[Ffmpeg-devel-irc] ffmpeg-devel.log.20190401

burek burek021 at gmail.com
Tue Apr 2 03:05:04 EEST 2019


[00:25:55 CEST] <cone-843> ffmpeg 03Lynne 07master:0cea3ca8943b: opusdsp: create and move deemphasis and postfiltering from opus_celt
[00:25:56 CEST] <cone-843> ffmpeg 03Lynne 07master:605e3303109e: x86/opusdsp: implement FMA3 accelerated postfilter and deemphasis
[01:02:12 CEST] <cone-843> ffmpeg 03Michael Niedermayer 07release/4.1:cb4768e7f20b: avcodec/truemotion2: Fix integer overflow in tm2_null_res_block()
[01:02:13 CEST] <cone-843> ffmpeg 03Michael Niedermayer 07release/4.1:ad0f4a7d10f9: avformat/mov: Fix potential integer overflow in entry check in mov_read_trun()
[01:02:14 CEST] <cone-843> ffmpeg 03Michael Niedermayer 07release/4.1:65f94b732a27: avcodec/mpegpicture: Check size of edge_emu_buffer
[01:02:15 CEST] <cone-843> ffmpeg 03Carl Eugen Hoyos 07release/4.1:daca529112d8: lavc/bmp: Avoid a heap buffer overwrite for 1bpp input.
[01:02:16 CEST] <cone-843> ffmpeg 03Michael Niedermayer 07release/4.1:f1ecebcdb72b: avcodec/hevcdec: Avoid only partly skiping duplicate first slices
[01:02:17 CEST] <cone-843> ffmpeg 03Michael Niedermayer 07release/4.1:1d720b37f0f3: Update for 4.1.3
[01:17:03 CEST] <cone-843> ffmpeg 03James Almer 07master:27c94c57dc84: avformat/movenc: free eac3 private data only when closing the stream
[01:22:39 CEST] <jamrial> michaelni: can i backport the above commit to the 4.1 branch, or are you in the middle of testing/tagging?
[01:36:00 CEST] <michaelni> jamrial, feel free to back port, i intend to do the release after sleeping
[01:36:09 CEST] <jamrial> ok, thanks
[01:37:18 CEST] <cone-843> ffmpeg 03James Almer 07release/4.1:58cd70201edc: avformat/movenc: free eac3 private data only when closing the stream
[02:21:01 CEST] <taliho> Hello, I'm running FATE regression tests, but have an error "Test api-seek failed. Look at tests/data/fate/api-seek.err for details."
[02:22:03 CEST] <taliho> The error message in the file is "Can't open file". Do I have download any test files separately, or is there a ./configure option I missed? 
[02:23:11 CEST] <jamrial> taliho: make fate-rsync will download the required samples to the folder you defined during configure with the --samples= option
[02:23:55 CEST] <jamrial> if you didn't use that option during configure, then just run make fate-rsync and make fate with the SAMPLES= option, set to the same folder
[02:27:37 CEST] <taliho> thanks jamrial
[10:32:24 CEST] <cone-536> ffmpeg 03Michael Niedermayer 07master:3a0ec1511e70: avcodec/rscc: Check that the to be uncompressed input is large enough
[10:39:27 CEST] <cone-536> ffmpeg 03Gyan Doshi 07master:7f9a81b11056: doc/general: make column key note more verbose
[11:41:13 CEST] <thardin> is anyone else getting NAB spam?
[11:42:39 CEST] <JEEB> the one about getting user lists or something? yea
[11:43:03 CEST] <thardin> I guess they farm the mailing list then
[11:43:21 CEST] <kierank> thardin: yes a lot
[11:43:24 CEST] <thardin> I got some IBC spam too, but I've been suspecting that's because it's linked to my old work email or something
[11:43:45 CEST] <kierank> they farm MLs and ring me asking to speak to jdarnley_obs 
[11:43:51 CEST] <kierank> even though he has nothing to do with expos
[11:44:11 CEST] <J_Darnley> :) my condolences
[11:44:24 CEST] <thardin> heh
[11:45:48 CEST] <thardin> on an unrelated note, does anyone in here have experience with aaf export in protools?
[11:53:07 CEST] <BradleyS> thardin: maybe
[11:53:27 CEST] <BradleyS> what's the issue
[11:54:06 CEST] <thardin> it seems mute automation doesn't get exported, and I'm wondering if there's any case where it would
[11:54:30 CEST] <thardin> while volume does get exported
[11:54:40 CEST] <BradleyS> hmm, not sure about that
[11:55:19 CEST] <BradleyS> noise gate instead?
[11:55:36 CEST] <BradleyS> are volume envelopes exported?
[11:56:34 CEST] <BradleyS> also, if you're not aware, this is very very useful software (no affiliation): https://www.aatranslator.com.au
[11:56:56 CEST] <thardin> yeah, volume is exported
[11:57:04 CEST] <thardin> like the control point stuff
[11:58:01 CEST] <thardin> kAAFParameterDef_Amplitude
[11:59:02 CEST] <thardin> bookmarking aatranslator, thanks for the tip
[11:59:40 CEST] <BradleyS> looks like the mute automation is a known gray area
[11:59:51 CEST] <BradleyS> if a track is fully muted it's omitted
[12:00:21 CEST] <BradleyS> otherwise mute points can still be missing, advice is to use clip gain / volume envelopes instead
[12:00:38 CEST] <thardin> yeah that's what I've gathered so far
[12:00:43 CEST] <BradleyS> sucks
[12:00:58 CEST] <thardin> could protools be scripted to convert mute automation to volume?
[12:01:24 CEST] <BradleyS> good question, sadly i never used mute automation much
[12:33:16 CEST] <cone-536> ffmpeg 03Michael Niedermayer 07release/4.1:6c75df556f8c: avcodec/rscc: Check that the to be uncompressed input is large enough
[12:33:17 CEST] <cone-536> ffmpeg 03Michael Niedermayer 07release/4.1:4154f8967820: Changelog: update
[13:24:31 CEST] <cone-536> ffmpeg 03Linjie Fu 07master:a8355eed3699: lavc/qsvenc: expose low_power as a common option for QSV encoder
[13:24:32 CEST] <cone-536> ffmpeg 03Zhong Li 07master:d071a1117daf: lavc/qsvenc: get vps extradata from MSDK
[13:59:39 CEST] <cone-536> ffmpeg 03Zhong Li 07n4.1.3:HEAD: lavc/qsvenc: get vps extradata from MSDK
[14:28:34 CEST] <BBB> so are the matrox people just going to resubmit their patch in a new thread until we forget to say it's closed source software, and then quickly apply it and now it can't be removed anymore
[14:29:02 CEST] <thardin> probably
[14:29:56 CEST] <thardin> they didn't seem too receptive to the idea that they are, in fact, a hardware company
[14:59:55 CEST] <kierank> thardin: the same with all of them
[15:00:02 CEST] <kierank> software is just this annoying burden for them
[15:01:08 CEST] <thardin> there's win-win potential there, if there were only willing to see it
[15:02:18 CEST] <thardin> RE: maybe having a vote, that heliosvoting requires being used by either google or facebook
[16:56:12 CEST] <cone-536> ffmpeg 03Lynne 07master:b43b8d337db0: x86/opusdsp: fix WIN64 return value
[17:05:32 CEST] <thebombzen> I sent a patch to the mailing list last october, it had issues, and I resent the fixed version a few days later and didn't get a response. I resent the patch in January and also didn't get a response. should I send it again? I don't really want to spam
[17:05:43 CEST] <thebombzen> this is the patch, FWIW: https://0x0.st/zXg0.patch
[17:05:49 CEST] <JEEB> you should ping it
[17:06:09 CEST] <JEEB> it's the ffmpeg.c frame rate for remux thing, right?
[17:06:48 CEST] <thebombzen> yes
[17:06:53 CEST] <thebombzen> so resend it again?
[17:06:55 CEST] <JEEB> no
[17:07:03 CEST] <JEEB> just reply to the thread with a ping
[17:07:11 CEST] <thebombzen> wdym "a ping"
[17:07:34 CEST] <thebombzen> unless you just mean "bump it"
[17:07:56 CEST] <JEEB> literally that. "ping? requesting reviews because I feel like being able to set a frame rate when remuxing is worthwhile"
[17:13:32 CEST] <thebombzen> done
[19:59:18 CEST] <cone-536> ffmpeg 03Gyan Doshi 07master:8eca42e6320e: lavf/mov: skip extradata check in esds for MPEG-1/2 audio
[20:04:17 CEST] <Fenrirthviti> Hey guys, does anyone know offhand of FFmpeg implements their own bitrate padding for NVENC CBR modes?
[20:04:28 CEST] <Fenrirthviti> offhand if, that is.
[20:05:41 CEST] <JEEB> I'd say no. the encoder should do any padding if it feels like it needs such :P
[20:06:19 CEST] <Fenrirthviti> Trying to diagnose what's going on. Noticing some odd bitrate fluctuations with CBR where low/no-motion video the bitrate tanks.
[20:06:54 CEST] <Fenrirthviti> I'm being told that this is expected by NVIDIA, but that seems... wrong for a CBR mode.
[20:07:21 CEST] <Fenrirthviti> Was curious if anyone had run in to that before.
[20:08:31 CEST] <Fenrirthviti> Found a post from BtbN back in 2014 that seems to corroborate that they don't use any kind of bitrate padding. Seems odd to call it CBR in that instance. Frustrating :\
[20:10:08 CEST] <nevcairiel> are you sure you a re setting cbr mode properly? In my recent tests, the bitrate is always spot on, in such a way that i can only achieve with padding in x264 etc
[20:10:50 CEST] <Fenrirthviti> Yes, quite certain. It doesn't happen in situations with medium/high motion.
[20:11:11 CEST] <Fenrirthviti> If you encode, say, a still image or extremely simple animation with very little motion
[20:11:46 CEST] <Fenrirthviti> Bitrate will fall close to 0
[20:12:55 CEST] <Fenrirthviti> Context here is latest OBS Studio, with the nvenc update we did. Happens in FFmpeg as well with default options.
[20:13:28 CEST] <Fenrirthviti> Just factfinding to see if I can get more details on what's going on.
[20:55:28 CEST] <BtbN> nvenc does automatically insert padding in CBR mode
[21:23:55 CEST] <Fenrirthviti> Hm, their engineers are saying they don't.
[21:24:10 CEST] <Fenrirthviti> I'll check again.
[00:00:00 CEST] --- Tue Apr  2 2019


More information about the Ffmpeg-devel-irc mailing list