michaelni changed the topic of #ffmpeg-devel to: Welcome to the FFmpeg development channel | Questions about using FFmpeg or developing with libav* libs should be asked in #ffmpeg | This channel is publicly logged | FFmpeg 7.1 has been released! | Please read ffmpeg.org/developer.html#Code-of-conduct
<michaelni> haasn, I think i have reviewed all pending sws patches, if i missed one, please ping it
Samillion has quit [Ping timeout: 244 seconds]
<haasn> michaelni: huge thanks
<haasn> Will merge/address them tomorrow
Samillion has joined #ffmpeg-devel
arch1t3cht4 has joined #ffmpeg-devel
iive has quit [Quit: They came for me...]
arch1t3cht has quit [Ping timeout: 265 seconds]
arch1t3cht4 is now known as arch1t3cht
thilo has quit [Ping timeout: 265 seconds]
thilo has joined #ffmpeg-devel
^Neo has quit [Ping timeout: 265 seconds]
zenmov has quit [Quit: Lost terminal]
<fflogger> [editedticket] 0x20z: Ticket #11362 ([avcodec] memory leaks error at __interceptor_malloc) updated https://trac.ffmpeg.org/ticket/11362#comment:3
Guest84 has joined #ffmpeg-devel
^Neo has joined #ffmpeg-devel
^Neo has quit [Changing host]
^Neo has joined #ffmpeg-devel
jamrial has quit []
^Neo has quit [Ping timeout: 252 seconds]
rvalue has quit [Read error: Connection reset by peer]
rvalue has joined #ffmpeg-devel
mkver has joined #ffmpeg-devel
Martchus_ has joined #ffmpeg-devel
Martchus has quit [Ping timeout: 265 seconds]
Guest64 has joined #ffmpeg-devel
mark4o has joined #ffmpeg-devel
markh has quit [Ping timeout: 252 seconds]
mark4o is now known as markh
Guest64 has quit [Quit: Client closed]
<fflogger> [editedticket] Balling: Ticket #11362 ([avcodec] memory leaks error at __interceptor_malloc) updated https://trac.ffmpeg.org/ticket/11362#comment:4
Marth64 has quit [Quit: Leaving]
Marth64 has joined #ffmpeg-devel
tguillem has quit [Ping timeout: 252 seconds]
<Marth64> would the world benefit from a decent and minimalist python script that converts between chapter marker formats (edl, matroska xml, ffmetadata) in tools/ folder?
<Marth64> i have one that I'm thinking to polish and share
Guest84 has quit [Quit: Client closed]
<fflogger> [newticket] mahozad: Ticket #11366 ([undetermined] Regression when prepending/concatenating still image to video) created https://trac.ffmpeg.org/ticket/11366
cone-130 has joined #ffmpeg-devel
<cone-130> ffmpeg Scott Theisen master:9da1d2e66ab1: libavcodec/v4l2_buffers.c: set AVFrame interlaced flags
<fflogger> [editedticket] MasterQuestionable: Ticket #11366 ([avfilter] Regression when prepending/concatenating still image to video) updated https://trac.ffmpeg.org/ticket/11366#comment:1
<Compn> Marth64, that would be interesting
mkver has quit [Ping timeout: 248 seconds]
av500 has quit [Remote host closed the connection]
Guest84 has joined #ffmpeg-devel
av500 has joined #ffmpeg-devel
<fflogger> [editedticket] mahozad: Ticket #11366 ([avfilter] Regression when prepending/concatenating still image to video) updated https://trac.ffmpeg.org/ticket/11366#comment:2
mkver has joined #ffmpeg-devel
<fflogger> [newticket] elie.mora: Ticket #11367 ([undetermined] Chroma artefacts on decoded VVC streams) created https://trac.ffmpeg.org/ticket/11367
<cone-130> ffmpeg Niklas Haas master:fe9bf7cd5235: checkasm/sw_scale: add assertion for hscale assumption
<cone-130> ffmpeg Niklas Haas master:57bbdb4fb1e1: checkasm/sw_scale: add test for yuv2nv12cX
<cone-130> ffmpeg Niklas Haas master:c601bb8df5ae: checkasm/sw_rgb: add tests for yuv2packed{1,2,X}
<cone-130> ffmpeg Niklas Haas master:a9ae2cc14d43: checkasm/sw_rgb: add alpToYV12 check
<cone-130> ffmpeg Niklas Haas master:c6bf7f664527: swscale/unscaled: correctly round yuv2yuv when not dithering
<cone-130> ffmpeg Niklas Haas master:77db7f9b8781: swscale/unscaled: correctly copy semiplanar formats
<cone-130> ffmpeg Niklas Haas master:6c9218d748e0: swscale/unscaled: allow semiplanar copies
<fflogger> [editedticket] Niklas Haas <git@haasn.dev>: Ticket #11239 ([swscale] Copy unscaled NV12 -> NV12 (and other format with 2 planes) broken.) updated https://trac.ffmpeg.org/ticket/11239#comment:7
<fflogger> [editedticket] haasn: Ticket #10136 ([swscale] Can't convert color range with sws_scale) updated https://trac.ffmpeg.org/ticket/10136#comment:6
tufei_ has joined #ffmpeg-devel
tufei has quit [Ping timeout: 264 seconds]
odrling has quit [Remote host closed the connection]
^Neo has joined #ffmpeg-devel
^Neo has quit [Changing host]
^Neo has joined #ffmpeg-devel
odrling has joined #ffmpeg-devel
^Neo has quit [Ping timeout: 246 seconds]
rvalue- has joined #ffmpeg-devel
rvalue has quit [Ping timeout: 248 seconds]
rvalue- is now known as rvalue
<cone-130> ffmpeg Niklas Haas master:b9dfe8138eb2: swscale/utils: check for supported color transfers
<cone-130> ffmpeg Niklas Haas master:a8d01dff9a24: swscale/utils: add HDR metadata to SwsFormat
<cone-130> ffmpeg Niklas Haas master:5b21b7f52c7a: swscale/utils: set static/implied HDR metadata
<cone-130> ffmpeg Niklas Haas master:7432fa19cdd0: swscale/utils: read HDR mastering metadata from AVFrame
<cone-130> ffmpeg Niklas Haas master:9084d581e8e5: swscale/utils: read dynamic HDR10+ metadata from AVFrame
<cone-130> ffmpeg Niklas Haas master:1f0c50078409: swscale/utils: add helper function to infer colorspace metadata
<cone-130> ffmpeg Niklas Haas master:7b7c32322d1f: swscale/utils: fix XYZ primaries tagging
<cone-130> ffmpeg Niklas Haas master:45f0a7ad338c: swscale: add ICC intent enum and option
<cone-130> ffmpeg Niklas Haas master:b7a233335719: avfilter/vf_scale: add colorspace and transfer property options
<cone-130> ffmpeg Niklas Haas master:4315cc943aa2: avfilter/vf_scale: make options alignment consistent (cosmetic)
<cone-130> ffmpeg Niklas Haas master:01ce9d796300: avfilter/vf_scale: strip metadata when changing colorspace
<cone-130> ffmpeg Niklas Haas master:2e674780b740: swscale/csputils: add internal colorspace math helpers
<cone-130> ffmpeg Niklas Haas master:dddf536d3d57: swscale/cms: add color management subsystem
<cone-130> ffmpeg Niklas Haas master:a57fe519b60a: swscale/lut3d: add 3DLUT dispatch system
<cone-130> ffmpeg Niklas Haas master:efff80c8f65e: swscale/graph: add color mapping pass
<cone-130> ffmpeg Niklas Haas master:6da940e1189a: swscale/graph: allow dynamically updating HDR metadata
<cone-130> ffmpeg Niklas Haas master:253b8977c020: swscale: remove primaries/trc change warning
<haasn> merry christmas
<JEEB> wohoo
<JEEB> and thank you very much for doing this rework so people don't need to figure out what exactly swscale used to do ("is this matrix coeffs, primaries or what?" 8)
<JEEB> and yea, zimg is 100% planar based, there's a separate libp2p (the author seems to have fun naming things), which does packed to planar and the other way
<JEEB> so us having direct packed support is probably faster
<haasn> even planar to planar is not always faster in zimg
<haasn> seems only in some cases
<haasn> I think my next plan for swscale is to pull all of the DSP routines out into a dsp functions sruct
<haasn> so we can e.g. do more graceful fallback to C
<cone-130> ffmpeg Niklas Haas master:5ca5bbd462bc: swscale/options: add -sws_dither none alias
<fflogger> [editedticket] Niklas Haas <git@haasn.dev>: Ticket #9192 ([swscale] Command line switch sws_dither doesn't allow "none") updated https://trac.ffmpeg.org/ticket/9192#comment:12
<haasn> I was actually thinking about going the other way and making nu-swscale 100% packed based
<haasn> since that tends to give better performance on GPUs, it may be the case for CPU as well
<haasn> although I guess that makes harder
<haasn> asm*
<JEEB> x264 did move to semi-packed NV12 like input I think, for SIMD benefits
<haasn> I mean we already carefully tune our value ranges to avoid overflow
<haasn> so in theory you can just do simd-within-a-register style ops on packed values
<JEEB> man the age of this commit makes me feel old
jamrial has joined #ffmpeg-devel
<JEEB> (since I still recall that being a new thing which happened around the same time as high bit depth support started getting added
<haasn> the downside ofc is needing more asm routines
<haasn> I guess it's easy enough to start with planar code and move to packed code later
<haasn> since planar code and packed monochrome are the same thing
<JEEB> yea
<fflogger> [newticket] RandomPerson: Ticket #11368 ([ffmpeg] Mediacodec encoding not working) created https://trac.ffmpeg.org/ticket/11368
^Neo has joined #ffmpeg-devel
^Neo has quit [Changing host]
^Neo has joined #ffmpeg-devel
<Lynne> haasn: delete the historical MSB packing for 10/12/14 bit while at it, pl0x, 16 bit LSB-packed all the way
<haasn> Lynne: which routine are you referring to?
<haasn> I’m not sure where we do msb packing aside from the input/output code for p010 etc
<BBB> elenril: oh the 1984 reference, I hope people understood that. lovely
<BBB> so good to see my fellow friendly devs appreciate good literature <3
<BBB> we do all need to tone down a bit. this tone on the ML is not good
<cone-130> ffmpeg Niklas Haas master:01084f3d16ae: avutil/frame: add AVSideDataDescriptor for AV_FRAME_DATA_VIDEO_HINT
<cone-130> ffmpeg Niklas Haas master:3428a8d83035: avutil/frame: add AV_SIDE_DATA_PROP_{SIZE,COLOR}_DEPENDENT
<cone-130> ffmpeg Niklas Haas master:b88944a8aa59: avutil/frame: add av_frame_side_data_remove_by_props()
<cone-130> ffmpeg Niklas Haas master:f22965b8823c: avfilter/vf_scale*: strip metadata on size change
<cone-130> ffmpeg Niklas Haas master:d1eadc43b91a: avfilter/vf_zscale: strip metadata on change
<cone-130> ffmpeg Niklas Haas master:0d468730bc4e: avfilter/vf_libplacebo: update metadata stripping logic
<cone-130> ffmpeg Niklas Haas master:eecdc25d4b7c: avfilter/vf_colorspace: strip color volume metadata
<cone-130> ffmpeg Niklas Haas master:d23f443dfcad: avfilter/vf_lut*: strip color volume metadata
<elenril> BBB: I do not appreciate being accused of fraud, you know
<BBB> I understand
<BBB> but we do need to tone down the volume a bit
rvalue- has joined #ffmpeg-devel
<BBB> I don't think the current tmosphere will get us anywhere productive in that thread
rvalue has quit [Ping timeout: 252 seconds]
<BBB> Compn: I really do hope you understand the criticism of people on your action to block that ML thread without announcing it. I think at the very, very least, you should have announced it. And then I'm side-stepping the question of whether ML admins should be allowed to moderate arbitrarily at their discretion (which I think they should not - that's exactly what the CC was designed for - but that's just MHO)
<elenril> BBB: I also don't think language policing is top priority when the basic rules are being flagrantly ignored
<BBB> I understand the need for ML admins to act immediately in cases of obvious spam, I've done that on IRC before also. but behavioural stuff within our community should be left to the CC
<BBB> elenril: while I'm still CC (hopefully for nothing but a few more days), can you please tone it down a bit anyway? I'm asking friendly and I'm not saying you're wrong in anything you say, I'm just asking that we tone it down a bit
<elenril> sorry, I cannot
<elenril> the outcome of this whether the project is an actual democracy, or an autocracy ruled by michaelni
<elenril> that is fundamental
rvalue- is now known as rvalue
___nick___ has joined #ffmpeg-devel
___nick___ has quit [Client Quit]
<another|> urg.
* another| closes Thunderbird again
___nick___ has joined #ffmpeg-devel
<courmisch> BBB: siding with the bully. fuck you
<courmisch> you should be ashamed of yourself.
<JEEB> haasn: feels good when `-vf scale=intent=perceptual:out_color_matrix=bt709:out_chroma_loc=left:out_primaries=bt709:out_transfer=bt709:w=1920:h=1080,format=yuv420p` just does what you expect :3
<haasn> would feel even better if we had full negotiation for the colorspace properties
<haasn> so codecs could expose what colorspaces they support
<haasn> and PQ would automatically downconvert when outputting e.g. JPG
<JEEB> right
<JEEB> I think I have seen how good dithered 8bit PQ looks with windows, so PQ to 8bit output could almost just be tonemapped to something
<JEEB> (sRGB? 2.2?)
<Lynne> haasn: we define yuv4xx1x as having padding in the MSBs, which means we have one pixel format for each bit depth, and sometimes different assembly for each bit depth
<Lynne> in a parallel world, we didn't go down that path and instead had padding in the LSBs with a separate bits field, like we do for audio
<haasn> oh, sure
<haasn> but that's fixed and not going to change any time soon
<haasn> for internal use we convert to reference levels
<Lynne> yeah, sadly, it would be a lot of work to convert all codecs and asm too
<haasn> and user code
courmisch has left #ffmpeg-devel [#ffmpeg-devel]
Sebastinas has left #ffmpeg-devel [#ffmpeg-devel]
<Compn> BBB, yes i agree, next time will announce moderation.
<Compn> BBB, but as you can see, i tried to moderate out the flames. while jamrial (as ml admin) has allowed the flames. and while the CC has allowed the flames. its just a matter of which moderation you want. do you want compn style moderation or do you want the current moderation? with developers saying "f you" to you?
<Compn> i'm not saying which is better, i side-step that question and leave it up to the viewer.
halloy9798 has joined #ffmpeg-devel
<halloy9798> Hi guys,
<Compn> whats up halloy9798
<halloy9798> thank you @Compn. I have a open source tools to help simplify run ffmpeg commands. can I introduce here and get feedback from users?
<Compn> this is development channel for ffmpeg itself. if you are looking for ffmpeg users they are in #ffmpeg channel
<putacho> halloy9798: no, this chan is for ffmpeg development only
<halloy9798> can I do it in #ffmpeg in channel?
<Compn> what link for your frontend ?
<halloy9798> thank you
<Compn> i want to look at it :D
<halloy9798> @Compn, now I will share it in #ffmpeg channel.
halloy9798 is now known as hsngerami
hsngerami has quit [Changing host]
hsngerami has joined #ffmpeg-devel
hsngerami has quit [Quit: hsngerami]
cone-130 has quit [Quit: transmission timeout]
IndecisiveTurtle has joined #ffmpeg-devel
<BBB> Compn: I don't mind people being annoyed at me, it's part of the job of being in the CC. as long as they hold up my request to behave nicely in their ML (and IRC) communications with other fellow devs, it's fine with me.
<Marth64> good morning/evening
<Marth64> doesn't election end today?
<fflogger> [editedticket] MasterQuestionable: Ticket #11368 ([ffmpeg] Mediacodec encoding not working) updated https://trac.ffmpeg.org/ticket/11368#comment:1
<fflogger> [editedticket] Balling: Ticket #11367 ([undetermined] Chroma artefacts on decoded VVC streams) updated https://trac.ffmpeg.org/ticket/11367#comment:3
kierank has joined #ffmpeg-devel
<kierank> Does anyone know discord?
<kierank> I want to make one for the asm lessons
<Traneptora> kierank: wdym by "know discord"
<Traneptora> I use it regularly, yea
<kierank> Like best practice for making a channel, moderation etc
<Traneptora> discord doesn't use channels, it uses "guilds" (often called servers) which contain several channels
<Traneptora> it's all relatively self-explanatory. a lot of the ideas are pretty similar. kick/ban/etc. bans from guilds. there's roles, just like irc's ~, &, @, etc. but they're a bit more flexible.
<Traneptora> and you can hide/show channels or grant permissions to roles
<Traneptora> basic practice is to allow anyone to do stuff that's communication-related (like react, send messages, view message history, etc.) but to only have moderation stuff (kick/ban/delete messages send by other users/etc) be available to mods
<Traneptora> if you're familiar with how permissions on IRC works and basic general vibes of what it means to be a moderator then you should be able to set one up yourself without too much trouble
<Marth64> anecdote: they gave me a hard time with my VOIP number just FYI you probably will need a real mobile number for registration
<Marth64> besides that its cool
<Traneptora> oh yea if you want 2FA you need to use an actual mobile number. I bought a burner sim from tmobile to do that
<Traneptora> once you set up 2FA you can then switch over to an auth app, but to set it up initially you need an actual phone
<Marth64> oh neat
<Traneptora> the big thing that isn't there by default is you should remove every user's ability to ping everyone
<Traneptora> you don't need a phone to register, just an email. but you do need an actual phone to turn on 2FA
Teukka` has joined #ffmpeg-devel
pross_ has joined #ffmpeg-devel
some02 has joined #ffmpeg-devel
haihao_ has joined #ffmpeg-devel
deus0ww_ has joined #ffmpeg-devel
Marth64 has quit [Quit: Leaving]
Teukka has quit [Quit: Not to know is bad; not to wish to know is worse. -- African Proverb]
deus0ww has quit [Ping timeout: 265 seconds]
pross has quit [Ping timeout: 265 seconds]
sudden has quit [Ping timeout: 265 seconds]
haihao has quit [Ping timeout: 265 seconds]
Marth64[m] has joined #ffmpeg-devel
<OctopusET> Hello, I'm Moon who helped organizing VDD24. There are 16 names on the VDD website and I need 4 more names, because my univerisity wants proof there were at least 20 people are attended. So, Can anyone provide your name
averne_ has joined #ffmpeg-devel
<OctopusET> It will be only wrote on this website https://www.videolan.org/videolan/events/vdd24/
averne has quit [Ping timeout: 246 seconds]
averne_ is now known as averne
<OctopusET> *Anyone who attended VDD24 and its workshop
<OctopusET> You can send your name on here or private message to me.
<Compn> OctopusET, sup moon! also ask in #videolan as there were videolan devs there
<OctopusET> Thank you
<Compn> OctopusET, did you ask in the signal group too ? :)
<JEEB> I think it's locked now since the event ended
<OctopusET> I can't write any message on signal group anymore
<Compn> i guess the afterparty signal group
<OctopusET> Ah no I didn't. I will, thanks
<wbs> OctopusET: I presume it's unnecessary to message you if one's name is already listed there (among the presenters)?
<OctopusET> Yes, if it's listed I don't need it
<OctopusET> Still thank you
<BBB> vdd24 was amazing, thanks for organizing OctopusET
<Compn> i hope the university isnt giving you too much trouble, moon. let us know if theres anything else we can do
<OctopusET> BBB: I really appreciate
<OctopusET> Compn: Ah, the university is trying to reimburse us for the fees for the posters and other things.
<OctopusET> Names will be really helpful, thanks
<OctopusET> That's why they asked 20 names
<JEEB> yeh, the event was great :) so really thanks for providing the venue and all the other many things.
<OctopusET> :D
<BBB> my name is already on the presenter's list, so nothing new there, sorry
<JEEB> OctopusET: anyways, I was there at VDD https://github.com/jeeb (profile should have my name if it's helpful :) )
<OctopusET> thank you, then I will add your name on website too
<JEEB> as long as it helps you get through some funky official requirements :)
<JEEB> (I would expect videolan to be able to confirm the full list of guests)
iive has joined #ffmpeg-devel
pross_ is now known as pross
mkver has quit [Ping timeout: 248 seconds]
Marth64[m] has quit [Quit: Leaving]
cone-137 has joined #ffmpeg-devel
<cone-137> ffmpeg James Almer master:f1ddba24a0cf: avutil/frame: add missing size dependent prop to LCEVC payload side data
<ePirat> OctopusET, I was there too, Marvin Scholz (https://github.com/ePirat)
<OctopusET> Thanks
Teukka` has quit [Quit: Not to know is bad; not to wish to know is worse. -- African Proverb]
Teukka has joined #ffmpeg-devel
Teukka has quit [Changing host]
Teukka has joined #ffmpeg-devel
___nick___ has quit [Ping timeout: 248 seconds]
<ePirat> Merry Christmas / Happy holidays to everyone
Marth64 has joined #ffmpeg-devel
System_Error has quit [Remote host closed the connection]
System_Error has joined #ffmpeg-devel
rvalue has quit [Ping timeout: 252 seconds]
cone-137 has quit [Quit: transmission timeout]