[Ffmpeg-devel-irc] ffmpeg-devel.log.20120502

burek burek021 at gmail.com
Thu May 3 02:05:03 CEST 2012


[00:03] <Shimmy> ubitux, interesting indeed
[00:03] <Shimmy> ubitux i really hope ffmpeg can create files playable on my device... i donno how to use mencoder
[00:04] <ubitux> mencoder isn't really maintained anymore you know
[00:17] <burek> fate tests paseed :))) woo hoo :))) I <3 ffmpeg :}
[00:17] <burek> passed*
[00:30] <Compn> Shimmy : there are a few ffmpeg forums, if thats what you want ?
[00:34] <Shimmy> compn, yes that's what i want, btw, i fully posted my issue here: http://ffmpeg-users.933282.n4.nabble.com/Converted-video-runs-on-device-in-fast-motion-td4598300.html
[00:35] <Shimmy> compn, relaxed, ubitux, mbradshaw, please vote here, it takes a second and is very important: http://area51.stackexchange.com/proposals/40292/ffmpeg
[00:35] <Shimmy> ask me if you're doubting or unsure
[00:36] <Shimmy> burek, thanks for voting! if you think im gonna make money from it then you're wrong. I actually think it's my tribute to the ffmpeg community if such a site will be opened
[00:36] <Shimmy> ITOH, it's completely free, so it's a shame not to vote
[00:41] Action: Compn thinks its pretty strange there have been 3 or 4 requests for an ffmpeg forum over the past 2 months
[00:41] <Compn> er not requests , but people starting forums and asking to link to them
[00:42] <burek> well the average user of ffmpeg (not a developer) is more likely to be familiar with forums, rather than mailing lists or trac...
[01:18] <CIA-17> ffmpeg: 03Justin Ruggles 07master * r6d5bf67f04 10ffmpeg/libswresample/ (rematrix.c swresample.c swresample_internal.h): 
[01:18] <CIA-17> ffmpeg: swr: add lfe_mix_level
[01:18] <CIA-17> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[01:18] <CIA-17> ffmpeg: 03Justin Ruggles 07master * r9b42653b22 10ffmpeg/libswresample/rematrix.c: 
[01:18] <CIA-17> ffmpeg: swr: update side to back mixing.
[01:18] <CIA-17> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[01:18] <CIA-17> ffmpeg: 03Michael Niedermayer 07master * r7bea0a0644 10ffmpeg/doc/APIchanges: 
[01:18] <CIA-17> ffmpeg: APIChanges: add entry for libavresample
[01:18] <CIA-17> ffmpeg: Appears this was forgotten
[01:18] <CIA-17> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[01:18] <CIA-17> ffmpeg: 03Michael Niedermayer 07master * rf2e799a3d0 10ffmpeg/libswresample/ (rematrix.c swresample.c): 
[01:18] <CIA-17> ffmpeg: swr: add double precision support to the rematrix code
[01:18] <CIA-17> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[01:19] Last message repeated 1 time(s).
[01:19] <CIA-17> ffmpeg: 03Michael Niedermayer 07master * r33f7033452 10ffmpeg/libswresample/swresample.c: 
[01:19] <CIA-17> ffmpeg: swr: automatically choose s16/flt/dbl to preserve input precision unless user overrides.
[01:19] <CIA-17> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[01:40] <Compn> hmm
[01:40] <Compn> so hacker news has no moderation guidelines and its moderators are secret and it has no help forums/contact?
[01:47] <burek> is this a serious problem ? "Test seek-lavf_mmf failed."
[01:48] <burek> http://pastebin.com/PTUqng8d
[01:48] <Compn> check http://fate.ffmpeg.org
[01:48] <Compn> maybe its broken
[02:00] <burek> ok, I think that everything is ok now :) I've setup an automatic cron job that will compile static 32/64 bit ffmpeg every day, after the successful "git pull" command
[02:00] <burek> that way we will have a way for users, that are not able to compile the latest git code and see if their bug still exists
[02:00] <burek> to just download and run the latest static version of ffmpeg: http://ffmpeg.gusari.org/static/
[02:07] <CIA-122> ffmpeg: 03Michael Niedermayer 07master * r29ec5c1102 10ffmpeg/tests/ref/seek/lavf_mmf: 
[02:07] <CIA-122> ffmpeg: fate: update mmf seek checksum, change caused by av_get_packet() useage
[02:07] <CIA-122> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[02:26] <saste> michaelni: can you have a look at tickets 1260/1267?
[02:26] <saste> they are the same problem, decoder returns a not empty data[3], which causes a crash
[02:27] <saste> i already encountered the problem on other occasions, but never figured out how to fix it
[02:27] <saste> the problem should be located in the mpegvideo code
[02:30] <michaelni> saste, ill look tomorrow
[02:30] <michaelni> need to go to bed now ...
[02:30] <saste> yep no hurry of course
[02:30] <michaelni> remind me in case i forget ...
[02:30] <saste> ok i'll do, good night!
[06:10] <taqattack> FFmpeg seems to get throttled when streaming to RTMP in Windows. I streamed the same file for same file with same parameters using FFmpeg build from same source and I get different results. http://i.imgur.com/u1sWy.png
[11:34] <CIA-122> ffmpeg: 03Jakub Stachowski 07master * r8501bed08c 10ffmpeg/libavcodec/wmalosslessdec.c: 
[11:34] <CIA-122> ffmpeg: Don't restrict reverse decorrelation to both coded channels. It is also used for mono data.
[11:34] <CIA-122> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[11:34] <CIA-122> ffmpeg: 03Michael Niedermayer 07master * rb197c1c1ac 10ffmpeg/ffmpeg.c: 
[11:34] <CIA-122> ffmpeg: ffmpeg: preserve NULL ptrs in get_buffer()
[11:34] <CIA-122> ffmpeg: Fixes Ticket1260, Ticket1267
[11:34] <CIA-122> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[11:42] <saste> michaelni: thanks :)
[13:53] <burek> however I configure ffserver
[13:53] <burek> to stream libx264 + aac (in flv)
[13:53] <burek> I get this:
[13:54] <burek> http://pastebin.com/TdYMNk23
[13:54] <burek> broken ffmpeg default settings detected
[13:54] <burek> is that a bug?
[18:36] <CIA-122> ffmpeg: 03Paul B Mahol 07master * r2cda0429aa 10ffmpeg/ (5 files in 3 dirs): 
[18:36] <CIA-122> ffmpeg: lavfi: remove screenshot libmpcodecs wrapper
[18:36] <CIA-122> ffmpeg: The filter is useless.
[18:36] <CIA-122> ffmpeg: Signed-off-by: Paul B Mahol <onemda at gmail.com>
[18:36] <CIA-122> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[19:46] <mancha> Hello folks. I wanted to report a bug with mkv playback on 0.10.2 (and 0.10.2 git from 2 days ago). 
[19:46] <mancha> it can be seen by playing http://people.opera.com/shwetankd/webm/sunflower.webm. the problem occurs at about second 8.2 or so and also at the end (if using ffplay but not if using mplayer)
[19:47] <mancha> the file plays without issue on vlc which uses libmatroska/libebml as primary mkv decoder.
[19:49] <mancha> both ffplay and mplayer have issues at second 8.2 but only ffplay has trouble with the video's ending. mplayer is also built with 0.10.2 ffmpeg code here.
[20:23] <ubitux> is there any kind of magic to store multiple packet at the same pts? or something to queue multiple subpackets in the same packet?
[20:23] <ubitux> (except merging the data packet and redo the demuxer work on decoder side)
[20:36] <michaelni> ubitux, what are you trying to do ?
[20:37] <ubitux> mainly subtitles
[20:38] <ubitux> a lot of sub formats can have subtitles events at the same time
[20:38] <ubitux> (with different markup for instance)
[20:49] <michaelni> what is the problem with multiple packets at the same pts ?
[20:49] <michaelni> non monotine error ?
[20:49] <michaelni> if so we should add a check for subs there
[20:50] <michaelni> i mean type == SUBTITLE
[21:08] <mancha> hi. any ffmpeg devs who work on the mkv decoder around?  :)
[21:09] <mancha> i gots me an issue to bring up if anyone's willing to discuss it
[21:15] <Compn> theres a bunch of devs here 
[21:15] <Compn> if you just ask, you may get answer :)
[21:15] <burek> he likes to ask questions ^^
[21:16] <mancha> this file (http://people.opera.com/shwetankd/webm/sunflower.webm) causes problems in ffmpeg 0.10.2 (and HEAD from a few days ago). the problems start around second 8.2
[21:16] <mancha> vlc with libmatroska/libebml plays it fine and i seem to recall earlier ffmpegs were ok too (though i can't be 100% certain now)
[21:18] <mancha> if any devs can confirm it's an issue and if it has already been noticed/reported it'd be great :)
[21:19] <mancha> mplayer (built around ffmpeg 0.10.2+) and ffmpeg 0.10.2 both struggle at second 8.2 though mplayer does exit gracefully after the clip ends. ffplay hangs infinitely.
[21:20] <nevcairiel> I would suggest to open a ticket on trac
[21:26] <CIA-122> ffmpeg: 03Justin Ruggles 07master * r4010d724e1 10ffmpeg/ (7 files in 2 dirs): lavr: use 8.8 instead of 10.6 as the 16-bit fixed-point mixing coeff type
[21:26] <CIA-122> ffmpeg: 03Mans Rullgard 07master * r4bf2e7c5f1 10ffmpeg/libavcodec/twinvq.c: 
[21:26] <CIA-122> ffmpeg: twinvq: fix out of bounds array access
[21:26] <CIA-122> ffmpeg: ModeTab.fmode has only 3 elements, so indexing it with ftype
[21:26] <CIA-122> ffmpeg: in the initialier for 'size' is invalid when ftype == FT_PPC.
[21:26] <CIA-122> ffmpeg: This fixes crashes with gcc 4.8.
[21:26] <CIA-122> ffmpeg: Signed-off-by: Mans Rullgard <mans at mansr.com>
[21:27] <CIA-122> ffmpeg: 03Mans Rullgard 07master * radebad07e0 10ffmpeg/libavutil/arm/intreadwrite.h: (log message trimmed)
[21:27] <CIA-122> ffmpeg: arm: intreadwrite: fix inline asm constraints for gcc 4.6 and later
[21:27] <CIA-122> ffmpeg: With a dereferenced type-cast pointer as memory operand, gcc 4.6
[21:27] <CIA-122> ffmpeg: and later will sometimes copy the data to a temporary location,
[21:27] <CIA-122> ffmpeg: the address of which is used as the operand value, if it thinks
[21:27] <CIA-122> ffmpeg: the target address might be misaligned. Using a pointer to a
[21:27] <CIA-122> ffmpeg: packed struct type instead does the right thing.
[21:27] <CIA-122> ffmpeg: 03Hendrik Leppkes 07master * r4b7fa553a9 10ffmpeg/libavcodec/indeo3.c: 
[21:27] <CIA-122> ffmpeg: indeo3: fix motion vector validation
[21:27] <CIA-122> ffmpeg: The index of the motion vector has to be checked before being
[21:27] <CIA-122> ffmpeg: multiplied by 2 for the array index.
[21:27] <CIA-122> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[21:27] <CIA-122> ffmpeg: Signed-off-by: Derek Buitenhuis <derek.buitenhuis at gmail.com>
[21:27] <CIA-122> ffmpeg: 03Mans Rullgard 07master * rababec7b95 10ffmpeg/libavutil/arm/intreadwrite.h: 
[21:27] <CIA-122> ffmpeg: arm: intreadwrite: disable inline asm for gcc 4.7 and later
[21:27] <CIA-122> ffmpeg: pcm_bluray: set bits_per_raw_sample for > 16-bit
[21:27] <CIA-122> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[21:27] <CIA-122> ffmpeg: Signed-off-by: Derek Buitenhuis <derek.buitenhuis at gmail.com>
[21:27] <CIA-122> ffmpeg: 03Michael Niedermayer 07master * r2330eb1474 10ffmpeg/: (log message trimmed)
[21:27] <CIA-122> ffmpeg: Merge remote-tracking branch 'qatar/master'
[21:27] <CIA-122> ffmpeg: * qatar/master:
[21:27] <CIA-122> ffmpeg:  arm: intreadwrite: disable inline asm for gcc 4.7 and later
[21:27] <CIA-122> ffmpeg:  arm: intreadwrite: fix inline asm constraints for gcc 4.6 and later
[21:27] <CIA-122> ffmpeg:  indeo3: fix motion vector validation
[21:27] <CIA-122> ffmpeg:  pcm_bluray: set bits_per_raw_sample for > 16-bit
[21:34] <ubitux> michaelni: yes iirc there is the non monotonic error (even with AVFMT_TS_NONSTRICT in some case, maybe for muxing in the ass format i don't remember)
[21:48] <michaelni> ubitux, does it work if you just disable the non monotone checks ? note there could be problems if things are really out of order
[22:21] <j-b> is Michael Bradshaw around?
[22:22] <ubitux> mbradshaw on #ffmpeg
[22:22] <j-b> nice
[22:22] <ubitux> michaelni: i'm looking at it
[22:22] <burek> j-b, check #ffmpeg
[22:24] <ubitux> michaelni: it's actually "well" encoded
[22:25] <ubitux> though i don't know if ass supports that kind of thing:
[22:25] <ubitux> Dialogue: 0,0:00:22.30,0:00:27.80,and different colors
[22:25] <ubitux> Dialogue: 0,0:00:24.00,0:00:27.80,all {\b1}{\i1}mixed{\b0} together{\r}.
[22:25] <ubitux> mmh wait.
[22:25] <ubitux> i meant:
[22:25] <ubitux> Dialogue: 0,0:00:22.30,0:00:27.80,Text may appear in different styles\N(Normal, {\b1}Bold{\r}, {\i1}Italic{\r})
[22:25] <ubitux> Dialogue: 0,0:00:22.30,0:00:27.80,and different colors
[22:26] <ubitux> so basically it works, but there are some warning on ass side
[22:26] <ubitux> not sure if it's really an issue
[22:27] <ubitux> michaelni: btw, do you have any more comment on the jacosub patch?
[22:43] <michaelni> ubitux,if the code works and you arent aware of big issues id say push it
[22:43] <ubitux> ok
[22:44] <ubitux> (did i missed any other message?)
[22:44] <ubitux> no more comment on the probing?
[22:51] <ux> ok after my isp, now my pc
[22:51] <ux> seriously.
[22:51] <ux> i'm cursed
[22:55] <pasteeater> people are still trying to compile ffmpeg via SVN...
[22:55] <burek> and they succeed!
[22:56] <ohsix> maybe they can't get off the pipe
[22:56] <ohsix> the world was told a lot of "build a newer version" before there were like, releases and stuff
[22:56] <burek> why don't you shut down svn if it's not used
[23:01] <ubitux> something is srsly fucked up in the world/
[23:02] <mancha> inorite
[23:28] <mbradshaw_> I've got a question about audio streams and PTS packet values... no one on the #ffmpeg channel seems to know so I'm hoping some experts here might have an idea
[23:30] <mbradshaw_> My audio stream has a sample rate and a time base of 1/44100. The first frame's packet's PTS value is 0 and the frame has 1024 samples. The second frame's packet's PTS value is 1014, and it too has 1024 samples. But the PTS of the second frame looks 10 too early, as it's overlapping the first frame... how can I handle this properly in my decoding library? If I pay attention to the PTS values, I'll overlap the audio when I pla
[23:40] <kierank> well you have to drop samples
[23:40] <kierank> you have no choice
[23:42] <mbradshaw_> Alright, I was just wondering if that was the normal thing to do. Thanks.
[00:00] --- Thu May  3 2012


More information about the Ffmpeg-devel-irc mailing list