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
System_Error has quit [Remote host closed the connection]
jarthur_ has joined #ffmpeg-devel
jarthur_ has quit [Client Quit]
jarthur has quit [Ping timeout: 260 seconds]
System_Error has joined #ffmpeg-devel
System_Error has quit [Ping timeout: 264 seconds]
thilo has quit [Ping timeout: 260 seconds]
thilo has joined #ffmpeg-devel
System_Error has joined #ffmpeg-devel
mkver has joined #ffmpeg-devel
aaabbb has quit [Quit: Reconnecting]
aaabbb has joined #ffmpeg-devel
aaabbb has quit [Changing host]
aaabbb has joined #ffmpeg-devel
^Neo has quit [Ping timeout: 252 seconds]
cone-259 has joined #ffmpeg-devel
<cone-259>
ffmpeg James Almer master:75ade2dd499f: avcodec/hevc_mp4toannexb: prepend extradata before the leading PS for an IRAP
<fflogger>
[editedticket] jamrial: Ticket #11458 ([avcodec] Input file .mkv with -c:v copy option generates corrupted output when remuxing to raw HEVC container) updated https://trac.ffmpeg.org/ticket/11458#comment:4
jamrial has quit []
Tanay has joined #ffmpeg-devel
Guest94 has joined #ffmpeg-devel
System_Error has quit [Remote host closed the connection]
System_Error has joined #ffmpeg-devel
zsoltiv has quit [Ping timeout: 268 seconds]
zsoltiv_ has quit [Ping timeout: 244 seconds]
Guest94 has quit [Quit: Client closed]
Martchus_ has joined #ffmpeg-devel
Martchus has quit [Ping timeout: 252 seconds]
psykose has quit [Remote host closed the connection]
psykose has joined #ffmpeg-devel
System_Error has quit [Remote host closed the connection]
System_Error has joined #ffmpeg-devel
cone-259 has quit [Quit: transmission timeout]
Teukka has quit [Read error: Connection reset by peer]
Teukka has joined #ffmpeg-devel
Teukka has quit [Changing host]
Teukka has joined #ffmpeg-devel
Tanay has quit [Ping timeout: 248 seconds]
Kei_N has joined #ffmpeg-devel
Kei_N_ has quit [Ping timeout: 260 seconds]
Tanay has joined #ffmpeg-devel
Tanay has quit [Ping timeout: 276 seconds]
Tanay has joined #ffmpeg-devel
sunny78 has joined #ffmpeg-devel
<sunny78>
Hello, I am trying to submit a path to ffmeg-devel@ffmpeg.org and I am receiving Mail Delivery error - User unknown in local recipient table from ffbox0-bg.ffmpeg.org. What could be wrong?
tufei__ has quit [Remote host closed the connection]
tufei_ has joined #ffmpeg-devel
meego has joined #ffmpeg-devel
meego has quit [Quit: Leaving...]
<frankplow>
sunny78: ffmeg-devel should be ffmpeg-devel
eslam has joined #ffmpeg-devel
HarshK23 has quit [Quit: Connection closed for inactivity]
Tanay has joined #ffmpeg-devel
j45_ has joined #ffmpeg-devel
j45 has quit [Ping timeout: 248 seconds]
j45_ is now known as j45
j45 has quit [Changing host]
j45 has joined #ffmpeg-devel
^Neo has joined #ffmpeg-devel
^Neo has quit [Changing host]
^Neo has joined #ffmpeg-devel
jamrial has joined #ffmpeg-devel
<frankplow>
I'm not sure if the captcha is working on the forgejo
<frankplow>
Unless my zoology is *really* bad
Tanay has quit [Ping timeout: 272 seconds]
<Lynne>
frankplow: if you give it invalid answers a few times, it'll let you in
<Lynne>
gave me a 404, and just clicking on sign it, it let me
<frankplow>
Lynne: It doesn't seem to. This is for registration btw, not for sign in.
<Lynne>
put in "ffmpeg" and "2" as the answers, and it'll give you a 404
<Lynne>
ah, yeah, that lets you sign in, not register
<Lynne>
still, it should be fixed (BtbN)
<sfan5>
is ffmpeg accepting contributions on forgejo now? I see a few PRs
<Lynne>
no, the repo is not synced up yet
<Lynne>
those are tests
<sfan5>
aw
VoteLeader has joined #ffmpeg-devel
<frankplow>
If I enter an incorrect answer, then I get a 404 on clicking continue, if I enter a correct answer (case-insensitive it seems), then when I click continue the page refreshes and the input boxes clear.
<BtbN>
What should be fixed exactly? It's not meant to protect the sign in
<BtbN>
You need to allow cookies for the captcha to work
<frankplow>
I have a cookie called "CaptchaPassed" with a value of "true", still no luck
<BtbN>
That's all it checks for for letting you through to the sign-in
<frankplow>
I'm trying to register, not sign in
j45 has quit [Read error: Connection reset by peer]
j45 has joined #ffmpeg-devel
j45 has quit [Changing host]
j45 has joined #ffmpeg-devel
<kierank>
doesn't work here either
System_Error has quit [Remote host closed the connection]
System_Error has joined #ffmpeg-devel
<VoteLeader>
let me try
<VoteLeader>
confirmed
<VoteLeader>
also the page interface is from 60th of previous century
<VoteLeader>
but couldnt you use sign in with github?
<Lynne>
if you're talking about accuracy as a sale point for your fft, you're doing it wrong
eslam has quit [Ping timeout: 260 seconds]
<BtbN>
How does it not work? I just tested it again, and it works fine for me. What's the failure mode? Do you not get the cookie? Does your browser not send the cookie?
<VoteLeader>
Lynne: its closed source, so you need to test it to report here how its power of 2 behaves versus tx
<BtbN>
I reproduced it with Chrome... It's a Browser-Side caching issue. The Browser decided "this page surely has to be identical to before". Hitting Ctrl+F5 once you land back there gets you the proper sign up page.
<JEEB>
local caching is fun
<kierank>
I'm using firefox
<BtbN>
works reliably for me in firefox
<BtbN>
only Chrome and Edge fail
Tanay has quit [Ping timeout: 276 seconds]
<JEEB>
local caching can be disabled with developer tools & in network tab check that the disable cache checkbox is set. and if it's still borked with that, then it's a separate issue
<kierank>
Even with GitHub I don't get any email
<kierank>
GitHub login mode
<BtbN>
I'm sending "do not cache this" headers now
<BtbN>
so hopefully it works now
<wbs>
BtbN: can you raise the quotas for the account mstorsjo?
<BtbN>
done
<wbs>
thanks!
<BtbN>
kierank: check your spam folder. It sends mail via the ffmpeg.org mail server. And specially Google and Microsoft hate it.
<VoteLeader>
"ffmpeg.org mail server"
<kierank>
I obviously checked my spam dolder
<kierank>
And resent twice
<kierank>
VoteLeader: ffmpeg, the best a man can get!
<VoteLeader>
hpkfft does not have C api at all
<VoteLeader>
how can i test this then?
<BtbN>
kierank: I sent you a test-mail via the admin interface, did that get through?
<kierank>
Yes
<BtbN>
strange, it means it must have not even tried sending you one then.
<BtbN>
Guess it hates specifically your account for some reason. Cause tons of other people signed up via github successfully before and two after you
<VoteLeader>
put on shadow list
E81l7HT8T7sF9JdA has joined #ffmpeg-devel
<BtbN>
Manually activated the account now
<VoteLeader>
how many wings chicken have?
<Lynne>
BtbN: I had the same issue when I signed up
<BtbN>
yes, but that was caused by the manual admin approval mode
<Lynne>
ah
<BtbN>
It actively prevents you from confirming your mail address when that's set. The admin is supposed to do that.
<BtbN>
So why it now does the same thing specifically for kierank is a bit strange.
<VoteLeader>
fixed vp9 support is USM demuxer!
<JEEB>
what's USM?
<VoteLeader>
game format
<JEEB>
huh, they had VP9? I guess it kinda makes sense
<VoteLeader>
afaik mpeg1, h264 and vp9, and adpcm_adx, hca on audio side
<VoteLeader>
vp9 is pretty good, dunno why that hate
<JEEB>
yea I mean I did encodes in VP9 back when dav1d didn't have x86_64 10bit SIMD
<JEEB>
because you got 10bit + browser playable with VP9 in MP4
Tanay has joined #ffmpeg-devel
<jamrial>
VoteLeader: what hate?
<jamrial>
youtube even uses it for live streams, with decent results
<nevcairiel>
but thats the thing, youtube is practically the only one that uses it
<Lynne>
they mainly use it on apple devices only, others still use webm dash
<Lynne>
apple are allergic to open standards
<JEEB>
if you are talking about MP4, that was actually open and freely available until 2017 or so, and Apple loves it just because it's originally an offspring of QTFF aka MOV. and the reason why it's no longer open is 100% ISO
<JEEB>
even the apple guy who chaired mp4-sys was not liking ISO's one-sided decision
* JEEB
sent an email regarding it
<JEEB>
due to how utilized mp4 in general is on the internet, that was actually a messed up move by ISO, but what do you expect from an org that only has CHF in its mind
<JEEB>
DASH actually became free again after 3GPP/ETSI got involved
<JEEB>
since 3GPP/ETSI provided their stuff into DASH under the understanding that the spec would be freely available
<JEEB>
MP4 doesn't seem to have any group within the working group yelling as hard
<VoteLeader>
are there anime torrents with vp9?
<JEEB>
probably someone did it
<Lynne>
JEEB: no, talking about vp9, opus, and specifically webm
<JEEB>
ah, ok
<frankplow>
H.26x would not be publicly available if it weren’t for the ITU I believe. I think ISO’s policy is that all standards they publish must be pay-to-play unless circumstances are exceptional
<JEEB>
yea
<JEEB>
basically requested. previously it was once, and then in 2017 they made it so that the org needs to separately request it for each edition
<JEEB>
mixed with ISO not having old versions of specs available (at least in their free section, not sure if you can still buy the old versions on the paid side)
<frankplow>
BtbN: Registration on Forgejo works for me now, thank you
<BtbN>
you basically just had to Ctrl+F5 after passing the captchs
<BtbN>
the browser cached the captcha page...
<frankplow>
Actually I might be having the same issue as kierank where I don't receive the confirmation email still
<frankplow>
Been a few minutes and it hasn't shown up and it isn't in my spam folder
Marth64 has quit [Remote host closed the connection]
Marth64 has joined #ffmpeg-devel
<VoteLeader>
Lynne: apparently there is ALD compressor for doubles/floats and it uses simple XOR on integer data and some FFOR algorithm, but i struggle to use/understand it
MyNetAz has quit [Remote host closed the connection]
<VoteLeader>
typo there: full name is : ALP: Adaptive Lossless Floating-Point Compression
MyNetAz has joined #ffmpeg-devel
<Lynne>
michaelni: ^
<Lynne>
it would be good if it generates ordered and equiproportional integer domain values to enable prediction
<VoteLeader>
i think it does nonsense divisions
<VoteLeader>
but i'm interested in XOR and FFOR part
<VoteLeader>
and see if just interpreting float as uint32_t can do it
E81l7HT8T7sF9JdA has quit [Quit: Leaving]
<ramiro>
is marton here on irc?
cone-755 has joined #ffmpeg-devel
<cone-755>
ffmpeg Araz Iusubov master:36985704420e: avcodec/amfenc: Fix max rate control log message
<ramiro>
USE_ONEPASS_SUBTITLE_RENDER in ffplay.c is hardcoded, I'm wondering if we can just remove the #if that uses it.
<michaelni>
Lynne, took a quick look at ALP, will look more closely, but it seems generic 64bit float compression not 2D images. in FFv1 ATM everything i have is 16bit floats (if someone has 32 or 64bit float images?)
<Lynne>
rather than using alp to encode, you can use alp to translate the coeffs info ints and use the same prediction method as ints
<michaelni>
yes but with 16bit floats converting to int while maintaining ordering just needs 8kb in worst case (to store every single 16bit value that is used)
<michaelni>
64 kbit == 8kbyte
<michaelni>
Lynne, do you stil want to be backup mentor for "VVC x86 simd optimization " ?
<michaelni>
so ALP encoding has to beat simple dumb <8kb table unless theres some deeper magic iam missing. Of course for 32bit / 64bit float its a different story and ALP would be more interresting
<Lynne>
yup, I've just added myself as a mentor
<Lynne>
alp may allow you to scale up to 32 bits
<michaelni>
thank you!
<michaelni>
about alp: yes
<Lynne>
tbh 16bit float pixfmts aren't really that useful since most intermediate VFX steps are done on 32bit floats
<michaelni>
but i have no 32bit int or lfoat images
<michaelni>
so i cant test anything with 32bit floats
<Lynne>
the tonemap filter spits out 32bit planar floats
<Lynne>
if you could use a single function to handle float->int conversion it would be better spec-wise
<michaelni>
maybe but allowing identity, RLE map, something that handles fixed point numbers stored in float, <--- this would have 3 methods each covering an area the others do not cover optimally
<michaelni>
if there was a universal method?
<michaelni>
IIUC ALP is not universal
<Lynne>
it would be implemented as a preprocess step, basically, you'd run alp when you load each line
<Lynne>
you already do that for RGB images, and I think it would be safe to assume any float images are RGB
<michaelni>
yes, i already implemented the generic remap table for 16bit floats in "avcodec/ffv1: flip half of float16 and Compactify floats"
<michaelni>
another remaping (ALP or other could be done at the same place as an alterantive)
VoteLeader has joined #ffmpeg-devel
<VoteLeader>
should h264 parser set timestamps if demuxer set parse timesamps flag ?
<kierank>
timestamps from where
<kierank>
if fixed_frame_rate it can set some kind of arbitrary timestamp
<VoteLeader>
where to set fixed frame rate?
<kierank>
there is fixed_frame_rate flag in vui
<VoteLeader>
i get yellow warnings about unsed timestamps
<VoteLeader>
and ffmpeg than generates new timestamps, but not also dts
<VoteLeader>
which means seeking is not correct any more