[Ffmpeg-devel-irc] ffmpeg-devel.log.20191119

[burek]

[00:01:04 CET] <ePirat> anyway I just tried to point out a possible quick solution, nevermind if you already knew about it :D
[00:01:28 CET] <BtbN> I'm aware of OBS being able to do it, but it needs to run as a Windows service, and "just work"
[00:01:37 CET] <BtbN> Since it'll be installed in PCs which are given to players
[00:01:53 CET] <ePirat> uh that sound tricky
[00:02:21 CET] <ePirat> given how tricky doing that even in OBS can be, and screen capture is kinda their primary focus :D
[00:02:50 CET] <BtbN> The ffmpeg desktop duplication thing looks to be in terrible shape btw.
[00:02:58 CET] <BtbN> it's kinda a mess of comments and commented out code
[00:03:30 CET] <BtbN> The reason for all this being that people want to play in 120Hz and more, but the entire video processing chain only does 60Hz, and has no idea about DisplayPort
[00:04:02 CET] <BtbN> So the PC needs some secondary output that just outputs the primary, and in our case only, screen as 60Hz over SDI
[00:17:24 CET] <cone-898> ffmpeg 03Paul B Mahol 07master:6c2f8663090c: avfilter/vf_bm3d: improve threshold scaling with different block_size and depth
[01:30:35 CET] <cone-898> ffmpeg 03James Almer 07master:4e2bef6a82b3: avcodec/cbs_av1: keep separate reference frame state for reading and writing
[01:30:49 CET] <jamrial> BBB: ^
[01:30:56 CET] <jamrial> the switch frame sample should remux fine now
[01:31:16 CET] <BBB> woohoo, cool
[07:38:53 CET] <cone-055> ffmpeg 03Gyan Doshi 07master:2ff444bd3ade: avfilter/Makefile: add missing dependency for scale_cuda
[09:37:51 CET] <cone-055> ffmpeg 03Gyan Doshi 07master:6e0461d8d446: doc/filters: correct psnr example
[09:39:22 CET] <cone-055> ffmpeg 03Gyan Doshi 07master:f394d7b382aa: doc/filters: correct ssim example
[09:40:30 CET] <cone-055> ffmpeg 03Gyan Doshi 07master:0321bde0a2e4: doc/filters: correct libvmaf example
[09:41:10 CET] <cone-055> ffmpeg 03Gyan Doshi 07master:ef479ee660e4: avfilter/Makefile: add missing dependency for chromashift
[09:41:43 CET] <cone-055> ffmpeg 03Gyan Doshi 07master:0cfda90b3484: avfilter/Makefile: add missing dependency for lut3d
[10:37:53 CET] <j-b> 'morning
[10:39:11 CET] <cone-055> ffmpeg 03Paul B Mahol 07master:c8f269f24f99: avfilter/vf_chromashift: remove unused header
[10:41:06 CET] <Lynne> I guess work doesn't wait for your jetlag to pass
[11:17:33 CET] <kierank> I have been waking up at 3am every day
[11:17:43 CET] <kierank> drank a lot of red bull so i didn't fall asleep at 7
[11:28:51 CET] <j-b> rcombs: ping
[11:50:34 CET] <j-b> is Nicolas on IRC?
[11:52:17 CET] Action: jdarnley does not know
[11:52:54 CET] <durandal_1707> Nicolas is always on IRC
[12:06:30 CET] <JEEB> j-b: I think he has at least previously noted that he dislikes IRC
[12:06:42 CET] <JEEB> so I would default him to only being on the ML
[12:06:55 CET] <j-b> ok
[12:23:15 CET] <cone-055> ffmpeg 03Paul B Mahol 07master:5ed6b735ab8b: avfilter/vf_blend: cosmetics: reindent
[12:37:45 CET] <cone-055> ffmpeg 03Paul B Mahol 07master:18d25ecede6a: avfilter/vf_weave: pal and hwaccel formats are not supported
[12:45:49 CET] <cone-055> ffmpeg 03Paul B Mahol 07master:258f66998fa9: avfilter/vf_deblock: add 12bit yuva formats
[12:52:35 CET] <kierank> durandal_1707: stop trolling
[12:52:48 CET] <durandal_1707> kierank: stop trolling
[13:15:02 CET] <cone-055> ffmpeg 03Paul B Mahol 07master:804fce8bc24f: avfilter/vf_midequalizer: add 16bit formats
[13:27:39 CET] <rcombs> j-b: hi
[13:27:47 CET] <rcombs> this about voting?
[15:49:04 CET] <j-b> rcombs: yes
[15:53:38 CET] <durandal_1707> voting should not be hidden
[15:54:27 CET] <j-b> nothing is hidden
[15:55:24 CET] <durandal_1707> vote cast should not be hidden
[15:55:55 CET] <j-b> like a normal election in a democratic country.
[15:56:24 CET] <durandal_1707> normal elections in democratic country are full of lies
[15:57:53 CET] <j-b> Yeah, a good dictature is always what we need
[15:59:47 CET] <durandal_1707> nothing is more transparent when everybody can see casted vote
[16:05:21 CET] <BBB> durandal_1707: feel free to propose that in the next meeting. in this meeting, it was decided by the group to vote like this, whether you like it or not
[16:06:15 CET] <j-b> If people don't come in person, how can they see the casted vote?
[16:07:49 CET] <rcombs> are we talking about publishing the ballots, or publishing who voted how?
[16:08:24 CET] <rcombs> publishing the ballots (or equivalent data) is excellent policy and I'd be surprised if anyone objected
[16:08:49 CET] <j-b> of course, publishing the ballots is normal
[16:09:04 CET] <rcombs> (most online setups for this kind of thing support it)
[16:09:13 CET] <rcombs> publishing anything associating a voter with their ballot is usually frowned upon, though
[16:09:45 CET] <rcombs> the only real advantage is that it provides proof that your ballot was, in fact, among those counted
[16:10:12 CET] <rcombs> but it means that you could potentially feel social pressure to vote in a particular way
[16:10:27 CET] <j-b> "potentially"
[16:10:41 CET] <rcombs> there's been some work to try to come up with a system that allows you to verify with certainty that your vote was counted without having to publicly associate yourself with it
[16:10:59 CET] <j-b> with a blockchain? :)
[16:11:03 CET] <rcombs> but none of those systems are particularly practical
[16:11:09 CET] <rcombs> and yes some involve that
[16:11:52 CET] <rcombs> as long as you trust whoever's running the software, it should be fine not to associate names with ballots
[16:12:22 CET] <rcombs> this isn't a safe assumption to make in, like, a political election, where great power is at stake and there's major incentive for corruption
[16:12:40 CET] <rcombs> but for an open-source project I would hope that it's fine
[16:56:51 CET] <durandal_1707> ubitux: are you against adding >8 bit support to nlmeans?
[16:57:22 CET] <JEEB> why would he be *against* it?
[17:00:12 CET] <durandal_1707> dunno, developers are weird birds
[17:00:19 CET] <ubitux> durandal_1707: JEEB: might be tricky to handle overflows if with the 32-bit integral
[17:00:42 CET] <ubitux> but sure go for it
[19:09:39 CET] <cone-055> ffmpeg 03James Almer 07master:f18a5efb5222: fate/cbs: add initial AV1 tests
[19:26:04 CET] <jamrial> michaelni: good news, i can upload samples with rsync now
[19:28:11 CET] <JEEB> nice
[19:28:15 CET] <JEEB> validation \o/
[19:41:04 CET] <kurosu> lol, connecting from this old client, last message (9 months ago) sent: "Justin Ruggles? Now that's a name I haven't heard in a long time"
[19:45:42 CET] <JEEB> :D
[19:47:23 CET] <michaelni> jamrial, great, just curious, what was the problem?
[19:49:12 CET] <jamrial> i have no idea, but i suspect it was an issue with the network interface from vmware or similar, becasue once i updated it the problem went away, despite the arch linux image being the same
[19:50:01 CET] <jamrial> i could ssh from windows just fine, it was only failing within the vm
[20:48:15 CET] <kurosu> "Note that some people who voted for this method are de facto excluded from voting because of this very method :)"
[20:48:22 CET] Action: kurosu high-fives JEEB
[20:49:27 CET] <durandal_1707> stop nonsense
[20:59:48 CET] <kurosu> actually, among the voters, kswanson, Thilo, kierank, jannau and lotharkript (it seems, and at least) were voted out according to the new rule
[21:00:53 CET] <kierank> Yes
[21:02:01 CET] <durandal_1707> so they can not vote at fosdem
[21:02:11 CET] <durandal_1707> this was on purpose
[21:03:10 CET] <BBB> it was the kgb
[21:03:16 CET] <j-b> You mean FSB
[21:03:29 CET] <j-b> kurosu: yes, exactly, I remember your case.
[21:03:44 CET] <BBB> kgb, fsb, cia, fbi, vp9
[21:03:46 CET] <BBB> all the same
[21:03:49 CET] <j-b> vp10
[21:04:03 CET] <kierank> av1
[21:04:06 CET] <kierank> av11
[21:04:12 CET] <durandal_1707> dav1d aka mosad
[21:04:20 CET] <BBB> o_O
[21:04:21 CET] <j-b> stop drinking
[21:04:37 CET] <BBB> that went downhill really quickly
[21:05:50 CET] <BradleyS> don't forget rle
[21:06:04 CET] <kurosu> What case? Anyway, more than half of the voters basically relinquished voting rights to other/new people
[21:06:35 CET] <kurosu> or stasi
[21:07:29 CET] <kurosu> nowadays, all these go by the cone handle
[21:07:34 CET] <BradleyS> snow, snowden, it's a conspiracy
[21:07:39 CET] <kurosu> (all hail our irc bot overlords)
[21:11:29 CET] <durandal_1707> you should all only maintain ansi codec, that is all your top reach
[22:49:58 CET] <Illya> j-b: thank you for your summary of the meeting on the mailing list
[22:59:50 CET] <BBB> pandora :-p
[23:00:10 CET] <BBB> that's kinda fun
[23:01:28 CET] <durandal_1707> wrong channel
[23:37:12 CET] <jamrial> BtbN: fun, the amf encoder was doing the same as nvenc :p
[23:39:25 CET] <BtbN> Wonder if that's pure coincidence or if there is shared code
[23:41:12 CET] <jamrial> coincidence, it was also using ff_alloc_packet2
[23:45:06 CET] <cone-157> ffmpeg 03James Almer 07master:fdf46b4a6b36: avcodec/amfnec: allocate packets using av_new_packet()
[23:45:42 CET] <BtbN> So it did not have the kinda unique situation of supporting encode2 and send/recv
[23:46:33 CET] <jamrial> no, only receive_packet in amf's case
[23:46:59 CET] <BtbN> I actually think nvenc is the only encoder doing that
[23:51:15 CET] <cone-157> ffmpeg 03James Almer 07release/4.2:57365f67a070: avcodec/amfnec: allocate packets using av_new_packet()
[23:51:52 CET] <cone-157> ffmpeg 03James Almer 07release/4.1:08d3cc2f1daf: avcodec/amfnec: allocate packets using av_new_packet()
[00:00:00 CET] --- Wed Nov 20 2019