[Ffmpeg-devel-irc] ffmpeg-devel.log.20120715

[burek]

[01:35] <CIA-41> ffmpeg: 03Janne Grunau 07master * rac29054fda 10ffmpeg/libavcodec/ (avcodec.h imgconvert.c version.h): 
[01:35] <CIA-41> ffmpeg: imgconvert: add avcodec_find_best_pix_fmt2()
[01:35] <CIA-41> ffmpeg: The number of pixel formats outgrew the number of available bits in
[01:35] <CIA-41> ffmpeg: the bitmask used in avcodec_find_best_pix_fmt().
[01:35] <CIA-41> ffmpeg: avcodec_find_best_pix_fmt2() uses a PIX_FMT_NONE terminated list
[01:35] <CIA-41> ffmpeg: of pixel formats instead.
[01:35] <CIA-41> ffmpeg: 03Janne Grunau 07master * r39bb27bf79 10ffmpeg/libavcodec/imgconvert.c: 
[01:35] <CIA-41> ffmpeg: imgconvert: avoid undefined left shift in avcodec_find_best_pix_fmt
[01:35] <CIA-41> ffmpeg: CC: libav-stable at libav.org
[01:35] <CIA-41> ffmpeg: 03Janne Grunau 07master * rf30260d32a 10ffmpeg/libavcodec/ (avcodec.h imgconvert.c version.h): imgconvert: deprecate avcodec_find_best_pix_fmt()
[01:35] <CIA-41> ffmpeg: 03Max Lazarov 07master * rcaac3ab6ef 10ffmpeg/ (libavutil/eval.c tests/ref/fate/eval): 
[01:35] <CIA-41> ffmpeg: eval: fix swapping of lt() and lte()
[01:35] <CIA-41> ffmpeg: CC: libav-stable at libav.org
[01:35] <CIA-41> ffmpeg: 03Janne Grunau 07master * r2d497c141d 10ffmpeg/ (libavutil/eval.c tests/ref/fate/eval): eval: add gt(), gte(), lt() and lte() fate tests
[01:35] <CIA-41> ffmpeg: 03Michael Niedermayer 07master * r39afcf1d7e 10ffmpeg/: (log message trimmed)
[01:35] <CIA-41> ffmpeg: Merge remote-tracking branch 'qatar/master'
[01:35] <CIA-41> ffmpeg: * qatar/master:
[01:35] <CIA-41> ffmpeg:  eval: add gt(), gte(), lt() and lte() fate tests
[01:35] <CIA-41> ffmpeg:  eval: fix swapping of lt() and lte()
[01:35] <CIA-41> ffmpeg:  imgconvert: deprecate avcodec_find_best_pix_fmt()
[01:35] <CIA-41> ffmpeg:  imgconvert: add avcodec_find_best_pix_fmt2()
[02:20] <burek> I didn't know that many things depend on libavutil.. I tried to uninstall something named "libavutil51:i386" on my debian machine and it gave me a list of a ton of packages that will be uninstalled too, including even xserver, kde, etc.. geez..
[02:20] <durandal_1707> xserver? something is fishy there....
[02:20] <burek> I also think I scr.. up something :X
[03:45] <CIA-41> ffmpeg: 03Michael Niedermayer 07master * rab46300078 10ffmpeg/: 
[03:45] <CIA-41> ffmpeg: Merge remote-tracking branch 'dwbuiten/master'
[03:45] <CIA-41> ffmpeg: * dwbuiten/master:
[03:45] <CIA-41> ffmpeg:  doc: Remove stray @item from git-howto
[03:45] <CIA-41> ffmpeg:  libfdk-aac: Port to ff_alloc_packet2
[03:45] <CIA-41> ffmpeg:  doxy: move av_guess_sample_aspect_ratio to correct place
[03:45] <CIA-41> ffmpeg: Merged-by: Michael Niedermayer <michaelni at gmx.at>
[03:45] <CIA-41> ffmpeg: 03Derek Buitenhuis 07master * r7ff318e2dd 10ffmpeg/libavcodec/libfdk-aacenc.c: 
[03:45] <CIA-41> ffmpeg: libfdk-aac: Port to ff_alloc_packet2
[03:45] <CIA-41> ffmpeg: Signed-off-by: Derek Buitenhuis <derek.buitenhuis at gmail.com>
[03:45] <CIA-41> ffmpeg: 03Derek Buitenhuis 07master * r86dd5fbf15 10ffmpeg/doc/git-howto.texi: 
[03:45] <CIA-41> ffmpeg: doc: Remove stray @item from git-howto
[03:45] <CIA-41> ffmpeg: Fixes warning:
[03:45] <CIA-41> ffmpeg:  ** @item outside of table or list (l. 362)
[03:45] <CIA-41> ffmpeg: Signed-off-by: Derek Buitenhuis <derek.buitenhuis at gmail.com>
[03:45] <CIA-41> ffmpeg: 03Derek Buitenhuis 07master * r29470bb940 10ffmpeg/libavformat/avformat.h: 
[03:45] <CIA-41> ffmpeg: doxy: move av_guess_sample_aspect_ratio to correct place
[03:45] <CIA-41> ffmpeg: It should not be a part of the RIFF FourCCs group.
[04:15] <CIA-41> ffmpeg: 03Paul B Mahol 07master * r6ad4560031 10ffmpeg/libavcodec/mss1.c: (log message trimmed)
[04:15] <CIA-41> ffmpeg: mss1: improve check if decoded pivot is invalid
[04:15] <CIA-41> ffmpeg: The pivot has to lie between 0 and base.
[04:15] <CIA-41> ffmpeg: Check of ==base is insufficient.
[04:15] <CIA-41> ffmpeg: Thus replace it by a proper check.
[04:15] <CIA-41> ffmpeg: Fixes out of array write.
[04:15] <CIA-41> ffmpeg: Fixes bug #1531.
[04:29] <CIA-41> ffmpeg: 03Paul B Mahol 07master * ra8f6df8a4f 10ffmpeg/libavcodec/libfdk-aacenc.c: 
[04:29] <CIA-41> ffmpeg: libfdk-aacenc: remove redundant log message if ff_alloc_packet2 fails
[04:29] <CIA-41> ffmpeg: The whole point of ff_alloc_packet2 is to not bloat code with
[04:29] <CIA-41> ffmpeg: duplicated error messages.
[04:29] <CIA-41> ffmpeg: Signed-off-by: Paul B Mahol <onemda at gmail.com>
[04:33] <CIA-41> ffmpeg: 03Paul B Mahol 07master * r1463bd902a 10ffmpeg/libavcodec/utils.c: 
[04:33] <CIA-41> ffmpeg: cosmetics: reindent after d3abbb1d1
[04:33] <CIA-41> ffmpeg: Signed-off-by: Paul B Mahol <onemda at gmail.com>
[04:54] <CIA-41> ffmpeg: 03Paul B Mahol 07master * r2c31ed3330 10ffmpeg/libavcodec/exr.c: 
[04:54] <CIA-41> ffmpeg: exr: make channel_offsets int instead of int8_t
[04:54] <CIA-41> ffmpeg: Prior to this change max number of channels for float data which was
[04:54] <CIA-41> ffmpeg: going to be correctly decoded was 32, which is rather small
[04:54] <CIA-41> ffmpeg: considering that exr allows multiple channel layers.
[04:54] <CIA-41> ffmpeg: Signed-off-by: Paul B Mahol <onemda at gmail.com>
[13:08] <ma11s> Any plans to support null packets in MPEG TS streams to achieve true CBR?
[13:09] <ohsix> "mpeg ts" already supports that, no? wouldn't what you're really asking mean a specific element like the mux or one of the encoders?
[13:10] <ma11s> ohsix: sorry, I mean the mpegts encoder. I am looking through mpegtsenc.c trying to figure out if it's implemented, but it doesn't look like it
[13:11] <ma11s> If I understand the specs correctly, when mpeg ts encoder is configured for CBR, it should create a new PID with ID 0x1fff it uses for padding the stream to achieve a constant bitrate
[13:12] <ohsix> oh crap i thought i was on a different channel
[13:12] <ohsix> i have no idea how ffmpeg does anything with regard to that, sorry
[13:13] <ma11s> thanks regardless :-)
[13:13] <ohsix> there's more than one way to do cbr isn't there? all along the way there's places to soak up some more bits, something like that sounds like bitstuffing when you are mixxing different elementary streams
[13:15] <ma11s> yep, but I need a transcoder that can conform to the pretty strict standards of DVB, so the tricks unfortunately fails in my case
[13:16] <JEEB> ma11s, you might want to look into the OBE
[13:16] <JEEB> open broadcast encoder
[13:16] <JEEB> it has, as far as I know, a spec-compliant MPEG-TS muxer
[13:17] <ma11s> JEEB: Thank you! I will take a look
[13:18] <ohsix> broadcast is a bit different from encoding
[13:18] <ma11s> oh man, how come a day worth of googling did not turn up this project
[13:18] <JEEB> ohsix, it should be able to encode transport streams for VOD as well as do live encoding
[13:18] <ohsix> the part where you get to pid 0x1fff is just before it leaves all your broadcast equipment, not when you're transcoding or whatever
[13:19] <ohsix> JEEB: right i just couldn't think of a way to explain it
[13:19] <ma11s> ohsix: ah, I see, then I am misunderstanding how it works
[13:20] <ohsix> not to say that ffmpeg couldn't do it, it's just a bunch of discrete things
[13:21] <JEEB> well, to be honest, if ffmpeg could do it, kierank would've not made his own mpeg-ts muxing library :)
[13:21] <JEEB> which is then one of the base components of OBE
[13:21] <ohsix> could doesn't mean appropriate or should :D
[13:21] <JEEB> the mpeg-ts output of ffmpeg is "good for most practical uses" I would guess
[13:22] <ohsix> ffmpeg as employed in broadcast equipment would surely need it, but file in file out type stuff it doesn't
[13:22] <ma11s> I work at a broadcaster, so probably why I am running into this issue
[13:23] <ohsix> what is it you're actually trying to do? their equipment should be able to beat any bitstream into shape
[13:23] <ma11s> I got a specification from our architectures that our VOD should conform to, and they pay lots of cash for proprietary encoders. So I thought I would give it a good stab and see if I could find an open source alternative
[13:24] <ohsix> ah
[13:24] <ohsix> those cheap boxes that eventually play them really do need stuff set up
[13:25] <ohsix> i don't even know enough about the rate control in ffmpeg to be able to tell you if you could get something like cbr without bitstuffing :o
[13:25] <JEEB> anyways, OBE sounds like something you'd like
[13:26] <ma11s> yep, going to take a good look at it. Thanks for the tip!
[13:27] <ohsix> if not obe there's probably something that lets you check for stream compliance and rejigger them with another mux
[13:41] <ma11s> I think OBE should fit the bill. I am going to encode some files and see if it passes the compliance checks
[13:41] <ma11s> thanks so much for the tip, maybe I can finally rid ourselves of the proprietary software
[14:42] <kierank> ma11s: you can ask questions in #obe if you want
[14:50] <CIA-41> ffmpeg: 03Piotr Bandurski 07master * r8d0ec6e060 10ffmpeg/libavformat/isom.c: 
[14:50] <CIA-41> ffmpeg: isom: add "MNG" FourCC
[14:50] <CIA-41> ffmpeg: sample:
[14:50] <CIA-41> ffmpeg: http://www.datafilehost.com/download-92da395c.html
[14:50] <CIA-41> ffmpeg: Reviewed-by: compn <tempn at twmi.rr.com>
[14:50] <CIA-41> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[14:50] <CIA-41> ffmpeg: 03Michael Niedermayer 07master * r209f92d94a 10ffmpeg/libavcodec/ (allcodecs.c h263dec.c): 
[14:50] <CIA-41> ffmpeg: lavc: add seperate AVCodec for CODEC_ID_H263P so demuxers can use that id.
[14:50] <CIA-41> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[14:50] <CIA-41> ffmpeg: 03Piotr Bandurski 07master * r96e2507363 10ffmpeg/ (libavcodec/rawdec.c libavformat/isom.c): 
[14:50] <CIA-41> ffmpeg: isom: add "NO16" FourCC
[14:50] <CIA-41> ffmpeg: samples:
[14:50] <CIA-41> ffmpeg: http://www.datafilehost.com/download-46d9a0fa.html
[14:50] <CIA-41> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[14:50] <CIA-41> ffmpeg: 03Piotr Bandurski 07master * r198c4ea200 10ffmpeg/libavformat/riff.c: 
[14:55] <CIA-41> ffmpeg: 03Piotr Bandurski 07master * ra8ff7e3f96 10ffmpeg/libavformat/riff.c: riff: Assign U263 FourCC to CODEC_ID_H263
[14:56] <truexfan81> just curious, has ffmpeg mt become stable?
[15:01] <sandsmark> truexfan81: I think it was merged already?
[15:02] <truexfan81> anyway to check if the one i have is mt?
[15:05] <CIA-41> ffmpeg: 03Carl Eugen Hoyos 07master * ra1e52c9441 10ffmpeg/libavcodec/ptx.c: 
[15:05] <CIA-41> ffmpeg: Simplify ptx decoding on big-endian.
[15:05] <CIA-41> ffmpeg: Reviewed-by: Reimar Döffinger
[15:05] <truexfan81> was looking through the mplayer configure script, it looks like the version of ffmpeg i have was pulled from git://git.videolan.org/ffmpeg.git by the script
[15:08] <CIA-41> ffmpeg: 03Stefano Sabatini 07master * rfda968aa8f 10ffmpeg/libavfilter/ (23 files): 
[15:08] <CIA-41> ffmpeg: lavfi: decrease logging level of message showing initial parameters
[15:08] <CIA-41> ffmpeg: Reduce log clutter, consistent with 1a49a169eb74a97.
[16:38] <CIA-41> ffmpeg: 03Stefano Sabatini 07master * r9a1963fbb8 10ffmpeg/ffprobe.c: ffprobe: reindent after the previous commit
[16:38] <CIA-41> ffmpeg: 03Stefano Sabatini 07master * r5626e812d2 10ffmpeg/ffprobe.c: 
[16:38] <CIA-41> ffmpeg: ffprobe: only decode frame if the codec is available
[16:38] <CIA-41> ffmpeg: Fix crash when a stream codec is unknown, in particular fix trac
[16:38] <CIA-41> ffmpeg: ticket #1243.
[17:37] <saste> burek: you there?
[17:46] <burek> yes
[17:46] <burek> saste, 
[17:47] <saste> http://ffmpeg.org/trac/ffmpeg/ticket/821
[17:47] <saste> burek: cannot reproduce the issue
[17:47] <burek> oh man.. that was 7 month ago :)
[17:47] <saste> maybe something changed in the last 7 months
[17:47] <burek> I don't even remember what was that all about
[17:47] <burek> probably :)
[17:48] <burek> just close it :)
[17:48] <saste> you was missing some X11 components required for enabling X11...
[17:48] <burek> yes
[17:48] <burek> I think I posted that to suggest creating a warning message during ./configure or so
[17:48] <saste> when i tried it i got an error message, while you reported that it was failing silently
[17:48] <burek> which will hint the user to install dependencies
[17:49] <burek> libXext
[17:50] <burek> I remember now, that was missing
[17:50] <burek> and libXfixes: -lXext -lXfixes
[17:50] <burek> I was on a non-gui shell so I didn't have X installed at all
[17:51] <burek> but was trying to compile ffmpeg statically with as much libs as possible, thus that situation :)
[17:52] <burek> btw, I've created a script to concat multiple media files into one, based on your yesterday's link to a faq item http://ffmpeg.gusari.org/mcat
[17:52] <burek> I don't know should I put it on the wiki or leave it at the forum
[17:52] <saste> burek: 4a584edad7cd9be8241e8b6fe1fadf8340cd292b
[17:53] <saste> ok so it seems i can close the ticket
[17:54] <burek> yes, I think so to
[18:33] <ma11s> kierank: thanks, I am already there!
[18:52] <Compn> truexfan81 : yes, ffmpeg-mt was merged a while back, right around the libav split...
[18:53] <truexfan81> ah ok
[21:22] <retrosnub> hello, i'm trying to record from a webcam using ffmpeg using "ffmpeg -v debug -f v4l2 -s 352x288 -i /dev/video0 -crf 20 out.mp4", but i keep getting the "Cannot find a proper format for codec_id 0, pix_fmt -1." error.
[21:22] <retrosnub> log: http://pastebin.com/KFTkaJTR
[21:22] <retrosnub> v4l-info: http://pastebin.com/EcUytyjN
[21:22] <retrosnub> i'm using a recent static build of ffmpeg. playing video with "vlc v4l2:///dev/video0" works just fine.
[21:22] <retrosnub> could someone verify if this is a bug?
[21:23] <durandal_1707> retrosnub: what pix_fmt you camera supports?
[21:23] <retrosnub> v4l-info says: "pixelformat             : 0x41495043 [CPIA]"
[21:26] <durandal_1707> that is some special format, you could upload sample of this somewhere and hope someone figures out how to decode it
[21:27] <ubitux> --enable-libv4l?
[21:27] <durandal_1707> actually http://www.mail-archive.com/linuxtv-commits@linuxtv.org/msg06817.html already have info how to decode this
[21:29] <durandal_1707> so you just need to get somene to port that code to lavc
[21:29] <ubitux> or you could just use libv4l directly using --enable-libv4l
[21:29] <ubitux> just like vlc is most likely doing
[21:30] <retrosnub> durandal_1707: can i submit a bug report for that somewhere?
[21:30] <durandal_1707> retrosnub: there is bug tracker, look at ffmpeg.org
[21:30] <durandal_1707> ubitux: libv4l is broken
[21:30] <retrosnub> ubitux: is there a static compiled version, so i can test, or do i have to compile by myself?
[21:31] <durandal_1707> so for libv4l one would need to use older version prior to breakage...
[21:31] <durandal_1707> or fix it
[21:32] <retrosnub> is there a way to get vlc to pipe the video to ffmpeg for further processing as a workaround?
[21:34] <ubitux> retrosnub: not that i know
[21:37] <Compn> vlc can output to a fifo
[21:38] <Compn> which ffmpeg can read from
[21:38] <durandal_1707> retrosnub: if you could get compressd cpia file and you upload it somewhere I (or others) may try to write cpia decoder for it....
[21:39] <retrosnub> Compn: thanks, will try
[21:39] <retrosnub> durandal_1707: i'd like to do that. is dd'ing from /dev/video0 ok?
[21:43] <durandal_1707> retrosnub: dunno, probably yes if dd gets anything, but i will know that only after i see file...
[21:51] <retrosnub> dd dump: http://ubuntuone.com/4nTRHLLdk3xEyNHk2fok3T
[21:51] <retrosnub> cat dump: http://ubuntuone.com/0LroWTzIkgpL0QfllLPlfB
[21:51] <retrosnub> the dd dump is pretty small, so probably useless
[21:54] <Compn> you cant dd it
[21:54] <Compn> cat works better
[21:55] <durandal_1707> rodgort: from quick looks it looks valid...
[21:55] <Compn> oh
[21:56] <Compn> retrosnub : unless the camera outputs something standard raw, you wont be able to read directly from /dev/video 
[21:56] <Compn> i think
[21:56] <Compn> retrosnub : just google 'your webcam mplayer' or 'yourwebcam ffmpeg'
[21:56] <Daemon404> humm
[21:56] <Compn> should find someone that did the same thing :)
[21:56] <Daemon404> would anyone be interested in a patchwork install for ffmpeg-devel?
[21:57] <retrosnub> Compn: did already, but the webcam is really old, could not find anyone using linux with the same problem
[21:57] <Daemon404> im contemplatign setting one up
[21:58] <Compn> oh
[21:58] <Compn> retrosnub : did you say it didnt work with v4l input ?
[21:58] <Compn> retrosnub : you said it works with vlc, vlc can encode things using ffmpeg, why not just use it ?
[21:59] <retrosnub> Compn: up until now i couldn't get the pipe with vlc working. i need ffmpeg for further processing the video
[22:00] <durandal_1707> retrosnub: hah. looks like first frame is uncompressed
[22:00] <Compn> what further processing ?
[22:00] <Compn> watermark ? cropping ? filters ? 
[22:00] <Compn> vlc can do all three (afair)
[22:00] <retrosnub> enconding and deshaking filter
[22:00] <Compn> ah
[22:01] <retrosnub> durandal_1707: is there hope?
[22:01] <Compn> retrosnub : well you should ask in #videolan how to pipe from vlc :)
[22:02] <durandal_1707> retrosnub: please upload cat sample to ffmpeg bug tracker, i may write decoder for this, but currently i'm busy doing other things, so just it does not get lost....
[22:02] <Compn> durandal_1707 : you think its bayer ?
[22:02] <retrosnub> durandal_1707: thanks, will do
[22:02] <durandal_1707> Compn: it is yuv in special containter and have trivial inter frame compression
[22:02] <Compn> ah
[22:02] <Compn> cool
[22:07] <j-b> who would be able to fix some issues with -mt in libavcodec?
[22:26] <Compn> j-b : astrange was maintainer of it
[22:26] <Compn> i mean , when it was being developed seperately
[22:26] <Compn> but there are some other people who worked on it , or other decoders to make them frame mt'd
[22:30] <saste> michaelni: ticket #218, first video frame is located at 15s, ffmpeg can't correctly detect the pixel format
[22:30] <saste> i tried to increase probesize but to no avail, any idea?
[22:37] <nevcairiel> the probe size as well as the maximum probe duration?
[22:54] <saste> nevcariel: indeed did the trick (-analyzeduration)
[22:54] <Daemon404> that is teh fix a lot of the time
[23:05] <retrosnub> durandal_1707: i uploaded the cat sample to ticket #1537
[23:06] <durandal_1707> retrosnub: ok, once that ticket get fixed you will be automatically notified
[23:07] <retrosnub> thanks
[00:00] --- Mon Jul 16 2012