[Ffmpeg-devel-irc] ffmpeg-devel.log.20140418

[burek]

[01:16] <cone-910> ffmpeg.git 03Alessandro Ghedini 07master:32d05934abc7: mp3dec: decode more data from Info header
[01:16] <cone-910> ffmpeg.git 03Michael Niedermayer 07master:cdc166acd80b: Merge commit '32d05934abc7427bb90380a4c1ab20a15fd7d821'
[01:16] <cone-910> ffmpeg.git 03Michael Niedermayer 07master:e48fb4e1a5a9: ffmpeg: fix mixed merged & split side data with the stream->packet hack
[02:12] <cone-910> ffmpeg.git 03Michael Niedermayer 07master:97751e1324de: avformat/replaygain: fix mixed declaration and statement
[04:39] <cone-910> ffmpeg.git 03Peter Ross 07master:27998451d86d: DSD Stream File (DSF) demuxer
[05:04] <cone-910> ffmpeg.git 03Lukasz Marek 07master:24e5f5cd3ec7: lavd/pulse_audio_common: rename variables and enums
[05:04] <cone-910> ffmpeg.git 03Lukasz Marek 07master:176046d2b59c: lavd/pulse_audio_common: add context helper functions
[05:04] <cone-910> ffmpeg.git 03Lukasz Marek 07master:10627e7e10f7: lavd/pulse_audio_enc: cosmetics
[05:04] <cone-910> ffmpeg.git 03Lukasz Marek 07master:eb9dee2d0d2d: lavd/pulse_audio_enc: replace simple API with async API
[05:04] <cone-910> ffmpeg.git 03Lukasz Marek 07master:bc31a57902b8: lavd/pulse_audio_enc: add nonblocking mode
[05:04] <cone-910> ffmpeg.git 03Lukasz Marek 07master:cef118032910: lavd/pulse_audio_enc: add buffer fullness messages
[05:04] <cone-910> ffmpeg.git 03Lukasz Marek 07master:4c0f36412abc: lavd/avdevice: add volume messages
[05:04] <cone-910> ffmpeg.git 03Lukasz Marek 07master:eb9136a7d60e: lavd/pulse_audio_enc: implement volume messages
[05:04] <cone-910> ffmpeg.git 03Lukasz Marek 07master:35bf7a491e4d: lavd/pulse_audio_enc: implement pause messages
[05:04] <cone-910> ffmpeg.git 03Michael Niedermayer 07master:a614f892a6cf: Merge remote-tracking branch 'lukaszmluki/master'
[06:46] <cone-910> ffmpeg.git 03Carl Eugen Hoyos 07master:941b2240f2ce: Do not overwrite VDPAU structures in ff_MPV_frame_start().
[07:37] <ubitux> crow: http://git.videolan.org/gitweb.cgi/ffmpeg.git/?p=ffmpeg.git;a=commitdiff;h=941b2240f2ce59c41f4a9ffec88c512f64c75613
[08:31] <crow> ubitux wow that looks geat thank you
[09:00] <ubitux> crow: you should thank Carl :p
[09:01] <crow> ubitux i dont know his nickname but sure thank to Carl, soon i will test it.
[10:17] <lucaswang> hi all. in ffmpeg, which pixel format equals y210 (Packed, 4:2:2, 10-bit) ? i only see 10bit planar yuv
[10:32] <wm4> lucaswang: it's not supported directly, but there's a raw codec to consume/produce this format
[10:33] <cone-220> ffmpeg.git 03Carl Eugen Hoyos 07release/2.2:26b6d70c72e4: Do not overwrite VDPAU structures in ff_MPV_frame_start().
[10:37] <lucaswang> ok. i see. thanks.
[11:15] <ubitux> don't we need d5c0036d4adcec7a38aad6e3aab8792ab9bc074a in avfundation code as well?
[13:03] <Compn> wm4 : hmm, adding y210 would be trivial for the right persons...
[13:04] <Compn> so many colorspaces. i think they are only added when users request them now
[13:04] <Compn> or sample exists
[13:04] <wm4> as usual you have no idea what you're talking about
[13:04] <wm4> maybe
[13:05] <wm4> hm he asked for y210, probably wanted v210
[13:06] <Compn> is v210 packed or planar ?
[13:06] <wm4> or maybe they're the same
[13:06] <wm4> packed
[13:12] <wm4> yeah, seems y210 == v210
[13:22] Action: funman never heard about y210
[13:23] <funman> http://msdn.microsoft.com/en-us/library/windows/desktop/bb970578%28v=vs.85%29.aspx -> uses y210
[13:29] <wm4> funman: I didn't compare the formats closely (because it's so fucked up), but one result said they're exactly the same
[13:30] <wm4> one strange thing is that ffmpeg has a v210 fourcc, but not a y210 one
[13:33] <funman> wm4: next to that URL: " Includes description of what must be V210, inexplicably named as Y210. Comments welcome."
[13:34] <wm4> funman: yep saw that too
[13:42] <muken> v210 is a quicktime thing.
[13:46] <Compn> it doesnt help that every single broadcast codec comes up with its own *210 fourcc
[13:46] <Compn> or isom
[13:46] <muken> wait. v210 != Y210
[13:47] <Compn> wm4 : its non trivial to add another 10bit colorspace ? 
[13:47] <wm4> muken: oh?
[13:47] <muken> Y210 == v216 with significantBits=10 of sgbt extension 
[13:47] <wm4> oh
[13:48] <Compn> there is a codec that has special cases for each colorspace, that would be difficult. i think kostya looked into it once :)
[13:48] <Compn> like reinventing swscale for one codec :D
[14:09] <BBB> oh man, scaling
[14:09] <BBB> oh joy
[14:48] <cone-220> ffmpeg.git 03Peter Ross 07master:0a13487b1a2d: avformat/dsddec: create audio stream before processing id3
[15:56] <cone-220> ffmpeg.git 03Michael Niedermayer 07master:b9831364de51: avcodec/vc1dec: use av_realloc_array()
[15:56] <cone-220> ffmpeg.git 03Michael Niedermayer 07master:bb0d9aedcbda: avcodec/tta: use av_malloc(z)_array()
[15:56] <cone-220> ffmpeg.git 03Michael Niedermayer 07master:71f8cdcc1648: avcodec/snow: use av_malloc(z)_array()
[18:33] <cone-220> ffmpeg.git 03James Almer 07master:968c44ce8950: configure: Remove deprecated incompatible_fork_abi option
[18:41] <cone-220> ffmpeg.git 03Justin Ruggles 07master:39ef000e19b5: mjpeg: Do not fail jpeg decoding on bad EXIF data.
[18:42] <cone-220> ffmpeg.git 03Justin Ruggles 07master:591c0c26b598: mjpeg: cosmetics: indentation
[20:23] <ubitux> http://wiki.multimedia.cx/index.php?title=H4M
[20:23] <ubitux> anyone worked on this?
[20:24] <JEEB> not that I have seen
[20:27] <ubitux> linkmauve1 has some samples from the pikmin game
[20:27] <ubitux> typically "sr_S.h4m: HVQM4 v1.3 GameCube movie, 640 x 480, 33367us, 48000Hz audio"
[20:27] <linkmauve1> Hi. o/
[20:28] <JEEB> the closest I've had to pikmin was the sample from pikmin2's opening or so
[20:28] <JEEB> which was in THP
[20:29] <JEEB> which IIRC contained some sort of MJPEG
[20:29] <JEEB> so yeah, that's something else :3
[20:29] <linkmauve1> I remember playing THP from SMS.
[20:29] <linkmauve1> Using ffmpeg.
[20:29] <linkmauve1> Actually I can check right now.
[20:29] <JEEB> yeah, the audio and video from THP is supported
[20:29] <linkmauve1> Ok.
[20:30] <JEEB> "ham" isn't yet, afaik
[20:31] <linkmauve1> Maybe vgmstream has some information already, Im going to check.
[21:09] <cone-220> ffmpeg.git 03Michael Niedermayer 07master:9b888fb159eb: avcodec/roqvideoenc: use av_malloc(z)_array()
[21:09] <cone-220> ffmpeg.git 03Michael Niedermayer 07master:683b6e3183ee: avcodec/smacker: use av_mallocz_array()
[21:09] <cone-220> ffmpeg.git 03Michael Niedermayer 07master:60c8c7144e1d: avcodec/qtrleenc: use av_mallocz_array()
[21:27] <linkmauve1> I love it when people leave debug info in the released game.
[21:28] <linkmauve1> http://linkmauve.fr/files/build.map.txt
[21:42] <ubitux> nice
[21:42] <ubitux> debug symbols are there too?
[21:42] <ubitux> :))
[21:43] <ubitux> >Decode[IPB]Pic
[21:43] <ubitux> heh.
[21:43] <ubitux> just force some mpeg* video codec to see if it outputs something
[21:44] <ubitux> linkmauve1: can you get some asm out of it or something?
[21:44] <ubitux> (you're using an emulator right?)
[21:44] <linkmauve1> Here is the file supposedly containing that code: http://linkmauve.fr/files/boot.dol
[21:44] <linkmauve1> I can only get it at runtime, the debugger seems fairly limited.
[21:45] <linkmauve1> I think its there mostly to help debug emulator issues, and not the game itself.
[21:46] <ubitux> >boot
[21:46] <ubitux> ??
[21:48] <linkmauve1> Its the only dol file in the disk, and the only file matching HVQM.
[22:21] <Daemon404> https://plus.google.com/+LinusTorvalds/posts/H9oSRqwSfaM
[22:23] <ubitux> yeah well, the rotate meta is available since a long while already
[22:23] <ubitux> i wonder if i should send a small patch for ffplay for the lulz
[22:26] <ubitux> does anyone have some rotated sample available?
[22:26] <beastd> lol. i had this very problem. i shot a video with my phone a week ago on my vacation. wanted to show it at work , but first i worked out the filter chain for mplayer because it was upside down ;)
[22:26] <iive> i have the feeling that he rotated it manually, using the rotate filter. this is why he is happy that it remembers the correct setting.
[22:26] <beastd> ubitux: you want mine ;)
[22:26] <ubitux> i think that's something like maybe 40 line of code for 99% of the common cases
[22:27] <Daemon404> lol
[22:27] <ubitux> i'm willing to give it a shot but i need 3 samples with the common 90, 180 and 270°
[22:28] Action: Daemon404 handles it at work!
[22:28] <Daemon404> i might have a 90
[22:28] <Daemon404> sec
[22:30] <Compn> Daemon404 : mplayer superiority....
[22:31] Action: iive takes cover
[22:33] <Daemon404> Compn, he is praising the ui
[22:33] <Daemon404> aka not mplayer :P
[22:33] <Daemon404> ubitux, i notice'd you a link
[22:33] <Daemon404> 7.4mb
[22:33] <Compn> :P
[22:33] <ubitux> Daemon404: thx
[22:34] <Compn> Daemon404 : good point, i'll mail ingo and see if we cant stuff that into gmplayer :P
[22:44] <beastd> ubitux: Mailed you a 6MB sample as attachment. Hope it got through.
[22:48] <ubitux> v
[22:48] <ubitux> beastd: heh
[22:48] <ubitux> beastd: got it, but i'm not using that mail anymore you know
[22:48] <ubitux> anyway thx
[22:50] <ubitux> the kb on this machine is going to drive me nuts
[22:52] <beastd> ubitux: sorry. i need to update to your new mail adress then
[23:05] <ubitux> ok seems i have a working PoC
[23:06] <ubitux> http://b.pkh.me/0001-ffplay-support-rotated-video.patch
[23:07] <ubitux> should be fine, right?
[23:13] <linkmauve1> Oh, I think I just found a bug in brstms playback (maybe due to mpv), if I seek back, playback will end earlier than the end of the stream, with a red [ffmpeg/audio] adpcm_thp: invalid number of samples in packet error on stdout.
[23:14] <Daemon404> what are the macros for benchmarking?
[23:14] <Daemon404> deccycles etc
[23:14] <linkmauve1> And if I seek forward, playback stops at about one minute.
[23:15] <jamrial> START_TIMER and STOP_TIMER
[23:15] <jamrial> check timer.h
[23:16] <Daemon404> right
[23:16] <Daemon404> thanks
[23:16] <ubitux> linkmauve1: if you can't reproduce with ffplay you should nag wm4 
[23:17] <ubitux> rotated video patch for ffplay on ml
[23:17] <jamrial> "trolled-by" lol
[23:19] <linkmauve1> With ffplay I get L04.brstm: error while seeking when I click.
[23:19] <linkmauve1> Sometimes seek goes back to the start.
[23:19] <linkmauve1> seeking*
[23:20] <ubitux> i see no seek code in lavf/brstm.c
[23:21] <J_Darnley> Is transpose slower than the two flips for 180 rotation?
[23:21] <linkmauve1> Oh, thats probably why then.
[23:22] <Daemon404> not sure, but i do know our transpose code is *SLOW*
[23:22] <ubitux> J_Darnley: can you 180 with transpose?
[23:22] <Daemon404> well you need a flip too
[23:22] <J_Darnley> I don't know
[23:23] <ubitux> i should use the special keys for transpose btw, should be better than "1" and "2"
[23:23] <J_Darnley> perhaps I don't have a good definition of "transpose"
[23:30] <ubitux> i wonder how better that will become with the overthought rotation api
[23:32] <Compn> lol @ 'clock' 
[23:32] <Compn> ubitux : you were reading #videolan ?
[23:32] <ubitux> nope
[23:33] <ubitux> Compn: can you sum up for me?
[23:33] <Compn> arguing between positive and negative numbers and angles and clocks. also left/right. basically what to call the rotation, transform, etc
[23:34] <Compn> and getting nowhere
[23:34] <ubitux> ah sure, whatever
[23:34] <ubitux> well i guess they're likely overthinking the issue like koda is doing with the new clean awesome side data api
[23:35] <ubitux> ...but hey, <1h to just forward the rotate metadata to lavfi and be done with that looks good enough to me
[23:35] <ubitux> anyone seeing a downside to that approach?
[23:36] <Compn> nope
[23:36] <Compn> making more api is bad :P
[23:36] <Compn> especially in bikeshed country
[23:42] <ubitux> i wonder how vlc does it btw
[23:43] <ubitux> i guess they don't use the "rotate" metadata, don't they?
[00:00] --- Sat Apr 19 2014