[Ffmpeg-devel-irc] ffmpeg-devel.log.20140627

[burek]

[00:12] <cone-973> ffmpeg.git 03Anton Khirnov 07master:6a927d7aaf56: output example: use a macro instead of a static variable
[00:12] <cone-973> ffmpeg.git 03Michael Niedermayer 07master:877716a79c9d: Merge commit '6a927d7aaf5625e83a674072913b9e292a303fd1'
[00:21] <cone-973> ffmpeg.git 03Anton Khirnov 07master:a7fcd4122b19: output example: store the scaling context in the stream context
[00:21] <cone-973> ffmpeg.git 03Michael Niedermayer 07master:9a2850c26af8: Merge commit 'a7fcd4122b19b0f934020f4e261d0c44c4c32e11'
[01:01] <cone-973> ffmpeg.git 03Anton Khirnov 07master:182d3f8221c2: output example: rewrite encoding logic
[01:01] <cone-973> ffmpeg.git 03Michael Niedermayer 07master:8469c73576f7: Merge commit '182d3f8221c23716ea4eafa460bdb94968f71f26'
[01:11] <cone-973> ffmpeg.git 03Anton Khirnov 07master:a5864e9a2531: output example: free the muxing format context properly
[01:11] <cone-973> ffmpeg.git 03Michael Niedermayer 07master:ad081e4def36: Merge commit 'a5864e9a253143437471dba61a533e7fa9acec1c'
[01:16] <cone-973> ffmpeg.git 03Kostya Shishkov 07master:422e14f721c2: indeo2: rename stride to pitch for consistency with other Indeo decoders
[01:16] <cone-973> ffmpeg.git 03Michael Niedermayer 07master:826726b83fab: Merge commit '422e14f721c22cf9c19a8e7aae051ba9d559f6b6'
[02:03] <jamrial> michaelni: can you check if this is faster for you? https://github.com/jamrial/FFmpeg/commit/53ea0eec06fe4117bf596470811e1f6224bc35d1
[02:03] <jamrial> also, how did you measure it the first time, or what video did you use?
[02:05] <michaelni> was one of the fate samples
[02:06] <michaelni> maybe vsynth1-flv.flv but iam nt sure i picked randomly after adding a abort() in mpegvideo_motion.c to find one that used the code
[02:07] <michaelni> jamrial, do you see a speed gain ?
[02:07] <michaelni> i mean with the original patch
[02:07] <michaelni> didnt yet look at the new one
[02:11] <jamrial> i saw a speed gain in emu_edge_hvar_ssse3, but not in the many hfixtbl_ssse3 functions. in some widths the latter seemed a bit slower, even
[02:11] <cone-973> ffmpeg.git 03Dirk Ausserhaus 07master:5ec6d152e26c: indeo4: B-frames decoding
[02:11] <cone-973> ffmpeg.git 03Michael Niedermayer 07master:357168bcf634: Merge commit '5ec6d152e26c570c0a16ec72c1f354db95708179'
[02:12] <jamrial> this using the timer macros inside emulated_edge_mc() around a single hfix call
[02:13] <michaelni> i had the timers around the emulated edge mc calls
[02:13] <jamrial> not sure how you measured things yourself
[02:14] <jamrial> ok
[02:20] <michaelni> tried the new patch with 1 timer around the 3 calls in mpeg_motion_internal
[02:20] <michaelni> i cant really meassure a difference
[02:22] <michaelni> for testing i created a flv from matrixbench (-i matrixbench_mpeg2.mpg  -an test.flv) benchmark was done with ./ffmpeg -threads 1 -i test.flv -f null -
[02:23] <michaelni> the number of calls to this code in the whole file are very few though
[02:23] <michaelni> so at least for this file it shouldnt matter at all how fast it is
[02:23] <michaelni> do you have a file that uses it alot ?
[02:25] <jamrial> no, tried a couple files and most would call it only a few times as well
[02:42] <cone-973> ffmpeg.git 03Diego Biurrun 07master:24f45c16224d: fate: Add dependencies for dct/fft/mdct/rdft tests
[02:42] <cone-973> ffmpeg.git 03Michael Niedermayer 07master:5a636383db9b: Merge commit '24f45c16224d4c5d482e928676714766ffdda4fc'
[03:07] <Timothy_Gu> michaelni: why is matrixbench so often used as benchmarking material?
[03:09] <michaelni> when theres no reason to pick something specific then i tend to pick matrixbench or for rawvideo foreman or raw image lena
[03:13] <Timothy_Gu> so it's just a personal preference? there isn't a reason behind it?
[03:19] <cone-973> ffmpeg.git 03Diego Biurrun 07master:d2869aea0494: dsputil: Move MMX/SSE2-optimized IDCT bits to the x86 subdirectory
[03:19] <michaelni> yes
[03:19] <cone-973> ffmpeg.git 03Michael Niedermayer 07master:09a7a4704e84: Merge commit 'd2869aea0494d3a20d53d5034cd41dbb488eb133'
[03:20] <jamrial> where can i find it for that matter? i don't see it in the fate samples folder
[03:25] <michaelni> jamrial, http://samples.ffmpeg.org/benchmark/testsuite1/
[03:26] <jamrial> thanks
[03:39] <cone-973> ffmpeg.git 03Gaullier Nicolas 07master:134beb9e0222: avformat/mxfenc: rename very recently added option to set/force channelcount in MXF D-10
[03:47] <Compnn> Timothy_Gu : for samples i use microsofts' video for windows sample.avi , its 15fps cinepak :P
[04:02] <Timothy_Gu> lol
[06:18] <cone-973> ffmpeg.git 03Michael Niedermayer 07master:b96c9513f35a: avcodec/snow: factor ff_snow_get_buffer() out
[08:31] <anshul> I reset my ffmpeg git to previous commit and found my repository is now making avplay and avconv, is it normal?
[08:34] <debianuser> That was probably a merge from livav, and you switched to its parent from libav.
[09:55] <ubitux> i have 2 mts in h264, one is interlaced, the other is progressive
[09:55] <ubitux> both mark frames as interlaced
[09:56] <ubitux> do we have any way of fixing this with ffmpeg without re-encoding?
[09:56] <ubitux> like a way to alter packets without the decode/encode pass
[09:57] <ubitux> with bsf maybe?
[10:46] <thardin> http://www.nasaspaceflight.com/2014/06/recovering-falcon-9-ocean-landing-video-done/
[10:47] <thardin> didn't know about this before
[12:53] <ubitux> http://blog.securitymouse.com/2014/06/raising-lazarus-20-year-old-bug-that.html
[13:15] <michaelni> ubitux, want to write news entry & update security page ?
[13:15] <ubitux> not really
[13:16] <ubitux> did you make a new release after the fix?
[13:16] <michaelni> yes
[13:16] <michaelni> d6af26c55c1ea30f85a7d9edbc373f53be1743ee
[13:17] <michaelni> need to logoff, will be bak soon
[13:18] <Compnn> whoa
[13:18] <Compnn> michael offline
[13:18] <Compnn> i like these security researchers, always saying how big a deal each bug is
[13:19] <Compnn> in reality, your operating system and all of its libs are full of issues like this
[13:19] <Compnn> yes each bug is important, but you do not have security and you must do something else to protect yourself :)
[13:20] <Compnn> virtual machine maybe
[13:38] <anshul> saste, regarding #153 did you got the the answer of ur own question, what is expected output
[13:43] <saste> anshul, do you mean this: https://trac.ffmpeg.org/ticket/153 ?
[13:44] <saste> not that I know
[13:45] <anshul> yes that ticket
[13:46] <saste> anshul, i'm not sure it is even a valid ticket
[13:47] <anshul> is disscussion regarding ticket  on devel-mailing list valid
[13:58] <cone-692> ffmpeg.git 03James Almer 07master:a6b8aef0dc70: configure: remove duplicate ape_decoder_select line
[13:58] <cone-692> ffmpeg.git 03James Almer 07master:607c88126c14: configure: fix g729 decoder dependencies
[13:58] <cone-692> ffmpeg.git 03James Almer 07master:5fef4afb8527: configure: add missing audiodsp dependency for ra144
[14:09] <saste> anshul, what thread?
[14:13] <anshul> saste, i dont know which thread it was,  I just wanted to know that the question you asked on that ticket #153, should i bring light on that stating new thread
[14:26] <anshul> What would be an better idea if subtiltle bitmap width is larger then video width. should be call an scale filter or should be remove extra alpha spcae in start and end
[14:36] <cone-692> ffmpeg.git 03Anshul Maheshwari 07master:263932c08409: avcodec/dvbsubdec: fix ignoring blank frame by updating got_output
[15:05] <saste> anshul, i was never answered my question
[15:05] <saste> i played the file, and the issue was not apparent to me, that's why I asked
[15:05] <saste> anshul, is your patch fixing the issue?
[15:07] <anshul> my patch is not regarding that issue #153 its regarding #3737 
[15:07] <saste> anshul, oh now i see, it is the "help" text message which doesn't look right at the beginning of the video, right?
[15:09] <anshul> I know about that issue its might be because not considering y =0 in yuv and trying to make it rgb 
[15:09] <anshul> in dvbsub y=0 have special meaning , but dont know that is the only issue
[15:11] <anshul> when i debug the ffplay my speaker volume is much higher than i do directly ffplay, is any one else got this problem 
[17:37] <michaelni> Timothy_Gu, launchpad.net/ffmpeg/ lists 2.2.3 as latest release, not sure who updated it in the past, but whoever did please update
[17:37] <Timothy_Gu> michaelni: it's automatic
[17:37] <michaelni> ahh, ok thats perfect, thanks
[17:57] <cone-849> ffmpeg.git 03Sergey 07master:6d2df3c00a78: ffmpeg: fix deadlock regression in threading error handing
[19:19] <cone-849> ffmpeg.git 03Diego Biurrun 07master:e63b818dbe60: dv: Properly split decoder and encoder initialization
[19:19] <cone-849> ffmpeg.git 03Michael Niedermayer 07master:3ac51145304f: Merge commit 'e63b818dbe6060b606ae10eb184859f8b97d2353'
[19:29] <cone-849> ffmpeg.git 03Diego Biurrun 07master:5dd8c08fd5e4: mpeg: Change ff_convert_matrix() to take an MpegEncContext parameter
[19:29] <cone-849> ffmpeg.git 03Michael Niedermayer 07master:1fd0e9a56d75: Merge commit '5dd8c08fd5e4c04d7a08d8934f0098a8a4a35c28'
[19:31] <Voicu> any idea why avcodec_decode_video2 would fail with "Invalid argument" ? AFAICT the parameters are correct
[19:32] <Voicu> the codec and packets are valid since I am already using them for something else and the frame is only suppposed to be allocated with av_frame_alloc, right?
[19:52] <Voicu> hmm, can it be from some weird timestamp?
[20:01] <thardin> woop, d10_channelcount pushed
[22:51] <cone-849> ffmpeg.git 03Michael Niedermayer 07master:8c22bea57447: avcodec/vp8: Fix multiple ff_thread_finish_setup() calls
[00:00] --- Sat Jun 28 2014