[Ffmpeg-devel-irc] ffmpeg-devel.log.20111109

[burek]

[00:18] <CIA-18> ffmpeg: 03Michael Niedermayer 07master * rbf0ac7aa69 10ffmpeg/libavformat/mpegvideodec.c: 
[00:18] <CIA-18> ffmpeg: mpegvideo_probe: reduce score for invalid streams
[00:18] <CIA-18> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[00:18] <CIA-18> ffmpeg: 03Michael Niedermayer 07master * r3bdfef31ac 10ffmpeg/ (libavcodec/fraps.c tests/ref/fate/fraps-v5): 
[00:18] <CIA-18> ffmpeg: fraps: check for overread
[00:18] <CIA-18> ffmpeg: Fixeds Ticket619
[00:18] <CIA-18> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[00:28] <maister> michaelni, killed 3 of the shifts
[00:28] <maister> 6623177 dezicycles in rgb16tobgr24, 2048 runs, 0 skips
[00:28] <maister> 6588559 dezicycles in rgb16tobgr24, 4096 runs, 0 skips
[00:28] <standesbeamter> hi, is tjoppen around?
[00:33] <michaelni> maister, great!
[00:34] <maister> http://pastebin.com/uAWGtnL6
[00:34] <maister> only tested rgb15tobgr24 so far
[00:39] <maister> but sure very tidy compared to shift/or version
[00:39] <maister> heh
[00:44] <michaelni> looks good from a quick glance, one of the pands could be avoided too but it could be done in a seprate patch later as well
[00:45] <michaelni> do you want to do more testing than  rgb15tobgr24 ? or should i take a deeper look and if its good apply ?
[00:45] <maister> would be nice if you could verify as well.
[00:45] <maister> I can test rgb15to32 as well
[00:48] <maister> rgb15to32 works
[00:53] <maister> michaelni, 16-bit is borked at least
[00:53] <maister> probably one of those magic constants
[01:02] <maister> michaelni, fixed it
[01:02] <maister> http://pastebin.com/nWAPZBrZ
[01:07] <maister> LexSfX, michaelni likes the current patch
[01:07] <maister> hopefully gets merged today
[01:07] <maister> wrong channel
[01:07] <maister> :P
[02:01] <CIA-18> ffmpeg: 03Themaister 07master * r0827222b9c 10ffmpeg/ (17 files in 3 dirs): 
[02:01] <CIA-18> ffmpeg: Use more accurate conversion for rgb15/16 to rgb24/32 (C/MMX).
[02:01] <CIA-18> ffmpeg: Fate update by michael.
[02:01] <CIA-18> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[03:31] <CIA-18> ffmpeg: 03Justin Ruggles 07master * rf9324d5add 10ffmpeg/avplay.c: 
[03:31] <CIA-18> ffmpeg: avplay: reset rdft when closing stream.
[03:31] <CIA-18> ffmpeg: this fixes a crash when cycling audio streams if the spectrograph is
[03:31] <CIA-18> ffmpeg: displayed.
[03:31] <CIA-18> ffmpeg: 03Anton Khirnov 07master * r8a691dfdab 10ffmpeg/doc/git-howto.txt: doc/git-howto: expand format-patch and send-email notes.
[03:31] <CIA-18> ffmpeg: 03Anton Khirnov 07master * r83daced0a3 10ffmpeg/libavformat/avformat.h: lavf: expand doxy for some AVFormatContext fields.
[03:31] <CIA-18> ffmpeg: 03Diego Biurrun 07master * r276b995d85 10ffmpeg/libavcodec/x86/ (cabac.h h264_i386.h): x86: drop pointless ARCH_X86 #ifdef from files in x86 subdirectory
[03:31] <CIA-18> ffmpeg: 03Diego Biurrun 07master * r9412f81138 10ffmpeg/libavcodec/indeo3data.h: indeo3data: add missing config.h #include for HAVE_BIGENDIAN
[03:31] <CIA-18> ffmpeg: 03Diego Biurrun 07master * rce33320b30 10ffmpeg/libavcodec/ (8 files in 3 dirs): 
[03:31] <CIA-18> ffmpeg: Remove redundant filename self-references inside files.
[03:31] <CIA-18> ffmpeg: Filenames are brittle across renames and add no useful information.
[03:31] <CIA-18> ffmpeg: 03Justin Ruggles 07master * r03e5d6118c 10ffmpeg/libavcodec/ra288.c: ra288: use memcpy() to copy decoded samples to output
[03:31] <CIA-18> ffmpeg: 03Justin Ruggles 07master * r2467d8d9ea 10ffmpeg/libavcodec/ra288.c: ra288: return error if input buffer is too small
[03:31] <CIA-18> ffmpeg: 03Anton Khirnov 07master * r3110ad8329 10ffmpeg/libavformat/mpegts.c: 
[03:31] <CIA-18> ffmpeg: mpegts: set stream id on just created stream, not an unrelated variable
[03:31] <CIA-18> ffmpeg: Bug introduced in 84ad31ff180fa089cd6bfd93c246336a16036455.
[03:31] <CIA-18> ffmpeg: Thanks to Uoti Urpala for finding it.
[03:31] <CIA-18> ffmpeg: 03Alex Converse 07master * r735e601be1 10ffmpeg/libavformat/mxfdec.c: 
[03:31] <CIA-18> ffmpeg: mxfdec: Fix comparison of unsigned expression < 0.
[03:31] <CIA-18> ffmpeg: 'size' is populated by functions returning int64_t and int that return
[03:31] <CIA-18> ffmpeg: negative error codes.
[03:31] <CIA-18> ffmpeg: 03Justin Ruggles 07master * r0131e70af5 10ffmpeg/libavcodec/ (ra288.c ra288.h): ra288: utilize DSPContext.vector_fmul()
[03:31] <CIA-18> ffmpeg: 03Martin Storsjö 07master * rf38f3b88a5 10ffmpeg/libavformat/tls.c: 
[03:31] <CIA-18> ffmpeg: tls: Use ERR_get_error() in do_tls_poll
[03:31] <CIA-18> ffmpeg: The return value ret isn't an error code that can be passed
[03:31] <CIA-18> ffmpeg: to ERR_error_string().
[03:31] <CIA-18> ffmpeg: This makes the error messages printed actually contain useful
[03:31] <CIA-18> ffmpeg: information.
[03:31] <CIA-18> ffmpeg: * qatar/master:
[03:31] <CIA-18> ffmpeg:  tls: Use ERR_get_error() in do_tls_poll
[03:31] <CIA-18> ffmpeg:  indeo3: Fix a fencepost error.
[03:31] <CIA-18> ffmpeg:  mxfdec: Fix comparison of unsigned expression < 0.
[03:31] <CIA-18> ffmpeg:  mpegts: set stream id on just created stream, not an unrelated variable
[03:31] <CIA-18> ffmpeg: 03Justin Ruggles 07master * rf50b6be57d 10ffmpeg/libavcodec/mace.c: mace: only calculate output buffer size once
[04:02] <CIA-18> ffmpeg: 03Andrey Utkin 07master * r7a032cb97f 10ffmpeg/libavformat/udp.c: 
[04:02] <CIA-18> ffmpeg: Fix segfault on UDP reading
[04:02] <CIA-18> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[04:02] <CIA-18> ffmpeg: 03Michael Niedermayer 07master * rb2bdca1b4a 10ffmpeg/libavformat/udp.c: 
[04:03] <CIA-18> ffmpeg: udp: simplify 7a032cb97f743a651138d2da6214c79e0ce089dd
[04:03] <CIA-18> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[04:23] <qrtt1> git.videolan.org is gone. no service there.
[04:39] <michaelni> qrtt1, tell that to j-b 
[04:39] <qrtt1> michaelni: sorry, who is j-b?
[04:41] <michaelni>  Jean-Baptiste KEMPF
[04:41] <Compn> j-b is our busy busy admin  :)
[04:42] <qrtt1> ha
[04:42] <michaelni> qrtt1, git://github.com/FFmpeg/FFmpeg.git
[04:42] <qrtt1> michaelni: Thank you, I use that now.
[04:43] <michaelni> it might be a few hours behind though
[04:44] <Daemon404> hmm
[04:44] <Daemon404> i want to use add_hfyu_median_prediction, but it's rather non-trivial to call...
[04:44] <Daemon404> an has a bunch of paramaters i dont think it should even for my purposes
[04:49] <Daemon404> oh well
[04:56] <CIA-18> ffmpeg: 03Michael Niedermayer 07master * r11e155c219 10ffmpeg/libavcodec/ra288.c: 
[04:56] <CIA-18> ffmpeg: ra288: try to fix 10l typo that breaks compile on some platforms
[04:56] <CIA-18> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[05:19] <Daemon404> michaelni, ive sent my decoder into libav first, mainly cause theyre bigger dicks about this sort of stuff (getting the least fun part done first). Would you like me to also send it ti ffmpeg-devel?
[05:19] <Daemon404> you've got my permission to s/libav/ffmpeg/i
[05:20] Action: Daemon404 doesnt care much for this fork business.
[05:32] <Compn> no one does
[05:32] <michaelni> Daemon404, iam not sure what you talk about  but yes, please submit patches to the ffmpeg-devel ML
[05:33] <Daemon404> i simply mean libav are bigger dicks about letting others' code in
[05:33] <Daemon404> i wasnt sure if it was a sane thing to do to send the same patch to both MLs
[05:34] <Daemon404> basically i wanted to get the bad part of the experience out of the way early.
[05:34] <Compn> oh
[05:34] <Compn> to put it another way, libav isnt merging from ffmpeg much
[05:35] <Daemon404> yes.
[05:35] <Compn> so if you want it in their repo, you better submit it there
[05:35] <Daemon404> that was my reasoning, since i wanted it in both.
[05:36] <Daemon404> both kshishkov and michaelni have helped me a fair bit, and i appreciate it
[05:36] <Compn> kostya's worked on a lot of codecs :)
[05:37] <michaelni> id wish kostya to return to ffmpeg
[05:37] <Daemon404> what i was worrying about, is that if i sent the patch to both mailing lists
[05:37] <Daemon404> each would specify different fixes
[05:37] <Daemon404> and it'd end up a mess
[05:37] <Daemon404> doing it serially seemed more sane
[05:38] <Compn> what people have done in the past is just merge changes from both lists into one patchset
[05:38] <Compn> rarely do they conflict
[05:38] <michaelni> yes
[05:39] <Daemon404> so send to both in parallel, merge all the fixes, and send the new patch to both
[05:39] <Daemon404> is what youre saying
[05:39] <Compn> yes
[05:39] <Daemon404> seems sane enough
[05:39] <Daemon404> though i still think having to do this is silly
[05:39] <Compn> the timing does not have to be exact or anything
[05:39] <michaelni> the fork was and is silly ...
[05:40] <Daemon404> my original plan was to get it in libav, then submit it to ffmpeg and backport any new fixes
[05:40] <Daemon404> but ill use youre way, Compn 
[05:41] <michaelni> if you prefer iam fine with your plan too
[05:41] <Daemon404> ok
[05:41] <Daemon404> i dont mean to offend anyone
[05:41] <Compn> i'm not sure what is best :)
[05:41] <Daemon404> <_<
[05:41] Action: Daemon404 is neutral.
[05:41] <Compn> its hard to be neutral around these parts
[05:41] <michaelni> best is to submit just to ffmpeg-devel, and let libav merge it if they liek or not if they dont
[05:41] <Compn> many people have tried and failed
[05:41] <Daemon404> well... there's only one person i dislike out of any of the devs on any of the projects
[05:42] <Daemon404> so im mostly-neutral
[05:42] <Daemon404> ;)
[05:43] <Daemon404> and id prefer to stay out of drama.
[06:04] <Daemon404> oh, libutvideo encoder is also done. just waiting on upstream to apply my patch to fix an encoder-blocking bug on linux.
[06:04] <michaelni> Daemon404, your plan has a problem i think
[06:04] <Daemon404> hmm?
[06:04] <michaelni> when your patch is approved in libav its merged in a few hours in ffmpeg
[06:05] <michaelni> so submiting to ffmpeg afterwards feels a bit odd
[06:05] <Daemon404> oh
[06:05] <Daemon404> so, er, you prefer is send it to ffmpeg-devel now? or not at all?
[06:05] <Daemon404> up to you.
[06:05] <michaelni> send to both is better than just libav IMHO
[06:05] <Daemon404> k
[06:05] <michaelni> thx
[06:06] <Daemon404> it's already sent to libav, so ill send it to ffmpeg now... once i s/libav/ffmpeg/io
[06:06] <Daemon404> -o
[06:06] <michaelni> ill go to bed now, its 6 o clock already :)
[06:06] <Daemon404> all right
[06:06] Action: Daemon404 boots his libavcodec dev VM
[06:11] <Daemon404> michaelni, actually ill wait to send to ffmpeg... as applying my VBLE patch now to ffmpeg would cause api/abi incompatibilities
[06:11] <Daemon404> since libav has oen extar entry in the codec id's in avcodec.h
[06:11] <Daemon404> they'd be in different orders.
[06:12] <michaelni> Daemon404, read the doc in avcodec.h about the codec ids
[06:12] <Daemon404> i thought the order of teh codec ids mattered
[06:12] <Daemon404> for compatibility
[06:13] <michaelni>  * 2. Give it a value which when taken as ASCII is recognized uniquely by a human as this specific codec.
[06:13] <michaelni>  *    This ensures that 2 forks can independantly add CodecIDs without producing conflicts.
[06:13] <Daemon404> ah
[06:13] <Daemon404> i missed the latter line.
[06:14] <michaelni> also see remap_deprecated_codec_id()
[06:14] <michaelni> but now i really should go to bed ;)
[06:14] <Daemon404> good night
[09:31] <syclopse> Hi
[09:31] <syclopse> can anybody help me
[09:32] <syclopse> how can I record audio + video using ffmpeg
[10:20] <ubitux> syclopse: looks like a usage question, you should ask #ffmpeg
[10:21] <ubitux> and i think what you want is answered here: http://ffmpeg.org/ffmpeg.html#Video-and-Audio-grabbing
[13:25] <teratorn> does ffmpeg have any functions for blending frames with an alpha channel? e.g. to overlay one image on top of another with some level of transparency?
[13:41] <michaelni> teratorn, look in libavfilter/vf_overlay.c
[14:06] <CIA-18> ffmpeg: 03Carl Eugen Hoyos 07master * r367468f148 10ffmpeg/libavcodec/flicvideo.c: 
[14:06] <CIA-18> ffmpeg: Check extradata_size before accessing extradata.
[14:06] <CIA-18> ffmpeg: Fixes ticket #627.
[14:13] <teratorn> michaelni: thanks
[14:23] <Compn> teratorn : look at vf_stereo3d.c in mplayer/libmpcodecs :D
[14:23] <Compn> i think ffmpeg also has a copy of the 3d filter, but i dont think it works yet
[14:24] <teratorn> thanks, what I reallly want is some way to turn a PNG with an alpha channel in to YUV420P data + an alpha channel
[14:24] <Compn> you just want a png watermark then ?
[14:25] <teratorn> I'm currently manaully producing the YUV420P with an ffmpeg command line, but I want it done with the C API
[14:25] <teratorn> Compn: to start with yeah... I'm going to add effects though, where the watermark fades in/out.. stuff like that
[14:26] <teratorn> I have the overlay working, I just need to get the alpha working... right not I inspect the luminocity values in my YUV overlay data and if == 16 then I don't set that pixel :)
[14:26] <teratorn> s/not/now/
[14:26] <teratorn> 16 is what the background just happens to be in the YUV data that I got from the PNG
[14:34] <Compn> hmm, i think someone had done similar things, but i dont recall atm
[14:35] Action: Compn goes afk
[17:10] <ubitux> Incompatible pixel format 'rgb24' for codec 'gif', auto-selecting format 'rgb8'
[17:10] <ubitux> [...]
[17:10] <ubitux> [gif @ 0xa6ba460] ERROR: gif only handles the rgb24 pixel format. Use -pix_fmt rgb24.
[17:10] <ubitux> mmmg.
[17:13] <Daemon404> logic!
[17:14] <ubitux> it seems the gif encoding got broken at some point
[17:14] <Daemon404> michaelni, thanks for teh comments. useful stuff from both you and kostya
[17:35] <michaelni> ubitux, the gif muxer takes rawvideo in rgb24 the gif encoder takes rgb8
[17:35] <michaelni> one gives animated gif one a single image if iam not mistaken
[17:35] <Daemon404> michaelni, i understand all your comments except one of then: if(lengths[p]) pix=get_bits(gb, lengths[p])
[17:36] <Daemon404> s/then/them/
[17:37] <michaelni> get_bits(0) doesnt work
[17:37] <michaelni> so you need the if
[17:37] <Daemon404> but you wrote it under a bit of code that doesnt even use get_bits
[17:37] <michaelni> :)
[17:37] <Daemon404> which is why i am confused
[17:38] <michaelni> currently you read 4 togather in a get_bitslong
[17:38] <michaelni> and extract the 4 by shift + mask
[17:38] <Daemon404> yes
[17:38] <michaelni> you could just read each woth get_bits()
[17:38] <michaelni> it may be faster or slower
[17:38] <michaelni> but probably will be simpler & cleaner
[17:38] <ubitux> michaelni: i remember doing sth like ffmpeg -i in.avi -vframes 5 out.gif a while ago, but maybe i'm mistaken
[17:39] <ubitux> maybe it was in two steps
[17:39] <Daemon404> so that comment is assuming ive implemented the above
[17:39] <Daemon404> alright
[17:39] <michaelni> Daemon404, yes
[17:39] <michaelni> ubitux, -f gif ?
[17:39] Action: Daemon404 has tons of fixes from michael and kostya now 
[17:39] Action: Daemon404 boots VM.
[17:40] <michaelni> Daemon404, you work in a linux in a VM in windows or something?
[17:40] <Daemon404> no
[17:40] <Daemon404> i just like havign segregated VMs for different projects
[17:40] <Daemon404> because im weird
[17:41] <ubitux> michaelni: mmh indeed, -f gif and -pix_fmt rgb24 do the trick
[17:41] <michaelni> ok, then you fit well in the ffmpeg projhect :)
[17:41] <gnafu> Daemon404: Or cool, depending on how you look at it :-).
[17:41] <ubitux> too bad it's not autodetected, but anyway thx
[17:41] <Daemon404> gnafu, plus it's OK if i trash the OS
[17:41] <Daemon404> <_<
[17:41] <michaelni> ubitux, was it ever autodetected ?
[17:42] <michaelni> if it was iam happy to think about how to fix it
[17:42] <ubitux> not sure it was
[17:42] <ubitux> i don't remember doing that at least, but i might be wrong
[17:44] <michaelni> -pix_fmt rgb24 seems enough
[17:45] <michaelni> at least until someone adds .gif to the image2 muxer extensions
[17:49] <Daemon404> who runs the mplayer archive atm?
[17:49] <Daemon404> Compn?
[18:01] <CIA-18> ffmpeg: 03Michael Niedermayer 07master * r26ae9a5d7c 10ffmpeg/ (libavcodec/pthread.c libavformat/utils.c): 
[18:01] <CIA-18> ffmpeg: pthreads: drop has_b_frames mangling, its buggy in many many ways,
[18:01] <CIA-18> ffmpeg: not to mention race conditions and that its used for stream copy, used to determine IPB type by
[18:01] <CIA-18> ffmpeg: applications and other things.
[18:01] <CIA-18> ffmpeg: Fixes various frame drop/timestamp issues with frame multithreading.
[18:01] <CIA-18> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[18:06] <Compn> what archive ?
[18:07] <Compn> Daemon404 : mike runs samples.mplayerhq.hu , j-b runs the samples.libav.org/samples/ mirror
[18:07] <Daemon404> oh
[18:07] <Compn> mike melanson
[18:07] <Daemon404> k
[18:07] <Daemon404> kostya wanted a binary + samples
[18:07] <Daemon404> so i figured it'd be nice to put em ther.e
[18:08] <Compn> ah, mail mike at multimedia.cx and attach or link to them
[18:08] <Compn> since his hosting doesnt have ftp access or anything :\
[18:08] <Daemon404> ah
[18:08] <Daemon404> weird.
[18:09] <Compn> thats what i've been doing anyways ;)
[18:09] <Compn> you can also upload binary + samples to ffmpeg trac or libav bugzilla of course 
[18:09] Action: Compn not understand why libav went bugzilla
[18:15] <ubitux> at least luca preferred it
[18:15] <ubitux> because roundup was too limited
[18:17] <ubitux> http://pastie.org/private/jaddkw5jfldowpokexs7a
[18:17] <ubitux> here is an old irc log
[18:18] <Compn> oh ok
[18:18] <Compn> i didnt think too many of us figured how to control roundup from email
[18:19] <Compn> ubitux : btw, bonus points for you if you get their bugzilla to be able to search for wildcards like the mplayer bugzilla (i wonder if this is just because mplayer bugzilla is old)
[18:19] Action: Compn looks for example
[18:20] <ubitux> i don't care at all, and don't like bugzilla :p
[18:20] <Compn> ah , i can just reuse the url > http://bugzilla.mplayerhq.hu/buglist.cgi?query_format=advanced&bug_status=NEW&bug_status=ASSIGNED&bug_status=REOPENED
[18:20] <Compn> libav buglist > http://bugzilla.libav.org/buglist.cgi?query_format=advanced&bug_status=NEW&bug_status=ASSIGNED&bug_status=REOPENED
[18:20] <Compn> haha
[19:17] <CIA-18> ffmpeg: 03Michael Niedermayer 07master * r8892d0492d 10ffmpeg/ffmpeg.c: 
[19:17] <CIA-18> ffmpeg: Revert commit 0bc391281ad80d4eb365f3d8978637b07003e8dc
[19:17] <CIA-18> ffmpeg:  ffmpeg: remove -psnr option.
[19:17] <CIA-18> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[19:20] <j-b> "mplayer & mplayer2 fall in this category possibly." ; vlc too :)
[19:27] <michaelni> j-b, you use has_b_frames to count threads ?
[19:27] <michaelni> instead of what is documented ?
[19:28] <j-b> ah, maybe not
[19:29] <j-b> no, we don't
[19:29] <j-b> I think we just set the thread_count all the time
[20:11] <CIA-18> ffmpeg: 03Michael Niedermayer 07master * r46082e6186 10ffmpeg/libavcodec/fraps.c: 
[20:11] <CIA-18> ffmpeg: fraps: fix memleak
[20:11] <CIA-18> ffmpeg: Found-by reimar
[20:11] <CIA-18> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[20:12] <CIA-18> ffmpeg: 03Michael Niedermayer 07master * r82a1d57575 10ffmpeg/libavcodec/fraps.c: 
[20:12] <CIA-18> ffmpeg: fraps: check overread per sample instead of per line
[20:12] <CIA-18> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[20:31] <Daemon404> michaelni, i changed my code to read in the vals 1 at a  time rather than 4 at once into a uint32_T
[20:31] <Daemon404> it seems to be much slower
[20:31] <Daemon404> (more calls/iterations?)
[20:35] <Daemon404> oh hmm.
[20:38] <michaelni> Daemon404, do whatever is fastest
[20:40] <Daemon404> yeah
[20:44] <michaelni> note reading 32bit needs get_bits_long() reading 25bits or less just needs get_bits() which is faster
[20:45] <michaelni> it might be worth to optimize get_bits_long() if its used in one of the speed critical loops
[20:51] <iive> Daemon404: reading one byte takes same time as reading one double word (4 bytes), and probably even quad word (at 64 bit).
[20:51] <Daemon404> yes thats what i figured.
[20:56] <michaelni> the thing is just that reading 4 needs it split up later with shift and & while reading 1 does not
[20:57] <michaelni> also out get_bits_log() does 2 get_bits() calls and combines them with shift and |
[20:57] <michaelni> get_bits_loNg
[20:58] <michaelni> so currently its doing 2 16bit get_bits combines in 32bit and then splits up to 4 again. 
[21:01] <CIA-18> ffmpeg: 03Michael Niedermayer 07master * rd8407ee2b1 10ffmpeg/ffplay.c: 
[21:01] <CIA-18> ffmpeg: ffplay: limit lowres to the maximum supported. Fixes Ticket591
[21:01] <CIA-18> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[21:01] <CIA-18> ffmpeg: Signed-off-by: Marton Balint <cus at passwd.hu>
[21:02] <CIA-18> ffmpeg: 03Michael Niedermayer 07master * r8cd210700b 10ffmpeg/: 
[21:02] <CIA-18> ffmpeg: Merge remote-tracking branch 'cus/stable'
[21:02] <CIA-18> ffmpeg: * cus/stable:
[21:02] <CIA-18> ffmpeg:  ffplay: limit lowres to the maximum supported. Fixes Ticket591
[21:02] <CIA-18> ffmpeg: Merged-by: Michael Niedermayer <michaelni at gmx.at>
[21:20] <CIA-18> ffmpeg: 03Nicolas George 07master * re7062a1e9f 10ffmpeg/libavutil/ (audioconvert.c audioconvert.h avutil.h): 
[21:20] <CIA-18> ffmpeg: audioconvert: make av_get_channel_layout accept composite names.
[21:20] <CIA-18> ffmpeg: Signed-off-by: Nicolas George <nicolas.george at normalesup.org>
[21:20] <CIA-18> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[21:23] <iive> i was wondering about writing asm version of the get_bit* routines.
[21:25] <michaelni> would surely be worth a try
[21:25] <iive> there are few tricks that gcc doesn't know, e.g. how to properly use Carry flag.
[21:28] <michaelni> btw if someone wants something easy (and useless) to optimize
[21:28] <michaelni> fraps.c contains a simple "pseudo" yuv->rgb loop
[21:28] <michaelni> would be very trivial SIMD work ...
[21:29] <CIA-18> ffmpeg: 03Sebastien Zwickert 07master * r701e5348d4 10ffmpeg/libavcodec/ (vda.c vda.h): 
[21:29] <CIA-18> ffmpeg: vda: set destination image buffer attributes.
[21:29] <CIA-18> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[23:51] <CIA-18> ffmpeg: 03ami_stuff 07master * rdea8f6e18b 10ffmpeg/libavcodec/frwu.c: 
[23:51] <CIA-18> ffmpeg: lowercase FRWU string
[23:51] <CIA-18> ffmpeg: 8556406d38dc2307dd3f187c4528d64f5c350d5f (lavc: lowercase the FRWU codec name)
[23:51] <CIA-18> ffmpeg: got partially reverted by ec6402b7c595c3ceed6d1b8c1b75c6aa8336e052
[23:51] <CIA-18> ffmpeg: (lavc: use designated initialisers for all codecs.)
[23:51] <CIA-18> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[00:00] --- Thu Nov 10 2011