 03Michael Niedermayer 07release/2.1:d45a724192d8: avcodec/msvideo1enc: fix SKIPS_MAX
[12:02] <durandal_1707> there should be .ffmpegrc because of -hide_banner
[12:13] <cone-206> ffmpeg.git 03Timothy Gu 07master:bba73f5574af: doc: remove encoder/muxer/sinks docs from ffplay- and ffprobe-all
[12:25] <ubitux> saste, wm4, so, should we add a -vf drawass?
[12:26] <wm4> +1
[12:26] <ubitux> i already love the name
[12:27] <saste> ubitux, probably, but the name is hilarious
[12:28] <durandal_1707> just ass
[12:28] <nevcairiel> I wish someone would finish the native win32 font backend for libass though, fontconfig is practically unusable on windows
[12:29] <j-b> I wish we had a library to draw text that is not linked to a format
[12:30] <j-b> aka libass without the ass part
[12:30] <nevcairiel> well you always need some kind of formating to tell it what to draw how
[12:31] <j-b> draw( this_text, format_struct_t )
[12:31] <nevcairiel> that gets pretty low level if you want every character in another color, for example
[12:32] <nevcairiel> inline formating has certain advantages
[12:33] <nevcairiel> not saying libass is the solution, but i see the advantage of inline style information over such an API
[12:34] <j-b> sure
[13:15] <durandal_1707> is there hex editor/viewer that allows to change color of certain bytes (range)?
[13:19] <saste> durandal_1707, probably you can do that with emacs and some emacs-lisp programming
[14:29] <cone-535> ffmpeg.git 03Anton Khirnov 07master:ebfe622bb1ca: mpegvideo: drop support for real (non-emulated) edges
[14:29] <cone-535> ffmpeg.git 03Michael Niedermayer 07master:2830f287e2f2: Merge commit 'ebfe622bb1ca57cecb932e42926745cba7161913'
[14:37] <cone-535> ffmpeg.git 03Anton Khirnov 07master:ef8c93e2f18c: vp8: drop support for real (non-emulated) edges
[14:37] <cone-535> ffmpeg.git 03Michael Niedermayer 07master:7904dfed7370: Merge commit 'ef8c93e2f18c624d0c266687e43ab99af7921dd3'
[14:47] <cone-535> ffmpeg.git 03Anton Khirnov 07master:ca96e3371690: vp9: drop support for real (non-emulated) edges
[14:47] <cone-535> ffmpeg.git 03Michael Niedermayer 07master:3773738e26c9: Merge commit 'ca96e337169093979d7c763064ad9dae12b3108c'
[14:48] <michaelni> BBB / ubitux, if you dont like the vp8/9 changes just say so and ill revert them
[14:50] <ubitux> > the gain is not large enough to justify the added complexity.
[14:50] <ubitux> those 2 lines?
[14:53] <nevcairiel> well the real diff is a bit larger
[14:54] <BBB> I bet they didn't measure it
[14:54] <BBB> anyway, work...
[14:54] <nevcairiel> iirc, the reason to remove it is that someone thought it was unreasonable to require edges to be allocated in get_buffer, since some players apparently didn't do that and it caused issues for them
[15:06] <cone-535> ffmpeg.git 03Anton Khirnov 07master:a246af86e470: FATE: add MSS2 tests
[15:06] <cone-535> ffmpeg.git 03Michael Niedermayer 07master:2b318f8ca916: Merge commit 'a246af86e47044dec243db2f708cc125d608ea78'
[15:24] <Compn> bik in mkv haha
[15:25] <Compn> why
[15:25] Action: Compn dares not read
[15:25] <JEEB> that thread is /still/ going on?
[15:25] <JEEB> oh great
[15:25] Action: JEEB blames Compn 
[15:25] <JEEB> she is saying that it should be supported because mplayer can decode the result :D
[15:36] <cone-535> ffmpeg.git 03Guillaume Martres 07master:e588615d938f: hevc: fix decoding of one PU wide files
[15:36] <cone-535> ffmpeg.git 03Michael Niedermayer 07master:c1362ca0479e: Merge commit 'e588615d938f8581f0d6f3771662d08cadfc00de'
[15:52] <cone-535> ffmpeg.git 03Anton Khirnov 07master:d48132b7de76: mjpegdec: apply flipping after decoding, not before
[15:52] <cone-535> ffmpeg.git 03Michael Niedermayer 07master:cab5315519ff: Merge remote-tracking branch 'qatar/master'
[15:54] <Compn> JEEB : you cant blame me because mplayer is awesome
[15:54] <Compn> although i would like to take credit
[15:54] <Compn> you could blame me if she was using mencoder , but my answer would be 'use ffmpeg'
[15:54] <Compn> and then we're here already
[18:21] <cone-535> ffmpeg.git 03Andre Anjos 07master:23ffc4c70d59: doc/examples/muxing: Fixes frame initialization.
[20:23] <cone-535> ffmpeg.git 03Diego Biurrun 07master:4b48201d4d9d: ac3tab.h: #include the correct headers
[20:23] <cone-535> ffmpeg.git 03Michael Niedermayer 07master:b4f97d4a0edc: Merge commit '4b48201d4d9d57a2a1ecf8504bc928aff83cb99a'
[20:32] <cone-535> ffmpeg.git 03Diego Biurrun 07master:b0be1ae792ac: x86: avcodec: Add a bunch of missing #includes for av_cold
[20:32] <cone-535> ffmpeg.git 03Michael Niedermayer 07master:390452bab627: Merge commit 'b0be1ae792ac8bbfb0fc7b9b9cb39eaf0feb489b'
[20:49] <cone-535> ffmpeg.git 03Diego Biurrun 07master:4c642d8d9870: x86: hpeldsp: Add missing av_cold attribute to init function
[20:49] <cone-535> ffmpeg.git 03Michael Niedermayer 07master:92b2404571bc: Merge commit '4c642d8d98703faf52983243098f35865e15b312'
[21:04] <ubitux> michaelni: in cab5315519ff8172f3b07dbcc4d4829ae567b1a9 shoudln't we use FF_CEIL_RSHIFT()?
[21:04] <ubitux> i'm afraid it will cause troubles with odd sizes
[21:04] <cone-535> ffmpeg.git 03Maxim Polijakowski 07master:e6f0bb65270b: ATRAC3+ decoder
[21:04] <cone-535> ffmpeg.git 03Michael Niedermayer 07master:268d93172460: Merge remote-tracking branch 'qatar/master'
[21:05] <ubitux> btw, i see some code remaining doing stuff about emu edge
[21:46] <cone-535> ffmpeg.git 03Michael Niedermayer 07master:af15c17daa5d: avcodec/mjpegdec: remove new flipping code
[21:47] <cone-535> ffmpeg.git 03Michael Niedermayer 07master:7c3700cd1d86: avcodec/mjpegdec: use the AVFrame width/height for flipping
[21:47] <cone-535> ffmpeg.git 03Michael Niedermayer 07master:ebfa1264aacf: avcodec/mjpegdec: allow lowres with amv
[21:48] <michaelni> ubitux, should be fixed
[21:48] <ubitux> awesome, thx!
[21:48] <michaelni> or did you see more emu edge stuff that should be removed updated ?
[21:48] <ubitux> i was just reading cvslog
[21:49] <ubitux> didn't look further
[21:57] <j-b> 1m
[21:57] <j-b> oops, sorry
[23:13] <wm4> ubitux: ^
[23:14] <wm4> ...
[23:14] <wm4> well, the bot is dead or so: http://trac.ffmpeg.org/ticket/3287
[23:15] <nevcairiel> its not dead, just slow
[23:51] <cone-535> ffmpeg.git 03Michael Niedermayer 07master:77d2a1ca595e: avformat/rmdec: when reading audio blocks, dont leave holes when reading fails
[00:00] --- Fri Jan 10 2014