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 10th February 2024, 12:18   #921  |  Link
FranceBB
Broadcast Encoder
 
FranceBB's Avatar
 
Join Date: Nov 2013
Location: Royal Borough of Kensington & Chelsea, UK
Posts: 2,915
Quote:
Originally Posted by birdie View Post
It's not about completeness, it's about a ton of C code which needs to be thoroughly verified for bugs
I see! Well I didn't mean the mailing list, I meant when one briefly mentioned it at FOSDEM, but anyway gotcha.
Still, it's also not main10 ready.
Anyway, I guess I'll try again with a benchmark after this summer when hopefully all coding tools are gonna be supported and there's gonna be manually written assemblies for everything and see whether it closed the gap with VVDec in the meantime.

Then, as decoders mature, the question remains: will we get x266 too this year? Eheheheh
FranceBB is offline   Reply With Quote
Old 10th February 2024, 13:49   #922  |  Link
benwaggoner
Moderator
 
Join Date: Jan 2006
Location: Portland, OR
Posts: 4,783
Quote:
Originally Posted by FranceBB View Post
Then, as decoders mature, the question remains: will we get x266 too this year? Eheheheh
That has certainly been the strong indication. And hopefully x265 3.6, as we approach three years since 3.5 was released.

Hopefully x266 will be able to reuse some of the great ARM optimizations added since 3.5. Master is more than twice as fast as 3.5 on Graviton and Apple Silicon.

VVC is around where x265 was in late 2014, taking over an hour to encode one minute of high quality 2160p24 on then high-end hardware.
__________________
Ben Waggoner
Principal Video Specialist, Amazon Prime Video

My Compression Book
benwaggoner is offline   Reply With Quote
Old 17th February 2024, 11:44   #923  |  Link
FranceBB
Broadcast Encoder
 
FranceBB's Avatar
 
Join Date: Nov 2013
Location: Royal Borough of Kensington & Chelsea, UK
Posts: 2,915
Nuo Mi just finished working on Intra Block Copy, which was the last coding tool not yet supported by the decoder.
I sent him 3 real life H.266 + AC3 .ts samples using it so that he can test locally and see whether they actually work or not.
If everything goes well, hopefully they're soon gonna be able to merge it upstream.
What does it mean for us?
Well, it means that very soon we're gonna have LWLibavVideoSource() and FFVideoSource() working and able to decode H.266 in Avisynth and VapourSynth!
What does it mean for the wider community beyond Avisynth and VapourSynth?
Well, it means it that as soon as it's merged upstream, it will be ported everywhere the likes of MPV, VLC etc and now that the decoder is ready we're quite literally one step away from it finally being adopted.
So, that one will hopefully be sorted soon, which means that if x266 gets released before the end of 2024, give it a few months of refinement/optimization/tests and hopefully 2025 (next year) will finally be the year of the widespread H.266 adoption!
FranceBB is offline   Reply With Quote
Old 17th February 2024, 16:02   #924  |  Link
birdie
Artem S. Tashkinov
 
birdie's Avatar
 
Join Date: Dec 2006
Posts: 345
Quote:
Originally Posted by FranceBB View Post
Nuo Mi just finished working on Intra Block Copy, which was the last coding tool not yet supported by the decoder.
I sent him 3 real life H.266 + AC3 .ts samples using it so that he can test locally and see whether they actually work or not.
If everything goes well, hopefully they're soon gonna be able to merge it upstream.
What does it mean for us?
Well, it means that very soon we're gonna have LWLibavVideoSource() and FFVideoSource() working and able to decode H.266 in Avisynth and VapourSynth!
What does it mean for the wider community beyond Avisynth and VapourSynth?
Well, it means it that as soon as it's merged upstream, it will be ported everywhere the likes of MPV, VLC etc and now that the decoder is ready we're quite literally one step away from it finally being adopted.
So, that one will hopefully be sorted soon, which means that if x266 gets released before the end of 2024, give it a few months of refinement/optimization/tests and hopefully 2025 (next year) will finally be the year of the widespread H.266 adoption!
Adoption is a strong word here.

H.265 is nowhere near as ubiquitous as H.264 still is. On the Internet H.265 is basically not used, the scene has also largely ignored it, most smartphones still default to H.264 despite sporting a HW H.265 encoder.

Unless H.266 shows very strong improvements over ... H.264 in 1080p and 4K resolutions, it still has a very strong chance of becoming ... relatively irrelevant and used only by aficionados of modern codecs.

Yeah, both H.265 and H.266 are used or about to be used behind the scenes in production and distribution (e.g. DVB) but end users couldn't care less, and at least for me "adoption" means something which end users are exposed to/dealing with.

And VP9/AV1 both trump H.265 by a wide margin, almost everyone watches youtube daily.
birdie is offline   Reply With Quote
Old 17th February 2024, 16:26   #925  |  Link
birdie
Artem S. Tashkinov
 
birdie's Avatar
 
Join Date: Dec 2006
Posts: 345
Performance Comparison of VVC, AV1, HEVC and AVC for High Resolutions

Quote:
For quality assessment, we used objective metrics, namely Peak Signal-to-Noise Ratio (PSNR), Structural Similarity Index (SSIM) and Video Multi-Method Assessment Fusion (VMAF), which belong to full-reference methods. In terms of the Bjøntegaard-Delta (BD) model, the results showed that H.266/VVC outperforms all other codecs, namely H.264/AVC, H.265/HEVC and AV1, respectively. Averaged bitrate savings were approximately 78% for H.266/VVC, 63% for AV1 and 53% for H.265/HEVC relative to H.264/AVC, 59% for H.266/VVC, 22% for AV1 compared to H.264/AVC and 46% for H.266/VVC relative to AV1, all for 8K resolution. The provided results also varied depending on the used test sequence and resolution – with higher resolution, the effectiveness of newly developed codecs such as H.266/VVC and AV1 is greater. This confirmed the fact that the H.266/VVC and AV1 codecs have been primarily developed for videos at high resolutions such as 8K and/or UHD.
Attachments Pending Approval
File Type: png codecs.png

Last edited by birdie; 17th February 2024 at 16:34.
birdie is offline   Reply With Quote
Old 18th February 2024, 05:23   #926  |  Link
benwaggoner
Moderator
 
Join Date: Jan 2006
Location: Portland, OR
Posts: 4,783
Quote:
Originally Posted by birdie View Post
Adoption is a strong word here.

H.265 is nowhere near as ubiquitous as H.264 still is. On the Internet H.265 is basically not used, the scene has also largely ignored it, most smartphones still default to H.264 despite sporting a HW H.265 encoder.
By "Internet" do you mean web? There's huge amounts of IP delivered HEVC, including most HDR and UHD premium content. But mostly to devices and apps, not browsers. Web browser support for DRM-protected HEVC is still pretty new, so will only get used for services that can support multiple codecs.

Quote:
Unless H.266 shows very strong improvements over ... H.264 in 1080p and 4K resolutions, it still has a very strong chance of becoming ... relatively irrelevant and used only by aficionados of modern codecs.
Outside of browser use, premium content is already largely HEVC with some AV1. Companies using those today will certainly be looking towards VVC and AV2 to get further compression efficiency improvements.

Quote:
Yeah, both H.265 and H.266 are used or about to be used behind the scenes in production and distribution (e.g. DVB) but end users couldn't care less, and at least for me "adoption" means something which end users are exposed to/dealing with.
If an end user cares about what codec is being used, someone has messed up . I doubt even 2% of the population knows what codec is being used in any particular thing they're watching.

Quote:
And VP9/AV1 both trump H.265 by a wide margin, almost everyone watches youtube daily.
There are certainly demographics that watch YouTube daily, which probably overlaps with Doom9 quite a bit. The bulk of people 25+ spend less time on YouTube than subscription streaming, TV, sat, cable, etc. YouTube certainly has a huge number of streams per user per day, but not nearly as many hours per user per day, as lean-back content tends to be longer form.

The user-generated content and the premium content markets are quite different, with a big gap in how much companies are willing to pay per title to encode. The average premium title gets a LOT more views than the average YouTube clip, so the cost of slower encoding and encoding to multiple codecs gets amortized in bandwidth cost savings and customer experience improvement over a LOT more views.
__________________
Ben Waggoner
Principal Video Specialist, Amazon Prime Video

My Compression Book
benwaggoner is offline   Reply With Quote
Old 21st February 2024, 00:08   #927  |  Link
nevcairiel
Registered Developer
 
Join Date: Mar 2010
Location: Hamburg/Germany
Posts: 10,349
Quote:
Originally Posted by benwaggoner View Post
Outside of browser use, premium content is already largely HEVC with some AV1. Companies using those today will certainly be looking towards VVC and AV2 to get further compression efficiency improvements.
Not that soon, as you still need decoders. STBs, Android devices, TVs, VVC adoption hasn't exactly moved in those.
__________________
LAV Filters - open source ffmpeg based media splitter and decoders
nevcairiel is offline   Reply With Quote
Old 21st February 2024, 15:55   #928  |  Link
kurkosdr
Registered User
 
Join Date: Aug 2009
Posts: 315
Quote:
Originally Posted by benwaggoner View Post
Outside of browser use, premium content is already largely HEVC with some AV1. Companies using those today will certainly be looking towards VVC and AV2 to get further compression efficiency improvements.
My guess is that broadcasters will eventually move to VVC, but when it comes to streaming it's not as clear cut, the companies behind the ISO/ITU formats will have to earn their royalty dollars by offering compression gains that justify the price compared to the AOM codecs. Also, keep in mind streaming isn't fighting for the last Mbps of multiplex space like the broadcasters do (internet speeds tend to get faster as time goes on), so they can make do with a format that's slightly worse as long as it's free. There is a reason Netflix is experimenting with AV1 but not VVC. Things are about to get interesting.

Last edited by kurkosdr; 21st February 2024 at 15:58.
kurkosdr is offline   Reply With Quote
Old 22nd February 2024, 18:28   #929  |  Link
benwaggoner
Moderator
 
Join Date: Jan 2006
Location: Portland, OR
Posts: 4,783
Quote:
Originally Posted by kurkosdr View Post
My guess is that broadcasters will eventually move to VVC, but when it comes to streaming it's not as clear cut, the companies behind the ISO/ITU formats will have to earn their royalty dollars by offering compression gains that justify the price compared to the AOM codecs. Also, keep in mind streaming isn't fighting for the last Mbps of multiplex space like the broadcasters do (internet speeds tend to get faster as time goes on), so they can make do with a format that's slightly worse as long as it's free. There is a reason Netflix is experimenting with AV1 but not VVC. Things are about to get interesting.
AV1 hit the market years before VVC, and is actually deployable today. Netflix, Prime Video, YouTube, and others have been delivering it for a while. Nothing about that suggests that Netflix and others aren't experimenting with VVC as well! And will be experimenting with AV2, H.267, and so on, forever.

All indications are that VVC will offer real-world improvements on top of both HEVC and AV1 sufficient to merit evaluation, and potentially good ROI for adoption. The business models to figure out when to introduce another codec for what content are pretty complex. You can imagine that with content that's viewed 1M times, a 30% bitrate reduction it the top UHD HDR bitrate can yield some pretty material bandwidth cost savings.

Also, compression efficiency improvements don't just save money. Better bang for the bit means better quality for bandwidth-constrained customers. Lots of people around the world aren't able to get the top bitrates, particularly on mobile devices.
__________________
Ben Waggoner
Principal Video Specialist, Amazon Prime Video

My Compression Book
benwaggoner is offline   Reply With Quote
Old 22nd February 2024, 21:09   #930  |  Link
birdie
Artem S. Tashkinov
 
birdie's Avatar
 
Join Date: Dec 2006
Posts: 345
I wonder what's the uptake for VVC by Qualcomm/Mediatek/Apple/NVIDIA/AMD/Intel going to be. Without a HW decoder, this standard is dead in the water.

Intel Lunar Lake has been rumored to have a HW VVC decoder but that's all I heard. Looks like VVC is at the very least 2-3 years away from a widespread use/adoption and that's only if we're lucky.
birdie is offline   Reply With Quote
Old 22nd February 2024, 23:31   #931  |  Link
ShortKatz
Registered User
 
Join Date: Aug 2018
Location: Germany
Posts: 119
Quote:
Originally Posted by birdie View Post
I wonder what's the uptake for VVC by Qualcomm/Mediatek/Apple/NVIDIA/AMD/Intel going to be. Without a HW decoder, this standard is dead in the water.
At least Qualcomm is definitely working to implement VVC.
https://www.qualcomm.com/news/onq/20...e-we-there-yet
https://www.qualcomm.com/news/onq/20...-fi-inventions
ShortKatz is offline   Reply With Quote
Old 23rd February 2024, 13:49   #932  |  Link
kurkosdr
Registered User
 
Join Date: Aug 2009
Posts: 315
Quote:
Originally Posted by benwaggoner View Post
All indications are that VVC will offer real-world improvements on top of both HEVC and AV1 sufficient to merit evaluation, and potentially good ROI for adoption. The business models to figure out when to introduce another codec for what content are pretty complex. You can imagine that with content that's viewed 1M times, a 30% bitrate reduction it the top UHD HDR bitrate can yield some pretty material bandwidth cost savings.

Also, compression efficiency improvements don't just save money. Better bang for the bit means better quality for bandwidth-constrained customers. Lots of people around the world aren't able to get the top bitrates, particularly on mobile devices.
AOM will not sit still waiting for VVC to catch on, AV2 is already under development, and unlike VVC, it can be implemented at no cost (no royalties) by utilizing GPU shaders and existing decoding hardware (exposed nicely by the driver as "hardware acceleration").

This is why things are about to get interesting: AOM's strategy to release codecs "mid-cycle" of ISO/ITU's release schedule means AV2 will compete with VVC, not ECM.
kurkosdr is offline   Reply With Quote
Old 23rd February 2024, 14:09   #933  |  Link
birdie
Artem S. Tashkinov
 
birdie's Avatar
 
Join Date: Dec 2006
Posts: 345
Quote:
Originally Posted by kurkosdr View Post
AOM will not sit still waiting for VVC to catch on, AV2 is already under development, and unlike VVC, it can be implemented at no cost (no royalties) by utilizing GPU shaders and existing decoding hardware (exposed nicely by the driver as "hardware acceleration").

This is why things are about to get interesting: AOM's strategy to release codecs "mid-cycle" of ISO/ITU's release schedule means AV2 will compete with VVC, not ECM.
Not a single widespread video codec has worked this way, suddenly AV2 will be capable of running off shaders. Yeah, really.
birdie is offline   Reply With Quote
Old 23rd February 2024, 15:04   #934  |  Link
kurkosdr
Registered User
 
Join Date: Aug 2009
Posts: 315
Quote:
Originally Posted by birdie View Post
Not a single widespread video codec has worked this way, suddenly AV2 will be capable of running off shaders. Yeah, really.
AMD had done it in the past, and anyway, hardware support can be added without an additional royalty expense. That alone is the reason AV1 was a thing in 2022 but VVC is not a thing in 2024, despite those two being 2 years apart.
kurkosdr is offline   Reply With Quote
Old 23rd February 2024, 16:36   #935  |  Link
birdie
Artem S. Tashkinov
 
birdie's Avatar
 
Join Date: Dec 2006
Posts: 345
Quote:
Originally Posted by kurkosdr View Post
AMD had done it in the past, and anyway, hardware support can be added without an additional royalty expense. That alone is the reason AV1 was a thing in 2022 but VVC is not a thing in 2024, despite those two being 2 years apart.
AV1 is not a thing in 2024.

1) Not a single midrange mobile Qualcomm SoC supports it. MediaTek fairs better, see below.
2) People nowadays rarely upgrade PCs, so it will take up to a decade to upgrade the entire PC fleet (2 billion devices) for it to support HW AV1 decoding.
3) The vast majority of TV sets presently sold lack a HW AV1 decoder. I bought a brand new 55" smart LG smart TV from their 2023 lineup for my parents six months ago and it doesn't support HW AV1 decoding, nor it cannot decode it in software (too feeble for that).

Overall the vast majority of consumer devices cannot play it effectively or at all.

This is a not a H.266 vs. AV1 topic. People here discuss H.266/VVC. I understand you're enamored by the AV1 codec which Google generously donated to the community but you could advertise it somewhere else. Thanks.

Edit: MediaTek Dimensity 8200 SoC indeed supports HW AV1 decoding (kudos to them), but there's nothing from Qualcomm. It only supports it in premium Snapdragon 8 Gen 2/3 SoCs.

Apple only has iPhone 15 Pro/Pro Max which support it.

Last edited by birdie; 23rd February 2024 at 16:43.
birdie is offline   Reply With Quote
Old 23rd February 2024, 18:31   #936  |  Link
kurkosdr
Registered User
 
Join Date: Aug 2009
Posts: 315
Quote:
Originally Posted by birdie View Post
AV1 is not a thing in 2024.

1) Not a single midrange mobile Qualcomm SoC supports it. MediaTek fairs better, see below.
2) People nowadays rarely upgrade PCs, so it will take up to a decade to upgrade the entire PC fleet (2 billion devices) for it to support HW AV1 decoding.
3) The vast majority of TV sets presently sold lack a HW AV1 decoder. I bought a brand new 55" smart LG smart TV from their 2023 lineup for my parents six months ago and it doesn't support HW AV1 decoding, nor it cannot decode it in software (too feeble for that).

Overall the vast majority of consumer devices cannot play it effectively or at all.

This is a not a H.266 vs. AV1 topic. People here discuss H.266/VVC. I understand you're enamored by the AV1 codec which Google generously donated to the community but you could advertise it somewhere else. Thanks.

Edit: MediaTek Dimensity 8200 SoC indeed supports HW AV1 decoding (kudos to them), but there's nothing from Qualcomm. It only supports it in premium Snapdragon 8 Gen 2/3 SoCs.

Apple only has iPhone 15 Pro/Pro Max which support it.
Well, at least AV1 has been present in PCs sold since 2022 (RTX 30 Series, most AMD RX 6000 series, and GPUs in the intel 13th Gen CPUs), and as you correctly say it's also supported by SoCs for high-end smartphones, so streaming services can already see gains by encoding to AV1. Where is VVC decoding support 2 years later? Nowhere except the 8K UHD TV niche.

That said, I do think VVC will eventually make its way to broadcasting, so TVs will eventually start supporting for it, but by then AV2 will also be a thing with hardware decoders available (and AV1 pretty widely supported by then). And that's why things are about to get interesting.

Last edited by kurkosdr; 23rd February 2024 at 18:45.
kurkosdr is offline   Reply With Quote
Old 24th February 2024, 16:40   #937  |  Link
Tiramisu
Registered User
 
Tiramisu's Avatar
 
Join Date: Feb 2024
Posts: 1
For what it's worth Mediatek Pentonic Series already support VVC decoding and are currently being used in some Hisense consumer TVs. Unfortunately I cannot link any sources better than this Youtube video because I am unable to find other reputable sources that clearly mention the CPU being used.
Tiramisu is offline   Reply With Quote
Old 24th February 2024, 19:21   #938  |  Link
birdie
Artem S. Tashkinov
 
birdie's Avatar
 
Join Date: Dec 2006
Posts: 345
Quote:
Originally Posted by Tiramisu View Post
For what it's worth Mediatek Pentonic Series already support VVC decoding and are currently being used in some Hisense consumer TVs. Unfortunately I cannot link any sources better than this Youtube video because I am unable to find other reputable sources that clearly mention the CPU being used.
That's very old news, https://en.wikipedia.org/wiki/Versat...oding#Hardware

SoCs have been available for many months already, but TV sets not so much.
birdie is offline   Reply With Quote
Old 25th February 2024, 09:34   #939  |  Link
birdie
Artem S. Tashkinov
 
birdie's Avatar
 
Join Date: Dec 2006
Posts: 345
H.266 IBC support was merged into ffmpeg yesterday along with a ton of new fixes. Hooray!
birdie is offline   Reply With Quote
Old 25th February 2024, 09:40   #940  |  Link
v0lt
Registered User
 
Join Date: Dec 2008
Posts: 1,976
Quote:
Originally Posted by birdie View Post
H.266 IBC support was merged into ffmpeg yesterday along with a ton of new fixes. Hooray!
MPC-BE 1.6.11.164 with updated ffmpeg.
v0lt 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 11:37.


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