[Ffmpeg-devel-irc] ffmpeg-devel.log.20160825

[burek]

[00:17:01 CEST] <durandal_1707> so I ported another wavelet denoiser which is several times faster than mplayer owdenoise
[00:33:39 CEST] <JEEB> hmm, anyone bench'd clang 3.8 vs gcc on ARMv7?
[02:40:47 CEST] <Compn> durandal_1707 : people use ffmpeg lavfi ? :)
[02:42:56 CEST] <Compn> j-b , bp0 , atomnuker , durandal_1707 : i'm not saying dont remove faac, but maybe , just MAYBE, put a map of -acodec faac to the ffmpeg aac encoder so removing faac does not break scripts
[02:43:13 CEST] <Compn> and maybe ... just maybe... print a little warning "faac not found, using fffaac instead"
[02:43:36 CEST] <Compn> because we like our users, and our users scripts , and this would be like 5 lines of code
[02:43:38 CEST] <Compn> right ?
[02:43:45 CEST] <Compn> doesnt hurt anyone...
[02:44:05 CEST] <Compn> except whoever uses faac and needs faac and faac alone. but you guys are ready to slap those people anyway
[02:46:40 CEST] <Compn> durandal_1707 : also carl doesnt know what your question means either
[02:46:49 CEST] <Compn> or why you didnt ask yourself
[02:51:19 CEST] <durandal_1707> Compn: question about what?
[02:52:46 CEST] <Compn> [12:12] <durandal_1707> Compn: you should really talk to Carl
[02:52:46 CEST] <Compn> [14:32] <durandal_170> Compn: about FFmpeg future
[02:54:48 CEST] Action: Compn stares into the abyss
[02:56:32 CEST] <durandal_1707> I was dreaming...
[02:59:25 CEST] <Compn> alrighty
[02:59:53 CEST] <TD-Linux> aren't the scripts already broken by nature of using libfaac? :^)
[03:00:43 CEST] <Compn> it was one of the only encoders available until the fdkaac one
[03:00:47 CEST] <Compn> for ... years
[03:01:02 CEST] <Compn> there is no logical reason not to do this simple workaround.
[03:03:33 CEST] <rcombs> there's no reason not to fix your bash scripts
[03:04:10 CEST] <rcombs> what, do we want some sort of deprecation period?
[03:04:52 CEST] <durandal_1707> no
[03:05:12 CEST] <Compn> and every faac example on the internet...
[03:05:17 CEST] <Compn> for people to copy and paste quickly
[03:05:43 CEST] <Compn> (i would also be ok with making a workaround for other encoders not found to go to the default ffmpeg encoder)
[03:09:36 CEST] <TD-Linux> are the built in encoder parameters a compatible superset of faac parameters?
[03:11:21 CEST] <Compn> "ignoring faac options , using defaults"
[03:11:23 CEST] <Compn> :D
[03:24:35 CEST] <nevcairiel> Feel free to send a patch to the ML to propose such a codec replacement mechanic
[03:32:16 CEST] <cone-761> ffmpeg 03Michael Niedermayer 07master:01aee8148d4f: avcodec/exr: Check tile positions
[04:10:05 CEST] <blb> they just showed using ffmpeg on mr robot
[04:21:48 CEST] <cone-761> ffmpeg 03Michael Niedermayer 07release/3.1:905372be8f74: avfilter/drawutils: Fix single plane with alpha
[04:21:49 CEST] <cone-761> ffmpeg 03Michael Niedermayer 07release/3.1:055e5c80ee07: avcodec/h2645: Fix NAL unit padding
[04:21:50 CEST] <cone-761> ffmpeg 03Michael Niedermayer 07release/3.1:7d42daeea2df: vcodec/h2645_parse: Clear buffer padding
[04:21:51 CEST] <cone-761> ffmpeg 03Michael Niedermayer 07release/3.1:afd57722e1a8: avcodec/rawdec: Fix bits_per_coded_sample checks
[04:21:52 CEST] <cone-761> ffmpeg 03Michael Niedermayer 07release/3.1:4770eac663da: avformat/swfdec: Fix inflate() error code check
[04:21:53 CEST] <cone-761> ffmpeg 03Michael Niedermayer 07release/3.1:77f978996bd5: avcodec/indeo2: check ctab
[04:21:54 CEST] <cone-761> ffmpeg 03Michael Niedermayer 07release/3.1:22a0c0e76427: avcodec/cfhd: Increase minimum band dimension to 3
[04:21:55 CEST] <cone-761> ffmpeg 03Michael Niedermayer 07release/3.1:93422bc92e94: avcodec/h264_parser: Factor get_avc_nalsize() out
[04:21:56 CEST] <cone-761> ffmpeg 03Hendrik Leppkes 07release/3.1:0ad4d4198a40: h2645_parse: don't overread AnnexB NALs within an avc stream
[04:21:57 CEST] <cone-761> ffmpeg 03Hendrik Leppkes 07release/3.1:fabc1c9e567d: h2645_parse: only read avc length code at the correct position
[04:21:58 CEST] <cone-761> ffmpeg 03Michael Niedermayer 07release/3.1:ec30a498e66a: avcodec/h2645_parse: Use get_nalsize() in ff_h2645_packet_split()
[04:21:59 CEST] <cone-761> ffmpeg 03Michael Niedermayer 07release/3.1:8003a5d23792: avcodec/h2645_parse: fix nal size
[04:22:00 CEST] <cone-761> ffmpeg 03Michael Niedermayer 07release/3.1:049d7677156a: avcodec/diracdec: Check numx/y
[04:22:01 CEST] <cone-761> ffmpeg 03Michael Niedermayer 07release/3.1:8c4a67183b07: avcodec/svq3: fix slice size check
[04:22:02 CEST] <cone-761> ffmpeg 03Michael Niedermayer 07release/3.1:4943abe05110: avcodec/adpcm: Fix adpcm_ima_wav padding
[04:22:03 CEST] <cone-761> ffmpeg 03Michael Niedermayer 07release/3.1:2f07937926e5: libavcodec/wmalosslessdec: Check the remaining bits
[04:22:04 CEST] <cone-761> ffmpeg 03Michael Niedermayer 07release/3.1:596513ca2ce9: avformat/wtvdec: Check pointer before use
[04:22:05 CEST] <cone-761> ffmpeg 03Michael Niedermayer 07release/3.1:ae893819620b: avcodec/aacenc: Tighter input checks
[04:22:06 CEST] <cone-761> ffmpeg 03Michael Niedermayer 07release/3.1:79f52a0dbd48: avcodec/exr: Check tile positions
[04:22:07 CEST] <cone-761> ffmpeg 03Michael Niedermayer 07release/3.1:949094a4cdd9: Update for 3.1.3
[04:31:41 CEST] <jamrial> michaelni: can i backport bba6a03 to release/3.1 or are you in the middle of tagging?
[04:33:10 CEST] <jamrial> since there was at least one downstream developer that mentioned the examples weren't being helpful regarding codecpar usage, i think it would be good to have it there
[04:39:37 CEST] <michaelni> jamrial, sure, iam not taging now, i wanted to give fate.ffmpeg.org a bit time to test first 
[04:39:49 CEST] <jamrial> michaelni: alright, thanks
[04:44:39 CEST] <cone-761> ffmpeg 03James Almer 07release/3.1:40ab55746e29: examples/demuxing_decoding: convert to codecpar
[04:44:40 CEST] <cone-761> ffmpeg 03James Almer 07release/3.1:c46d22a4a584: Changelog: update after last commit
[06:16:40 CEST] <durandal_1707> should I call lut2, lut2d instead?
[09:07:54 CEST] <Guest_84853> allah is doing
[09:07:59 CEST] <Guest_84853> sun is not doing allah is doing
[09:08:04 CEST] <Guest_84853> moon is not doing allah is doing
[09:08:10 CEST] <Guest_84853> stars are not doing allah is doing
[09:08:16 CEST] <Guest_84853> planets are not doing allah is doing
[09:08:23 CEST] <Guest_84853> galaxies are not doing allah is doing
[09:08:46 CEST] <Guest_84853> oceans are not doing allah is doing
[09:08:52 CEST] <Guest_84853> mountains are not doing allah is doing
[09:09:02 CEST] <Guest_84853> trees are not doing allah is doing
[09:09:07 CEST] <Guest_84853> mom is not doing allah is doing
[09:09:16 CEST] <Guest_84853> dad is not doing allah is doing
[09:10:02 CEST] <Guest_84853> boss is not doing allah is doing
[09:10:08 CEST] <Guest_84853> job is not doing allah is doing
[09:10:15 CEST] <Guest_84853> dollar is not doing allah is doing
[09:10:23 CEST] <Guest_84853> degree is not doing allah is doing
[09:10:32 CEST] <Guest_84853> medicine is not doing allah is doing
[09:10:39 CEST] <Guest_84853> customers are not doing allah is doing
[09:10:51 CEST] <Guest_84853> you can not get a job without the permission of allah
[09:11:05 CEST] <Guest_84853> you can not get married without the permission of allah
[09:11:38 CEST] <Guest_84853> nobody can get angry at you without the permission of allah
[09:11:44 CEST] <shadow42085> I hope I am in the right room I have something to add to the ffmpeg wiki
[09:11:45 CEST] <Guest_84853> light is not doing allah is doing
[09:12:08 CEST] <Guest_84853> fan is not doing allah is doing
[09:15:47 CEST] <shadow42085> I have something to add to the trac.ffmpeg.org/wiki/CompilationGuide/Ubuntu
[09:16:25 CEST] <JEEB> it will hopefully get edited if you're doing it wrong
[09:17:11 CEST] <shadow42085> i've never used trac wiki setup only used wikipedia model
[09:18:07 CEST] <shadow42085> because its not a bug
[09:19:05 CEST] <c_14> shadow42085: just create an account, navigate to the page and click edit
[11:16:19 CEST] <michaelni> atomnuker, about "[FFmpeg-devel] [PATCH] avcodec/aaccoder: Limit sf_idx difference for all cases", should this be in 3.1.3 ? if so it needs a review now
[11:20:45 CEST] <cone-556> ffmpeg 03Paul B Mahol 07master:5d7743019b32: avfilter: add lut2 filter
[11:20:46 CEST] <cone-556> ffmpeg 03Paul B Mahol 07master:feb2ea6e09bf: avfilter: add yuvtestsrc source filter
[11:51:21 CEST] <atomnuker> michaelni: give me a few hours, I want to make some comparisons
[11:53:20 CEST] <michaelni> atomnuker, ok
[12:47:04 CEST] <atomnuker> I don't see any difference but it's still probably masking the real problem
[12:47:19 CEST] <atomnuker> where can I get the sample from?
[12:49:17 CEST] <atomnuker> michaelni ^^
[17:07:58 CEST] <flux> I wonder if stripping the FFMPEG_CONFIGURATION from config.h (and putting it behind an extern to some .c-file) would make ffmpeg more compatible with ccache :)
[18:41:41 CEST] <cone-556> ffmpeg 03Michael Niedermayer 07master:e1ced854f2fb: avfilter/af_acrusher: Rename sun/sov
[18:41:42 CEST] <cone-556> ffmpeg 03Steven Liu 07master:500662784341: tests/audiomatch: Fix compile warning
[18:42:57 CEST] <durandal_1707> lol, please remove sun from fate
[18:46:22 CEST] <iive> btw, if you need name for the next release, you can consider Mr Robot, Free Society or something else that pays homage to the serial.
[18:47:22 CEST] <iive> they used ffmpeg in it.
[18:54:05 CEST] <Compn> yes! 
[18:54:09 CEST] <Compn> they used ffmpeg a couple of times :)
[18:54:29 CEST] <Compn> also the filenames of the show are most interesting
[18:56:38 CEST] <Compn> they have some good open source guy on that show to provide some knowledge anyway
[19:29:32 CEST] <Compn> er, its the soundtrakc names i like more
[19:29:39 CEST] <Compn> 1.2_3-therealshayla.wav,
[19:29:39 CEST] <Compn> 1.1_5-illusionofchoice.mp3,
[19:29:39 CEST] <Compn> 1.4_6-n0execution.au,
[19:29:39 CEST] <Compn> 1.3_2-everyrev0lution.ra,
[19:30:01 CEST] <Compn> would be more fun if they were really encoded in each format like that :D
[19:31:08 CEST] <jamrial> that's a great way to get people to ask for a refund :p
[19:38:54 CEST] <D`K_C> Hi folks, my application crashes when using av_parser_parse2 (which uses the mpegaudio parser) because the input buffer is read beyond its end, is that me badly using this function (should the input buffer be padded somehow?) or is that a real bug. I managed to dump the audio and wrote a simpler test case to hilight the issue
[19:45:17 CEST] <JEEB> D`K_C: are you using av_malloc to allocate your buffer?
[19:46:32 CEST] <D`K_C> nope, the input buffer is allocated through malloc
[19:48:30 CEST] <JEEB> try switching to av_malloc/av_free
[19:48:41 CEST] <D`K_C> JEEB: actually I just changed the allocation to av_malloc and issue is still there
[19:48:58 CEST] <JEEB> ok
[19:49:22 CEST] <JEEB> I remember there was a constant that let you set some extra padding that was used in some internals... lemme look if I can find it
[19:49:31 CEST] <D`K_C> is that okay if I push my test case on a github repo for you to take a look
[19:49:44 CEST] <D`K_C> yeah AV_INPUT_BUFFER_PADDING_SIZE
[19:50:11 CEST] <JEEB> yeah, I use that in the utvideo encoder f.ex.
[19:50:12 CEST] <D`K_C> but still, the function is not respecting the buf_size parameter
[19:51:04 CEST] <JEEB> anyways, I've not used parsers myself - only demuxers tbqh
[19:51:47 CEST] <D`K_C> I'll push that to a repo with valgrind output
[20:16:25 CEST] <D`K_C> here we go: https://github.com/d-k-c/ffmpeg-av_parser_parse2-bug
[21:09:47 CEST] <durandal_1707> I gonna apply both mov patches soon
[21:15:35 CEST] <durandal_1707> is that fine?
[21:20:04 CEST] <ubitux> the edit list thing?
[21:22:10 CEST] <durandal_1707> ubitux: 2 patches from Derek
[21:22:20 CEST] <ubitux> ok
[21:22:33 CEST] <ubitux> my mails are broken since a few days... :(
[21:22:46 CEST] <ubitux> and ovh seems to have derp to not help anything
[21:22:59 CEST] <ubitux> i guess i should have my own dns at some point
[21:32:29 CEST] <cone-556> ffmpeg 03Derek Buitenhuis 07master:04f8d312877f: mov: Remove ancient heuristic hack
[21:32:31 CEST] <cone-556> ffmpeg 03Derek Buitenhuis 07master:425be3c810e0: mov: Remove old b-frame/video delay heuristic
[22:12:49 CEST] <atomnuker> durandal_1707: can the filmgrain filter be used as a generic dither filter too?
[22:14:50 CEST] <durandal_1707> it just blends simplex noise to source
[22:16:09 CEST] <durandal_1707> dithering is smarter, cares about sourounding
[22:17:09 CEST] <durandal_1707> I think doing it as source may be better, than you can apply different blends
[22:18:14 CEST] <durandal_1707> it may not produce film grain at all, so test it and report back
[22:25:00 CEST] <durandal_1707> atomnuker: hmm, it is fine for dithering, removes banding too, I need to try highbitdepth
[22:28:31 CEST] <cone-556> ffmpeg 03Vignesh Venkatasubramanian 07master:42a635ddae21: webm_chunk: Set pts precision to milliseconds
[23:33:38 CEST] <DSM__> when does the next major bump going to happen?
[23:34:02 CEST] <nevcairiel> undetermined
[23:35:08 CEST] <DSM__> nevcairiel: okay
[00:00:00 CEST] --- Fri Aug 26 2016