[Ffmpeg-devel-irc] ffmpeg-devel.log.20160622

burek burek021 at gmail.com
Thu Jun 23 02:05:03 CEST 2016

[03:05:32 CEST] <cone-475> ffmpeg 03Michael Niedermayer 07master:e8931d79cc4d: avformat/flvdec: Accept last size if its off by 1
[03:05:32 CEST] <cone-475> ffmpeg 03Michael Niedermayer 07master:3f380b9b40f8: avformat/flvdec: Also print sum_flv_tag_size on size mismatches
[03:05:32 CEST] <cone-475> ffmpeg 03Michael Niedermayer 07master:e0faad837cd5: avformat/flvdec: Do not check last size if it could not be read due to EOF
[10:22:24 CEST] <cone-782> ffmpeg 03Matthieu Bouron 07master:df3484f87c4d: lavc/videotoolbox: fix avcC extradata creation
[10:34:39 CEST] <andrey_turkin> is there any plans on when 3.1 comes out?
[10:34:45 CEST] <nevcairiel> "soon"
[10:35:17 CEST] <andrey_turkin> I gathered that much from initial mail )
[10:35:53 CEST] <andrey_turkin> there was a link to some bugs in trac. If those are blockers then maybe not that soon
[10:35:59 CEST] <ubitux> andrey_turkin: not sure if it's blocking, but we are not in sync with libav yet
[10:36:21 CEST] <ubitux> they suddenly started to push many things
[10:36:21 CEST] <andrey_turkin> that would be nice to have, I guess
[10:36:32 CEST] <ubitux> so i started with 180 commits to merge
[10:36:35 CEST] <ubitux> i merged about 50
[10:36:39 CEST] <ubitux> and now there are 220 to merge
[10:36:40 CEST] <ubitux> :(
[10:36:45 CEST] <nevcairiel> you work backwards
[10:36:59 CEST] <ubitux> :(
[10:37:07 CEST] <andrey_turkin> did you managed to work through those H264 changes?
[10:37:15 CEST] <ubitux> the first wave yes
[10:37:20 CEST] <andrey_turkin> that's cool
[10:37:20 CEST] <ubitux> there are at least 2 more
[10:37:41 CEST] <ubitux> not really
[10:37:45 CEST] <andrey_turkin> IIRC there are also some hwcontext changes which would be really nice to have (device creation at least)
[10:37:50 CEST] <mateo`> nevcairiel: don't make him even sadder :p
[10:38:21 CEST] <andrey_turkin> before making API "stable"
[10:39:10 CEST] <andrey_turkin> and then there are bunch of changes to avconv which seemed pretty good to have. I wonder if someone can manage to port those over to ffmpeg
[10:42:14 CEST] <andrey_turkin> ubitux: what's the last merged libav commit?
[10:43:24 CEST] <ubitux> git log --oneline --reverse ..libav/master for what's left
[10:43:37 CEST] <ubitux> last merged was b7e64fba7f37cc0399beae844f0a5dbef9219376
[10:44:57 CEST] <andrey_turkin> ah. yeah, that's a lot of h264 stuff
[10:45:30 CEST] <ubitux> maybe i'm going to continue merging right now
[10:45:45 CEST] <ubitux> i have better to do but fuck life
[10:46:14 CEST] <andrey_turkin> )
[10:46:39 CEST] <andrey_turkin> you have some nice pile of non-h26* related commits before second wave though. Maybe those are easier
[10:47:15 CEST] <ubitux> some are not difficult but very annoying anyway
[10:47:29 CEST] <andrey_turkin> I bet
[10:47:43 CEST] <ubitux> 41ed7ab45fc693f7d7fc35664c0233f4c32d69bb took me several hours yesterday typically
[10:48:42 CEST] <andrey_turkin> hopefully most commits would be more local than that
[10:49:09 CEST] <ubitux> so, question, i'm merging 763d69bfb2f0094e99f43e657cfd2b0471361f6b
[10:49:30 CEST] <ubitux> is psnr() really supposed to go away in the next deprecation?
[10:49:37 CEST] <ubitux> or we're keeping it?
[10:50:29 CEST] <ubitux> (it doesn't seem so to me)
[10:50:40 CEST] <nevcairiel> don't overthink those merges, this just applies the ifdef to a presumably unused function, if its equally only used in a deprecated block in our code then just add it, if its used in other blocks then don't
[10:50:42 CEST] <andrey_turkin> maybe its only use in avconv was already under same guard?
[10:51:58 CEST] <nevcairiel> looks like the psnr function is used in ffmpeg.c more than in avconv.c, so just dont apply the guard
[10:52:13 CEST] <ubitux> yep
[10:52:19 CEST] <andrey_turkin> it uses other, non-deprecated, means to get same value
[11:02:16 CEST] <omerjerk> I've a init function implemented in one of my C files. I've put av_cold before the function definition. I need to define that function in the header file too. do I need to put av_cold there as well ?
[11:05:59 CEST] <cone-782> ffmpeg 03Diego Biurrun 07master:763d69bfb2f0: Add some more deprecation guards
[11:06:00 CEST] <cone-782> ffmpeg 03Clément BSsch 07master:e6247a22c023: Merge commit '763d69bfb2f0094e99f43e657cfd2b0471361f6b'
[11:12:06 CEST] <cone-782> ffmpeg 03Clément BSsch 07master:0491d6993aef: lavc/error_resilience: remove unused skip_last_mv label
[11:12:07 CEST] <cone-782> ffmpeg 03Clément BSsch 07master:85185578ce30: lavc/qtrleenc: simplify FF_API_CODED_FRAME deprecation guard
[11:12:56 CEST] <ubitux> 257f00e aaaah i hate these commits :(
[11:14:58 CEST] <andrey_turkin> seems straightforward enough. 
[11:15:16 CEST] <andrey_turkin> except you have to basically redo the same thing and not merge
[11:16:23 CEST] <ubitux> yeah
[11:27:38 CEST] <andrey_turkin> well at least you can noop nvenc and avconv patches. that's about 2 dozens of patches not to worry about
[11:29:04 CEST] <ubitux> what's the status about nvenc?
[11:29:30 CEST] <andrey_turkin> should be up to date with somehow recent libav
[11:29:31 CEST] <cone-782> ffmpeg 03Diego Biurrun 07master:257f00ec1ab0: Split global .gitignore file into per-directory files
[11:29:32 CEST] <cone-782> ffmpeg 03Clément BSsch 07master:5558ff3a9f11: Merge commit '257f00ec1ab06a2a161f535036c6512f3fc8e801'
[11:29:46 CEST] <andrey_turkin> I think BtbN was happy with current state of affairs anyway
[11:34:39 CEST] <ubitux> hey
[11:34:50 CEST] <ubitux> no one wants to merge a6a750c7?
[11:37:36 CEST] <andrey_turkin> I could give it a try but not right now
[11:38:06 CEST] <ubitux> ok i'll do it in about 20 min then
[11:47:06 CEST] <j-b> BBB: crash in vp9.c, http://pastebin.com/H4yRJGyS, rings a bell?
[11:52:44 CEST] <BtbN> andrey_turkin, ubitux, yeah, I like the current state of nvenc in ffmpeg, I think most remaining differences should be merged on the libav side of things. At least I'm not aware of anything that's missing in ffmpeg.
[11:53:23 CEST] <ubitux> i'll skip the commits then
[11:54:08 CEST] <andrey_turkin> BtbN: maybe 3399a26d3f57d462e839c0ee51223ae9aca20852 should be merged though?
[11:54:57 CEST] <BtbN> Isn't that patch, from Agatha Hu, long included in ffmpeg? At least I remember merging something like this before.
[11:55:19 CEST] <andrey_turkin> I don't see any uses of avctx->slices
[11:55:21 CEST] <nevcairiel> all those patches appear on our side first
[11:55:29 CEST] <andrey_turkin> sliceModeData is forced to 1
[11:55:50 CEST] <BtbN> http://git.videolan.org/?p=ffmpeg.git;a=commit;h=f1a889737532ad8696574cdb95b844371f74c864
[11:56:30 CEST] <BtbN> that's probably the patch they are referring to.
[11:56:47 CEST] <BtbN> What exactly is the point of using more than one slice with a hardware encoder anyway?
[11:57:01 CEST] <nevcairiel> maybe you need blu-ray compat or something that needs 4 slices
[11:57:08 CEST] <BtbN> I don't have a problem with making it configurable though
[11:58:45 CEST] <nevcairiel> in any case I would say ubitux should just skip all the nvenc patches and if there is something you want specifically, you patch that manually?
[11:59:09 CEST] <BtbN> yep, just no-op them. If something seems interesting that I missed, feel free to give me a ping.
[11:59:37 CEST] <ubitux> alright
[12:27:30 CEST] <cone-782> ffmpeg 03Michael Bradshaw 07master:989a614b707d: libavformat/matroskaenc: omit segment UID for webm
[12:42:31 CEST] <michaelni> andrey_turkin, ubitux about 3.1, i think its better to make it and not wait, there will be 3.2, waiting with 3.1 until 3.2 would otherwise be released would just mean fewer releases
[12:42:51 CEST] <ubitux> can i have pnm in 3.1?
[12:43:04 CEST] <ubitux> i'm waiting for a message from carl
[12:43:26 CEST] <ubitux> michaelni: btw, if we make a release, can we take the time to make it properly with release notes?
[12:43:27 CEST] <michaelni> sure, theres still a odd crash i saw with mixed 3.0 3.1 libs i want to at least understand why it happened
[12:43:37 CEST] <ubitux> instead of like previous time
[12:43:48 CEST] <ubitux> like it was thrown away like a dirty pos
[12:43:57 CEST] <michaelni> it requires someone to write release notes though
[12:44:42 CEST] <ubitux> we need someone to put into words the content of the Changelog
[12:45:09 CEST] <ubitux> i'm sure non-dev ppl can help with that
[12:45:20 CEST] <michaelni> yes
[12:45:30 CEST] <ubitux> also mention what api changes are in
[12:45:39 CEST] <ubitux> and eventually directions
[12:46:06 CEST] <ubitux> i mean i can probably do it but... i'm kind of busy on equally import things for the project
[12:46:23 CEST] Action: ubitux $help_message
[12:46:26 CEST] <michaelni> ubitux, you have more important things to do than the release notes
[12:46:50 CEST] <ubitux> depends on the PoV
[12:47:30 CEST] <ubitux> i'm sure users prefer to read a nice release notes than the project having all tests programs randomly shuffled accross the codebase :P
[12:48:06 CEST] <michaelni> do we have a volunteer to write release notes ?
[12:48:53 CEST] <michaelni> that is anyone reading this, do you have time to write release notes ?
[12:51:13 CEST] <ubitux> Timothy_Gu, llogan, reynaldo, Compn, ... ^
[12:52:09 CEST] <ubitux> ppl who wrote stuff present in the Changelog can tell "stories" too for the writers
[12:52:39 CEST] <ubitux> like "we removed libdcadec because the maintainer is now maintaining it in ffmpeg" or stuff like that
[12:55:44 CEST] <mateo`> speaking of 3.1, i'd like to have the latest mediacodec patchset (improve handling of eos) in it
[12:57:36 CEST] <Compn> yes im sure we can write something
[12:57:38 CEST] Action: Compn runs afk
[13:05:21 CEST] <mateo`> michaelni: when are you planning to make the release ?
[13:06:38 CEST] <michaelni> i think it should be done sooner than later
[13:06:49 CEST] <michaelni> we are already behind shedule
[13:07:39 CEST] <Compn> btw send a mail to llogan, he is not on irc all the time
[13:08:45 CEST] <mateo`> michaelni: by sooner, whatdo you mean ? a few hours / a day / ... ?
[13:09:15 CEST] <michaelni> before weekend hopefully
[13:10:57 CEST] <michaelni> so Thursday or Friday maybe
[13:22:09 CEST] <ubitux> michaelni: #include "time.h"  this should be <time.h>, no?
[13:22:24 CEST] <ubitux> swresample.c sorry
[13:22:34 CEST] <ubitux> libswresample/swresample-test.c
[13:23:02 CEST] <ubitux> or is it libavutil?
[13:25:59 CEST] <nevcairiel> there is no time.h in that folder and it wouldnt find one from libavutil
[13:26:03 CEST] <nevcairiel> so probabl <time.h>
[13:26:40 CEST] <ubitux> my first guess yes
[13:26:44 CEST] <ubitux> ok
[13:47:21 CEST] <cone-782> ffmpeg 03Diego Biurrun 07master:a6a750c7ef24: tests: Move all test programs to a subdirectory
[13:47:22 CEST] <cone-782> ffmpeg 03Clément BSsch 07master:5d48e4eafa6c: Merge commit 'a6a750c7ef240b72ce01e9653343a0ddf247d196'
[13:49:48 CEST] <ubitux> note on this last one: we can probably move many of these tests to tests/api since a large portion of them only use the public api
[13:53:04 CEST] <ubitux> complete list to check: find lib* -name tests -exec find {} -name '*.c' \;
[13:53:42 CEST] <ubitux> oh, libav doesn't have api tests at all, that's why they did it that way
[13:55:24 CEST] <nevcairiel> i think those api tests were build during outreachy or some similar program, not from libav
[13:57:46 CEST] <ubitux> cosmetics merge are the easiest but most annoying commits to merge& :(
[13:57:53 CEST] <ubitux> i spent 2 hours again with the last one
[13:59:43 CEST] <cone-782> ffmpeg 03Clément BSsch 07master:a9eda4b26069: swr: fix time.h include
[14:08:19 CEST] <durandal_1707> kierank: do you know about rtmd fourcc in mp4?
[14:14:57 CEST] <Compn> rtmd ? 
[14:15:06 CEST] <Compn> dont remember that isom/fourcc
[14:15:17 CEST] <Compn> although there are weird .mov tags
[14:21:00 CEST] <cone-782> ffmpeg 03Michael Niedermayer 07master:4fecb97fd36c: avformat/utils: change bit_rate to 64 bit in estimate_timings_from_bit_rate()
[14:21:01 CEST] <cone-782> ffmpeg 03Michael Niedermayer 07master:99c3c3237bb5: avformat/utils: Use internal->avctx for bitrate in estimate_timings_from_bit_rate() if needed
[14:41:14 CEST] <ubitux> i like how there is a label "out" in h264_decode_frame() which is actually... on top of the function :p 
[15:21:04 CEST] <ubitux> durandal_1707: are you trolling CE with the bz2?
[15:21:11 CEST] <ubitux> if not, you can use -D with git
[15:22:02 CEST] <durandal_1707> can't post via elinks some patches
[15:22:09 CEST] <nevcairiel> -D patches cannot be applied, which would probably result in even worse complaints
[15:22:16 CEST] <durandal_1707> so I need to bzippid
[15:23:58 CEST] <durandal_1707> I don't want to start chromium to upload patch, and I have git sm disabled due security
[15:27:23 CEST] <cone-782> ffmpeg 03Clément BSsch 07master:48ac4532d437: lavf/img2dec: add pnm pipe demuxers
[15:37:21 CEST] <ubitux> http://cinemetrics.fredericbrodbeck.de/
[15:57:34 CEST] <fritsch> ubitux: nice one. Do you know about "constant action movie"?
[15:58:36 CEST] <fritsch> it's used by security camera manufacturers, if nothing happens -> videos play fast, if a lot happens -> videos play normal speed. helps to watch lengthy recordings that were done over night.
[15:59:35 CEST] <fritsch> a measure based on this is the optical flow
[15:59:45 CEST] <fritsch> which can also be used to do what that guy does with these graphs
[16:18:51 CEST] <ubitux> we need a filter for that
[16:18:53 CEST] <ubitux> ;)
[16:28:23 CEST] <andrey_turkin> there's gsoc to implement some motion estimation algorithm, right? so... why not? )
[17:17:30 CEST] <kierank> durandal_1707: no
[17:23:22 CEST] Last message repeated 1 time(s).
[17:23:22 CEST] <kierank> oh lol
[17:36:51 CEST] <cone-782> ffmpeg 03Diego Biurrun 07master:11de006babf7: Combine deprecation guards where appropriate
[17:36:52 CEST] <cone-782> ffmpeg 03Clément BSsch 07master:a444a50510e7: Merge commit '11de006babf735aafa3462d43dd2c02bb6ac6e2f'
[17:39:07 CEST] <ubitux> is e62ff72fc1052273deb708ba715f73e5187281d4 really ok?
[17:39:36 CEST] <ubitux> michaelni: opinion on this one ^ ?
[17:42:26 CEST] <ubitux> (it doesn't pass fate)
[17:42:54 CEST] <ubitux> it fails fate-libavcodec-options as it now errors out
[17:52:47 CEST] <ubitux> after this one i'll merge up to dxva
[17:53:13 CEST] <ubitux> or maybe i'll do them as well but i'll need testing
[18:22:27 CEST] <nevcairiel> i can do the dxva one if you want
[18:24:45 CEST] <nevcairiel> not sure we require the timebase, the crash doesnt apply to us
[18:29:24 CEST] <AnonBaite> https://www.sendspace.com/file/e05gkx
[18:29:28 CEST] <AnonBaite> this file is not supported by ffmpe
[18:29:31 CEST] <AnonBaite> *ffmpeg
[18:29:48 CEST] <AnonBaite> [xvag @ 04c67a60] codec 24 is not implemented. Update your FFmpeg version to the newest one from Git. If the problem still occurs, it means that your file has a feature which has not been implemented.
[18:30:10 CEST] <nevcairiel> not all files are, please open an issue on trac if you believe the file is valid and can be played with other software
[18:31:45 CEST] <AnonBaite> well, on trac.ffmpeg.org you need to regisster
[18:32:23 CEST] <AnonBaite> besides, I need to update my ffmpeg windows build
[18:32:33 CEST] <AnonBaite> just to see if the error still exist
[18:34:06 CEST] <AnonBaite> yep, even with the latest windows build I got this error still exists
[18:37:14 CEST] <AnonBaite> I tried to register in trac.ffmpeg.org and I got this:
[18:37:19 CEST] <AnonBaite> Erro: Not Found
[18:37:21 CEST] <AnonBaite> No handler matched request to /register_anonbaiter_two
[18:37:25 CEST] <AnonBaite> am I doing something wrong?
[18:40:20 CEST] <AnonBaite> oh I got it now
[18:40:24 CEST] <BtbN> you failed the bot-check.
[18:42:35 CEST] <AnonBaite> oh
[18:47:42 CEST] <AnonBaite> how can I upload into FTP?
[18:49:22 CEST] <BtbN> upload via ftp to upload.ffmpeg.org/incoming
[18:49:52 CEST] <AnonBaite> are there any pre-requesites?
[18:50:02 CEST] <BtbN> an ftp client helps a lot.
[18:50:11 CEST] <AnonBaite> ftp client?
[18:51:23 CEST] <AnonBaite> FileZilla?
[18:52:05 CEST] <cone-782> ffmpeg 03Michael Niedermayer 07master:70d48accd855: avcodec/tests/options: Set timebase
[18:53:00 CEST] <AnonBaite> can I upload files with FileZilla Client?
[18:53:09 CEST] <BtbN> any client is fine
[18:53:19 CEST] <michaelni> ubitux, that commit should not fail our tests anymore, if you want to merge it
[18:57:01 CEST] <AnonBaite> is there any password to get there?
[18:57:06 CEST] <AnonBaite> I mean, to the ftp server
[18:57:06 CEST] <AnonBaite> ?
[19:06:01 CEST] <durandal_1707> AnonBaite: anonymous?
[19:09:01 CEST] <AnonBaite> okay
[19:12:18 CEST] <AnonBaite> it didn't work for me
[19:12:27 CEST] <AnonBaite> I logged in using anonymous
[19:12:37 CEST] <AnonBaite> I tried to connect to the host "upload.ffmpeg.org"
[19:12:55 CEST] <AnonBaite> and instead it's not connected to me
[19:13:06 CEST] <durandal_1707> perhaps its down
[19:13:35 CEST] <durandal_1707> just upload somewhere and post link here
[19:13:57 CEST] <AnonBaite> I already did
[19:13:58 CEST] <AnonBaite> https://www.sendspace.com/file/e05gkx
[19:16:03 CEST] <AnonBaite> wait
[19:16:11 CEST] <AnonBaite> the filename is not describing the situation
[19:16:46 CEST] <durandal_1707> I assume vgmstream decodes files fine?
[19:17:03 CEST] <AnonBaite> except not really
[19:17:16 CEST] <AnonBaite> you have to make a few steps to make it work
[19:18:10 CEST] <AnonBaite> besides, vgmstream can play .xvag files that are mono/stereo, but with .xvag files that has more than 4 channels...
[19:18:46 CEST] <durandal_1707> you can decode with ffmpeg -f mp3 -i ...
[19:18:56 CEST] <AnonBaite> it doesn't play very well
[19:21:18 CEST] <ubitux> michaelni: cool, thanks :)
[19:21:49 CEST] <AnonBaite> huh
[19:22:44 CEST] <durandal_1707> AnonBaite: what's wrong?
[19:23:06 CEST] <AnonBaite> the decoding result doesn't sound very good to me
[19:23:17 CEST] <AnonBaite> I mean, the file is supposed to be 4 channels, right?
[19:24:26 CEST] <durandal_1707> hmm, the only way to get extra channels is by postprocessing
[19:24:42 CEST] <AnonBaite> post processing?
[19:29:28 CEST] <pfelt> a bit off topic, but a bit on.  based on the most recent email in my inbox on the -devel list, where is the "official" git repo?  I've seen the FFmpeg one on github and one on vlc's servers.   
[19:30:26 CEST] <AnonBaite> by the way, how can I do post processing?
[19:31:36 CEST] <cone-782> ffmpeg 03Anton Khirnov 07master:e62ff72fc105: lavc: make avcodec_open2() fail when the timebase is not set for encoding
[19:31:37 CEST] <cone-782> ffmpeg 03Clément BSsch 07master:d4b1b3b06c21: Merge commit 'e62ff72fc1052273deb708ba715f73e5187281d4'
[19:33:27 CEST] <cone-782> ffmpeg 03Anton Khirnov 07master:8996515b137f: avpacket: fix setting AVPacket.data in av_packet_ref()
[19:33:28 CEST] <cone-782> ffmpeg 03Clément BSsch 07master:17fdffc9f80e: Merge commit '8996515b137f962e3a8691241c285c96437b6c53'
[19:35:58 CEST] <jkqxz> pfelt:  The official repo is source.ffmpeg.org/ffmpeg.  It is hosted by videolan.
[19:37:12 CEST] <AnonBaite> durandal_1707: do you mean postprocessing from the ffmpeg itself or...?
[19:40:59 CEST] <durandal_1707> AnonBaite: yes
[19:42:58 CEST] <AnonBaite> but how?
[19:43:05 CEST] <AnonBaite> do I need to use some filters or something?
[19:43:23 CEST] <cone-782> ffmpeg 03Henrik Gramner 07master:ba3eb745cc23: x86inc: Fix AVX emulation of scalar float instructions
[19:43:24 CEST] <cone-782> ffmpeg 03Anton Mitrofanov 07master:8d02579fae75: x86inc: Fix AVX emulation of some instructions
[19:43:25 CEST] <cone-782> ffmpeg 03Anton Mitrofanov 07master:300fb0df84e9: x86inc: Improve handling of %ifid with multi-token parameters
[19:43:26 CEST] <cone-782> ffmpeg 03Anton Mitrofanov 07master:2fb1d17a5a6b: x86inc: Enable AVX emulation in additional cases
[19:43:27 CEST] <cone-782> ffmpeg 03Clément BSsch 07master:a24cbbbfc413: Merge commit 'ba3eb745cc236765c916b74b39c1874da5c3aefc'
[19:43:28 CEST] <cone-782> ffmpeg 03Clément BSsch 07master:3d1b9383b711: Merge commit '8d02579fae756336ae2a88d521e8cf2f6b436a2f'
[19:43:29 CEST] <cone-782> ffmpeg 03Clément BSsch 07master:1c85027afe8d: Merge commit '300fb0df84e9a563ed3b8a9b3d365abb6cf34600'
[19:43:30 CEST] <cone-782> ffmpeg 03Clément BSsch 07master:93f5a27313af: Merge commit '2fb1d17a5a6b6ff8da2434cde0fda821f06f608c'
[19:51:52 CEST] <AnonBaite> durandal_1707: can you at least give me some pointers in how should I do post-processing on ffmpeg?
[19:54:40 CEST] <cone-782> ffmpeg 03Anton Khirnov 07master:31f6a4b4b83a: lavc: mark the old audio/video encoding API as deprecated
[19:54:41 CEST] <cone-782> ffmpeg 03Clément BSsch 07master:541e1c2c3b7c: Merge commit '31f6a4b4b83aca1d73f3cfc99ce2b39331970bf3'
[19:57:51 CEST] <cone-782> ffmpeg 03Alexandra Hájková 07master:e4b38878da6b: Move check_marker() from get_bits to mpeg4videodec
[19:57:52 CEST] <cone-782> ffmpeg 03Clément BSsch 07master:575be7311cff: Merge commit 'e4b38878da6b38ef5ca54ce14be985d6ae088834'
[20:00:19 CEST] <durandal_1707> AnonBaite: I don't think it currently possible, besides how would you know how much big is each frame?
[20:01:32 CEST] <AnonBaite> I guess I don't
[20:09:35 CEST] <cone-782> ffmpeg 03Alexandra Hájková 07master:d338abb664fe: mpeg4videodec: Call av_log() in check_marker() with AVCodecContext instead of NULL
[20:09:36 CEST] <cone-782> ffmpeg 03Clément BSsch 07master:7475a9351f57: Merge commit 'd338abb664febbc2c7266af7818aab1f12dbc161'
[20:09:37 CEST] <cone-782> ffmpeg 03Clément BSsch 07master:6c98398b0be7: lavc/get_bits: add a logging context to check_marker()
[20:10:18 CEST] <ubitux> nevcairiel: so do you want to do the dxva2 commits or i do them in a branch now and you test?
[20:13:35 CEST] <nevcairiel> i can do them later tonight, i wrote that code afterall
[21:13:21 CEST] <cone-782> ffmpeg 03Michael Niedermayer 07master:b0b3676e136a: avcodec/flac_parser: Raise threshold for detecting invalid data
[21:13:22 CEST] <cone-782> ffmpeg 03Michael Niedermayer 07master:d0b21b28a3d3: avformat/flacdec: Fix seeking close to EOF
[22:31:07 CEST] <ubitux> nevcairiel: if it's not too late i'll continue the merging just after the dxva2 
[22:50:07 CEST] <cone-782> ffmpeg 03Anton Khirnov 07master:c46db38cde8e: hwcontext: add a dxva2 implementation
[22:50:08 CEST] <cone-782> ffmpeg 03Hendrik Leppkes 07master:3689efea3d59: Merge commit 'c46db38cde8e8fd8ecb1c6602f10ec0e002f29a8'
[22:51:31 CEST] <nevcairiel> ubitux: unfortunately merging the next one might take a while since the libav versiond oesnt support 10-bit decoding
[22:55:13 CEST] <ubitux> how is it a problem?
[22:55:24 CEST] <nevcairiel> need to add 10-bit support to the hwcontext module
[22:59:42 CEST] <ubitux> i don't understand yet, but can i help?
[23:04:39 CEST] <nevcairiel> nah, will be done shortly i hope
[23:08:07 CEST] <cone-782> ffmpeg 03Petru Rares Sincraian 07master:610c0770e3d1: fate: add test for amerge
[23:29:06 CEST] <cone-782> ffmpeg 03Hendrik Leppkes 07master:0c312a45629f: hwcontext_dxva2: support P010 10-bit surfaces
[23:30:13 CEST] <ubitux> nevcairiel: nice, but why was it blocking?
[23:30:30 CEST] <ubitux> like, couldn't error out saying "not implemented yet" in case of 10-bit?
[23:30:40 CEST] <nevcairiel> but why would I introduce a regression
[23:30:57 CEST] <nevcairiel> (the next merge needs it)
[23:30:57 CEST] <ubitux> isn't it a new feature?
[23:31:18 CEST] <nevcairiel> the next merge moves existing code to use the new hwcontext thing
[23:31:28 CEST] <nevcairiel> but the old code supports 10-bit, hwcontext didnt
[23:31:28 CEST] <ubitux> aah wrt to old dxva ok
[23:31:36 CEST] <ubitux> get it, sorry, thx
[23:38:15 CEST] <j-b> So, P010 is the default 10bits DxVA output?
[23:38:19 CEST] <nevcairiel> yes
[23:39:06 CEST] <cone-782> ffmpeg 03Anton Khirnov 07master:b23613268c6b: avconv_dxva2: use the hwcontext API
[23:39:07 CEST] <cone-782> ffmpeg 03Hendrik Leppkes 07master:e799e2270cb8: Merge commit 'b23613268c6b56a8f9de7859562d82b4b88353d9'
[23:40:15 CEST] <ubitux> nevcairiel: thx!
[23:40:21 CEST] <ubitux> can i go back to merging?
[23:40:32 CEST] <nevcairiel> sure
[23:40:41 CEST] <nevcairiel> let me know when you need a break, i could do some as well
[23:41:06 CEST] <ubitux> it's ok for now :)
[23:57:11 CEST] <iive> it's kind of news to me that there is already 10bit hardware decoding support
[23:58:36 CEST] <nevcairiel> has been for a while for hevc now
[23:58:44 CEST] <nevcairiel> vp9 will likely follow once new hardware is there
[00:00:00 CEST] --- Thu Jun 23 2016

More information about the Ffmpeg-devel-irc mailing list