[Ffmpeg-devel-irc] ffmpeg-devel.log.20130226

[burek]

[02:56] <cone-850> ffmpeg.git 03Michael Niedermayer 07master:bfcc21a4725d: libavcodec/motion-test: set the bitexact flag
[02:56] <cone-850> ffmpeg.git 03Michael Niedermayer 07master:5bcb379ffe54: motion-test: fix warning: dsp_mask is deprecated
[02:56] <cone-850> ffmpeg.git 03Michael Niedermayer 07master:f6fff8e54697: ac3enc_template: silence may be used uninitialized in this function warnings
[03:39] <cone-850> ffmpeg.git 03Matt Wolenetz 07release/1.1:bc9d341be84e: x86: h264: Don't use redzone in AVX h264_deblock on Win64
[03:39] <cone-850> ffmpeg.git 03Reinhard Tartler 07release/1.1:88ae77cea4c1: update Changelog
[03:39] <cone-850> ffmpeg.git 03Anton Khirnov 07release/1.1:b786ddc0f2ea: loco: check that there is data left after decoding a plane.
[03:39] <cone-850> ffmpeg.git 03Anton Khirnov 07release/1.1:488ffb813514: mov: use the format context for logging.
[03:39] <cone-850> ffmpeg.git 03Anton Khirnov 07release/1.1:8bce2c60b8eb: lagarith: avoid infinite loop in lag_rac_refill()
[03:39] <cone-850> ffmpeg.git 03Anton Khirnov 07release/1.1:612b28194b4c: flicvideo: avoid an infinite loop in byte run compression
[03:39] <cone-850> ffmpeg.git 03Anton Khirnov 07release/1.1:48fd461977f7: av_memcpy_backptr: avoid an infinite loop for back = 0
[03:39] <cone-850> ffmpeg.git 03Anton Khirnov 07release/1.1:e2cf32ca5f58: mlpdec: do not try to allocate a zero-sized output buffer.
[03:39] <cone-850> ffmpeg.git 03Anton Khirnov 07release/1.1:a6403a3b6917: qtrle: add more checks against pixel_ptr being negative.
[03:39] <cone-850> ffmpeg.git 03Anton Khirnov 07release/1.1:bb3f1cad171b: 4xm: check the return value of read_huffman_tables().
[03:39] <cone-850> ffmpeg.git 03Anton Khirnov 07release/1.1:77493bfd9762: cavs: initialize various context tables to 0
[03:39] <cone-850> ffmpeg.git 03Reinhard Tartler 07release/1.1:a991c0673f71: update Changelog
[03:39] <cone-850> ffmpeg.git 03Reinhard Tartler 07release/1.1:670128ff13da: Prepare for 9.3 Release
[03:39] <cone-850> ffmpeg.git 03Vicente Jimenez Aguilar 07release/1.1:b6ae41e7f425: doc: filters: Correct BNF FILTER description
[03:39] <cone-850> ffmpeg.git 03Diego Biurrun 07release/1.1:dc745b76aa9a: doc: developer: Allow tabs in the vim configuration for Automake files
[03:39] <cone-850> ffmpeg.git 03Michael Niedermayer 07release/1.1:f64e4a8c9a19: Merge remote-tracking branch 'qatar/release/9' into release/1.1
[05:00] <cone-850> ffmpeg.git 03Michael Niedermayer 07release/1.1:6e8ed38fabb5: aac: reconfigure output on pop
[05:00] <cone-850> ffmpeg.git 03Michael Niedermayer 07release/1.1:3348e66e2eb3: doc/APIchanges: fix odd .01 versions
[05:00] <cone-850> ffmpeg.git 03Michael Niedermayer 07release/1.1:ece16d91ee43: apichanges: fix date
[05:00] <cone-850> ffmpeg.git 03Michael Niedermayer 07release/1.1:4bde8c1369e7: apichanges: Use , instead of / to seperate multiple hashes
[05:00] <cone-850> ffmpeg.git 03Michael Niedermayer 07release/1.1:98e96652f11d: apichanges: fix 2 wrong hashes
[05:00] <cone-850> ffmpeg.git 03Michael Niedermayer 07release/1.1:50ebb524cd54: doc/APIchanges: List merge commit hashes and version numbers
[11:22] <durandal_1707> michaelni: the strange split for alpha is present in RGB48 case to
[11:22] <durandal_1707> is that bug?
[11:25] <durandal_1707> michaelni: also i set wrong alpha value, you did not notice it
[11:32] <durandal_1707> i changed alpha to 1 in yuv2rgb and now i get segv
[11:34] <durandal_1707> hmm, i think i get why segv happens, lets prove it...
[11:35] <durandal_1707> michaelni: will add swscale rewrite task as gsoc and will mentor it if noone will not
[12:01] <durandal_1707> i think i'm missing something in ff_yuv2rgb_c_init_tables
[12:37] <cone-435> ffmpeg.git 03Diego Biurrun 07master:b58b00aeca21: configure: Separate "ln" command line arguments
[12:37] <cone-435> ffmpeg.git 03Michael Niedermayer 07master:2b277f2992d6: Merge commit 'b58b00aeca21de00ab6da2944684f784d9d6bc47'
[12:43] <cone-435> ffmpeg.git 03Diego Biurrun 07master:b2d688ea9f9c: configure: Identify icc compiler with a less ambiguous pattern
[12:43] <cone-435> ffmpeg.git 03Justin Ruggles 07master:e951b6d94c44: vorbisdec: cosmetics: rename variable avccontext to avctx
[12:43] <cone-435> ffmpeg.git 03Michael Niedermayer 07master:13fa07417326: Merge commit 'e951b6d94c441d46b396ef12da1428297d77251d'
[12:56] <cone-435> ffmpeg.git 03Justin Ruggles 07master:09031b463966: vorbisenc: cosmetics: rename variable avccontext to avctx
[12:56] <cone-435> ffmpeg.git 03Justin Ruggles 07master:699d02b839ff: libschroedinger: cosmetics: rename variable avccontext to avctx
[12:56] <cone-435> ffmpeg.git 03Justin Ruggles 07master:e8da807537e3: cmdutils: only use libavresample when it is enabled
[12:56] <cone-435> ffmpeg.git 03Michael Niedermayer 07master:6fbddc80d63d: Merge commit 'e8da807537e314d74cb6d93598f1dcfb891fa655'
[13:02] <cone-435> ffmpeg.git 03Justin Ruggles 07master:d925cca95f5d: avconv: remove an unused variable
[13:02] <cone-435> ffmpeg.git 03Diego Biurrun 07master:76b19a398435: Fix a number of incorrect intmath.h #includes.
[13:02] <cone-435> ffmpeg.git 03Michael Niedermayer 07master:9b1a0c2ee82f: Merge commit '76b19a3984359b3be44d4f7e4e69b7b86729a622'
[13:13] <cone-435> ffmpeg.git 03Diego Biurrun 07master:c242bbd8b693: Remove unnecessary dsputil.h #includes
[13:13] <cone-435> ffmpeg.git 03Michael Niedermayer 07master:a984efd104cc: Merge commit 'c242bbd8b6939507a1a6fb64101b0553d92d303f'
[13:19] <cone-435> ffmpeg.git 03Diego Biurrun 07master:3a02b6884cda: configure: icc: Drop nonsense adding of cpuflags to LDFLAGS
[13:19] <cone-435> ffmpeg.git 03Michael Niedermayer 07master:e907aa98cd27: Merge remote-tracking branch 'qatar/master'
[13:35] <durandal_1707> i cant't find any posterize filter that saste is talking about
[13:59] <saste> durandal_1707, elbg
[14:02] <durandal_1707> and why it was not accepted?
[14:02] <saste> http://thread.gmane.org/gmane.comp.video.ffmpeg.devel/101397/focus=101443
[14:03] <saste> durandal_1707, ^^
[14:04] <saste> Random-generator solution with flling palette and output with random entries would be even faster but produces images with worse quality
[14:04] <saste> kostya reply :)
[14:05] <durandal_1707> idea about sws is limited
[14:05] <durandal_1707> pal8 use only 256 colors
[14:05] <durandal_1707> and there is now way to control that to higher/lower number
[14:06] <durandal_1707> or do different kind of posterization
[14:06] <wm4> or do processing that yields output optimal for multiple frames?
[14:07] <saste> durandal_1707, basically it needs to be updated to support more formats
[14:07] <saste> but I suppose even an updated version of the patch would be acceptable in its current form
[14:07] <saste> another thing that i wanted to implement
[14:08] <saste> the possibility to select the palette
[14:08] <saste> right now it computes the palette at each frame IIRC
[14:08] <durandal_1707> it is implemeneted in swscale, something i'm not interested
[14:09] <wm4> swscale is a rewrite candidate, patches welcome
[14:17] <durandal_1707> wm4: want to help?
[14:17] <wm4> patches welcome
[14:20] <wm4> hm there was some other library which can do scaling and pixel conversion
[14:20] <wm4> but I can't find it
[14:20] <wm4> oh, there: http://forum.doom9.org/showthread.php?t=166504
[14:21] <durandal_1707> it is vapoursynth plugin
[14:28] <durandal_1707> on positive side it is wtfpl license
[14:30] <durandal_1707> wm4: have any other library in mind?
[14:30] <wm4> not really... I also found pixman (has some new code for yuv conversions) and gstreamer shit (uses some sort of dynamic compiler...)
[14:35] <durandal_1707> dynamic compiler?
[14:36] <durandal_1707> i found babl
[14:37] <durandal_1707> then there is pixfc-sse
[14:39] <durandal_1707> babl lacks dithering and not multiple of 8 support
[14:39] <wm4> durandal_1707: this is the dynamic compiler thing: http://liboil.freedesktop.org/ (link doesn't seem to work right now)
[14:39] <durandal_1707> dunno how hard would be to add that to babl
[14:40] <wm4> lol pixfc-sse employs similar techniques as swscale
[14:42] <wm4> also it's GPL3
[14:45] <wm4> gstreamer also has a very old copy of some pre-swscale ffmpeg format conversion (?)
[14:47] <Compn> wm4 : there should obviously be one in imagemagick
[14:49] <wm4> durandal_1707: oh, here's gstreamer's conversion stuff: http://cgit.freedesktop.org/gstreamer/gst-plugins-base/tree/gst/videoconvert/gstvideoconvertorc.orc
[15:11] <durandal_1707> babl use lgpl 3
[15:25] <durandal_1707> wm4: "patch welcome" is all what you get here
[15:25] <wm4> durandal_1707: patches welcome to fix that
[16:56] <cone-485> ffmpeg.git 03Michael Niedermayer 07master:c6d3b3be1555: aacsbr: Silence warning: max_qmf_subbands may be used uninitialized in this function
[16:56] <cone-485> ffmpeg.git 03Michael Niedermayer 07master:e4eedb983dbc: swscale-test: fix 3 pointer type warnings
[19:11] <Compn> whoa
[19:11] <Compn> xvba support?!?
[19:12] <Compn> xbmc guys are working hard :)
[19:18] <nevcairiel> why does linux not manage to create one hw interface that everyone supports? now we have vdpau, vaapi, xvba, and probably more :p
[19:20] <Compn> nevcairiel : erm, you try getting intel, nvidia and ati to place nicely in this thing where everyone has patents on every little bit of 3d video tech from 20 years ago
[19:21] <nevcairiel> works for windows
[19:21] <Compn> i bet somewhere, some company has those 3dfx patents
[19:21] <nevcairiel> nvidia bought 3dfx
[19:21] <Compn> there you go
[19:24] <nevcairiel> its still in their interest to support one API so more applications support their hardware
[19:32] <Compn> well then you're talking about 80% windows and 10% linux
[19:32] <Compn> so they spend time on 80% windows
[19:32] <Compn> ignore 10% linux and move on
[19:32] <Compn> drop some patches and run
[19:32] <Compn> nvidia dedicated a guy to linux support, thats most of the reason vdpau was accepted in most applications
[19:33] <Compn> plus a lot of devs had nvidia hardware ...
[19:34] <Compn> like carl, uau and others
[19:40] <cone-485> ffmpeg.git 03Michael Niedermayer 07master:997a36238ff0: mpeg12: Detect MXF essence stuff at the end of frames
[20:15] <iive> for a long time xvba had no official documentation. I've heard the vaapi wrapper was written under NDA
[00:00] --- Wed Feb 27 2013