[Ffmpeg-devel-irc] ffmpeg-devel.log.20130831

[burek]

[00:53] <BBB> michaelni: so if you look at emulated_edge_mc, it takes a single stride argument and assumes that the image and emu buffer hold the same stride. what if the image is 1 mb by 1 mb (16x16 in mpeg, for example) and the MV is subpel, so we need a 17x17 temp buffer? the stride is only 16 so this wouldn't work
[00:54] <BBB> (if -flags +emu_edge is set)
[00:54] <BBB> michaelni: so ... does emulated_edge_mc need 2 stride arguments?
[00:54] <BBB> this affects codecs other than mpeg also ofc - in fact this affects all codecs equally
[01:05] <michaelni> BBB, hmm, stride =16 isnt supported atm, even with 2 stride emu , at least some of the  MC code itself also assume src_stride == dst_stride if iam not mistaken
[01:05] <michaelni> and the MC code couldnt read 17x17 with a stride=16 source
[01:06] <BBB> it's an issue for vp9 (and potentially hevc) where the base block unit is larger
[01:06] <BBB> I mean, nobody sane would make a 16x16 video
[01:06] <BBB> but a 64x64 video...
[01:06] <BBB> I mean, that's how realnetworks got big
[01:06] <BBB> I can just say it's unsupported but it doesn't feel quite right, I'm not 100% sure about it
[01:07] <michaelni> iam not against adding a 2nd stride parameter to emu
[01:07] <michaelni> the alternative would be maybe to enforce a larger stride and copy that 64x64 image to 64 stride
[01:08] <michaelni> but if a 2nd stride for emu is all thats needed then thats probably simpler
[01:08] <BBB> depends on the codec
[01:08] <BBB> mc needs 2 stride arguments
[01:08] <michaelni> otoh if all MC asm needs to be changed ...
[01:08] <BBB> and emulated_edge_mc needs 2 stride arguments
[01:08] <BBB> right
[01:08] <BBB> :)
[01:08] <BBB> I'm not looking forward to that
[01:09] <BBB> for a corner case that nobody ever uses...
[01:10] <BBB> vp8 mc asm has 2 stride arguments
[01:10] <BBB> so it's not unprecedented
[01:11] <BBB> I was thinking of whether to use 1 or 2 stride arguments for vp9 mc functions
[01:11] <BBB> I just changed it all to 1 to be consistent with h264 etc
[01:11] <BBB> but now thinking of going back to 2
[01:11] <BBB> not sure yet what's best
[01:11] <michaelni> well, i dont know either what is best. You work on the code you are in a better position to decide that
[01:12] <BBB> probably for future codecs 2 stride arguments is better, since a 64x64 block size and a 64x64 image just isn't completely unfeasible
[01:12] <BBB> whereas 16x16 just won't ever happen
[01:12] <BBB> I mean, nobody makes 16x16 videos
[01:15] <michaelni> the internet is big, theres probably little that wont ever happen, luckly we dont have to support every insane thing someone creates
[03:16] <cone-84> ffmpeg.git 03Michael Niedermayer 07master:227a0eb5a924: avformat/avidec: match first index and first packet size=0 handling
[04:54] <cone-84> ffmpeg.git 03Michael Niedermayer 07release/1.1:890c36d7ff1a: avcodec/ffv1dec: check global header version
[04:54] <cone-84> ffmpeg.git 03Michael Niedermayer 07release/1.1:df2fc635439d: ffv1dec: check that global parameters dont change in version 0/1
[04:54] <cone-84> ffmpeg.git 03Michael Niedermayer 07release/1.1:f78a3868fd3d: ffv1dec: Check bits_per_raw_sample and colorspace for equality in ver 0/1 headers
[04:54] <cone-84> ffmpeg.git 03Michael Niedermayer 07release/1.1:558d0b9483d0: avcodec/dsputil: fix signedness in sizeof() comparissions
[05:37] <cone-84> ffmpeg.git 03Michael Niedermayer 07release/1.1:86bf0a8871df: update for 1.1.6
[05:58] <cone-84> ffmpeg.git 03Michael Niedermayer 07fatal: ambiguous argument 'refs/tags/n1.1.6': unknown revision or path not in the working tree.
[05:58] <cone-84> Use '--' to separate paths from revisions
[05:58] <cone-84> refs/tags/n1.1.6:HEAD: avformat/avidec: match first index and first packet size=0 handling
[09:48] <cone-345> ffmpeg.git 03Michael Niedermayer 07release/0.10:49f434ce2e5e: avformat/avidec: match first index and first packet size=0 handling
[09:48] <cone-345> ffmpeg.git 03Michael Niedermayer 07release/0.11:d3ef17740d93: avformat/avidec: match first index and first packet size=0 handling
[09:48] <cone-345> ffmpeg.git 03Michael Niedermayer 07release/0.9:cf701b008fd3: avformat/avidec: match first index and first packet size=0 handling
[09:48] <cone-345> ffmpeg.git 03Michael Niedermayer 07release/1.0:6b5c4be9e0b5: avformat/avidec: match first index and first packet size=0 handling
[09:48] <cone-345> ffmpeg.git 03Michael Niedermayer 07release/1.1:f743c4ece8a8: avformat/avidec: match first index and first packet size=0 handling
[09:48] <cone-345> ffmpeg.git 03Michael Niedermayer 07release/1.2:32cff0d3f096: avformat/avidec: match first index and first packet size=0 handling
[09:48] <cone-345> ffmpeg.git 03Michael Niedermayer 07release/2.0:1821c849da5b: avformat/avidec: match first index and first packet size=0 handling
[10:54] <compn> yo from vdd
[10:55] <ubitux> ah you there?
[10:55] <compn> yes
[10:55] <compn> saw michael time out, whered he go?
[10:56] <ubitux> not at vdd i believe
[11:01] <compn> i know that
[11:03] <compn> kostya is here
[11:05] <compn> hi carl on irc log
[11:06] <compn> i met merbanan and wbs. not sure about daemon...
[11:07] <Daemon404> yes you did
[11:08] <compn> ok, then i did
[11:08] <compn> hehe
[11:10] <compn> oh michael is on irc im blind
[11:10] <JEEBsv> :)
[11:10] <compn> hollywood studios will never let dcp without drm....
[11:11] <compn> they'd rather lose the entire country revenue than let that happen
[11:44] <ubitux> compn: so where are you now?
[12:24] <compn> in the conf
[12:24] <compn> in a theatre with a bunch of devs
[12:25] <compn> insulting speakers, j-b just insulted mplayers edl
[12:25] <ubitux> i know im here
[12:25] <compn> oh lol , im near the top.
[12:25] <ubitux> and i laught, sry ;)
[12:26] <compn> where you at ubi? raise your hand
[12:26] <ubitux> we are half way, dtarting left
[12:26] <ubitux> no i wont raise my hand ;)
[12:26] <compn> haha ok
[12:26] <compn> vp9!
[12:27] <kierank> ubitux: compn is the guy with the huge beard
[12:29] <compn> i have my beard tucked, its hiding...
[12:32] <compn> beastd said it was only him and me from mplayer
[12:32] <compn> have to update him..
[13:46] <cone-497> ffmpeg.git 03Christian Schmidt 07master:0d6549d8b57a: pcm_bluray: rename pcm-mpeg.c to pcm-bluray.c
[13:46] <cone-497> ffmpeg.git 03Michael Niedermayer 07master:4273236e2f63: Merge commit '0d6549d8b57aaa5be8efb794154cdbe9a645910f'
[13:54] <cone-497> ffmpeg.git 03Christian Schmidt 07master:21fd2f84ce21: pcm_bluray: cosmetics
[13:54] <cone-497> ffmpeg.git 03Michael Niedermayer 07master:b043d80be4d8: Merge remote-tracking branch 'qatar/master'
[14:11] <cone-497> ffmpeg.git 03Carl Eugen Hoyos 07master:e7da19e9fa45: Fix make checkheaders for vdpau_internal.h
[14:11] <cone-497> ffmpeg.git 03Carl Eugen Hoyos 07master:b970f85564ba: Fix make checkheaders for wavpackenc.h
[14:11] <cone-497> ffmpeg.git 03Michael Niedermayer 07master:4fa00c567cd5: Merge remote-tracking branch 'cehoyos/master'
[14:15] <ubitux> i would like to say thank you free for providing some spaces for vdd
[14:15] <ubitux> but i would also like to say fuck you free for my fucking home connection
[14:15] <ubitux> isp my ass ffs
[14:17] <Daemon404> http://www.speedtest.net/my-result/2935767733
[15:33] <ubitux> what do you want to tell me compn ?
[15:33] <compn>  do you want to spend 1 year on a genric edl or apply mp4 patch and seperate mkv patch? 
[15:33] <ubitux> nope
[15:33] <ubitux> :)
[15:34] <compn> its multiple choice
[15:34] <compn> not yes no
[15:34] <ubitux> none"
[15:34] <ubitux> "none"
[15:34] <compn> none of those?]
[15:35] <compn> whats the alt?
[15:35] <ubitux> none of everything
[15:35] <ubitux> i suggest to wait for jesus
[15:35] <compn> is baptiste going to merge ffmbc?
[15:36] <compn> with ffmpeg i mean
[15:40] <mateo`> compn: that would be great :)
[21:13] <cone-497> ffmpeg.git 03Michael Niedermayer 07master:336982a0ce15: avcodec/utils: avcodec_string: make the colorspace printing code more extendible.
[21:13] <cone-497> ffmpeg.git 03Michael Niedermayer 07master:ea07dbdeaa66: avcodec/utils: avcodec_string: Print color_range
[21:13] <cone-497> ffmpeg.git 03Michael Niedermayer 07master:8da23be458f4: avcodec/utils: avcodec_string: print colorspace type
[00:00] --- Sun Sep  1 2013