18:43 <michaelni> :)

18:43 <michaelni> it works: https://libera.irclog.whitequark.org/ffmpeg-devel/2023-10-11

18:44 <JEEB> them logs

18:44 <JEEB> when was it last that we had a publicly available logger here

18:44 <Lynne> 2019

18:44 <Lynne> july

18:45 <JEEB> :)

18:45 <durandal_1707> ban the logs!

18:45 <durandal_1707> kick the logger!

18:47 <Lynne> everything you say will be used to train a markov chain bot to provide random quotes when prompted

18:52 <jamrial> ok cool

18:52 <jamrial> and it updates in real time too

18:54 <cone-512> ffmpeg John Mather master:7251dfdcee2f: avcodec/libkvazaar: Bump minimum version to 2.0.0

18:54 <cone-512> ffmpeg John Mather master:a2175ca8615d: avcodec/libkvazaar: Respect codec context color settings.

18:54 <JEEB> ugh... forgot one thing of course. should have adjusted the initial letter to lowercase .-.

18:59 <JEEB> not sure if that version check bump should be backported, that's been broken for quite a while :D

19:00 <JEEB> granted, it is telling that no-one seems to have noticed

19:14 <mkver> A logger! Is it christmas!?

19:19 <Traneptora> LogChamp

20:38 <cone-512> ffmpeg Andreas Rheinhardt master:c995311bcfda: avcodec/h261dec: Don't set write-only picture_number

20:38 <cone-512> ffmpeg Andreas Rheinhardt master:b561dafd561c: avcodec/h261dec: Discard whole packet when discarding

20:38 <cone-512> ffmpeg Andreas Rheinhardt master:ab95338a20b2: avcodec/mpeg4video_parser: Don't set write-only current_picture_ptr

21:00 <haasn> 20:10 <elenril> clearly runtime kernel generation is the way to go <- works for libplacebo

21:16 <Son_Goku> hey michaelni 👋

22:01 <durandal_1707> Lynne: cant get correct number from that pdf

22:02 <Lynne> durandal_1707: what's wrong with the output?

22:02 <Lynne> suboptimal, or?

22:03 <durandal_1707> Lynne: wrong number for angle, sin()/cos() of angle should give 0.5 for R = 2 * L

22:03 <durandal_1707> but it gives here 0.2

22:04 <durandal_1707> it should give angle, so when R/L is rotated it will be M/S with M max energy and S lowest

22:14 <Lynne> do you compute the input to the equation as they do?

22:15 <durandal_1707> yes

23:01 <Lynne> durandal_1707: does it give expected results for full and no correlation?

23:01 <durandal_1707> Lynne: i think i get correct result once i multiply sin(angle)/cos(angle) with M_SQRT2

23:36 <Lynne> they must've forgotten to specify normalization then

23:36 <durandal_1707> already calculate mean