[Ffmpeg-devel-irc] ffmpeg-devel.log.20140926

[burek]

[00:16] <cone-763> ffmpeg.git 03Carl Eugen Hoyos 07master:8de7bdd756b6: Show correct pix_fmt for vp9 videos != AV_PIX_FMT_YUV420P.
[00:16] <cone-763> ffmpeg.git 03Carl Eugen Hoyos 07master:475e3799cdcc: Support libopenjpeg 2.x via libopenmj2.
[00:16] <cone-763> ffmpeg.git 03Carl Eugen Hoyos 07master:88c937fdc8a8: Autodetect jpg images.
[00:29] <cbsrobot_> ubitux: do you mind if I add to the opw Misc Libavfilter extension something like: add >8bit support to libavfilter/drawutil.c ?
[00:54] <Timothy_Gu> Hey all, why does https://lists.ffmpeg.org/mailman/listinfo/ffmpeg-devel/ return 404?
[00:54] <reynaldo> Timothy_Gu: llogan said michaelni disabled mailman due to some problem.bug
[00:55] <reynaldo> problem and/org bug I mean
[00:57] <Timothy_Gu> reynaldo, oh
[00:58] <Timothy_Gu> reynaldo, llogan, michaelni, do you know when it will be back up?
[01:03] <reynaldo> Timothy_Gu: no idea
[01:03] <llogan> Timothy_Gu: i'm not sure
[01:17] <wm4> ubitux, BBB: so, as you know, vp9 packets in webm sometimes contain two frames
[01:18] <wm4> but libavcodec wants them neatly separated, and the parser usage in libavformat takes care of that
[01:18] <wm4> but this also means that the decoder does not return a frame sometimes, and that the PTS for 1 frame will be lost
[01:19] <wm4> that the decoder doesn't return a frame is very odd, I don't think any other decoder behaves this, as long as the packets are not corrupted, and frame skipping is disabled
[01:22] <llogan> ...it's not deleted though. just disabled temporarily.
[01:39] <rcombs> don't quite a lot of decoders sometimes not return a frame? E.g. if running multithreaded?
[01:41] <wm4> only at start of decoding
[01:41] <wm4> once you get the first video frame, decoding should always return a frame
[01:41] <wm4> unless you direct it to drop frames, or maybe if an error of some kind happened
[01:51] <iive> hum... doesn't the decoder return pts in the avframe?
[01:51] <iive> timestamp that he gets as dts.
[01:51] <wm4> the pts/dts is unset
[01:52] <wm4> because the returned frame originates from the parsed packet, which doesn't have and can't have a pts
[01:52] <iive> ok
[01:52] <wm4> I'm not quite sure what vp9 issue this was, something about golden frames
[01:52] <iive> still, afair, the decoder signals when frame is dropped and when there is no frame available.
[01:52] <wm4> and how this is not an issue for libvpx, because it can handle multiple frames in a packet
[01:53] <wm4> iive: dunno, does it?  but makes no difference here anyway, because the pts association is gone
[01:53] <iive> the code that handles the initial lag should handle this case...
[01:54] <iive> didn't you just said that there is pts/dts ?
[01:54] <iive> there is no pts/dts
[01:56] <wm4> yes, possible the decoder could reorder it correctly, or something
[01:58] <iive> oh, you mean that you are loosing the last frame
[01:58] <iive> this is an issue with all b-frame lag formats, afair, it was workarounded by calling decode without new packet
[01:59] <wm4> no
[01:59] <wm4> that's not my issue
[02:00] <iive> oh..
[02:07] <iive> so golden frames are frames that are not displayed, but later may be used for prediction with mv00 and cbp=0, aka emulation of b-frames.
[02:11] <iive> so these hidden frames don't have pts.
[02:12] <iive> i'm off. have fun.
[02:27] <michaelni> about the ML, just the web interface is disabled, mailman should be working fine through the email interface
[02:39] <Compn> so moderation is down then ?
[02:40] <Compn> or can be moderated via email? never tried that
[02:54] <michaelni> Compn, the webinterface should be up again in a day or 2
[03:00] <Compn> no worries
[04:49] <cone-964> ffmpeg.git 03Thilo Borgmann 07master:a69c70e148d9: lavd/avfoundation: Split adding a device and getting the device configuration into separate functions.
[05:13] <cone-964> ffmpeg.git 03Thilo Borgmann 07master:92827e186999: lavd/avfoundation: Using the actual stream index instead of hardcoded value.
[05:13] <cone-964> ffmpeg.git 03Thilo Borgmann 07master:1ea7a3e04e42: lavd/avfoundation: Simplify debug message generation.
[07:29] <rcombs> ^ we have confirmation that the numbers-in-usernames issue is fixed
[08:39] <Ritika> reynaldo: Hi!
[09:53] <ubitux> wm4: i don't know for the vp9 parser
[09:55] <nevcairiel> the timestamp thing is just a design limitation of the generic parsing code
[09:55] <nevcairiel> redesign welcome? :p
[09:55] <wm4> is that so
[09:55] <nevcairiel> that the vp9 decoder swallows soem frames is just a property of the codec otherwise
[09:56] <nevcairiel> well, if the parser detects two frames in one packet, it splits the packet and assigns the timestamp to the first frame
[09:56] <nevcairiel> in the vp9 case, this first frame is the one frame that gets swallowed later, and the timestamp lost
[09:57] <nevcairiel> but of course the parser has no way to know this stuff, since its design is limited
[09:57] <durandal_1707> what happened with mailing list archive?
[10:15] <ubitux> wm4: can you send me a sup file?
[10:15] <ubitux> would be nice to have the muxer though
[10:15] <ubitux> so i could extract some weird pgs and try stuff with them
[10:17] <durandal_1707> there is already patch for muxer
[10:17] <ubitux> yes but pending
[10:18] <ubitux> still waiting for an iteration, right?
[10:18] <ubitux> i'll push the sup demuxer in a moment but i'd like a file first
[10:19] <wm4> ubitux: can we try DCC?
[10:19] <ubitux> go ahead
[10:19] <wm4> wow it works
[10:19] <ubitux> :)
[10:19] <wm4> I just notice the file is WTFish huge though
[10:20] <ubitux> awesome speed
[10:20] <wm4> yeaaah
[10:22] <wm4> ubitux: there's also this one http://samples.mplayerhq.hu/sub/BluRay/title03_track2.sup
[10:22] <ubitux> ah, thanks
[10:23] <ubitux> 36M meh
[10:38] <cone-409> ffmpeg.git 03wm4 07master:7dd2005e1ac3: avformat: add SUP/PGS subtitle demuxer
[10:43] <rcombs> blimey, was that not there before?
[10:48] <ubitux> blimey?
[10:48] <av500> https://www.google.com/search?q=blimey
[10:48] <nevcairiel> he must be english
[10:49] <rcombs> not even
[10:49] <rcombs> while that expression isn't particularly common in the US, I thought it was pretty well-known internationally just because of its use in British literature and TV
[10:50] <ubitux> oh, ok
[10:51] <ubitux> that sup demuxer really feels simpler that the vobsub one
[10:57] <wm4> nice
[11:14] <cone-409> ffmpeg.git 03Clément BSsch 07master:9ae83ece8a8e: avformat/srtdec: simpler and more lenient probing
[12:23] <cone-409> ffmpeg.git 03Thilo Borgmann 07master:f2254e36c048: lavd/avfoundation: Use microseconds as common timebase.
[12:29] <cone-409> ffmpeg.git 03Diego Biurrun 07master:4d5b99dacdbb: caf: Give context structure a consistent name
[12:29] <cone-409> ffmpeg.git 03Michael Niedermayer 07master:b0458e62c8fc: Merge commit '4d5b99dacdbb56373e5abe2ad35fc87b5c0c5bd2'
[12:40] <cone-409> ffmpeg.git 03Diego Biurrun 07master:b97f6ef95629: pcm-dvd: Move a variable to a smaller scope
[12:40] <cone-409> ffmpeg.git 03Michael Niedermayer 07master:4c57be124f42: Merge commit 'b97f6ef956293fe06ae1e99bd295f42613081f37'
[12:59] <cone-409> ffmpeg.git 03Michael Niedermayer 07master:ee2e5acdb13f: avcodec/dxtory: remove unused variables
[12:59] <cone-409> ffmpeg.git 03Peter Ross 07master:1964251be7df: libavformat/iff: print error message when DSDIFF compression type is not supported
[20:54] <cone-646> ffmpeg.git 03Tim Walker 07master:b5fe13448b13: vc1: Use logical instead of bitwise or for coded_inter
[20:54] <cone-646> ffmpeg.git 03Michael Niedermayer 07master:bde8eb7fa33d: Merge commit 'b5fe13448b13bfdd48a75b7e12387e11e3526d7c'
[21:00] <cone-646> ffmpeg.git 03Michael Niedermayer 07master:a97328afef0c: vc1: Use logical instead of bitwise or for twomv
[21:00] <cone-646> ffmpeg.git 03Michael Niedermayer 07master:ecf552657889: Merge commit 'a97328afef0ccebfc8c3d9f9fdb8e93cbf1058ab'
[22:06] <cone-646> ffmpeg.git 03Tim Walker 07master:b39ebcddd47d: fate: Add VC-1 interlaced twomv test
[22:06] <cone-646> ffmpeg.git 03Michael Niedermayer 07master:b7caf8940123: Merge commit 'b39ebcddd47daf37659796aaa7d068668086507a'
[22:18] <cone-646> ffmpeg.git 03Tim Walker 07master:4c2fd4b26234: ac3enc: allow Dolby Pro Logic II as a preferred downmix mode.
[22:18] <cone-646> ffmpeg.git 03Michael Niedermayer 07master:15fda8bcf1ea: Merge commit '4c2fd4b262347273afe97865ba451a1abde43ae6'
[22:27] <cone-646> ffmpeg.git 03Tim Walker 07master:cc4992aaf3db: ac3enc: allow Dolby Pro Logic IIz as the Dolby Surround EX mode.
[22:27] <cone-646> ffmpeg.git 03Michael Niedermayer 07master:c895fa7f66c0: Merge commit 'cc4992aaf3dbb0af88d9727983d75636baf1f8cc'
[22:39] <cone-646> ffmpeg.git 03Tristan Matthews 07master:7703995a2e87: flac: Remove unused headers
[22:39] <cone-646> ffmpeg.git 03Michael Niedermayer 07master:7618d360ca0b: Merge commit '7703995a2e8760e66ff35148bf029e0f9e30dbe6'
[23:54] <cone-646> ffmpeg.git 03Reynaldo H. Verdejo Pinochet 07master:e3e051c42c48: libavfilter/af_join: avoid derreferencing possible null
[23:54] <cone-646> ffmpeg.git 03Reynaldo H. Verdejo Pinochet 07master:32288234a200: libavfilter/af_channelmap: avoid derreferencing possible null
[23:54] <cone-646> ffmpeg.git 03Reynaldo H. Verdejo Pinochet 07master:a9ea79bb4976: libavfilter/af_amix: avoid derreferencing possible null
[00:00] --- Sat Sep 27 2014