Welcome to Doom9's Forum, THE in-place to be for everyone interested in DVD conversion.

Before you start posting please read the forum rules. By posting to this forum you agree to abide by the rules.

 

Go Back   Doom9's Forum > Video Encoding > New and alternative video codecs

Reply
 
Thread Tools Search this Thread Display Modes
Old 5th July 2020, 18:04   #261  |  Link
ksec
Registered User
 
Join Date: Mar 2020
Posts: 117
This makes me wonder...... If MulticoreWare will be interested in x266.
ksec is offline   Reply With Quote
Old 6th July 2020, 06:03   #262  |  Link
hajj_3
Registered User
 
Join Date: Mar 2004
Posts: 1,120
Quote:
Originally Posted by ksec View Post
This makes me wonder...... If MulticoreWare will be interested in x266.
Their best developer left a few years ago to work for a video encoding server company which is why x265 doesn't do so well compared to proprietary h265 encoders unlike x264 which beats proprietary encoders.
hajj_3 is offline   Reply With Quote
Old 6th July 2020, 07:54   #263  |  Link
Jamaika
Registered User
 
Join Date: Jul 2015
Posts: 697
On the Astra 2E (28.2 ° E) satellite emission tests appeared in VVC compression, the future successor of the HEVC / H.265 standard. These experiments are conducted by the satellite operator SES and the French technology company ATEME specializing in solutions for the supply of video content.
At the moment, popular software is not ready to handle VVC streams. Therefore, it is not easy to preview the content. Satkurier.pl editors have managed to determine so much that the image itself is emitted at a rate of 20.55 Mbit / s - probably at a resolution of 8K. Although the audio track is reporting, it is not broadcast.
Technical parameters:
Astra 2E (28.2 ° E) tp. 14 (11.973 GHz, pol. V, SR: 31000, FEC: 9/10; DVB-S2 / 8PSK)

New codec 2020.06.04
https://www.sendspace.com/file/3l7d6z

Last edited by Jamaika; 6th July 2020 at 08:04.
Jamaika is offline   Reply With Quote
Old 6th July 2020, 11:38   #264  |  Link
ksec
Registered User
 
Join Date: Mar 2020
Posts: 117
Quote:
Originally Posted by hajj_3 View Post
Their best developer left a few years ago to work for a video encoding server company which is why x265 doesn't do so well compared to proprietary h265 encoders unlike x264 which beats proprietary encoders.
You mean Beamr? ( And BEAMR 5 is a god damn good encoder ).

In hindsight x264 was sort of the outliner, it sets the bar and expectation of a free encoder way too high. Dark Shikari was driving it with burning passion.

And MulticoareWare are now more focused in ML, Al, DataScience type of business. ( Cant blame them since that is where the money are right now )
ksec is offline   Reply With Quote
Old 6th July 2020, 12:20   #265  |  Link
FranceBB
Broadcast Encoder
 
FranceBB's Avatar
 
Join Date: Nov 2013
Location: Royal Borough of Kensington & Chelsea, UK
Posts: 2,883
Speaking of which, when do you think work on x266 will begin?
FranceBB is offline   Reply With Quote
Old 6th July 2020, 13:02   #266  |  Link
LigH
German doom9/Gleitz SuMo
 
LigH's Avatar
 
Join Date: Oct 2001
Location: Germany, rural Altmark
Posts: 6,752
Not before the specs are "final"? ... Are they?
_

PS: Oh, I just learned it is. Newsletter

The licensing is promoted to be FRAND = "fair, reasonable and non-discriminatory" ... we will see.
__________________

New German Gleitz board
MediaFire: x264 | x265 | VPx | AOM | Xvid

Last edited by LigH; 6th July 2020 at 15:46.
LigH is offline   Reply With Quote
Old 6th July 2020, 16:18   #267  |  Link
nevcairiel
Registered Developer
 
Join Date: Mar 2010
Location: Hamburg/Germany
Posts: 10,342
Quote:
Originally Posted by LigH View Post
The licensing is promoted to be FRAND = "fair, reasonable and non-discriminatory" ... we will see.
Except thats not up to them, but the hundreds of patent holders, who will manage to ruin it again, I'm sure. Getting every patent holder into one pool that they can actually guarantee fair terms for is unlikely.
__________________
LAV Filters - open source ffmpeg based media splitter and decoders
nevcairiel is online now   Reply With Quote
Old 6th July 2020, 16:39   #268  |  Link
ksec
Registered User
 
Join Date: Mar 2020
Posts: 117
Quote:
Originally Posted by nevcairiel View Post
Except thats not up to them, but the hundreds of patent holders, who will manage to ruin it again, I'm sure. Getting every patent holder into one pool that they can actually guarantee fair terms for is unlikely.
In case anyone missed it, that is the purpose of MC-IF. If you look at the members page linked below [1], pretty much all patent holder are there. ( Represented by their respective group, such as Qualcomm by Velos Media)

So the single pool ( first part of the equation ) is done, getting out there terms is another matter though.

Personally I think a decent terms would be $1 per devices for hardware encoder and decoder with "NO CAP". And free software implementation for decoding. ( Preferably with Image Format exemption. VVC as an image is ridiculously good at low bpp )



[1] https://www.mc-if.org/our-members
ksec is offline   Reply With Quote
Old 6th July 2020, 16:47   #269  |  Link
nevcairiel
Registered Developer
 
Join Date: Mar 2010
Location: Hamburg/Germany
Posts: 10,342
Quote:
Originally Posted by ksec View Post
In case anyone missed it, that is the purpose of MC-IF. If you look at the members page linked below [1], pretty much all patent holder are there. ( Represented by their respective group, such as Qualcomm by Velos Media)
And that originally was also the plan of the MPEG-LA, and then HEVC Advance happened.

30 something companies seems pretty low to cover all the patents, which would have large overlap with HEVC and even H.264 still. MPEG-LA had over 40 companies in the HEVC patent pool, and thats not including those that jumped to HEVC Advance.

I hope they manage to make a fair and resonable situation happen, but knowing how many of those companies have behaved in the past, i'm not entirely confident.
__________________
LAV Filters - open source ffmpeg based media splitter and decoders
nevcairiel is online now   Reply With Quote
Old 6th July 2020, 18:08   #270  |  Link
ksec
Registered User
 
Join Date: Mar 2020
Posts: 117
Quote:
Originally Posted by nevcairiel View Post
And that originally was also the plan of the MPEG-LA, and then HEVC Advance happened.

30 something companies seems pretty low to cover all the patents, which would have large overlap with HEVC and even H.264 still. MPEG-LA had over 40 companies in the HEVC patent pool, and thats not including those that jumped to HEVC Advance.

I hope they manage to make a fair and resonable situation happen, but knowing how many of those companies have behaved in the past, i'm not entirely confident.
Well MPEG-LA started HEVC licensing with only ~20 companies. i.e They never reached an agreement and went ahead. That was how HEVC advance started.

And for the listed 3x Members in MC-IF, HEVC Advance is in there. As a matter of fact all of the current HEVC Pool are there. ( Velos, HEVC Advance, Fraunhofer) Along with Sisvel and United Patents, together there are well over a hundred companies represented and more.

AFAIK, there has never been a larger list of companies joining together for video codec.

Again, that is not to say they will agree on something. Or if the agreement will really be "reasonable" by other's definition.

Finger Crossed.
ksec is offline   Reply With Quote
Old 6th July 2020, 18:56   #271  |  Link
NikosD
Registered User
 
Join Date: Aug 2010
Location: Athens, Greece
Posts: 2,901
An official press release of VVC/ H.266 from Fraunhofer HHI

50% less bit rate for same quality of H.265

https://newsletter.fraunhofer.de/-vi...t/V44RELLZBp/1
__________________
Win 10 x64 (19042.572) - Core i5-2400 - Radeon RX 470 (20.10.1)
HEVC decoding benchmarks
H.264 DXVA Benchmarks for all
NikosD is offline   Reply With Quote
Old 6th July 2020, 20:21   #272  |  Link
hajj_3
Registered User
 
Join Date: Mar 2004
Posts: 1,120
Quote:
Originally Posted by NikosD View Post

50% less bit rate for same quality of H.265
The Mpeg-LA said that HEVC had 50% better compression than h264 so i would take this claim with a pinch of salt. I'd be happy with 35% improvement.

VVC VTM 9.3 is out: https://vcgit.hhi.fraunhofer.de/jvet...VTM/-/releases

Hopefully someone will do some extensive benchmarks using V9.3 comparing it to other codecs in terms of compression now that it has been ratified.

Last edited by hajj_3; 6th July 2020 at 20:25.
hajj_3 is offline   Reply With Quote
Old 7th July 2020, 01:31   #273  |  Link
benwaggoner
Moderator
 
Join Date: Jan 2006
Location: Portland, OR
Posts: 4,750
Quote:
Originally Posted by hajj_3 View Post
The Mpeg-LA said that HEVC had 50% better compression than h264 so i would take this claim with a pinch of salt. I'd be happy with 35% improvement.
It was "up to 50%" and there are certainly cases where that is met. I would say the best HEVC encoders are about 2x better than the best H.264 encoders of 2012 with the HEVC spec was finished. H.264 encoders have gotten better in that time frame as well, although at a slower pace. In the same way, we can expect HEVC encoders to keep getting better well into the VVC era. Heck, Beamr has shown substantial H.264 encoder efficiency improvements in the last 18 months. We've even seen big MPEG-2 encoder improvements in the last five years. Moore's Law means we're always getting more MIPS/pixel and can figure out new ways to turn a lot more compute into incremental quality gains.

Quote:
Hopefully someone will do some extensive benchmarks using V9.3 comparing it to other codecs in terms of compression now that it has been ratified.
Alas, we really can't compare codecs. We can compare encoders. That's a big reason why the 1st encoders of a new codec are far from 2x better than the mature encoders of the prior generation codec. It take a while before implementations catch up with older encoders, and even longer for encoder vendors to figure out how to optimally use all the new tools in the much more complex newer codec.

One advantage of VVC over, say, AV1, is that a VVC encoder can start with a HEVC encoder without that much tweaking, and then VVC-specific features can be added. That's how x265 started.
__________________
Ben Waggoner
Principal Video Specialist, Amazon Prime Video

My Compression Book

Last edited by benwaggoner; 7th July 2020 at 16:35. Reason: Added more details about older codecs' evolution.
benwaggoner is offline   Reply With Quote
Old 7th July 2020, 15:09   #274  |  Link
Jamaika
Registered User
 
Join Date: Jul 2015
Posts: 697
New codec 2020.07.07 v9.3
https://www.sendspace.com/file/i9uta6
Jamaika is offline   Reply With Quote
Old 9th July 2020, 17:32   #275  |  Link
soresu
Registered User
 
Join Date: May 2005
Location: Swansea, Wales, UK
Posts: 196
Quote:
Originally Posted by benwaggoner View Post
One advantage of VVC over, say, AV1, is that a VVC encoder can start with a HEVC encoder without that much tweaking, and then VVC-specific features can be added. That's how x265 started.
It would have been an advantage if VVC was in the same position that HEVC was in over VP9.

In that case HEVC came out a year earlier than VP9, and VP8 had gained little traction beyond Youtube at that point.

This time AV1 has a 2 year lead, and several encoders both open and proprietary available with ASIC decoders trickling into the consumer marketplace.

Despite not being at quality parity with libaom, SVT AV1 can do much faster than real time HD on a single socket system - I would expect that to improve even further well before performant and feature complete software VVC encoders are in the mainstream.
soresu is offline   Reply With Quote
Old 9th July 2020, 18:02   #276  |  Link
benwaggoner
Moderator
 
Join Date: Jan 2006
Location: Portland, OR
Posts: 4,750
Quote:
Originally Posted by soresu View Post
It would have been an advantage if VVC was in the same position that HEVC was in over VP9.

In that case HEVC came out a year earlier than VP9, and VP8 had gained little traction beyond Youtube at that point.

This time AV1 has a 2 year lead, and several encoders both open and proprietary available with ASIC decoders trickling into the consumer marketplace.
It's an advantage in that an equivalent maturity VVC encoder will be available sooner after standardization than has been true with AV1. Given that VVC is fundamentally better technoology, I imagine we'll see VVC encoders that beat the then-best available AV1 at quality @ perf @ bitrate by the end of 2021.

Quote:
Despite not being at quality parity with libaom, SVT AV1 can do much faster than real time HD on a single socket system - I would expect that to improve even further well before performant and feature complete software VVC encoders are in the mainstream.
But the SVT AV1 still isn't competitive with commercial live HEVC encoders with quality @ perf @ bitrate. And in offline enocding, the best AV1 still isn't consistantly meeting let alone beating the best HEVC today.
__________________
Ben Waggoner
Principal Video Specialist, Amazon Prime Video

My Compression Book
benwaggoner is offline   Reply With Quote
Old 10th July 2020, 05:03   #277  |  Link
ksec
Registered User
 
Join Date: Mar 2020
Posts: 117
We will see what happens.

I waited 5 years to prove my point TSMC will Fab 5nm SoC in 2020, I could wait another 3 for AV1.
ksec is offline   Reply With Quote
Old 13th July 2020, 00:22   #278  |  Link
soresu
Registered User
 
Join Date: May 2005
Location: Swansea, Wales, UK
Posts: 196
Quote:
Originally Posted by benwaggoner View Post
It's an advantage in that an equivalent maturity VVC encoder will be available sooner after standardization than has been true with AV1. Given that VVC is fundamentally better technoology, I imagine we'll see VVC encoders that beat the then-best available AV1 at quality @ perf @ bitrate by the end of 2021.
I guess that depends on your definition of maturity.

The main HEVC encoder in use publicly is x265 - it's 8 years old now, and doesn't come close to the original MPEG promises for the codec, albeit HW265 does so clearly someone or several someones at MCW weren't giving it their A game.

That is with them using x264 as a base to speed up initial development.

Sure x265 can definitely do very well at low bitrates, but at high bitrates? Not so much. I've seen recent high bitrate HD encodes using both x264 and x265 at 10GB, and x265 definitely still has problems there.

Meanwhile at 8 years old x264 had already categorically beaten the incumbent best codec Xvid. (or is that just my memory playing tricks on me?)

Given that, either through laziness or lack of investment the x265 codec has actually shown slower development than x264.

Even if your prediction comes true, and I am fairly pessimistic about those chances given the history of HEVC implementations and the different nature of the ML based techniques - even then, if AOM choose to use SVT as a base for AV2 development, it could launch far faster and more performant than AV1/VP9/VP8 did, and make VVC's reign far shorter than that of HEVC.

From what I have seen in the experimental branch commits so far, AOM seem to be taking a similar ML guided/augmented take to VVC for AV2, and at least Google's talks at encoding summits certainly support that direction.

So I don't think VVC will remain in the lead for fundamentally better technology for very long, to say nothing of the less than likely outcome of a salient licensing platform for VVC in the near future - and that isn't even getting into the mess of uncertainty that the EVC pushers have made on top of all of this.

It seems almost like the proprietary codec pushers are actually trying to self destruct that path at this point, either through incompetence or greed, take your pick.

To me it makes more sense for those still pushing the proprietary angle to pre pool their patents to license at a fixed rate per year to AOM so that this post standardisation licensing musical chairs can be cut off at the knees - that way everybody that wants to get paid can still be paid.

I'm sure between the huge conjoined cash reserves of the main AOM members that they can work out an equitable fixed rate rather than waiting until the last year or 2 of a proprietary codec development to engage in a free for all.

Last edited by soresu; 13th July 2020 at 00:26.
soresu is offline   Reply With Quote
Old 13th July 2020, 00:31   #279  |  Link
soresu
Registered User
 
Join Date: May 2005
Location: Swansea, Wales, UK
Posts: 196
Quote:
Originally Posted by benwaggoner View Post
And in offline enocding, the best AV1 still isn't consistantly meeting let alone beating the best HEVC today.
Do you mean to say that Visionular's results for Aurora AV1 are false?

Because they purported to be exceeding libaom quality at faster speeds than x265 all the way back at the Big Apple event last year.

If they weren't lying, I would assume that they would probably have improved it further still by now.
soresu is offline   Reply With Quote
Old 13th July 2020, 04:47   #280  |  Link
benwaggoner
Moderator
 
Join Date: Jan 2006
Location: Portland, OR
Posts: 4,750
Quote:
Originally Posted by soresu View Post
Do you mean to say that Visionular's results for Aurora AV1 are false?

Because they purported to be exceeding libaom quality at faster speeds than x265 all the way back at the Big Apple event last year.

If they weren't lying, I would assume that they would probably have improved it further still by now.
I very much doubt they are lying. But in my experience 85% of comparisons like that have failed due to some mismatch or another. It is very common for an encoder company to choose a metric and a set of test content they focus on, and then report results on that which aren't borne out using a broader set of content and scenarios.

I'm not suggesting there is anything specific to that vendor along those lines. But I personally haven't seen real-world longer-form content where AV1 is reliably superior to HEVC. AV1 encoders just don't (yet) have the breadth of psychovisual optimization across the huge range of sources and scenarios of the best HEVC encoders.

One likely exception to that is very grainy/noisy content, as AV1 can remove grain on encoding and resynthesize it on on the decoder side. I've not seen a mature degrain-and-paramterize solution integrated into an AV1 encoder. But I anticipate a good one could beat HEVC for grainy content at moderate-low bitrates. Grain just takes so many bits, and HEVC simply doesn't have an equivalent tool.

H.264 for HD-DVD also had a film grain sythesis feature. But the MIPS/pixel available 14 years ago were tiny compared to now - just degraining and parametrizing was a R&D thing, not production ready. And the synthesis model was a lot more primative, so the results couldn't be as fine tuned.
__________________
Ben Waggoner
Principal Video Specialist, Amazon Prime Video

My Compression Book
benwaggoner is offline   Reply With Quote
Reply

Thread Tools Search this Thread
Search this Thread:

Advanced Search
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off

Forum Jump


All times are GMT +1. The time now is 12:42.


Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2024, vBulletin Solutions Inc.