[Ffmpeg-devel-irc] ffmpeg-devel.log.20150603

[burek]

[01:31:24 CEST] <cone-756> ffmpeg 03Michael Niedermayer 07master:02915602d931: swresample: Check the return value of resampler->init()
[01:52:10 CEST] <cone-756> ffmpeg 03Donny Yang 07master:7495e728f5a2: avcodec/apng: Remove blending support for AV_PIX_FMT_ARGB
[02:24:26 CEST] <llogan> after 1f47c23 the "packed b-frames" message vanishes for all my my samples (my sample size is a staggering 2, including the one from #2913). is there a known sample where the message still appears?
[02:26:39 CEST] <cehoyos_> llogan: I still see the warning for the pegasus sample from 2913.
[02:28:32 CEST] <cehoyos_> ... But only with ffplay, I will open a ticket.
[02:29:17 CEST] <llogan> i didn't test with ffplay...didn't even watch the video
[02:51:40 CEST] <rcombs> so lavf should probably expose a loop point if the file provides one
[02:53:25 CEST] <rcombs> not sure where that belongs (AVFormatContext? AVStream? [this seems to usually be present on formats with only one stream, but no need to lock ourselves in]), or in what format (stream time base? [what if there are multiple?] AV_TIME_BASE?)
[03:29:09 CEST] <cone-756> ffmpeg 03Michael Niedermayer 07release/2.4:bf6ba4a03281: swresample: Check the return value of resampler->init()
[03:29:11 CEST] <cone-756> ffmpeg 03Andreas Cadhalpun 07release/2.4:1047c286fa20: libopenjpegenc: add NULL check for img before accessing it
[03:56:49 CEST] <cone-756> ffmpeg 03Donny Yang 07n2.4.10:HEAD: avcodec/apng: Remove blending support for AV_PIX_FMT_ARGB
[04:22:25 CEST] <llogan> i shouldn't drunk tweet...
[04:22:44 CEST] <llogan> "FFmpeg 2.5.10 is now available."
[04:23:00 CEST] <llogan> </fixed>
[04:46:53 CEST] <Timothy_Gu> llogan: what do you think about writing a tweetbot for releases?
[04:48:50 CEST] <llogan> Timothy_Gu: if you think it is practical then feel free to do so
[05:16:55 CEST] <Timothy_Gu> llogan: I'll see what i can do
[05:39:43 CEST] <llogan> Timothy_Gu: thanks.
[10:30:10 CEST] <rcombs> https://gist.github.com/31d41ffacb44c11d5757 <-- anyone wanna play with a not-quite-done BFSTM demuxer?
[10:30:30 CEST] <rcombs> http://puu.sh/iaEW8/3cc50a620d.bfstm short sample
[10:31:42 CEST] <rcombs> something always goes screwy at the end of the stream lavc decodes just short of a packet worth of data
[10:32:40 CEST] <rcombs> so it tries to decode the remaining few bytes and spits out `adpcm_thp: invalid number of samples in packet`
[10:33:22 CEST] <nevcairiel> i have never heard of neither brstm or bfstm
[10:33:44 CEST] <rcombs> nevcairiel: Gamecube/Wii/WiiU ADPCM audio format
[10:35:59 CEST] <cehoyos_> nevcairiel: Und ich erst! (The number of recent aac reports...)
[10:40:15 CEST] <nevcairiel> unfortunately it doesn't fix similar cases with LATM audio
[11:00:53 CEST] <cehoyos_> nevcairiel: Too bad (I had not tested yet)...
[11:49:50 CEST] <nevcairiel> to be precise, it improves a specific latm sample (#4544), but doesn't fix it entirely, so there may be another bug hidden somewhere
[11:57:23 CEST] <cone-331> ffmpeg 03Ganesh Ajjanagadde 07master:196b885a5f0a: swresample/dither: check memory allocation
[12:05:37 CEST] <cehoyos_> nevcairiel: That is the sample I was thinking about.
[12:09:32 CEST] <nevcairiel> the "channelelement x.x missing" errors go away with the patch, the "channel element 0.2 is not allocated" errors remain, so its not quite fixed yet
[12:13:47 CEST] <cehoyos_> What about the output duration?
[12:14:50 CEST] <nevcairiel> i assume thats still too short, since the errors it does produce will skip over audio frames
[12:16:26 CEST] <nevcairiel> but i didnt check
[12:26:01 CEST] <cone-331> ffmpeg 03Michael Niedermayer 07master:d7a762553c6f: avformat/mxfenc: Accept MXF D-10 with 49.999840 Mbit/sec
[12:27:30 CEST] <cehoyos_> They don't test what they push, do they?
[12:28:41 CEST] <cehoyos_> I can confirm that the sample from 4544 is not fixed by the patch;-(
[12:29:26 CEST] <nevcairiel> a lot of the warning messages go away though, so its an improvement at least, and it does fix the sample the author linked, so yay for that
[12:32:34 CEST] <cehoyos_> Yes, that is good
[12:32:43 CEST] <cehoyos_> I don't see warnings for that sample though (only clicks)
[12:33:28 CEST] <nevcairiel> i think the messages are on verbose output, so default log level may not show them
[12:35:38 CEST] <cehoyos_> Yes, you are right.
[13:40:50 CEST] <kierank> lglinskih: how's it going?
[13:40:55 CEST] <kierank> do you have a github
[14:16:03 CEST] <cone-331> ffmpeg 03Shivraj Patil 07master:ce1761db19af: avcodec/mips: MSA (MIPS-SIMD-Arch) optimizations for HEVC uniw mc functions
[14:19:59 CEST] <lglinskih> kierank: oh, veeeryyy sloooooow 
[14:20:23 CEST] <lglinskih> yes, I'm lglinskih there
[14:40:33 CEST] <cone-331> ffmpeg 03Paul B Mahol 07master:b8c0cf7880cf: avfilter/vf_blend: add linearlight and glow mode
[14:47:24 CEST] <kierank> av500: hmm mips hevc simd
[14:47:25 CEST] <kierank> weird
[14:48:02 CEST] <kierank> and "Director Software and Tools, Meta at Imagination Technologies" cc'd in
[15:04:05 CEST] <wm4> I'm still struggling to understand why you'd want to decode hevc on mips CPUs
[15:04:34 CEST] <wm4> even if you do, I'd imagine you'd have an awful SoC POS with a hw accelerated decoder
[15:06:56 CEST] <kierank> exactly
[15:07:02 CEST] <kierank> I can understand realmedia 
[15:11:34 CEST] <av500> you would always want a SW fallback
[15:12:52 CEST] <kierank> why
[15:13:06 CEST] <kierank> oh 10-bit or other things
[15:13:11 CEST] <av500> sometimes he decoders are wierd
[15:13:21 CEST] <av500> also for stuff like thumbnails
[15:13:26 CEST] <av500> that you do in the background
[15:13:29 CEST] <kierank> ahhh
[15:13:33 CEST] <av500> we use lavc for thumbs always
[15:13:42 CEST] <kierank> good job!
[15:13:52 CEST] <av500> much better than hoping the hwdec is properly multiinstance
[15:14:05 CEST] <av500> also you dont want backgnd jobs to kill the android media stack
[15:14:08 CEST] <av500> been there, done that :)
[15:17:34 CEST] <wm4> didn't someone want to implement schannel support?
[15:25:48 CEST] <wm4> debian is so classy
[15:26:01 CEST] <wm4> update -> suddenly everything that merely link gnutls crashes on start
[15:31:08 CEST] <nevcairiel> debians decision to remove public symbols from all the ssl libraries in the name of security is so extremely stupid
[15:35:25 CEST] <Compn> this is the same debian with the openssl patch that killed the random number generator
[15:36:23 CEST] <Compn> So was this problem a backdoor, purposefully introduced? It seems unlikely. The maintainer who made the change, Kurt Roeckx, was later made Secretary of the Debian Project, suggesting that he’s a real and trustworthy person and probably not a fake identity made up by the NSA to insert a vulnerability.
[15:36:38 CEST] <Compn> lol
[16:30:38 CEST] <wm4> would anyone happen to know how mediainfo determines the refresh rate of matroska files?
[16:45:34 CEST] <cone-331> ffmpeg 03Stefano Sabatini 07master:771537edcf70: ffmpeg_dxva2: call GetDesktopWindow() in place of GetShellWindow()
[16:54:10 CEST] <nevcairiel> wm4: is badly a valid answer?
[16:59:56 CEST] <wm4> the code is pretty weird
[17:00:16 CEST] <wm4> but it does make an exact guess even if the framerate header is wrong, and even though matroska timestamps are rounded
[17:00:42 CEST] <wm4> it might just picking a "standard framerate" if the estimate is close enough, but I don't see where it does this
[17:48:40 CEST] <cone-331> ffmpeg 03Michael Niedermayer 07master:9614df4b1529: avformat/mov: Retry same packet on IO failure to avoid loosing a packet
[18:24:30 CEST] <cone-331> ffmpeg 03Michael Niedermayer 07master:cdc0087969b0: avutil/ppc/cpu: Our code assumes vsx implies altivec, thus assert this
[18:37:55 CEST] <cone-331> ffmpeg 03Martin Storsjö 07master:76eb0950b8d8: configure: Don't run "export $e" for an empty string
[18:37:56 CEST] <cone-331> ffmpeg 03Michael Niedermayer 07master:c08a64bb2b12: Merge commit '76eb0950b8d82b1e9617b9c854f091e13bb23131'
[18:51:29 CEST] <cone-331> ffmpeg 03Martin Storsjö 07master:b9875b497a38: configure: Remove unnecessary tests for d3d11va
[18:51:30 CEST] <cone-331> ffmpeg 03Michael Niedermayer 07master:511638151102: Merge commit 'b9875b497a38f494acb599083b0d004545958697'
[19:00:45 CEST] <cone-331> ffmpeg 03Martin Storsjö 07master:e97446e60057: configure: Check for DXVA2_ConfigPictureDecode instead of LPDIRECT3DSURFACE9
[19:00:46 CEST] <cone-331> ffmpeg 03Michael Niedermayer 07master:ee08beb7d921: Merge commit 'e97446e600578392549ef6d0191f2020dae7f8bf'
[19:18:50 CEST] <cone-331> ffmpeg 03Luca Barbato 07master:7f596368a404: bink: Factorize bink put_pixel
[19:18:51 CEST] <cone-331> ffmpeg 03Michael Niedermayer 07master:6034b4bb88c1: Merge commit '7f596368a404363d72b1be6d16c51420a71bc523'
[19:49:45 CEST] <rcombs>  <wm4> didn't someone want to implement schannel support? <-- I wanted someone else to :P
[19:53:46 CEST] <Daemon404> we all want someone else to
[20:13:36 CEST] <cone-331> ffmpeg 03nu774 07master:0289f81241e7: aac: Correctly map multichannel ADTS AAC with non-zero channel_config + PCE
[20:13:37 CEST] <cone-331> ffmpeg 03Michael Niedermayer 07master:120028968749: Merge commit '0289f81241e720452b5a77713488d54d3ec252d7'
[20:14:35 CEST] <Compn> can you correctly map multichannel audio? thought everyone uses different standards (in different containers no less)?
[20:25:28 CEST] <rcombs> aww, I was hoping that commit was for the encoder
[21:18:36 CEST] <cone-331> ffmpeg 03Shivraj Patil 07master:88188f55a2b4: avcodec/mips: MSA (MIPS-SIMD-Arch) optimizations for HEVC biw mc functions
[21:42:40 CEST] <cone-331> ffmpeg 03Shivraj Patil 07master:c96c73b0b052: avcodec/mips: MSA (MIPS-SIMD-Arch) optimizations for HEVC mc epel functions
[21:42:41 CEST] <cone-331> ffmpeg 03Shivraj Patil 07master:aef34ab95048: avcodec/mips: MSA (MIPS-SIMD-Arch) optimizations for HEVC uni mc epel functions
[22:12:43 CEST] <anoop_r> hi
[22:13:12 CEST] <anoop_r> Is it possible to filter the output of ffmpeg
[22:13:32 CEST] <anoop_r> means ffmpeg shows lots of info in command prompt + progress
[22:13:57 CEST] <anoop_r> is it possible to customise it 
[22:14:25 CEST] <llogan> -loglevel
[22:33:08 CEST] <philipl> cheers
[22:34:29 CEST] <anoop_r> loglevel is good
[22:34:53 CEST] <anoop_r> but i want to only display info like progress only
[22:35:01 CEST] <llogan> ask in #ffmpeg
[22:35:03 CEST] <anoop_r> is that possibel
[22:35:10 CEST] <anoop_r> ok
[22:35:59 CEST] <anoop_r> does ffmpeg libx265 can encode 10bit output
[22:37:39 CEST] <llogan> wrong channel.
[22:38:40 CEST] <anoop_r> sorry
[22:39:25 CEST] <Daemon404> (yes)|
[22:46:25 CEST] <cone-331> ffmpeg 03Michael Niedermayer 07master:3e34b7498f14: Revert "avformat/rtpenc: check av_packet_get_side_data() return, fix null ptr dereference"
[23:00:10 CEST] <cone-331> ffmpeg 03Nedeljko Babic 07master:1a06d0935fb2: libavutil: Add new fixed dsp functions.
[23:56:42 CEST] <cone-331> ffmpeg 03Anshul Maheshwari 07master:e91debf8f978: fate: Add Closed caption test
[00:00:00 CEST] --- Thu Jun  4 2015