[Ffmpeg-devel-irc] ffmpeg-devel.log.20111120

burek burek021 at gmail.com
Mon Nov 21 02:05:04 CET 2011

[00:07] <CIA-41> ffmpeg: 03Michael Niedermayer 07master * r4ebbcdb3eb 10ffmpeg/libavfilter/vsrc_mandelbrot.c: 
[00:07] <CIA-41> ffmpeg: vsrc_mandelbrot: avoid a | in interpol()
[00:07] <CIA-41> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[00:07] <CIA-41> ffmpeg: 03Michael Niedermayer 07master * ra5dfedd381 10ffmpeg/libavfilter/vsrc_mandelbrot.c: 
[00:07] <CIA-41> ffmpeg: vsrc_mandelbrot: detect edges in interpol()
[00:07] <CIA-41> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[00:07] <CIA-41> ffmpeg: 03Michael Niedermayer 07master * r90c02ae142 10ffmpeg/libavfilter/vsrc_mandelbrot.c: 
[00:07] <CIA-41> ffmpeg: vsrc_mandelbrot: increase maxiter to 7189
[00:07] <CIA-41> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[00:07] <CIA-41> ffmpeg: 03Michael Niedermayer 07master * r690860d0d5 10ffmpeg/libavfilter/vsrc_mandelbrot.c: 
[00:07] <CIA-41> ffmpeg: vsrc_mandelbrot: Use threshold to detect cycles.
[00:07] <CIA-41> ffmpeg: This way cycles are detected much earlier.
[00:07] <CIA-41> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[02:41] <michaelni> Compn, before your commit the censorship logo was over the heading of the news entry now its not anymore
[02:42] <Compn> michaelni : yeah, should be fixed
[02:42] <pasteeater> Compn: you need to add a space between " and /.
[02:43] <pasteeater> and the img also needs an alt tag
[02:43] <pasteeater> then you can probably remove the <p> and </p<
[02:43] <pasteeater> >
[02:43] <Compn> i'll let you :)
[02:44] Action: Compn brain fried
[02:45] <pasteeater> i still haven't given michaelni my public ssh key
[02:45] <pasteeater> been out of town too much lately
[02:45] <michaelni> Compn, why did you remove the style="position: absolute; z-index:5000" ?
[02:45] <michaelni> was there a problem with it ?
[02:46] <michaelni> it nicely displayed over the text in firefox as it was ...
[02:51] <michaelni> can someone test my original version in some other browser than firefox ?
[02:53] <pasteeater> is it supposed to be over the news entry title?
[02:54] <Compn> i didnt like it over the news entry title
[02:54] <Compn> i just made it a normal image link
[02:54] <Compn> maybe its just me 
[02:54] <Compn> i thought you wanted to post it as an entry, not censoring the news page :P
[02:54] <gnafu> I thought it was supposed to be accross the website's actual logo :-P
[02:55] <Compn> gnafu : it was supposed to be on the 16th, but i didnt see it til the 18th :P
[02:55] <Compn> so we are repositioning it 
[02:55] <michaelni> i liked it over the news entry :(
[02:55] <Compn> well feel free to revert then
[02:55] <michaelni> over the logo looks odd half a week after the thing
[02:55] Action: Compn using opera
[02:56] <michaelni> did it work in opera?
[02:56] <Compn> yes
[02:56] <michaelni> over the heading or text ?
[02:56] <Compn> the sopa bill is still awaiting a vote, isnt it ?
[02:56] <michaelni> iam no politican, no idea
[02:56] <Compn> i mean, just because that website says somethings over.... i didnt get the impression that the fight was over 
[02:56] <Compn> in fact, the meeting that occured on the 16th was really bad
[02:57] <michaelni> the fight is surely never over, these things will be retried and retried ...
[02:57] <Compn> but i mean... sopa isnt dead yet, it just had a meeting on the 16th
[02:59] <michaelni> Compn, ok if i try to put it back over the heading ?
[02:59] <Compn> sure, put it where you want it :)
[02:59] <Compn> i just thought you did a bad copy paste :P
[02:59] <Compn> as evidenced by your removing the <a> tag on template_head2
[02:59] <Compn> er , not replacing the <a> tag that is
[03:00] <michaelni> iam sloppy & scatterbrained :)
[03:00] <Compn> i have no objections to where it goes , i fixed it without asking :)
[03:01] <michaelni> well if theres some way i could improve it iam surely interrested in doing that
[03:01] <michaelni> but IMHO putting it over the heading is a eye catcher
[03:02] <michaelni> so more likely people would look & click & support
[03:03] <CIA-41> ffmpeg: 03Thierry Foucu 07master * rc693aa6f71 10ffmpeg/libavutil/imgutils.c: 
[03:03] <CIA-41> ffmpeg: imgutils: Fix illegal read.
[03:03] <CIA-41> ffmpeg: Found with address sanitizer.
[03:03] <CIA-41> ffmpeg: Signed-off-by: Alex Converse <alex.converse at gmail.com>
[03:03] <CIA-41> ffmpeg: 03Justin Ruggles 07master * r86d3730360 10ffmpeg/libavcodec/options.c: avcodec: remove pointless AVOption, internal_buffer_count
[03:03] <CIA-41> ffmpeg: 03Justin Ruggles 07master * rfa9aeb82e2 10ffmpeg/libavcodec/options.c: 
[03:03] <CIA-41> ffmpeg: avcodec: fix some const warnings
[03:03] <CIA-41> ffmpeg: libavcodec/options.c:583: warning: assignment discards qualifiers from pointer
[03:03] <CIA-41> ffmpeg: target type
[03:03] <CIA-41> ffmpeg: libavcodec/options.c:589: warning: initialization discards qualifiers from
[03:03] <CIA-41> ffmpeg: pointer target type
[03:03] <CIA-41> ffmpeg: 03Justin Ruggles 07master * rf3a29b750a 10ffmpeg/ (12 files in 2 dirs): 
[03:03] <CIA-41> ffmpeg: avcodec: move some AVCodecContext fields to an internal struct.
[03:03] <CIA-41> ffmpeg: A new field, AVCodecContext.internal is used to hold a new struct
[03:03] <CIA-41> ffmpeg: AVCodecInternal, which has private fields that are not codec-specific and are
[03:03] <CIA-41> ffmpeg: used by general libavcodec functions.
[03:03] <CIA-41> ffmpeg: Moved internal_buffer, internal_buffer_count, and is_copy.
[03:03] <CIA-41> ffmpeg: 03John Stebbins 07master * r09d243ddd0 10ffmpeg/libswscale/swscale_unscaled.c: 
[03:03] <CIA-41> ffmpeg: swscale: Fix stack alignment for SSE
[03:03] <CIA-41> ffmpeg: Although gcc guarantees 16 byte stack alignment, threads under WinXP
[03:04] <ubitux> Compn: you might want to add a alt="censorship..." to the img so it is a bit more valid
[03:04] <ubitux> there is also a <pre> issue that might be fixed by taking out the nested <pre> in the <p>
[03:05] <ubitux> anyway, 'night ppl
[03:08] <Compn> michaelni : ubitux and pasteeater want you to add an alt tag too :P
[03:09] <michaelni> ill add the alt tag
[03:10] <michaelni> Compn, what exactly should i put in it ?
[03:11] <Compn> stop censorship logo
[03:11] <Compn> alt tag is what pops up when you put your mouse cursor over an image
[03:12] <Compn> its used for screen readers so blind people get an audio description of an image
[03:13] <michaelni> added
[03:14] <Compn> ffmpeg logo alt tag should be something like 'ffmpeg logo containing zigzag motion search. created by herve ...'
[03:14] <michaelni> "..." ?
[03:15] <Compn> ... being herve's last name i've forgotten
[03:15] <michaelni> "ffmpeg logo containing zigzag motion search. created by herve being herve's last name i've forgotten" ?
[03:15] <pasteeater> Flores
[03:16] <Compn> heh
[03:17] <michaelni> what is "zigzag motion search" ?
[03:17] <Compn> FFmpeg logo displaying the zigzag motion search algorithm. logo is copyright Hervé Flores.
[03:17] <Compn> isnt that what the zigzag is ?
[03:17] <Compn> i dont actually remember what it is
[03:17] <michaelni> idct scantable
[03:18] <Compn> ah
[03:18] <michaelni> btw, isnt "ffmpeg logo" enough ?
[03:18] <Compn> well would a blind person know what it looks like?
[03:18] <Compn> of course, idct scantable is advanced idea too :D
[03:19] Action: Compn runs afk
[03:19] <Compn> bbl
[03:19] <ohsix> how many blind people are interested in this information heh
[03:20] <michaelni> i dont know, but id imagine they would be more interrested in the content of the page, and the links to other pages
[03:21] <michaelni> ill leave the alt logo text bikeshed to others, iam fine with anything compn likes
[03:28] <pasteeater> "FFmpeg" is fine with me
[03:29] <Compn> you guys are no fun, bikesheds are important.
[03:31] <pasteeater> give me a good mandelbrot example instead
[03:31] <michaelni> pasteeater, the defaults ?
[03:32] <pasteeater> "ffplay -f lavfi mandelbrot"?
[03:32] <pasteeater> that's all i've tried
[03:39] <CIA-41> ffmpeg: 03Anton Khirnov 07release/0.8 * r47953c33ea 10ffmpeg/ (7 files in 3 dirs): (log message trimmed)
[03:39] <CIA-41> ffmpeg: lavc: introduce avcodec_open2() as a replacement for avcodec_open().
[03:39] <CIA-41> ffmpeg: Adds support for decoder-private options and makes setting other options
[03:39] <CIA-41> ffmpeg: simpler.
[03:39] <CIA-41> ffmpeg: (cherry picked from commit 0b950fe240936fa48fd41204bcfd04f35bbf39c3)
[03:39] <CIA-41> ffmpeg: Conflicts:
[03:39] <CIA-41> ffmpeg:  libavcodec/avcodec.h
[03:40] <CIA-41> ffmpeg: 03Baptiste Coudurier 07release/0.8 * r23f0d0f16b 10ffmpeg/libavcodec/utils.c: 
[03:40] <CIA-41> ffmpeg: lavc: fix parentheses placement in avcodec_open2().
[03:40] <CIA-41> ffmpeg: Signed-off-by: Anton Khirnov <anton at khirnov.net>
[03:40] <CIA-41> ffmpeg: (cherry picked from commit 1d36fb13b088f55ece155153fb6ca8ea278fc837)
[03:40] <CIA-41> ffmpeg: Signed-off-by: Anton Khirnov <anton at khirnov.net>
[03:40] <CIA-41> ffmpeg: 03Anton Khirnov 07release/0.8 * rafe2726089 10ffmpeg/ (3 files in 2 dirs): 
[03:40] <CIA-41> ffmpeg: lavf: add avformat_find_stream_info()
[03:40] <CIA-41> ffmpeg: It supports passing options to codecs.
[03:40] <CIA-41> ffmpeg: (cherry picked from commit a67c061e0f3b55ffcc96f336fc0998e44b86c8e4)
[03:40] <CIA-41> ffmpeg: Conflicts:
[03:40] <CIA-41> ffmpeg:  libavformat/utils.c
[03:42] <michaelni> pasteeater, ./ffplay -video_size 600x600 -f rawvideo /dev/zero -vf nullsink,mandelbrot  -noframedrop
[03:43] <pasteeater> thanks
[03:43] <michaelni> np
[03:44] <michaelni> it should be made simpler & documented of course
[09:55] <ohsix> heh "Ignite" the racing game comes with swscale and avcodec
[12:19] <ubitux> <@Compn> alt tag is what pops up when you put your mouse cursor over an image // no this is the title=... afaik, but blind ppl "see" the alt=... 
[12:19] <ubitux> this is also what is printed when the pic is not yet downloaded too
[12:35] <CIA-41> ffmpeg: 03Reimar Döffinger 07master * r03f1caec61 10ffmpeg/libavutil/x86/x86inc.asm: 
[12:35] <CIA-41> ffmpeg: HACK: fix compilation with NASM.
[12:35] <CIA-41> ffmpeg: This issue is actually already fixed in NASM, however it will be a
[12:35] <CIA-41> ffmpeg: bit before it arrives everywhere.
[12:35] <CIA-41> ffmpeg: This should work as a temporary work-around.
[12:35] <CIA-41> ffmpeg: Signed-off-by: Reimar Döffinger <Reimar.Doeffinger at gmx.de>
[13:23] <Compn> ohsix : is it just for fmv clips ?
[13:45] <kkk> hello
[13:46] <kkk> I have problems with libaacplus
[13:46] <kkk> [libaacplus @ 0x2d66900] libaacplus doesn't support this output format!
[13:46] <kkk> libaacplus: bad aac setting: br:64000, AACch:65536, AACsr:22050
[13:46] <kkk> ffmpeg -i file.dat -re -threads 2 -acodec libaacplus -ac 2 -ab 64k -ar 22050 -vcodec libx264 -b 600k ....
[13:46] <kkk> and with libvo_aacenc
[13:47] <kkk> Press [q] to stop, [?] for help
[13:47] <kkk> ffmpeg: bitbuffer.c:269: WriteBits: Assertion `hBitBuf->cntBits <= (hBitBuf->pBi
[13:47] <kkk> tBufEnd - hBitBuf->pBitBufBase + 1) * 8' failed.
[13:47] <kkk> can someone help?
[14:12] <Compn> kkk : it looks like the aacch = channels? in which case, 65536 is incorrect
[14:12] <Compn> not sure why the channels is incorrect like that
[14:12] <Compn> probably a typo in the code
[14:14] <kkk> yeah i just want to report
[14:14] <kkk> because  -ac 2 is not recognized Compn
[14:14] <kkk> thats kind of bad :/
[14:21] <michaelni> kkk, -ar 44100 works here
[14:23] <kkk> ?
[14:23] <kkk> michaelni what ar?
[14:24] <kkk> you mean same setting with 44100?
[14:25] <michaelni> yes just replace 22050 by 44100
[14:25] <michaelni> it works with that here with git master
[14:25] <michaelni> if it doesnt, maybe tipok has an idea ...
[14:27] <kkk> i'll try
[14:29] <kkk> @michaelni nope
[14:30] <kkk> http://pastie.org/2892663
[14:30] <kkk> ffmpeg -i file -re -threads2 -acodec libaacplus -ac 2 -ab 64k -ar 44100 -vcodec libx264 -b 600k  -f flv 
[14:30] <kkk> dont work
[14:31] <michaelni> Compn, about the channels the wraper checks that they are 1 or 2 at line 41
[14:31] <kkk>    Metadata:
[14:31] <kkk>       variant_bitrate : 1424293
[14:31] <kkk>     Stream #0:1: Audio: aac, 44100 Hz, 2 channels, s16, 64 kb/s
[14:31] <kkk>     Metadata:
[14:31] <kkk>       variant_bitrate : 1424293
[14:32] <kkk> Is this maybe because of the variable bitrate?
[14:34] <michaelni> kkk, your command line is working for me with a random input file
[14:34] <michaelni> are you using git master of ffmpeg?
[14:34] <kkk> ffmpeg.zeranoe.com
[14:34] <michaelni> and which libaacplus version ?
[14:35] <kkk> ehm the last
[14:35] <kkk> i think
[14:35] <michaelni> ok
[14:35] <michaelni> maybe its a windows specific bug, i dunno
[14:35] <kkk> windows?
[14:35] <kkk> No, i'm using it under linux
[14:37] <michaelni> kkk you are using a windows build from zeranoe under linux ? ;)
[14:38] <ubitux> zeranoe seems to provide source snapshots too
[14:38] <ubitux> but maybe they are patched?
[14:38] <kkk> windows builds unter linux ?
[14:39] <kkk> No? How i should ? :D
[14:39] <kkk> I'll compile from the source
[14:39] <ubitux> ffmpeg.org also provides snapshots, why are you using zeranoe ones?
[14:42] <michaelni> kkk, also make sure you have matching headers & binaries from libaacplus
[14:42] <kkk> michaelni
[14:42] <kkk> i'm using the same sources
[14:42] <kkk> for month now
[14:42] <kkk> never got any problems
[14:43] <kkk> (libaacplus)
[14:47] <kkk> well, i'll try with the latest git ffmpeg.org code
[14:47] <kkk> but i dont think this is working
[14:52] <kkk> lol @ ubitux && michaelni : now while compiling ffmpeg.org code i got following error
[14:52] <kkk> yasm: FATAL: (cglobal:8) `%else': no matching `%if'
[14:52] <kkk> make: *** [libavcodec/x86/ac3dsp.o] Error 1
[14:53] <michaelni> kkk "git checkout HEAD^"
[14:53] <michaelni> reimar broke it
[14:53] <kkk> git checkout HEAD <---- just this command in the ffmpeg folder?
[14:53] <michaelni> HEAD^
[14:54] <michaelni> -> ^ <- this too
[14:54] <kkk> HEAD is now at 90c02ae... vsrc_mandelbrot: increase maxiter to 7189
[14:54] <kkk> i see, thanks
[14:54] <ubitux> i checked the snapshot from zeranoe, it seems to be the same, so no worry
[14:54] <michaelni> and "git checkout master" will get you back to the master once its fixed
[14:55] <kkk> ok thx
[15:00] <kkk> the same!
[15:00] <kkk> libaacplus: bad aac setting: br:64000, AACch:65536, AACsr:22050
[15:00] <kkk> [libaacplus @ 0x295faa0] libaacplus doesn't support this output format!
[15:00] <kkk> I'm having problems to set the channels
[15:01] <kkk> and 
[15:01] <kkk> libvo_aacenc:
[15:01] <kkk> [SWR @ 0x284b5a0] Value 1.000000 for parameter 'och' out of range
[15:01] <kkk> ffmpeg: bitbuffer.c:269: WriteBits: Assertion `hBitBuf->cntBits <= (hBitBuf->pBi
[15:01] <kkk> tBufEnd - hBitBuf->pBitBufBase + 1) * 8' failed.
[15:01] <kkk> Aborted
[15:04] <CIA-36> ffmpeg: 03Reimar Döffinger 07master * re8ab4bac0c 10ffmpeg/libavutil/x86/x86inc.asm: 
[15:04] <CIA-36> ffmpeg: Fix compilation with YASM.
[15:04] <CIA-36> ffmpeg: YASM does not support %ifempty even though it is in the official
[15:04] <CIA-36> ffmpeg: NASM documentation and supported since NASM 2.02.
[15:04] <CIA-36> ffmpeg: Signed-off-by: Reimar Döffinger <Reimar.Doeffinger at gmx.de>
[15:10] <michaelni> kkk try to recompile and disable one of the 2 libs, might be a conflict
[15:11] <michaelni> also try to recompile & install libaacplus if this doesnt help and make sure you have no old headers or versions of libaacplus laying around
[15:11] <michaelni> and a make distclean is a good idea too
[15:12] <michaelni> ubitux, av_opt_set_int(ost->swr, "och", enc->channels, 0);
[15:12] <michaelni> and min allowed is 1 
[15:15] <ubitux> oh sorry about that, should be change the min limit to 0?
[15:15] <ubitux> or add a condition maybe?
[15:16] <ubitux> sth like chl = av_get_channel_layout...(); if (chl) av_opt_set_int(...)
[15:16] <CIA-36> ffmpeg: 03Michael Niedermayer 07master * ra9011623ea 10ffmpeg/libavutil/opt.c: (log message trimmed)
[15:16] <CIA-36> ffmpeg: avopts: Fix random values being printed in error messages:
[15:16] <CIA-36> ffmpeg: Bug introduced in:
[15:16] <CIA-36> ffmpeg: commit 7bb1807c2dd9623842f1c6d454e7f2892b5bc8d7
[15:16] <CIA-36> ffmpeg: Author: Anton Khirnov <anton at khirnov.net>
[15:16] <CIA-36> ffmpeg: Date: Tue Aug 23 12:58:49 2011 +0200
[15:16] <CIA-36> ffmpeg:  AVOptions: refactor set_number/write_number
[15:18] <michaelni> ubitux, iam fine with min=0 iam also fine if the =0 is skiped, whatever you prefer
[15:20] <ubitux> how did you see the issue btw?
[15:26] <michaelni> ubitux, its printed each time you run ffmpeg :) for example in kkks report above (just with the wrong value of 1.0000 that ive just fixed)
[15:50] <ubitux> michaelni: http://blackhole.ubitux.fr/0001-swr-change-minimum-I-O-channel-count-to-zero.patch
[15:50] <ubitux> ok with this, and the commit messagE?
[16:01] <CIA-36> ffmpeg: 03Martin Storsjö 07master * r66e9c0b6ab 10ffmpeg/libavformat/Makefile: 
[16:01] <CIA-36> ffmpeg: httpproxy: Specify the object files needed in the Makefile
[16:01] <CIA-36> ffmpeg: Signed-off-by: Martin Storsjö <martin at martin.st>
[16:01] <CIA-36> ffmpeg: 03John Stebbins 07master * rc40e1757a1 10ffmpeg/configure: 
[16:01] <CIA-36> ffmpeg: httpproxy: Fix dependencies
[16:01] <CIA-36> ffmpeg: Fix building with --disable-network.
[16:01] <CIA-36> ffmpeg: Signed-off-by: Martin Storsjö <martin at martin.st>
[16:01] <CIA-36> ffmpeg: 03Andreas Öman 07master * r4d9d9a443f 10ffmpeg/libavcodec/ (Makefile allcodecs.c wmalosslessdec.c): wmall: Working bitstream parser
[16:01] <CIA-36> ffmpeg: 03Mashiat Sarker Shakkhar 07master * rfa3e0190f9 10ffmpeg/libavcodec/wmalosslessdec.c: Add missing ff_ before the AVCodec struct for wmalossless
[16:01] <CIA-36> ffmpeg: 03Mashiat Sarker Shakkhar 07master * r374582acfe 10ffmpeg/libavcodec/wmalosslessdec.c: Cosmetics
[16:01] <CIA-36> ffmpeg: 03Mashiat Sarker Shakkhar 07master * rd58db7207d 10ffmpeg/libavcodec/wmalosslessdec.c: Add buffers to context necessary for reverting cdmls and mclms filter
[16:01] <CIA-36> ffmpeg: 03Mashiat Sarker Shakkhar 07master * ra80a10bd73 10ffmpeg/libavcodec/wmalosslessdec.c: Use avpriv_copy_bits() instead of ff_copy_bits()
[16:01] <CIA-36> ffmpeg: 03Mashiat Sarker Shakkhar 07master * r87b9c23b9a 10ffmpeg/libavcodec/wmalosslessdec.c: Implement clear_codec_buffers()
[16:01] <CIA-36> ffmpeg: 03Mashiat Sarker Shakkhar 07master * r79b16e780d 10ffmpeg/libavcodec/wmalosslessdec.c: Add .recent syntax element to cdlms struct
[16:01] <CIA-36> ffmpeg: 03Mashiat Sarker Shakkhar 07master * r269a0dee19 10ffmpeg/libavcodec/wmalosslessdec.c: Implement lms_update()
[16:01] <CIA-36> ffmpeg: 03Mashiat Sarker Shakkhar 07master * r93d2cabd61 10ffmpeg/libavcodec/wmalosslessdec.c: Add missing syntax elements to WmallDecodeCtx
[16:01] <CIA-36> ffmpeg: 03Mashiat Sarker Shakkhar 07master * r03684c8ff3 10ffmpeg/libavcodec/wmalosslessdec.c: Implement reset_codec()
[16:01] <CIA-36> ffmpeg: 03Mashiat Sarker Shakkhar 07master * rf3d349f697 10ffmpeg/libavcodec/wmalosslessdec.c: Replace placeholders with actual calls to clear_codec_buffers() and reset_codec()
[16:01] <CIA-36> ffmpeg: 03Mashiat Sarker Shakkhar 07master * rd5aef477e6 10ffmpeg/libavcodec/wmalosslessdec.c: Output decoded residues
[16:01] <CIA-36> ffmpeg: 03Mashiat Sarker Shakkhar 07master * ra7c31f1595 10ffmpeg/libavcodec/wmalosslessdec.c: Log tile size
[16:01] <CIA-36> ffmpeg: 03Mashiat Sarker Shakkhar 07master * r1e0db1c52a 10ffmpeg/libavcodec/wmalosslessdec.c: Initialize num_logged_tiles and remove unnecessary codes
[16:02] <CIA-36> ffmpeg:  dump_int_buffer() to dump samples from a buffer
[16:02] <CIA-36> ffmpeg:  Implement revert_cdlms()
[16:02] <CIA-36> ffmpeg:  Doxy for reset_codec()
[16:02] <ubitux> mmh maybe also a micro dump
[16:02] <michaelni> ubitux, LGTM
[16:02] <CIA-36> ffmpeg: 03Mashiat Sarker Shakkhar 07master * r3f3d2b5f65 10ffmpeg/libavcodec/wmalosslessdec.c: Implement use_high_update_speed() and use_normal_update_speed()
[16:02] <CIA-36> ffmpeg: 03Mashiat Sarker Shakkhar 07master * rd47782df26 10ffmpeg/libavcodec/wmalosslessdec.c: Doxy for reset_codec()
[16:02] <CIA-36> ffmpeg: 03Mashiat Sarker Shakkhar 07master * r3e4901b24e 10ffmpeg/libavcodec/wmalosslessdec.c: Store transient state and position of transient area
[16:02] <CIA-36> ffmpeg: 03Mashiat Sarker Shakkhar 07master * r7a9807dea4 10ffmpeg/libavcodec/wmalosslessdec.c: Clean-up
[16:06] <CIA-36> ffmpeg: 03Clément BSsch 07master * rbdd22e3523 10ffmpeg/libswresample/swresample.c: 
[16:06] <CIA-36> ffmpeg: swr: change minimum I/O channel count to zero.
[16:06] <CIA-36> ffmpeg: This will silence the warning of av_set_opt_int() in swr for
[16:06] <CIA-36> ffmpeg: input/output channels when the layout is not yet defined (or supported).
[17:23] Action: Daemon404 pokes nevcairiel again
[17:24] <nevcairiel> hm?
[17:25] <Daemon404> nevcairiel, im not sure if this is an ffdshow bug or a lav splitter bug
[17:25] <Daemon404> but ffdshow + lav splitter refuse to connect for COOK audio
[17:25] <Daemon404> (when splitting rmvb)
[17:25] <Daemon404> lav audio connects fine
[17:27] <CIA-36> ffmpeg: 03Michael Niedermayer 07master * r50d6f81956 10ffmpeg/libavcodec/mpegvideo_common.h: 
[17:27] <CIA-36> ffmpeg: mpegvideo: dont use ff_mspel_motion() for vc1
[17:27] <CIA-36> ffmpeg: Fixes Ticket655
[17:27] <CIA-36> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[17:27] <nevcairiel> lav splitter doesnt output the full RealAudio header, and i suppose ffdshow expects that
[17:27] <Daemon404> that might be why
[17:28] <nevcairiel> ffmpeg doesnt offer the full header, and i didnt really bother to try to re-create it
[17:29] <Daemon404> i see
[17:29] <Daemon404> ffdshow connects when cook audio is in mkv, and haali is splitting it
[17:29] <Daemon404> so you can check what it does?
[17:30] <nevcairiel> mkv could potentially work with lav splitter as well
[17:30] <nevcairiel> its only the rm demuxer in avformat that doesn't export the full RealAudio header in extradata
[17:30] <Daemon404> i see
[17:31] <Daemon404> cause i tried lav splitter, only cause the internal rm splitter in mpc-hc is busted as hell
[17:32] <nevcairiel> I wouldnt know, i really dont watch any rm
[17:32] <Daemon404> neither do i
[17:32] <Daemon404> but my SO does
[17:32] <Daemon404> thus -> my problem
[17:36] <kkk> michaelni you we're right, libaacplus is not working at all
[17:36] <kkk> maybe because of libvo_aacenc. I'll try it later
[17:43] <Daemon404> interesting nick
[18:24] <Compn> Daemon404 : haha, never noticed that
[18:24] <Compn> the KKK being the name of a racist organization started in the USA
[18:26] <Daemon404> indeed.
[18:34] <kkk> Well sorry
[18:34] <kkk> i never noticed that too but i'm not a racist ^^
[18:35] <kkk> I mean I'm german, but the nick kkk comes from kaiza my real nick. often i shorten it with kkk ^^
[18:39] <iive> well, we should be happy you didn't use kaida instead.
[18:39] <kkk> haha :D
[20:35] <pasteeater> anyone getting a bunch of chinese "out of office" replies from lindachuan at sina.com?
[21:02] <CIA-36> ffmpeg: 03Michael Niedermayer 07master * r15a686b714 10ffmpeg/libavcodec/mpegvideo.c: 
[21:02] <CIA-36> ffmpeg: h263: make default color black, like flv
[21:02] <CIA-36> ffmpeg: Fixes Ticket126
[21:02] <CIA-36> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[21:03] <michaelni> pasteeater, yes, my spam filter loves them :)
[21:46] <NoorulIslaam> im encoding video on android using ffmpeg+x264
[21:46] <NoorulIslaam> but even on the fastest settings for x264, i cant seem to get more than 6 or 7 FPS
[21:46] <NoorulIslaam> im recording in realtime from the camera
[21:47] <NoorulIslaam> the camera frames are processed by an opengl renderer
[21:47] <NoorulIslaam> and i do glReadPixels in RGB565 format to get the video frames
[21:47] <NoorulIslaam> without video encoding, the rendering is as fast as the camera can supply frames
[21:47] <NoorulIslaam> any ideas on how to make the process go faster?
[21:48] <NoorulIslaam> i considered storing the raw frames and encoding them after the user presses stop recording
[21:48] <NoorulIslaam> but at 640x480, RGB565, 24 FPS, for 60 secs... that right there is more than 800 MB of space
[21:48] <iive> NoorulIslaam: can the camera capture in yuv mode?
[21:49] <NoorulIslaam> of course i am using YUV mode
[21:49] <NoorulIslaam> but once renderered to the screen it becomes RGB565
[21:49] <NoorulIslaam> then i read back the pixels using glReadPixels
[21:49] <iive> i don't understand what are you doing?
[21:49] <NoorulIslaam> before passing the frame onto the x264 encoder, i use sws_scale to convert it to YUV420P
[21:50] <NoorulIslaam> sorry, let me explain step by step:
[21:50] <NoorulIslaam> 1 - camera supplies frame in NV21 format (a yuv format)
[21:50] <NoorulIslaam> 2 - opengl renderer runs some effects on the frame, and renders to framebuffer in RGB565 format
[21:51] <NoorulIslaam> 3 - glReadPixels is called to readback the rendered frame in RGB565 format
[21:51] <NoorulIslaam> 4 - sws_scale called to convert RGB565 frame to YUV420P
[21:51] <NoorulIslaam> 5 - YUV420P frame encoded via ffmpeg+x264 and written to mp4 file
[21:52] <NoorulIslaam> thats what im doing
[21:52] <iive> why don't you store the video directly from the camera?
[21:52] <NoorulIslaam> because i need to apply effects on it
[21:52] <iive> what kind of effects?
[21:52] <pasteeater> what are your x264 settings?
[21:52] <NoorulIslaam> anything from warping the image, to sepia tinting, to outline detection, etc etc
[21:53] <iive> are you doing something like simulated reality?
[21:53] <NoorulIslaam> i dont know if you would call it simulated reality
[21:53] <iive> is the opengl done by gpu unit?
[21:53] <NoorulIslaam> i call it realtime effects
[21:53] <NoorulIslaam> yes opengl runs on the GPU
[21:53] <NoorulIslaam> if i turn off video recording, im doing only steps 1 & 2
[21:53] <NoorulIslaam> then i get up to 30 FPS or however fast the camera can deliver frames
[21:53] <iive> NoorulIslaam: virtual reality overlayed over real footage in real time.
[21:54] <NoorulIslaam> iive no thats not what im doing
[21:54] <iive> aka, adding 3d objects that are not present in the reality.
[21:54] <NoorulIslaam> no
[21:54] <NoorulIslaam> im taking the frame from the camera, running a opengl vertex/fragment shader program on it, and the output of that is shown on the screen
[21:55] <NoorulIslaam> to record it to video, i have to read back from the screen and encode
[21:56] <iive> can you benchmark the functions that take most time?
[21:56] <NoorulIslaam> pasteeater: im doing a lot of experimenting, trying to get it to ultrafast. i saw a marginal improvement from 4 FPS to 6~7 FPS, but thats the most i seem to be getting
[21:56] <iive> something like `perf top`
[21:56] <NoorulIslaam> pasteeater: which settings would you like to know about? there are a lot
[21:56] <iive> how much do you get if you send the raw frames to /dev/null.
[21:56] <iive> aka, what framerate.
[21:57] <NoorulIslaam> okay, let me see
[21:57] <NoorulIslaam> what i will do is go up to the glReadPixels call
[21:57] <NoorulIslaam> but not send the frame to the encoder at all
[21:57] <NoorulIslaam> give me 1 min
[22:00] <NoorulIslaam> okay, i was getting a much smoother FPS that way
[22:00] <NoorulIslaam> doing steps 1 to 3, but not steps 4 & 5
[22:00] <NoorulIslaam> dont know exactly what FPS it was since no frames were written to file
[22:01] <NoorulIslaam> let me add step 4 and see how much it slows down
[22:04] <michaelni> If you would do the filtering to the yuv then you would not have to read back the rgb and rgb->yuv
[22:05] <NoorulIslaam> with step 4 its a little more slower, it looks to me around 15 FPS but not sure because i dont have any FPS measuring code in there right now
[22:06] <NoorulIslaam> michaelni thats true but part of the shaders rely on RGB data
[22:06] <NoorulIslaam> so i cant change that part
[22:06] <NoorulIslaam> on the iphone 4 i can do this at about 20 to 25 FPS
[22:07] <NoorulIslaam> on the iphone 4s at the full framerate of the camera
[22:07] <NoorulIslaam> right now im testing on the galaxy note
[22:07] <NoorulIslaam> the resolution is 320x512
[22:07] <NoorulIslaam> on the iphone its 320x480
[22:08] <NoorulIslaam> of course on the iphone i use hardware accelerated encoding
[22:08] <NoorulIslaam> dont have access to that on android
[22:09] <NoorulIslaam> let me try now just encoding the frames but not writing them to the mp4 file...
[22:09] <michaelni> you can probably do some NEON optims here and there that might help a bit (and would be very welcome if you submited them to ffmpeg-dev)
[22:11] <NoorulIslaam> hmm yeah, there were some NEON optims added a few days ago i think
[22:11] <NoorulIslaam> i should update my ffmpeg source tree and recompile and see if they make a difference
[22:22] <NoorulIslaam> hmm okay so now im doing everything except av_interleaved_write_frame. framerate has dropped significantly
[22:22] <NoorulIslaam> but is noticeably faster than if im also writing frames to disk
[23:04] <CIA-36> ffmpeg: 03Michael Niedermayer 07master * r3cc04b6a9a 10ffmpeg/doc/git-howto.txt: 
[23:04] <CIA-36> ffmpeg: git-howto: update git url.
[23:04] <CIA-36> ffmpeg: git.videolan.org is confusing to some users as videolan is a seperate project from ffmpeg.
[23:04] <CIA-36> ffmpeg: the new name points to git.videolan.org but avoids the confusion.
[23:04] <CIA-36> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[23:30] <NoorulIslaam> something's wrong here... i just got the 0.8.6 release
[23:30] <NoorulIslaam> and after i ./configure
[23:30] <NoorulIslaam> the Makefile is a symbolic link pointing to ./Makefile
[23:30] <NoorulIslaam> im on mac os x
[23:50] <ubitux> http://ubitux.fr/pub/shots/lifesrc-Changelog.png hehe
[23:51] <ubitux> (./ffplay -f lavfi "lifesrc=f=Changelog:s=1000x1000, boxblur=2:2" after the patch i just sent)
[23:55] <ubitux> another nice option would be to link edges
[23:59] <ubitux> oh, it looks like there is a huge leak in boxblur
[00:00] --- Mon Nov 21 2011

More information about the Ffmpeg-devel-irc mailing list