Av1

>av1
I hope this isn't the future

Attached: .png (573x143, 16K)

Other urls found in this thread:

github.com/xiph/rav1e
en.wikipedia.org/wiki/VP9#Hardware_implementations
bugzilla.mozilla.org/show_bug.cgi?id=1461268
my.mixtape.moe/jjsgrv.webm
github.com/OpenVisualCloud/SVT-AV1
twitter.com/SFWRedditImages

>brand new codec
>gets surprised when no hardware encoding and decoding
god just fucking wait retard

Rust will fix it: github.com/xiph/rav1e

IT'S BEEN
F O U R
Y E A R S

>four years
>old
It hasn't even been a year since the spec hit 1.0

binaries got frozen in october of last year iirc

How many years is that?

x264 has had 15 years of optimization

x264 is being phased out as we speak tho
vp9 too half the time it took vp8 to get to an acceptable state
now if only they stopped trying to push webp like a square peg through a round hole that'd be great

AMD has no hardware encoding and decoding for fucking VP8.

Attached: September 13, 2008.png (469x224, 28K)

and that's a good thing

>x264 is being phased out as we speak tho
That doesn't contradict what I said, and even if you are correct x264 is still the best all-round codec currently
x265 only wins at extremely low bitrate, content with no grain like anime, and HDR
AV1 is still too slow to be relevant and there's no hwdec currently

he's using ffmpeg. it still only has experimental encoder.
bloat

>Amount of video lost: 3.6s
>out of 2.6 seconds video
WELL
OK THEN

Attached: .png (640x118, 13K)

That's ridiculous the codec should be able to calculate an average bitrate based in filesize constraints

J-Just Wait!

AYYMD HOUSEFIRES don't even do VP9 hardware decoding

-an -c:v libaom-av1 -crf 30 -strict experimental -minrate:v 10000k -b:v 10000k -maxrate:v 10000k -cpu-used 2 -row-mt 1 -slices 4 -lag-in-frames 16 -auto-alt-ref 1 -bufsize 6000k -rc_init_occupancy 40000k -qcomp 0 -fs 2988441
these are the constraints I'm using btw

>Four years
The spec was finalized last June.

>AV1 Intel hardware acceleration initial release: 2028
>AV1 AMD hardware acceleration initial release: 2038
Just wait bro.

Because it's fucking pointless? They have for VP9 and HEVC, which actually matter. And they will for AV1 as well.

>AMD
>VP9
Pick one :)

Because it's fucking pointless? They have for HEVC, which actually matter. And they will for AV1 as well.

Are you serious? Decoding exists since Polaris
en.wikipedia.org/wiki/VP9#Hardware_implementations

At least mpv with libdav1d will be fast on older CPUs after they merge SSSE3 CDEF MR.

Hybrid since Polaris and full h/w since RR.

Do you even READ?

AYYMD does not support VP9 hardware decoding at all, not on Poolaris or Pooga

Intel and Nvidia both have REAL VP9 hardware decoding

>Intel Kaby Lake, Coffee Lake, Whiskey Lake, Amber Lake, Apollo Lake and Gemini Lake CPU families and Nvidia Maxwell GM206, Pascal, Volta & Turing GPU families have full fixed function VP9 hardware decoding for highest decoding performance and power efficiency.

Your reading comprehension is ZERO

RR has full h/w decoding.
So do Picasso and Navi.

You literally removed "AMD Raven Ridge APU" family from your quote.

Anyway, shitpost away.

Raven Poo is not Poolaris or Pooga

>AMD h/w decoding

Attached: AMD VP9 2.png (845x361, 63K)

Google is already rolling out av1 encoded videos on Youtube.

They have been doing it since September

>hurr AMD has no h/w VP9 decode
>they do
>durr poo
Based rajeesh.

>Replying to bait

This entire thread is bait, starting with the OP.

>2018+1
>buy AMD
>broken VP9 decoding
>no VP9 encoding
>but they will have one for AV1 as well
Just wait.

it's not

>CDEF
It literally doesn't matter for Youtube videos. They disabled it precisely for being heavy on the CPU. Their own decoder in libaom struggles with it.
It only matters if you want to play the typical benchmark videos like Chimera streams or non Youtube encodings.

bugzilla.mozilla.org/show_bug.cgi?id=1461268

>Recent AMD driver releases should have removed the Hybrid VP9 support already.

AYYMD doesn't even have VP9 decoding, they removed it from their drivers, top kek

If you bought AYYMD garbage, you deserve it for being a cuck

>AYYMD doesn't even have VP9 decoding
How does my 2400G do it then?

slowly

Doesn't seem so.

>using hardware decoding on PCs
lol
it's like you don't care about video quality

>new codec
>software

Why bother?

>buying Ryzen for multitasking
>CPU usage spikes to 60% while playing a video

my.mixtape.moe/jjsgrv.webm
your browser should be able to render this

man, mixtape.moe is going down the drain
at the same time, streamable can't handle av1 at all since it reencodes everything

Attached: .png (390x255, 15K)

Why are you using crf in combination with the syntax for a constant bitrate?
Why are you trying to use a 10Mbps bitrate, when you want to hit a 3MB limit?

It does and very smoothly too thanks to dav1d.

because the file is and it's about 2 seconds long, captured at 1080p/60 10Mbps
also I don't know why -crf stuck around, I thought I didn't copy it over but apparently I did
Still, I'm not gonna retry, 159 frames took well over 10 minutes to render

That's because you've never tried anything better

>libaom

Oh. So the 3.6 seconds cut-off is just wrongly reported information?

is there a better encoder available currently?
It SHOULD be, unless it's counting it from keyframe instead of absolute frame, or it's just a off-by-one error, since the duration IS off by one second (2.6 vs 3.6)

That's sounds more like some ffmpeg fuckery to be honest.

I'm sure it is desu
Still, taking 10 minutes to encode 2 seconds of video is not feasible

The new Intel encoder seems promising.

SVT-AV1 is two orders of magnitude faster, but it's also pretty hectic as far as development goes
github.com/OpenVisualCloud/SVT-AV1
20 commits a week is way too fast, and half of the commits break compilation
granted, it's been on github for a total of less than 20 days
I'm excited to see if they can manage to at least match a third of the encode speed that VP9 has, gonna be a no-brainer if they can