Let's discuss the AV1 codec

Regarding video codecs, it's clear that h264 is senile and should be put out of its misery.
HEVC (h265) is slowly gaining interest (you see hardware acceleration being implemented and some pirate release groups also publishing in it), but it is heavily shackeled by the shekelseeking MPEG.

Brand spanking new is the Alliance for Open Media's AV1 codec. Codebase should (finally) be frozen Q4 2017, hardware acceleration coming in 2018.
en.wikipedia.org/wiki/AOMedia_Video_1
fosdem.org/2017/schedule/event/om_av1/

Do you think we will finally see a truly popular successor to h264 in 2018?

Other urls found in this thread:

bitmovin.com/bitmovin-supports-av1-encoding-vod-live-joins-alliance-open-media/
elecard.com/page/aom_av1_vs_hevc
medium.com/@mbebenita/av1-bitstream-analyzer-d25f1c27072b
en.wikipedia.org/wiki/Alliance_for_Open_Media#Members
en.wikipedia.org/wiki/MPEG_LA)
streamingmediaglobal.com/Articles/ReadArticle.aspx?ArticleID=118062
en.wikipedia.org/wiki/High_Efficiency_Video_Coding#Patent_License_Royalties
mpegla.com/Lists/MPEG LA News List/Attachments/91/n-14-09-29.pdf
twitter.com/SFWRedditVideos

I was researching some better lossy formats for my new build and from what I can tell H.265 for videos, aac for audio and pjg for images is the way to go at the moment.

Latest comparison by bitmovin (April 18th, 2017):
bitmovin.com/bitmovin-supports-av1-encoding-vod-live-joins-alliance-open-media/

Says it all. I here welcome our new codec overlord, AV1.

Also check the member list on wiki. It's obvious it will be the 'one codec to rule them all' situation within the next few years.

Also check other comparisons:
elecard.com/page/aom_av1_vs_hevc (April 10, 2017)
medium.com/@mbebenita/av1-bitstream-analyzer-d25f1c27072b (February 10, 2017)

I don't know. It all depends on the decision of big companies and pirate groups.

>The AV1 codec has its roots in the codebase of Google’s VP9/VP10 codec with an additional 77 experimental coding tools that have been added and are under consideration. Out of that 77 experimental coding tools, only 8 are currently enabled by default (adapt_scan, ref_mv, filter_7bit, reference_buffer, delte_q, tile_groups, rect_tx, cdef), but the performance of the codec is already appealing. The final goal is to get as many promising coding tools into the final version of the codec and afterwards freeze the bitstream specification.

>only 8 out of 77 proposed futuristic coding tools enabled now
>already beating the shit out of competition

Just freeze the bitstream already.

>Do you think we will finally see a truly popular successor to h264 in 2018?
2018 seems a bit early since we'll still need a high-performance encoder to make AV1 viable.
But I have no doubt that it will eventually replace h264.

will Daala ever be worth a damn thing?

>now. It all depends on the decision of big companies and pirate groups.

en.wikipedia.org/wiki/Alliance_for_Open_Media#Members

LITERALLY EVERYONE THAT MATTERS.

They basically ganged up as a big fuck you to the MPEG group after the botched job they did with hevc/h.265

All the good parts of Daala have already been implemented in AV1.

Because HEVC is so fucking complicated theoretically I wonder if this has any chance of achieving similar results?

Daala has been swallowed by AV1. It's features are part of these '77 experimental coding tools'. One that's Daala's has been explicitly stated in the link that I provided as implemented (namely PVQ).

see
TLDR: Yes, it aims for a 50% improvement over HEVC with only reasonable increases in complexity/computation power.

>They basically ganged up as a big fuck you to the MPEG group after the botched job they did with hevc/h.265

MPEG LA* of course (en.wikipedia.org/wiki/MPEG_LA)

Except the main thing about Daala can't be accomplished in AV1. Daala just lent some intra frame coding stuff, the lapping thing can't be done with AV1 or any other motion-prediction based coding standard.
If you think Daala isn't coming, you're most likely wrong. It's a step further than H265 or AV1.

A walking idiot. If it was only MPEG LA dictating licensing terms, then H265 would probably have a chance to be as widespread as H264 on the future internet. But there are more licensing parties involved, namely HEVC Advance and some Toshiba shit, which means more complex licensing and more fees.

already dead on arrival. Way too late.

>implying autistic encoders will drop x264 support

As far as I understand Daala is now a research project that will occasionally ship new tech to the AOM, which makes me doubt that it will ever be a standalone product. Their lapping tech might end up in AV2 as far as we know.

Why? Couldn't be much more dead than HEVC anyway.

You brain is fried. Both HEVC and AV1 are trying to improve over AVC, not one another. 50% improvement is the target, but it's only viable to low bitrate content (up to 4Mbps) and QHD and up content.

HEVC will be used in broadcast TV and is already used on Netflix.

Baseline h264 is like mp3. It works on everything. You wont see it disappear anytime soon.

Netflix is an AV1 partner and said they'll be switching over once the bitstream is frozen and Intel/Nvidia/Amd have hardware acceleration.

Google also announced YouTube would be making the switch as well.

It took a shitload of time for Intel to finally implement hardware acceleration for hevc. Hopefully now that they're part of this cartel they're gonna do it faster.

>HEVC
>dead

Tell that to my huge TV and movies collection.
It's been a great journey to replace old shitty AVC encodes by way superior HEVC ones.

If AV1 actually comes one day (and delivers on their promises), great. Until then, HEVC is the only worthwile codec.

Sadly AV1 hardware decoding won't be available in Volta GPUs since the bitstream won't be finalized until Q4 2017

The first top to bottom GPU family with full fixed function AV1 hardware decoding will be post Volta

New decoders are sometimes introduced mid generation. 980 doesn't have hevc but 960 has. 1080/1060 doesn't have vp9 10-bit but 1050 has

Why don't manufacturers have a FPGA type device on the die so they can introduce hardware decoding on existing chips

Because FPGAs aren't exactly power efficient compared to fixed function ASICs

Intel is doing this for custom fabbed high end Xeons for big customers (think Amazon, Google, etc).

We might see on die integrated FPGAs by 2025 for consumer chips

They'll probably stick with H264, AV1/H265 are too slow to encode via CPU

>no harmonic
>no ericsson
The companies that really matter are the ones producing equipment for the broadcast industry like encoders/decoders and IRDs.

>Intel/Nvidia/Amd have hardware acceleration
Yeah, let's talk about that.

Google gets more video uploaded to their platforms every hour than the entire broadcast industry combined broadcasts every day.

The future belongs to internet based media and its tech companies.

What's to talk about? H264 hardware acceleration didn't exist at all when it was introduced in 2003 to the public. It wasn't until 2005 nvidia and AMD had mainstream GPUs with hardware decoding.

>Way too late.
while it's an improvement with 1080p, it's going to make the difference in 4k which is not common still nowadays, some 4k stuff that I have ist at 20-25 Mbps, which is good but with AV1 could be evene better

>implying autists care about CPU time
That's what Amazon AWS is for.

I have 1080p content at over 25mbps.

4k bluray specification says average bitrate can be up to 140mbps. I doubt AV1 would comfortably make that sub 20mbps without decent amount of quality loss.

sorry, more specifically I meant 4k HEVC content from TV, 27 Mbps was a really good improvement over the probably 120 Mbps with h.264; if they reach that 50%, I can see everyone playing 4k content in the near future

"broadcast industry" earns more in a week than google on youtube in whole year

Are you pretending to be retarded or are you the real thing?

half of all scientific contributions are made by the square root of the total number of scientific contributors.

the square root of 77 is 8.7.

roughly speaking 50% of it's potential has been squeezed out by those 8 tools. Don't have your hopes that high. They're going to freeze the standard anyway

there was little need for it, the two major factors in hardware supports are;
a. internet streaming
b. consumer media playback

in 2003, internet streams were so low resolution/bitrate, that cpus had no trouble decoding them on their own, and nobody wanted to watch video on portable devices with their miniscule size and resolution screens
and dvd was the latest consumer media format, which used MPEG2 video (which there were hardware decoders for, though by 2003 most cpus could decode that just fine as well)

nowadays, (or rather, since ~2006) we have much faster internet, and bluray uses H.264, so hardware support for that makes sense

Yea from the AV1 talk I watched earlier in the year one of the main devs essentially said the majority of the improvements are in the 8 tools already part of the standard and that the rest would be tested but only a few if any would remain by the bitfreeze.

This AV1 shit has been "happening" since like 2015 and I still don't see it anywhere

So wht does that make AV1 dead without IMMEDIATE hardware acceleration? Just means another 2-3 years of H264 until AV1 hardware acceleration is more mainstream.


I doubt you'll see too many people switching to h265 given the royalty fees I've seen.

yes and no, i'm just saying H.264 was a different story, in 2003 there wasn't really a need for it to be hardware accellerated, as until bluray came out, it was only used for low quality media

with AV1, it will need hardware support to take off properly (that is, replace HEVC completely), as people will want high quality media on low power devices, it's just how the world has changed since 2003

I almost only see consumer electronics and software companies
Seems like it doesn't get a lot of traction with content creators and production hw/sw companies
I'll wait for IBC, but I'd be surprised if I saw it there
The broadcast industry is very slow with accepting standards (AVB, Ravenna support is still lacking in most hardware and software and h.265 flopped because of licensing issues. I only saw a few Chinese hardware manufacturers advertise it last year)

It isn't going to replace HEVC and it isn't ment to. HEVC will be used in established industry (tv and movie production) while AV1 will be used for online production and streaming services.

HEVC royalties just won't allow for broad use online whereas AV1 is open source. Even if AV1 were worse than HEVC you'd still see widespread use online.

>and some pirate release groups also publishing in it

who does this other than yify tier poo in loo p2p encoders

if they make it at least as good as HEVC, why would any company choose HEVC over AV1?
they are aiming to make it easily implementable in hardware, so there should be little to no reason why it couldn't be used in consumer media formats or digital TV
small, independant, personal, and internet streaming purposes are basically a given, just like how most of those already use VP8/VP9

How can it get traction with content creators if it's still only a lab standard? It would be stupid and pointless for them to speak out in favor of AV1 as they have fuck all to do with it right now.

You just named them.
I didn't say they were good release groups, just that you see it appear here and there.

I'll just leave this here

AWS CPU time costs money.

You know there's like every company behind this standard right? And that it already swallowed 3 standards

AV1 is basically the opposite of that
instead of;
>H.264? old and non-free
>H.265? expensive
>VP10? daala? thor?, merged into AV1

AV1 is a combination of all the major free alternatives to MPEG, and is patent-free, if it goes as planned, it'll be the one no-brainer codec for everyone to use

XKCD is wrong about a lot of things

All devices I own encode to MPEG. I'm not going to degrade quality switching format.

and in the future devices might start encoding to AV1, what's your point?

most people don't convert their media in general

The day I dropped that shit comic.

this is a post that displays massive ignorance. in this case, the makers of AV1 (google) know specifically that they're competing, and are trying to be the top choice. and they will be eventually.

>stick figures are all white

Wow racist much? literally shaking right now

Comments on AV1 or GTFO.

avxenc must faster than x265 at comparable quality. Nobody used VP8/VP9 because the encoder was so fucking slow.

There are a lot of influential software companies on the Alliance for Open Media and AMD and Nvidia are on it too.

No one wants to pay 3 seperate competing HEVC patent pools with uncapped fees to release content.

Everything costs money.

It's not unlimited and technically, everything is fragmented now on the HEVC licensing front. But tell me that isn't just pants on head retarded compared to even AVC (H.264). When AV1 even is remotely somewhat better or close to HEVC and you can use it for free, who wouldn't hop on?

>Nobody used VP8/VP9
what do you exactly mean with that?

Yeah, that's a pretty fucking stupid thing to say, considering youtube uses it.

>Do you think we will finally see a truly popular successor to h264 in 2018?
VP9 is now getting some widespread use thanks to Google serving it to everything except old Android phones and Netflix adopting it. Similarly, VP9 hardware decoding is now showing up on GPUs.

It took like ~3 years since the bitstream was frozen to get to this point. What do you think will happen in 2018, barely a few months after AV1 bitstream is frozen?

Youtube isn't lest profitable product for Google. it's basically on life-support.

Individuals, I mean. Try to encode regular HD video on your PC with vpxenc. It's really slow compared to x264 or x265.

Whatever happened to that Eve VP9 encoder?

libvpx is just as fast as x264, libvpx-vp9 is the slow one, but not so much if you use -speed

What you think you posted:
>"It's basically on life-support because it isn't profitable."

What you really posted:
>"How Do I Loss Leader??? What are Google AdWord?? What are Moneytization??? What are Content id?"

>libvpx is just as fast as x264

wrong

streamingmediaglobal.com/Articles/ReadArticle.aspx?ArticleID=118062

>I then chatted with David Ronca, director of encoding technologies. Discussing Netflix's current workflows, he mentioned that they weren't using the stock open-source VP9 encoder available from Google. Rather, they were encoding with technology from a company called Two Orioles, which was founded by Ronald Bultje, a lead VP9 developer.

Making bank, it sounds like.

No AV1 ASIC for Apple device would be problematic.

it is, what version are you using? there's also -threads affecting the time encoding

It's not like Apple has HEVC, either. I guess we can keep on making H.264 fallbacks for fruit products until the end of time.

They own the container and sell H.264 videos at iTunes.

>it's clear that h264 is senile and should be put out of its misery
no it's not, h.264 is super versatile and is built into pretty much every gadget made in the last decade

fuck this weeb memeshit

hurr durr why do we need advancements in tech, i wanna keep watching my comfy animu with a 20 year old encoding standard - the post

Fuck you, we need more compression for 4k, and AV1 is gonna bring it and EVENTUALLY (when you have died from jacking it tot your animu) replace h264

>advancements
not every development is an advancement

also this shit where they promise to have running code and then define the spec as the running code like google did with vp8

that's nigger tier

>The AV1 codec has its roots in the codebase of Google’s VP9/VP10 codec
it's google up to their usual tricks

they're tired of paying their license fees lol

en.wikipedia.org/wiki/High_Efficiency_Video_Coding#Patent_License_Royalties

No one wants to pay these ridiculous licensing & royalties, that's why HEVC is dead on arrival & AV1 ascendancy is pretty much guaranteed

>AV1 will make Daiz commit suicide
Can't wait.

Since Apple is one of the licensors of H.264, their hardware wouldn't support hardware acceleration of AV1.

I honestly couldn't care less about what video format I use. Does anyone else feel the same way? My collection is in fuckloads of different formats and it doesn't bother me at all. Even on laptops, software decoding takes so little power that hwdec isn't necessary at all. Even on phones software decoding is fine.

H.265, too. huh

mpegla.com/Lists/MPEG LA News List/Attachments/91/n-14-09-29.pdf

just like youtube will never use vpx/av1 because google is in that list, right?

Google didn't obtain the patents of H.265 after they bought On2 for VP8. Apple is different. They are still a part of MPEG-LA.

meh, who gives a fuck. apple can't implement hardware encoders for shit.

Hardware decoder is more important. Streaming sites would hesitate to support AV1 if Jamal and Tyron can't play videos on their? stolen iPhone.

Remember that it's Apple that forces youtube to use both h264 and Google's own codecs. Like it or not, they have a lot of power, and they're sticking with MPEG-LA (FaceTime is already h265).

>bitmovin.com/bitmovin-supports-av1-encoding-vod-live-joins-alliance-open-media/
>The new video coding format AOMedia Video 1 (AV1) is meant to replace Google’s VP9 and compete with HEVC/H265 from MPEG. The Alliance is targeting an improvement of about 50% over VP9/HEVC with only reasonable increases in encoding and playback complexity.

facetime?

Daily reminder that people who think h264 is enough are just NEETs who are upset that they need to buy new hardware to support new standards.

Feels good having income and actually being excited to buy nice hardware rather than depend on a 10 year old laptop like NEETs do.

Going to wait until AV1 4K is natively supported before I buy my sexy new desktop to replace my i5 2500k.

I highly doubt that they wouldn't write a hardware decoder for it. they'd be stupid not to. hardware decoders are simple things compared to their encoder brethren.

besides, apple has nothing to lose and everything to gain from supporting AV1. they'd be literally retarded to not have a hardware decoder for it when it starts taking center-stage. they'd literally be pissing in to the wind trying to force everyone to use shitty H265 when literally no one else would want to do it due to royalty fees. I realize that apple's pretty retarded, but I'm a bit skeptical that they're that retarded. besides, companies (like youtube) would just implement both if it came down to it, and apple would be kind of fucked in that case anyway.