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
lexano has quit [Remote host closed the connection]
lexano has joined #ffmpeg-devel
iive has quit [Quit: They came for me...]
nitroxis has quit []
nitroxis has joined #ffmpeg-devel
thilo has quit [Ping timeout: 252 seconds]
thilo has joined #ffmpeg-devel
<Lynne>
Daemon404: both
aaabbb has quit [Ping timeout: 276 seconds]
<Lynne>
the reference implementation is written very heuristically and not that optimally, surprisingly
<Lynne>
the spec sometimes gives up
Martchus_ has joined #ffmpeg-devel
Martchus has quit [Ping timeout: 272 seconds]
aaabbb has joined #ffmpeg-devel
Sean_McG has quit [Quit: leaving]
arch1t3cht8 has joined #ffmpeg-devel
arch1t3cht has quit [Ping timeout: 252 seconds]
arch1t3cht8 is now known as arch1t3cht
jamrial has quit []
Martchus has joined #ffmpeg-devel
Martchus_ has quit [Ping timeout: 248 seconds]
Krowl has joined #ffmpeg-devel
Krowl has quit [Client Quit]
Krowl has joined #ffmpeg-devel
arbitercoin has joined #ffmpeg-devel
rvalue- has joined #ffmpeg-devel
rvalue has quit [Ping timeout: 255 seconds]
rvalue- is now known as rvalue
Krowl has quit [Read error: Connection reset by peer]
chocolate8 has joined #ffmpeg-devel
ngaullier has joined #ffmpeg-devel
ngaullie has joined #ffmpeg-devel
chocolate8 has quit [Ping timeout: 265 seconds]
ngaullier has quit [Ping timeout: 260 seconds]
Krowl has joined #ffmpeg-devel
<frankplow>
Stefano Sabatini hasn't posted on the ML in nearly three months. Has he said anything regarding this? I have a reimbursement request which he hasn't approved and could forward it to him directly, but don't want to pester him if he's relinquished that responsibility.
<elenril>
I'd say CC him directly
<elenril>
I didn't hear anything about him resigning from that position
<frankplow>
elenril: Ok, thanks
mkver has quit [Ping timeout: 248 seconds]
vr5 has joined #ffmpeg-devel
vr5 has quit [Remote host closed the connection]
Traneptora_ has joined #ffmpeg-devel
Traneptora has quit [Ping timeout: 276 seconds]
microlappy has joined #ffmpeg-devel
microlappy has quit [Client Quit]
rvalue- has joined #ffmpeg-devel
rvalue has quit [Ping timeout: 276 seconds]
tybu has joined #ffmpeg-devel
rvalue- is now known as rvalue
___nick___ has joined #ffmpeg-devel
tufei__ has joined #ffmpeg-devel
tybu has joined #ffmpeg-devel
tybu has quit [Changing host]
tufei_ has quit [Ping timeout: 260 seconds]
j45 has quit [Ping timeout: 276 seconds]
j45 has joined #ffmpeg-devel
j45 has joined #ffmpeg-devel
j45 has quit [Changing host]
Krowl has quit [Read error: Connection reset by peer]
arbitercoin has quit [Ping timeout: 252 seconds]
___nick___ has quit [Ping timeout: 245 seconds]
___nick___ has joined #ffmpeg-devel
tybu has quit [Quit: Client closed]
tybu has joined #ffmpeg-devel
tybu has quit [Ping timeout: 256 seconds]
tybu has joined #ffmpeg-devel
<Daemon404>
Lynne, the "gives up" bit of the spec - is it the part where it just goes full latex? lul
<Lynne>
nah, it gives good enough-ish instructions on how to decode given individual tasks, but gluing them all together - nah
<JEEB>
:)
Krowl has joined #ffmpeg-devel
sunyuechi has joined #ffmpeg-devel
sunyuechi has quit [Remote host closed the connection]
sunyuechi has joined #ffmpeg-devel
<Daemon404>
ah
sunyuechi has quit [Remote host closed the connection]
<Daemon404>
i wrote a decoder only using the spec
<Daemon404>
it was iffy
sunyuechi has joined #ffmpeg-devel
sunyuechi has quit [Remote host closed the connection]
sunyuechi has joined #ffmpeg-devel
tybu has quit [Quit: Client closed]
tybu has joined #ffmpeg-devel
sunyuechi has quit [Remote host closed the connection]
sunyuechi has joined #ffmpeg-devel
<sunyuechi>
Is it possible to send patches to ffmpeg-devel@ffmpeg.org now? I noticed that emails sent to others are received, but those sent to ffmpeg aren't.
<Daemon404>
it has always been
<Daemon404>
as long as you are signed up
sunyuechi has quit [Remote host closed the connection]
<Daemon404>
rip
sunyuechi has joined #ffmpeg-devel
tybu has quit [Ping timeout: 256 seconds]
<Daemon404>
sunyuechi, it has always possible. did you sign up?
sunyuechi has quit [Remote host closed the connection]
sunyuechi has joined #ffmpeg-devel
sunyuechi has quit [Remote host closed the connection]
sunyuechi has joined #ffmpeg-devel
<sunyuechi>
Yes, I submitted some patches to ffmpeg-devel@ffmpeg.org. It was working fine two weeks ago, but today there seems to be an issue.
<Daemon404>
did you actually sign up to the list, is what i meant
<Daemon404>
if so, my best guess is spam filter issues
sunyuechi has quit [Remote host closed the connection]
sunyuechi has joined #ffmpeg-devel
sunyuechi has quit [Remote host closed the connection]
sunyuechi has joined #ffmpeg-devel
sunyuechi has quit [Remote host closed the connection]
<Lynne>
Daemon404: oh, right, it was written in Go
<Lynne>
I remember it was around the time you posted a one-liner entropy decoder in perl
<Lynne>
(the range coder variant that ffv1/snow use is very dreadful is what I thought when writing)
<Lynne>
the mechanism through which the very first byte after initialization is handled was heavy-handed
<Lynne>
the data types didn't have much thought put into them other than "yeah, I guess an int will do"
<Lynne>
for example, the "low" value is deceptively bound to 16 bits, but it may under certain conditions jump up beyond that
<Lynne>
which means you need a yet another branch in your coder
<Lynne>
in the name of speed, I have 2 encoders, one which can handle first byte, and one which can't, and I use the one which can't for all data past the slice header
<Lynne>
I got renorm down to a couple of cmovs (GPUs have cmov instructions, thankfully)
<Daemon404>
that all soujds correct
<Daemon404>
sounds*
tybu has joined #ffmpeg-devel
tybu has quit [Changing host]
tybu has joined #ffmpeg-devel
<Lynne>
I'd rather write gpu shaders in cobol
<Lynne>
glsl is shit, and the only viable replacement is made by nvidia and still changes pretty much daily
<Lynne>
those few lines take 60 odd seconds to compile because the only glsl to spirv implementation is horrible
<IndecisiveTurtle>
1 minute for compile time is insane. And I thought 10+ms was not acceptable...
jamrial has joined #ffmpeg-devel
<Daemon404>
id love to try this but i dont have anything with a gpu atm :D
<Marth64[m]>
ffv1 vulkan brewing over here?
<Daemon404>
yeah Lynne linked it a few days ago
<Marth64[m]>
very cool
<Daemon404>
yea
<Daemon404>
reminds me michaelni, i did see you poke me on those ffv1 spec colorspace gh issues, i will have to redo those. result long lost to ssd failure.
<Daemon404>
i wonder if RDLS patents have expired yet
Krowl has quit [Read error: Connection reset by peer]
Lypheo has quit [Quit: Ping timeout (120 seconds)]
Lypheo has joined #ffmpeg-devel
Krowl has joined #ffmpeg-devel
tybu has quit [Quit: Client closed]
tybu has joined #ffmpeg-devel
tybu has quit [Ping timeout: 256 seconds]
tybu has joined #ffmpeg-devel
tybu has quit [Ping timeout: 256 seconds]
<jamrial>
haasn: ping, do you have a sample for av1 fg in itu t.35 to test the hevc implementation?
<haasn>
jamrial: yes, but on my home computer
<jamrial>
ah
<haasn>
Which I can’t currently access because I’m in the process of moving
<haasn>
Iirc it’s also confidential
<haasn>
Maybe ask j-b for access, he sent it to me
<jamrial>
we can always just inject the sei into an unrelated sample
tybu has joined #ffmpeg-devel
tybu has quit [Changing host]
tybu has joined #ffmpeg-devel
microchip_ has quit [Quit: There is no spoon!]
microchip_ has joined #ffmpeg-devel
microchip_ has quit [Client Quit]
microchip_ has joined #ffmpeg-devel
microchip_ has quit [Client Quit]
microchip_ has joined #ffmpeg-devel
microchip_ has quit [Remote host closed the connection]
microchip_ has joined #ffmpeg-devel
kmikita has quit [Ping timeout: 260 seconds]
kmikita has joined #ffmpeg-devel
kmikita has quit [Ping timeout: 260 seconds]
tybu has quit [Quit: Client closed]
tybu has joined #ffmpeg-devel
tybu has quit [Quit: Client closed]
tybu has joined #ffmpeg-devel
tybu has quit [Client Quit]
Krowl has quit [Read error: Connection reset by peer]
tybu has joined #ffmpeg-devel
kmikita has joined #ffmpeg-devel
<Lynne>
Daemon404: also I figured out how to parallelize golomb encoding
tybu has quit [Ping timeout: 256 seconds]
<JEEB>
nice
<Lynne>
it requires 32x32 aligned slices, but the idea is that since what you write only depends on the run index at the start of a line, you can walk amongst each row in parallel and construct a input run_index->output run index map
<Lynne>
then it takes barely a few ops to figure out the run index for each sample
<Lynne>
how does frame threading work with video encoders?
<Lynne>
does each thread get its own avcodeccontext?
ssfl has quit [Remote host closed the connection]
sunyuechi has joined #ffmpeg-devel
<sunyuechi>
Daemon404: It seems that the sending failure was due to an SMTP server issue. A few hours later, I was prompted to resend, and it was successful.
sunyuechi has quit [Client Quit]
Krowl has joined #ffmpeg-devel
Workl has quit [Ping timeout: 252 seconds]
<jamrial>
Lynne: yes. the main thread (user facing) spawns the requested amount of worker threads, each being their own avctx
<Lynne>
can encoders disable threading on init if a setting flag is set?
<jamrial>
as in -threads 1?
Workl has joined #ffmpeg-devel
<Lynne>
yes, but programatically
<Lynne>
I can trivially frame thread ffv1_vulkan to fully saturate a GPU, but only if frame threading is enabled
<Lynne>
but I'd also like to support the level 4 inter-frames
<Lynne>
so if gop > 2 -> disable threading, otherwise keep enabled
<jamrial>
i don't think that's possible, but you can either fail or warn about it
Krowl has quit [Ping timeout: 244 seconds]
<jamrial>
the encoder's init function is called after generic setup has been done, and is also called for all worker threads
mkver has joined #ffmpeg-devel
<Lynne>
fair enough
<Lynne>
the contents of avctx at init are duplicated to all threads, right? so I don't need to do any initialization
Krowl has joined #ffmpeg-devel
ngaullie has quit [Ping timeout: 276 seconds]
<jamrial>
Lynne: there's a field called is_copy that lets you run worked thread specific code (or prevent main thread code from run)
Workl has quit [Ping timeout: 246 seconds]
<Lynne>
ah, I see, I can just check if it isn't a copy and do my init as normal, I guess
Krowl has quit [Read error: Connection reset by peer]
tybu15 has joined #ffmpeg-devel
tybu15 has quit [Ping timeout: 256 seconds]
Teukka` has joined #ffmpeg-devel
Teukka has quit [Read error: Connection reset by peer]
Traneptora_ has quit [Quit: Quit]
Traneptora has joined #ffmpeg-devel
System_Error has quit [Remote host closed the connection]
System_Error has joined #ffmpeg-devel
vipyne has quit [Quit: Leaving.]
arbitercoin has joined #ffmpeg-devel
arbitercoin has quit [Ping timeout: 252 seconds]
System_Error has quit [Remote host closed the connection]
System_Error has joined #ffmpeg-devel
System_Error has quit [Remote host closed the connection]