Untitled

people.xiph.org/~negge/NAB2019.pdf

Attached: 1280px-AV1_logo_2018.svg.png (1280x710, 37K)

Other urls found in this thread:

github.com/joedrago/avif
x0.at/x5Q.pdf
twitter.com/SFWRedditGifs

BTFO MPEG.
Billons Dav1d in chrome next month!

Hey, that propaganda piece is actually pretty convincing. I hope AV1 does well.

4.xx times the cpu times sounds better than whatever the hell it was before.
>progress.

feels good man
anyway, there's only the reference encoder, right? they do not intend on creating an optimizer encoder like they did with decoder?

>nu-mozilla behind damage control
figures, its all they can do lately, and even that turns out poorly as can be demonstrated by that pdf
just fix the bitstream so hardware implementers can do their thing, until hardware picks this up in a frozen spec this shit is dead. so yeah, all that pdf is just shit since point #1 is moot
also, if any xiphdotorg retard is reading this, stop giving nu-mozilla the microphone to talk on your behalf, it reflects bad on your shit and people stop taking you seriously

>than whatever the hell it was before
it was about 780 times the cpu before

The spec was frozen just a few months ago, do you expect hardware developers to release anything in such a short time period?
More importantly, the software decoder (dav1d) is fast enough to enable av1 even on weak hardware.

Look at slide 43. Mozilla and Xiph develop rav1e. Probably our only hope for a optimized high quality encoder.
Intel's SVT-AV1 already provides a quite impressive encoding speed (realtime 720p encoding is possible with modern hardware), but they focus on live encoding.

>The spec was frozen just a few months ago
was. but hardware vendors introduced changes required for hard implementations. right now the specs is still not frozen, regardless of what damage control is being posted behind "safe-walls"

Oh shoot, didn't see that there. Yep, as long as it's as slow as x265 I'm cool.

They initially released the spec in June 2018. The updated version was published in January 2019.
What version are you waiting for now?

AV1 encoding will never be as fast as HEVC encoding (it's considerably more complex after all). Expect a similar step-down from HEVC to AV1 as with AVC to HEVC after a few years of encoder development.

what about for pictures though

The AVIF spec is finished. Right now they don't plan to release any special reference implementation though and rely on non-official FOSS projects like github.com/joedrago/avif
This presentation x0.at/x5Q.pdf also takes a look at AVIF (next to a lot of general AV1 technicalities). For some reason Xiph deleted it from their servers.

Thanks, OP. Here is to AVIF replacing WebM.

Attached: zen.png (1190x663, 205K)

AVIF is an image format based on AV1. WebM is a container that can hold AV1 streams. They aren't competing with each other, unless you're the "WebP hack" user.

Attached: encode.png (1196x679, 198K)

I meant WebP, but I type "WebM" so much more often than "WebP" that I typed it automatically. The "WebP hack" user hasn't posted for a while.

Probably. AVIF suffers AV1's slow encoding speed, but that's not as relevant for pictures as it is for videos. And at the end of the day it offers a lot of advantages over WebP (no forced chroma subsampling, several bit depths, higher compression efficiency, no completely different algorithm for lossless encoding, etc.).
>The "WebP hack" user hasn't posted for a while.
Good. He has ruined enough perfectly fine images already.

Oh shit

Attached: fpga.png (1161x658, 133K)

Pik will dab on avif

Imagine you lived in 1919 and were presented this sentence as an example of what the English language would be like a century later.

Imagine you were living in 2019 and just weren't familiar with new image codecs and dumb memes when you read that.

The extrapolations would you then make wouldn't be quite as fun.