[Ffmpeg-devel-irc] ffmpeg-devel.log.20130220

[burek]

[00:17] <wm4> is there a bug report yet for VDA failing with recent ffmpeg?
[00:18] <wm4> I think I heard something about this, but I haven't found a ticket
[00:28] <cone-841> ffmpeg.git 03Michael Niedermayer 07master:c69315a5deb0: cook: check js_subband_start for validity
[01:26] <cone-841> ffmpeg.git 03Michael Niedermayer 07release/0.8:1fd86f9a2136: vqavideo: fix return type
[01:26] <cone-841> ffmpeg.git 03Michael Niedermayer 07release/0.8:8aedb7515674: tiff: check bppcount
[01:26] <cone-841> ffmpeg.git 03Michael Niedermayer 07release/0.8:5af2fd317d84: wma: check byte_offset_bits
[01:26] <cone-841> ffmpeg.git 03Michael Niedermayer 07release/0.8:17966ae5bb43: qdm2: increase noise_table size
[01:28] <cone-841> ffmpeg.git 03Michael Niedermayer 07release/0.8:0a57df38f468: Update for 0.8.14
[04:15] <cone-841> ffmpeg.git 03Ronald S. Bultje 07master:2b10d414648b: h264: add videodsp dependency.
[04:15] <cone-841> ffmpeg.git 03Michael Niedermayer 07master:f5c00b347dc7: avcodec_align_dimensions2: Ensure cinepak has large enough buffers.
[04:15] <cone-841> ffmpeg.git 03u-bo1b at 0w.se 07master:4e635e10e29f: cinepak: Use correct colorspace.
[04:15] <cone-841> ffmpeg.git 03Michael Niedermayer 07master:a3adbedf9b63: cinepak: simplify
[04:15] <cone-841> ffmpeg.git 03u-bo1b at 0w.se 07master:8d1dd5bd5040: cinepak: More correct Cinepak decoder.
[07:48] <highgod> Hi, I want to ask a question, I want to move the opencl init function to filter module. Can I call this function at avfilter_init_filter? because I this function will compile all opencl kernel functions and create a bin file, when ran ffmpeg next time, opencl will read the bin file and do not compile the kernel again
[07:49] <highgod> it can save time
[10:22] <durandal_1707> how i pad both inputs with -filter_complex ?
[10:26] <durandal_1707> got it: they are named: [0:v], [1:v], ...
[10:46] <durandal_1707> michaelni: calling setdar/setsar with no arguments causes segv
[11:06] <cone-768> ffmpeg.git 03Ronald S. Bultje 07master:f859678f3f27: vp56: Remove clear_blocks call, and clear alpha plane U/V DC only
[11:06] <cone-768> ffmpeg.git 03Ronald S. Bultje 07master:0c0828ecc565: x86: Use simple nop codes for <= sse (rather than <= mmx)
[11:06] <cone-768> ffmpeg.git 03Ronald S. Bultje 07master:e5ffffe48d20: h264chroma: Remove duplicate 9/10 bit functions
[11:06] <cone-768> ffmpeg.git 03Michael Niedermayer 07master:536d1ed7934d: Merge commit 'e5ffffe48d20642acc079166f0fa7d93a6a9f594'
[11:21] <cone-768> ffmpeg.git 03Ronald S. Bultje 07master:2ed008204d54: h264: Add add_pixels4/8() to h264dsp, and remove add_pixels4 from dsputil
[11:21] <cone-768> ffmpeg.git 03Michael Niedermayer 07master:088f20a33251: Merge commit '2ed008204d5467be03a0a3af1e293b2f7038d0a0'
[11:31] <cone-768> ffmpeg.git 03Stefano Sabatini 07master:d709a18ac227: lavfi/hue: simplify parsing by making use of option shorthands
[11:31] <cone-768> ffmpeg.git 03Stefano Sabatini 07master:dd3732634528: lavfi/hue: reindent after previous commit
[11:49] <cone-768> ffmpeg.git 03Ronald S. Bultje 07master:9918f57dcfc2: get_bits: Return pointer to buffer that is the result of the alignment
[11:49] <cone-768> ffmpeg.git 03Ronald S. Bultje 07master:7ebfb466aec2: h264: Don't store intra pcm samples in h->mb
[11:49] <cone-768> ffmpeg.git 03Michael Niedermayer 07master:98f63e1f34b3: Merge commit '7ebfb466aec2c4628fcd42a72b29034efcaba4bc'
[11:51] <cone-768> ffmpeg.git 03Ronald S. Bultje 07master:fae6fd5b8739: h264/svq3: Stop using draw_edges
[11:51] <cone-768> ffmpeg.git 03Michael Niedermayer 07master:8c6d9ca6387f: Merge commit 'fae6fd5b873911cfcd7f7b3d72de6598af5e003a'
[11:58] <cone-768> ffmpeg.git 03Martin Storsjö 07master:26ee5ca03d76: h264: Get rid of unnecessary casts
[11:58] <cone-768> ffmpeg.git 03Diego Biurrun 07master:8a11ce43d083: build: Ensure that output directories for header objects are created
[11:58] <cone-768> ffmpeg.git 03Michael Niedermayer 07master:39b03932c2ed: Merge commit '8a11ce43d08352f7a290355ebb5b29c495ad9609'
[12:03] <cone-768> ffmpeg.git 03Stefano Sabatini 07master:271f4112670f: lavf/matroskaenc: avoid assert failure in case of cuepoints with duplicated PTS
[12:05] <cone-768> ffmpeg.git 03Diego Biurrun 07master:c9ad5f59aa84: intreadwrite: tomi: Add missing attributes.h #include for av_always_inline
[12:05] <cone-768> ffmpeg.git 03Michael Niedermayer 07master:f7b46d86dee8: Merge remote-tracking branch 'qatar/master'
[12:36] <ubitux> any hungarian here?
[12:39] <ubitux> mmh hungarian won't do actually
[12:41] <ubitux> Compn: would you happen to have some russian subtitles?
[12:41] <ubitux> or with a weird codepage/charset
[12:41] <ubitux> (to add a fate test)
[12:45] <cone-768> ffmpeg.git 03Michael Niedermayer 07master:951d39fecd1f: avfilter: fix segfault with setsar/dar
[12:55] <ubitux> btw, is there a "is codec id pcm" function helper somewhere?
[12:56] <durandal_1707> nope
[12:56] <ubitux> ok thx
[13:08] <Compn> ubitux : hmmmmmmm, try opensubtitles.org for any big hollywood movie, then look for different languages and hope they arent uft ? 
[13:08] <Compn> but i'll look around
[13:10] <Compn> utf*
[13:11] <Compn> ubitux : http://samples.ffmpeg.org/sub/bg.sub
[13:12] <Compn> thats probably hungarian or bulgarian
[13:12] Action: Compn has no idea :)
[13:12] <Compn> strange we didnt keep samples for different languages
[13:14] <Compn> diff subcp that is
[13:16] <ubitux> Compn: nice
[13:16] <ubitux> any idea what's the charset?
[13:16] <Compn> no
[13:16] <Compn> subcp=cp1251
[13:16] <Compn> try that
[13:17] <ubitux> i tried already
[13:17] <ubitux> koi8-r and various other cp
[13:17] <Compn> ah
[13:19] <Compn> -subcp ISO-8859-1 ?
[13:19] <Compn> i'm not actually sure what its supposed to look like :)
[13:20] <Compn> oh, -subcp windows-1251
[13:20] <Compn> that looks much better
[13:20] <ubitux> doesn't seem so, unless i'm doing something wrong
[13:21] <Compn> you try in mplayer ?
[13:21] <ubitux> no :)
[13:21] <Compn> :)
[13:21] <Compn> mr testing subtitles in a untested project
[13:21] <JEEB> subs in various text encodings remind me of a guy telling me that his English subs wouldn't show up correctly and instead showed random symbols
[13:21] <JEEB> I get a sample
[13:22] <JEEB> and I get EUC-KR Korean subtitles :D
[13:22] <Compn> lol
[13:22] <ubitux> :D
[13:22] <Compn> people think whatever subs that come with the movie are for their language
[13:22] <Compn> but most new releases are only the language that the movie is in...
[13:22] <Compn> korean movie, korean subs
[13:36] <ubitux> mmh actually it worked
[13:37] <ubitux> thx Compn 
[13:50] <Compn> ubitux : so subtitle filter converts it to ass ?
[13:51] <ubitux> subtitles filter is for hardcoding
[13:51] <ubitux> not sure to understand your question
[13:52] <Compn> uhhh
[13:52] <Compn> i guess i dont know what my question was indending to ask
[13:52] <Compn> ehe
[13:52] <ubitux> @_@
[13:52] <Compn> intending*
[13:52] Action: Compn kinda distracted
[13:54] <Compn> but thats cool. subtitle encoding support
[13:54] <Compn> whats next on your list ?
[13:54] <ubitux> world domination
[13:55] <ubitux> there are some important changes to do to subtitles
[13:56] <ubitux> notably fix the ass packets (which is far from being obvious)
[13:56] <ubitux> the StyledSubtitles API (to avoid the internal representation in ASS)
[13:56] <ubitux> subtitles in lavfi...
[13:56] <ubitux> teletext
[13:56] <ubitux> and other related crazy shit :)
[14:01] <Compn> are vobsub in mkv supported ?
[14:01] <Compn> i remember something like that wasnt
[14:01] <nevcairiel> should work fine
[14:01] <Compn> ok :)
[14:12] <ubitux> <@Compn> are vobsub in mkv supported ? // you can mux and demux vobsub yes
[14:13] <ubitux> (into mkv)
[14:18] <ubitux> hey, it seems that in haskell "f $ g x $ h y $ a + b" is equivalent to "f (g x (h y (a + b)))"
[14:18] <ubitux> who is up to add this awesome syntax to ffeval ?
[14:20] <saste> ubitux, tools/ffeval -h | head -n
[14:20] <saste> ubitux, tools/ffeval -h | head -n1
[14:21] <ubitux> saste: i mean, in eval() api
[14:21] <ubitux> of course
[14:21] <ubitux> so everyone can enjoy it
[14:27] <Compn> ubitux : i thought i gave you a "windows-1251" subcp but i see "cp1251" in your fate test...
[14:27] <Compn> arent they different?
[14:27] <av500> 1251 is 1251
[14:27] <ubitux> Compn: i believe it's the same
[14:28] <av500> 850 was the non-windows one
[14:28] <Compn> oh , i guess so.
[14:29] Action: Compn should probably take a break from computer now
[15:50] <durandal_1707> saste: eval is slow
[15:51] <saste> durandal_1707, one evaluation per frame?
[15:51] <saste> it is slow in geq, when you evaluate an expression for each pixel
[15:51] <durandal_1707> that usecase is fine
[15:52] <saste> but one/few evaluations per frame are hardly measurable
[15:52] <saste> (unless you do really fancy stuff in the expression of course)
[16:04] <ubitux> saste: what do you think of making filter request window sizes?
[16:04] <ubitux> (to auto-insert asetnbsamples filters)
[16:06] <saste> ubitux, implies auto-reconfiguration
[16:06] <saste> which is still missing
[16:06] <ubitux> auto reconfig? how is that different that auto inserting scale filter etc?
[16:07] <saste> ubitux, which use case do you have in mind?
[16:07] <saste> i don't usually do things just because it is possible to do them
[16:08] <ubitux> i need a fixed granularity of samples if ebur128
[16:08] <ubitux> in*
[16:08] <ubitux> so i want ebur128 to receive fixed frames of 100ms
[16:09] <ubitux> otherwise i would need to introduce a fifo in ebur128, add a request_frame, etc and basically reproduce the asetnbsamples code
[16:09] <ubitux> which is a pain
[16:11] <cone-137> ffmpeg.git 03Michael Niedermayer 07master:2ccaab9709aa: mov: fix self referencing timecode tracks
[16:17] <ubitux> saste: so anyway, i was thinking of having such thing in the format negociation (maybe)
[16:17] <ubitux> which might be semantically incorrect
[16:18] <ubitux> the other solution might be to let the current filter responsible from auto inserting the filter itself
[16:18] <ubitux> i don't think we do that already, that might be a bit tricky
[16:18] <ubitux> and another one, call the filter internally
[16:19] <saste> ubitux, what if the filter is not compiled?
[16:19] <ubitux> error out at configuration?
[16:19] <saste> or if the filter changes its syntax?
[16:19] <ubitux> it's not a hard dep
[16:19] <saste> another option would be to factorize the code at the C-api level
[16:19] <ubitux> it's just needed/used in a particular context
[16:19] <saste> but otoh should not be so hard to use audio queue internally
[16:21] <ubitux> asetnbsamples filter code is not that obvious, duplicating its code might be pretty ugly
[16:23] <saste> you could have a filter hook which auto-inserts a filter at the configuration stage
[16:24] <ubitux> how would you do that?
[16:26] <ubitux> <@saste> another option would be to factorize the code at the C-api level // basically export all the asetnbsamples symbols? :P
[16:27] <saste> ubitux, no, i mean create an internal API which provides the correct level of abstraction
[16:27] <saste> in case this is doable
[16:28] <ubitux> btw, this may be very useful with show{spectrum,waves}
[16:28] <ubitux> to avoid flickering
[16:30] <saste> ubitux, showwaves should not need it, as it auto-adjusts itself already
[16:30] <saste> but from what i see it, the solutions are two: C-level internal API, or per-filter hooking
[16:31] <ubitux> can you detail the per-filter hooking?
[16:31] <saste> the latter could address other use cases as well, even if I can't say at the moment
[16:31] <saste> you define a callback in your filter, which is executed at some point during the configuration stage
[16:31] <saste> this may do something such as inserting a filter or such
[16:32] <ubitux> mmh sounds good, will check
[16:32] <saste> for example, earwax may auto-insert an aresample
[16:32] <saste> the problem in that case is that you don't have control over the many params of aresample
[16:33] <saste> unless you pass them to earwax, which may be ugly (and imply the usual escaping problems)
[16:33] <ubitux> what kind of aresample feature non handled by the format negociation earwax needs?
[16:33] <saste> an audiophile (the kind of guy who may want to use earwax) may be picky about the resampling algo
[16:33] <mateo`> i'm looking for some interlaced j2k/mxf samples, is there a place where i can find this kind of files ?
[16:34] <durandal_1707> saste: you picked broken filter for your explanation
[16:34] <saste> durandal_1707, then fix it! :-)
[16:34] <ubitux> mateo`: maybe ask kierank or TimNich :p
[16:35] <durandal_1707> saste: thats why there is ERROR for reson if sample rate is wrong
[16:35] <ubitux> saste: mmh well yeah ok why not
[16:36] Action: durandal_1707 gonna write another showaudio filter
[16:36] <ubitux> NIH!
[16:37] <durandal_1707> ubitux: are you saying if wrong frame size is given to ebu you get useless output?
[16:38] <ubitux> durandal_1707: nope, but the metadata will if i add the feature
[16:38] <ubitux> since you need to get information on very small slice of audio
[16:39] <ubitux> OTOH, i can also just put the normalization inside ebur128 filter...
[16:40] <ubitux> but loosing volume optimizations would be too bad
[16:41] <durandal_1707> ubitux: play ape with insane profile with ebur128
[16:41] <durandal_1707> each frame have bunch of samples
[16:42] <ubitux> we could also have a asetnbsamples=random :)
[16:42] <ubitux> that would be nice to do some testing
[16:44] <ubitux> an eval in the n field should do the trick
[16:45] <durandal_1707> i added asetnsamples (there is no asetnbsamples filter, want to write one :-/ ) and it does not help
[16:46] <ubitux> yes sorry asetnsamples
[16:46] <ubitux> what do you mean it does not help?
[16:46] <durandal_1707> i added asetnsamples=n=1024,ebur.... 
[16:47] <ubitux> wait, you actually have issues with ebur128?
[16:48] <durandal_1707> yes
[16:48] <saste> ubitux, also i never loved that name "asetnsamples", what about asetframesize?
[16:48] <ubitux> saste: oh please, we had this discussion thousand times :D
[16:48] <ubitux> durandal_1707: how to reproduce?
[16:49] <saste> ubitux, i always forget
[16:49] <saste> because to me it is a natural association: asetnsamples -> no good
[16:49] <saste> will send a patch then soon or later
[16:51] <Compn> that ebur128 sound sreally complicated
[16:53] <durandal_1707> ubitux: http://samples.ffmpeg.org/A-codecs/lossless/luckynight.ape
[16:56] <ubitux> durandal_1707: ffplay -f lavfi amovie=luckynight.ape looks broken
[16:56] <ubitux> sounds like lavfi has some trouble with it
[16:57] <durandal_1707> lol
[16:57] <ubitux> it doesn't even play the same thing
[16:57] <ubitux> wtf
[16:57] <durandal_1707> http://en.wikipedia.org/wiki/Goniometer_%28audio%29
[16:59] <ubitux> anyway, doesn't look like a problem in ebu
[17:00] <saste> anyone knows what's the problem with that file?
[17:00] <durandal_1707> how does agoniometer sounds?
[17:01] <ubitux> saste: the format
[17:01] <saste> i suppose may depend on the fact that it contains *huge* packets
[17:02] <saste> so may be a problem in movie or in the lavfi device
[17:02] <saste> but i bet on movie
[17:41] <durandal_1707> how this one actually works? By just ploting dots in 2dimensions when L is x-axis and R is y-axis?
[18:24] <durandal_1707> ahh i found one that use jack audio system
[18:33] <durandal_1707> you don't like my filters?
[18:50] <wm4> so uh how do I specify the video decoder with ffmpeg
[18:50] <wm4> it seems -codec:v sets the encoder?
[18:51] <durandal_1707> you set it before -i ?
[18:52] <wm4> no
[18:58] <Compn> ubitux : that subtitle seeking -ss  bug is kinda important :P
[18:58] <ubitux> it's not a bug
[18:58] <ubitux> it's a feature request
[18:58] <ubitux> and not easy to solve
[18:58] <Compn> i dont think its a tough fix, dont just pass the current timepos to the sub code and it goes from there ?
[19:00] <ubitux> you can't pass it easily
[19:13] <cone-561> ffmpeg.git 03Stefano Sabatini 07master:49c8fe304f69: doc/filters: apply misc fixes to drawbox documentation
[19:13] <cone-561> ffmpeg.git 03Stefano Sabatini 07master:a2a1e20fa2c8: lavfi/decimate: use named options
[19:13] <cone-561> ffmpeg.git 03Stefano Sabatini 07master:7f07c61c2f6d: doc/filters: rework introductory description for decimate
[19:21] <cone-561> ffmpeg.git 03Stefano Sabatini 07master:3d77a275487b: lavfi/overlay: implement shortest option
[19:21] <cone-561> ffmpeg.git 03Stefano Sabatini 07master:2db0056f9680: lavfi/overlay: apply minor consistency fixes
[19:37] <saste> ubitux: libavfilter/vf_subtitles.c:313:14: warning: assignment discards const qualifier from pointer target type [enabled by default]
[19:42] <cone-561> ffmpeg.git 03Stefano Sabatini 07master:22b06103f019: lavfi/overlay: add format option
[19:46] <ubitux> huh strange i didn't spot it
[19:48] <cone-561> ffmpeg.git 03Clément BSsch 07master:9b9d996b34ae: lavfi/subtitles: fix recently introduced warning about discarded const qualifier.
[19:49] <ubitux> saste: thx
[20:05] <cone-561> ffmpeg.git 03Michael Niedermayer 07master:039f55c80333: h264: call alloc_scratch_buffers() only once linesize is known
[20:05] <cone-561> ffmpeg.git 03Michael Niedermayer 07master:da7672b20a74: h264: Fix handling of changing reference counts between slices
[21:04] <durandal_1707> ubitux: why you do not use table in swf?
[21:05] <ubitux> i could
[21:05] <ubitux> but the 16-bits bit flag is not part of the passed format
[21:06] <ubitux> i could construct one and extrand the tab
[21:06] <ubitux> not sure that's a better idea
[21:06] <ubitux> afaict the s16 in the tab are supposed to be "pcm" by the specs
[21:07] <ubitux> and the bit flag i'm using somehow describe it
[21:08] <durandal_1707> ubitux: you could use ff_get_pcm_codec_id()
[21:09] <ubitux> mmh interesting
[21:09] <ubitux> maybe a bit overkill
[21:11] <durandal_1707> ubitux: yea, not really important...
[21:11] <durandal_1707> how you found flag?
[21:12] Action: durandal_1707 wonders why this was not already done
[21:12] <ubitux> web & specs
[21:12] <ubitux> i'll re-send a patch soon though
[21:16] <ubitux> durandal_1707: i wonder if it doesn't affect the adpcm
[21:16] <ubitux> but i don't have any adpcm 8bit in swf samples
[21:16] <ubitux> only pcm 8bit, and very few
[21:19] <durandal_1707> ubitux: you cant encode one?
[21:20] <ubitux> no idea how i'm supposed too
[21:20] <ubitux> to*
[21:21] <ubitux> sounds like i likely need some windows-specific black magic tools
[21:21] <ubitux> i really don't like the dup of code in swfdec
[21:22] <ubitux> not sure to understand it very well too..
[21:22] <durandal_1707> ubitux: adpcm not mentioned in spec?
[21:24] <ubitux> the specs says (about the 16-bit flag): "This parameter only pertains to uncompressed formats"
[21:25] <ubitux> do we consider adpcm uncompressed?
[21:32] <durandal_1707> no
[21:39] <ubitux> ok
[21:39] <ubitux> i'll send a new patch set soon
[22:06] <cone-561> ffmpeg.git 03Clément BSsch 07master:9a0076f50cfc: lavf/swfdec: factorize the creation of a new stream.
[22:06] <cone-561> ffmpeg.git 03Clément BSsch 07master:624fb5f96544: lavf/swfdec: support 8-bits PCM audio.
[22:44] <jojva> Hi, I'm a student of a french university, and I'd be happy to know about how far the need to implement the MVC decoder has been brought to. Is there any decoder ? A partly functional, or even a beginning of implementation of MVC ? Cheers.
[22:47] <ubitux>  D.VIL. mvc1                 Silicon Graphics Motion Video Compressor 1
[22:47] <ubitux>  D.VIL. mvc2                 Silicon Graphics Motion Video Compressor 2
[22:47] <ubitux> ?
[22:48] <ubitux> (Decode, Video, Intra-only, Lossy)
[22:48] <jojva> Oh, sorry I should have been more precise. It's about H.264 / MVC (Multiview Video Codec), not Motion Video Compressor.
[22:50] <jojva> I can see on this thread for example, that I'm not the only one hoping for this : http://ffmpeg.org/pipermail/ffmpeg-devel/2012-June/126139.html
[22:56] <cone-561> ffmpeg.git 03Carl Eugen Hoyos 07master:c4dc6c4c86a0: Fix bits_per_coded_sample when encoding png with frame-level multithreading.
[22:56] <burek> im not sure if this is a bug, so i dont know should i put it on the bug tracker or what..
[22:56] <burek> when using -acodec aac (without any other audio encoding options), and when the audio input is like this: Stream #0:1: Audio: pcm_u8 ([1][0][0][0] / 0x0001), 8000 Hz, mono, u8, 64 kb/s
[22:57] <burek> ffmpeg i guess adds default -ab 128k
[22:57] <burek> and final stream ends up like this: Stream #0:1: Audio: aac, 8000 Hz, mono, fltp, 128 kb/s
[22:57] <burek> which causes: [aac @ 0360f720] Too many bits per frame requested
[23:00] <burek> for more reference: http://ffmpeg.gusari.org/viewtopic.php?f=11&t=825&p=1627
[23:09] <ubitux> is there a way, from a muxer, to get the packet pts exactly as it was stored originally in the demuxer?
[23:10] <ubitux> it seems i'm able to get it back with something like av_rescale(ts, stream->time_base, (AVRational){1,25})
[23:10] <ubitux> but the 1/25 is a lucky guess
[23:11] <nevcairiel> shouldnt the stream time base match the timebase used for storage in the source format?
[23:12] <nevcairiel> depends on the demuxer, i guess
[23:12] <ubitux> where is the "timebase used for storage" stored?
[23:13] <ubitux> (did i misunderstood what you said?)
[23:13] <nevcairiel> for some files, its always the same, say mpegts is always 1/90000, mkv has it in its headers somewhere, and in both cases its put in the avstream timebase and the file timestamp just read and not changed at all
[23:14] <nevcairiel> i'm sure there are exceptions that do change the timestamps
[23:37] <burek> geez, forum really got big..
[23:37] <burek> a lot of people is using it now
[23:38] <burek> i wasnt there for a couple of days and now i need to spend couple of hours to reply to all the questions :/
[23:38] <ubitux> (ok get it, i needed the tb from the stream->codec)
[23:39] <ubitux> burek: it will only get worse!
[23:39] <burek> that's the (murphy's) spirit :)
[23:41] <wm4> so, you don't have a machine where you'd be able to run a VDA FATE test or so?
[23:41] <wm4> same for other hwaccel stuff
[23:41] <wm4> though these would be harder
[23:41] <wm4> as the vda pseudo decoder works like a normal decoder
[23:52] <burek> what is ffmbc?
[23:53] <ubitux> a gpl fork of ffmpeg for broadcasting
[23:53] <burek> something like embedded ffserver or worse? :)
[23:53] <ubitux> just a ffmpeg specialized for broadcasters needs
[23:54] <burek> ok, thx :)
[00:00] --- Thu Feb 21 2013