[Ffmpeg-devel-irc] ffmpeg-devel.log.20140202

[burek]

[00:21] <cone-542> ffmpeg.git 03Michael Niedermayer 07master:4aa4533ee8b5: avdevice/v4l2: only use frame period from v4l2 if valid
[00:21] <cone-542> ffmpeg.git 03Michael Niedermayer 07master:0997c2504b25: avdevice/v4l2: only use average frame rate if set
[00:22] <michaelni> ubitux, hypothetical case fixed, thanks for pointig it out
[00:22] <ubitux> cool, thx :)
[00:23] <michaelni> (hypothtical as our code is as yu say different, ....)
[02:42] <cone-542> ffmpeg.git 03Michael Niedermayer 07master:a744064c4155: avcodec/tiff: reset geotag_count in free_geotags()
[02:42] <cone-542> ffmpeg.git 03Michael Niedermayer 07master:96c4ba2392b9: avcodec/hevc: propagate error code from hls_coding_quadtree()
[02:48] <cone-542> ffmpeg.git 03Lukasz Marek 07master:ffe46e5d13f8: lavd/opengl_enc_shaders: make shaders code static and const
[02:48] <cone-542> ffmpeg.git 03Lukasz Marek 07master:c72f916c151e: lavd/opengl_enc: create format description table
[02:49] <cone-542> ffmpeg.git 03Lukasz Marek 07master:c9074b992e31: lavd/opengl_enc: rename GL_RED_COMPONENT define
[02:49] <cone-542> ffmpeg.git 03Lukasz Marek 07master:4afe1c117855: lavd/opengl_enc: factorize create/release window functions
[02:49] <cone-542> ffmpeg.git 03Lukasz Marek 07master:c8275331bac5: lavd/opengl_enc: factorize deinit_context function
[02:49] <cone-542> ffmpeg.git 03Lukasz Marek 07master:1fb1ddf64749: lavd/opengl_enc: simplify opengl_load_procedures usage
[02:49] <cone-542> ffmpeg.git 03Lukasz Marek 07master:69b19f2642c3: lavd/opengl_enc: fix window caption
[02:49] <cone-542> ffmpeg.git 03Lukasz Marek 07master:5059d8dfe3c8: lavd/opengl_enc: remove inlines
[02:49] <cone-542> ffmpeg.git 03Michael Niedermayer 07master:549f052239d9: Merge remote-tracking branch 'lukaszmluki/master'
[03:29] <funman> michaelni: are you at FOSDEM?
[03:29] <michaelni> no
[03:30] <funman> ah that's a shame, why not
[03:30] <funman> it's very crowded but on the other hand it's nice to meet some people i only know from irc or mailing lists
[03:31] <michaelni> iam too much a geeky computer guy to travel around and meet people in real life
[03:32] <wm4> did you enter a deathmatch with someone from libav yet?
[03:32] <funman> well it's only geeky computer people anyway
[03:32] <wm4> (mandatory joke)
[03:32] <funman> no but i trolled libav people hard
[03:32] <wm4> oh?
[03:33] <funman> well I would tell you how if you come to FOSDEM next year ;D
[03:34] <wm4> too many scary people
[03:34] <funman> usually I do an audio recording of every VideoLAN meeting we make but this year I forgot to bring my recorders
[03:37] <funman> anyway this is your choice but you're missing on an occasion of trolling ;)
[04:16] <BBB> funman: I sat in one of those meetings when I was still on that side of the fork - but all those meetings are just useless
[04:16] <BBB> funman: see, most people just sit there, say their fancy word, then go on do whatever they did before - as if the meet never happened or nothing was said
[05:14] <cone-542> ffmpeg.git 03Luca Barbato 07master:7d56f2fffc14: mxf: Handle negative edit_rate
[05:14] <cone-542> ffmpeg.git 03Michael Niedermayer 07master:8fb67c102655: Merge commit '7d56f2fffc14ad02001ee0689bee76a759da9ede'
[05:30] <cone-542> ffmpeg.git 03Matthieu Bouron 07master:c71662485333: mxf: Read field dominance flag
[05:30] <cone-542> ffmpeg.git 03Michael Niedermayer 07master:23e48cd01fbd: Merge commit 'c716624853339ea4b1086a3957b334829370204c'
[05:31] <michaelni> mateo`, can you take a look at c71662485333 / 23e48cd01fbd, is it correct ? do we have something to test this ?
[05:39] <cone-542> ffmpeg.git 03Luca Barbato 07master:09e2203b8ba6: hevc: Consider first quantization group any reference to 0, 0
[05:39] <cone-542> ffmpeg.git 03Michael Niedermayer 07master:493f750e7a50: Merge commit '09e2203b8ba6943d5c0fe6d73b65b145c3fdf98e'
[05:46] <cone-542> ffmpeg.git 03Luca Barbato 07master:816e5b997028: hevc: Reject impossible slice segment
[05:46] <cone-542> ffmpeg.git 03Michael Niedermayer 07master:d9bb7fc3cb18: Merge commit '816e5b997028c8215c804b1e58b2388592ed612b'
[06:01] <cone-542> ffmpeg.git 03Michael Niedermayer 07master:ef034cbf185e: fate/xvid: fix idct & bitexact flag
[06:19] <cone-542> ffmpeg.git 03Reynaldo H. Verdejo Pinochet 07master:95d180998193: libavformat/mtv: add test for audio magic
[11:39] <JEEB> I'm somewhat surprised by the fact that I still can't find where exactly libavformat/mpeg.c decides the amount of streams
[12:08] <cone-959> ffmpeg.git 03Stefano Sabatini 07master:a270f63e832b: lavfi/overlay: add yuv422 format mode
[12:08] <cone-959> ffmpeg.git 03Stefano Sabatini 07master:5fc4dea39caf: examples: add avio_reading.c example
[12:20] <J_Darnley> Wow.  Spam on the bug tracker
[12:20] <JEEB> yeah, tracs are among the things to get random spam on them
[12:21] <JEEB> also do you happen to know mpeg.c ?
[12:22] <J_Darnley> Yeah, anything where the internet can post comments will get spam.  Still, it is the first time I saw it there
[12:22] <J_Darnley> and was that question directed at me?
[12:22] <JEEB> the FFmpeg one has gotten spam every now and then
[12:22] <JEEB> and yes
[12:23] <J_Darnley> No, not really.  I've only made one change relating to mpeg encoding and that was to make ffmpeg not error when encoding jpeg images with constant qualtiy at a low framerate
[12:24] <JEEB> ok
[12:24] <JEEB> I am just kind of... overwhelmed with the MPEG-PS parser
[12:24] <JEEB> wanted to find out why libavformat wouldn't pick up a stream at all
[12:26] <J_Darnley> ugh good luck
[13:16] <cone-959> ffmpeg.git 03Stefano Sabatini 07master:60251c13e3c7: examples/avio_reading: remove stray empty line
[13:16] <cone-959> ffmpeg.git 03Stefano Sabatini 07master:290da7e75461: ffmpeg: tweak debug_ts logging and add more of it
[13:16] <cone-959> ffmpeg.git 03Stefano Sabatini 07master:a535d3952c66: lavf/avformat.h: fix typos
[13:16] <cone-959> ffmpeg.git 03Stefano Sabatini 07master:5871ee5072f7: lavf: add output_ts_offset option to AVFormatContext
[13:16] <cone-959> ffmpeg.git 03Stefano Sabatini 07master:bf96e547b738: doc/APIchanges: fix weird reindent
[13:18] <saste> why AVFrame.type is deprecated?
[13:19] <saste> what am I supposed to use to get the frame type?
[13:21] <nevcairiel> whats a frame type?
[13:30] <BBB> I think it's keyframe/nonkeyframe
[13:30] <BBB> saste: use pict_type?
[13:30] <BBB> (P, B, I, etc)
[13:30] <BBB> where I is key
[13:30] <nevcairiel> there is even a seperate key_frame thing if thats really it
[13:32] <michaelni> saste, AVFrame.pict_type doesnt lok deprecated
[13:32] <BBB> why do you need to know any of that in raw frames?
[13:32] <BBB> each frame is a keyframe once it's decoded
[13:34] <nevcairiel> its kinda useful to know if it was a keyframe
[13:34] <nevcairiel> ie. you can decide to suppress frames until the first key_frame comes along
[13:34] <nevcairiel> not all decoders do that for you and happily output garbage
[13:36] <nevcairiel> its all just informational, not crucial flags in any way
[13:36] <nevcairiel> somtimes helps debugging an issue though
[13:36] <JEEB> I guess I'll make an issue of this sample... I really seem to be out of my element with lavf/mpeg.c
[13:37] <nevcairiel> maybe you should just tell the user that gave it to you to go away! :D
[13:37] <JEEB> :D
[13:37] <JEEB> I just decided to try it because libavcodec now has decoders for all the related formats
[13:37] <JEEB> and then just tried prying it open with something that parses MPEG-PS
[13:37] <JEEB> it's UMD Video, a sample I had around for a while
[13:38] <saste> michaelni, BBB: AVFrame.type, not pict_type
[13:38] <saste> it is used for example in movie to detect the *media* type of the frame
[13:38] <saste> A/V/S etc
[13:39] <nevcairiel> S doesn't use AVFrame, and for  A/V do what all the other code does, nb_samples is set, its A, otherwise V :P
[13:40] <michaelni> saste, if that worked thats just by coincidence
[13:40] <michaelni> type isnt the media type
[13:40] <michaelni> its buffer type like FF_BUFFER_TYPE_INTERNAL
[13:40] <saste> michaelni, ah!
[14:24] <BBB> I don't understand why video and audio share AVFrame, I mean, they share 1 or 2 variables like timestamp and ... that'sa bout it, but they're entirely different, why not make a AVFrame for video data and AVSamples for audio data?
[14:24] <BBB> to say that width applies to audio or samplerate to video is something only an engineer can come up with
[14:26] <nevcairiel> at least this way allocation and stuff related can be the same function
[14:26] <BBB> so common code can't be shared out of their structures being identical? how odd
[14:27] <anshul_> It would require an major bump, lots of thing will crash because of it
[14:27] <BBB> int allocate_video() { ... allocate_common(); .. } int allocate_audio() { .. allocate_common(); .. }
[14:27] <nevcairiel> just to get a few variables out of the context?
[14:28] <BBB> I think it's misdesign
[14:28] <BBB> but hat do i care i dont write spps, just codecs
[14:28] <BBB> s/spps/apps/
[14:29] <nevcairiel> I probably wouldn't have build it like this, but I also wouldn't change all the APIs to change it now
[14:30] <BBB> I can understand that part, I suppose I'm just trying to irritate the person that came up with this originally to crumble in humiliation for this suckiness
[14:30] <BBB> it's indeed probably not worth fixing now
[15:09] <saste> BBB: AVFrame design was defined in libavcodec ages ago
[15:09] <saste> it was video only at some point
[15:09] <saste> then the avcodec_encode/decode API was extended, and AVFrame was used to contain audio data as well
[15:10] <saste> i believe the idea was to have an unified API for both audio and video
[15:10] <saste> a for what regards AVSubtitles, it is still a different struct
[15:11] <saste> a possible idea would be to put all the A/V related fields in a contained structure
[15:55] <cone-959> ffmpeg.git 03Stefano Sabatini 07master:e756effd5858: lavfi/movie: fix display of pushed frame information
[16:22] <cone-959> ffmpeg.git 03Andrey Utkin 07master:6d7119dbd26a: lavfi/drawtext: fix reinit command
[17:54] <BBB> saste: let me just repeat myself, it's a terrible idea; audio and video are nothing alike
[17:55] <BBB> saste: and no, subtitles isn't any different - it has nothing alike with either video or audio
[17:55] <BBB> your body is a good example of how different they are - it chose to create different organs (eye vs ear) to perceive such senses. we should not try to be smarter than our body
[17:55] <nevcairiel> did you hear the brilliant plans to get rid of AVPacket and use AVFrame for compressed as well? :D
[17:59] <wm4> everything shall be a graph
[18:09] <saste> wm4, everything shall be a pony
[18:09] <nevcairiel> yay pony
[18:12] <JEEB> AVVideoPony
[18:12] <JEEB> AVAudioPony
[18:12] <JEEB> AVSubtitlePony
[18:16] <saste> BBB, yeah probably it's a bad idea
[18:23] <wm4> subtitle ponies will probably happen, though
[18:23] <wm4> because lavfi is hardcoded to take AVFrame only
[18:23] <wm4> I wonder how rendered subtitles will be handled... another AVFrame media type?
[18:23] <wm4> welcome to hell
[18:24] <nevcairiel> rendered subtitles are just images in need of blending, so they are  video avframes?
[18:24] <wm4> well, many subtitle things return lists of images, not a single frame
[18:25] <wm4> at least the PGS decoder returns multiple subimages, and libass outputs a huge list of alpha bitmaps which have to be blended on top of each other
[18:26] <nevcairiel> whatever does the rendering can do that :P
[18:26] <wm4> you mean to blend them into a RGBA frame?
[18:26] <saste> wm4, an intermediary step would involve to separate audio and video fields in substructures
[18:27] <saste> even if i think a "media data" generalization concept (stuff containing A or V or S data) it's useful at some point, for example when filtering media
[20:04] <llogan> michaelni: can you check to see if http:bl plugin is active? I see no mention of it in admin/monitor, but i don't have plugins permissions, AFAIK
[20:05] <llogan> the spammers are good at striking when i'm afk
[20:22] <michaelni> llogan, ive given you trac_admin 
[20:22] <michaelni> so you should be able to access all things now
[20:24] <llogan> michaelni: thanks
[20:27] <llogan> "You can enable or disable these filters from the General ’ Plugins panel of the web administration interface." appears to be non-existant
[20:31] <llogan> oh, thats probably disabled in trac.ini
[20:31] <llogan> ignore my self-talking
[20:54] <cone-977> ffmpeg.git 03Clément BSsch 07master:7f42bfad5d76: lavfi/ebur128: add sample and true peak metering.
[20:54] <cone-977> ffmpeg.git 03Jean First 07master:6ef2315aafe3: lavfi/ebur128: print peak metering in dBFS
[20:54] <cone-977> ffmpeg.git 03Jean First 07master:2384cada120c: lavfi/ebur128: add true peak metering per frame
[22:27] <cone-977> ffmpeg.git 03Lukasz Marek 07master:c81a9d13f427: lavd/opengl_enc: fix parentheses in if condition
[23:25] <michaelni> llogan, if theres anything i should change or check in trac.ini, just say what
[23:31] <JEEB> if anyone feels like it, feel free to see why lavf/mpeg.c ignores the audio track that seems to be on ID 0xBD ? http://fushizen.eu/u/jeeb/derp/00002.MPS
[23:31] <JEEB> it should contain either ATRAC3 or ATRAC3+
[00:00] --- Mon Feb  3 2014