Versatile Video Coding (VVC) / H.266: HEVC successor

[ksec]

I think it would be fair to say if HEVC decoder had similar optimisation of dav1d it would be faster / lower resources. The thing is dav1d has crazy amount of Assembly code in it and still getting even more assembly optimisation as of now. There are some cases dav1d is even faster than AVC decode.

Anyway back to VVC. I cant wait to see Brazil’s ambitious TV 3.0 using VVC + LCEVC together. I gather the rest of the Broadcasting world ( Europe and Japan ) are waiting for their results as well before moving forward. As I believe this is the first time a system has been designed with both OTA and OTT together. ( Not OTA plus OTT, but both having same weight. ) Latest VVC+LCEVC shows additional 40% Bitrate reduction on top of VVC on 4K materials. Large field test in 2025 and Full commercial deployment in 2026.

[kurkosdr]

Originally Posted by ksec

Latest VVC+LCEVC shows additional 40% Bitrate reduction on top of VVC on 4K materials. Large field test in 2025 and Full commercial deployment in 2026.

So, what bitrates are we talking about for VVC+LCEVC? Assuming 25Mbps for HEVC 4K HDR10.

[Z2697]

Originally Posted by ksec

I think it would be fair to say if HEVC decoder had similar optimisation of dav1d it would be faster / lower resources. The thing is dav1d has crazy amount of Assembly code in it and still getting even more assembly optimisation as of now. There are some cases dav1d is even faster than AVC decode.

Anyway back to VVC. I cant wait to see Brazil’s ambitious TV 3.0 using VVC + LCEVC together. I gather the rest of the Broadcasting world ( Europe and Japan ) are waiting for their results as well before moving forward. As I believe this is the first time a system has been designed with both OTA and OTT together. ( Not OTA plus OTT, but both having same weight. ) Latest VVC+LCEVC shows additional 40% Bitrate reduction on top of VVC on 4K materials. Large field test in 2025 and Full commercial deployment in 2026.

Yeah, so I restricted it to actual implementations specifically, because we all know how the theoretical performance compares.

[benwaggoner]

Originally Posted by ksec

I think it would be fair to say if HEVC decoder had similar optimisation of dav1d it would be faster / lower resources. The thing is dav1d has crazy amount of Assembly code in it and still getting even more assembly optimisation as of now. There are some cases dav1d is even faster than AVC decode.

Yeah, software HEVC was never that big a thing as HEVC adoption was driven by hardware DRM & decoding scenarios. AV1 has had a much slower adoption ramp and is intrinsically a lot more complex, and so it's gotten a lot of sustained optimization over years.

Anyway back to VVC. I cant wait to see Brazil’s ambitious TV 3.0 using VVC + LCEVC together. I gather the rest of the Broadcasting world ( Europe and Japan ) are waiting for their results as well before moving forward. As I believe this is the first time a system has been designed with both OTA and OTT together. ( Not OTA plus OTT, but both having same weight. ) Latest VVC+LCEVC shows additional 40% Bitrate reduction on top of VVC on 4K materials. Large field test in 2025 and Full commercial deployment in 2026.

It is exciting indeed!

I do wonder how Brazil is dealing with the IP licensing of VVC.

[john33]

Originally Posted by benwaggoner

....

I do wonder how Brazil is dealing with the IP licensing of VVC.

I understand from a Brazilian friend that in order for patents, etc., to be recognised/acknowledged in Brazil, they have to be filed in Brazilian Portuguese. If they are not, they are ignored!

[kurkosdr]

Originally Posted by john33

I understand from a Brazilian friend that in order for patents, etc., to be recognised/acknowledged in Brazil, they have to be filed in Brazilian Portuguese. If they are not, they are ignored!

That's not unique to Brazil, Chinese patents have to be filed in Chinese, Japanese patents in Japanese etc. Some patent offices allow patents to be first filed in English (and the translation to follow some months later), but not exclusively in English.

As an aside, patents are only valid in the country that issues them, outside that country, they aren't worth the paper they are printed on. So, if you want your invention to be patented in a certain number of countries, you have to file on the patent office of every one of those countries separately. There are of course the EP patents, which apply to more than one country, but that's just only for some countries. And even then, some EPO countries (such as Greece) require a full translation within three months of the grant date to the country's official language for an EP patent to be considered valid in the country.

Lawyer firms have people that do the translation and also know how to file on the patent offices of all "important" countries and the EPO. It's also why MPEG LA's patent lists are so long btw, because one invention may be listed as several patents (of different countries).

Now, how will Brazil deal with the licensing requirements? They will pay a royalty to each licensing entity, duh.

[john33]

I think we both know the answer to that one!

[kurkosdr]

Originally Posted by john33

I think we both know the answer to that one!

Huh? I don't understand your post. But anyway, my point is, there is no magic way around SEP royalties just because you are not in the US. Companies that contribute inventions to standards organizations are smart enough to file in every "important" country.

Even software such as VLC exists because French law has a specific carve-out for computer software, other stuff still pays royalties.

[LigH]

New uploads: [Windows][GCC 14.2.0][64 bit]

Fraunhofer VVC Encoder ver. 1.13.1-rc1 9169430

Fraunhofer VVC Decoder ver. 3.0.0 ea7d0af

[birdie]

Recent takes on VVC by various companies and individuals:

Jean-Baptiste Kempf, VLC
Q: A subject near and dear to both of us is VVC. We spoke about a year ago, and you said “VVC was dead.” I was surprised the latest version of FFmpeg had a VVC decoder in it. So, all jokes aside, what are you seeing from the industry regarding VVC adoption in the companies you work with now?
A: It’s the same, really. Unfortunately, I think the VVC use cases are too small and not interesting enough to become a massive thing. And the patent situation is even worse than the last one (HEVC), so for me it’s not compelling enough for mass adoption. But at the same time – as you know – I’m actually the one who sponsored the VVC decoder in FFmpeg, because the goal of FFmpeg (and VLC) is to support everything. That doesn’t mean I believe the codec is good or not; that’s not my problem. As a technical person, as head of VLC and an active member of MPEG, we support everything new. We need to support it. We support things like VP6, VP7, VP8, even VP9. (Those codecs are far less useful!) But I still think we’ll see a dual track of H.264 and AV1 in many cases. I feel that VP8 and VP9 will slowly give way to AV1, and that HEVC will stick around for broadcasting. Outside of broadcast, it’s just going to be H.264 and AV1. I’m not very optimistic about mass adoption of VVC or other newer codecs.

Thierry Fautier, Your Media Transformation
Q: In 60 seconds or less, what are your predictions about VVC?
A: VVC: great technology, poor adoption — and I cannot fix that. I’m not going to get into the patent discussion, but it’s a typical example. You saw my post recently on LinkedIn saying that for mobile devices it was AVC and HEVC for encoders. VVC… nobody’s touching it because it’s too complicated. The good news is that a VVC decoder on mobile does work: Alibaba presented a very decent implementation up to 4K. But if I want to encode, can I do it? No. (And I know your next question is going to be about the next codec. If we can’t sell VVC, how are we going to sell the next generation?) It’s remarkable to me that they’re even pursuing it, but I don’t want to go there.
Q: So you don’t see VVC having an impact? Is it never going to succeed?
A: You cannot say “never” — and don’t quote me on that — but I’d say VVC had high expectations and not too much results so far.
Q: So far, or ever?
A: Not so far. I think Brazil will launch it. Japan will make an 8K channel with it, but it’s not impactful. What I’d like to hear is the Chinese community saying, “We cannot use American technology, so let’s lean on VVC.” VVC would then be a huge success in China and maybe beyond China… maybe.

Will Law, Akamai
Q: How much interest are you seeing in VVC?
A: To be honest, not much, especially on the CDN side. Everyone’s practical. H.264 is still the dominant codec we deliver. There’s some uptake in HEVC, especially for 4K content now, but VVC is very, very little. More AV1 would be the third one in the mix. As a CDN, we’re codec-agnostic; we see MP4 containers, but we don’t know what’s inside. A lot of people ask for stats — surely we know who’s playing what. We don’t. We would have to dig into customer data and parse their objects to figure out the codecs they’re using. It’s a privacy issue and computationally too expensive to perform that kind of analysis.

In related news:

NHK develops ‘world first’ VVC-compatible real-time multi-layer video encoder
The encoder can compress sub-content in real-time, using multi-layer encoding, which results in two videos being able to play via a single broadcast channel

[Z2697]

How is VVC not American technology? I mean it probably should be a "global technology" but using it for "no uncle sam" sounds like not practical.
China has AVS mainly for broadcasting but there's no real use case outside of broadcasting.

[ksec]

Originally Posted by Z2697

How is VVC not American technology? I mean it probably should be a "global technology" but using it for "no uncle sam" sounds like not practical.
China has AVS mainly for broadcasting but there's no real use case outside of broadcasting.

Somewhat counterintuitively VVC is pretty much Chinese. Most of the contribution in H.266 came from China, either Bytedance, Tencent, Alibaba or some other Chinese companies. With some input from Qualcomm and a few other EU companies. There are still US companies representatives in the H.266 development, but in terms of actual R&D it is pretty much Chinese + Japanese + EU.

Most of the US companies went with AOM.

Interestingly, may be Thierry Fautier doesn't know. Tencent is already trialing VVC streaming. Just not widely adopted yet

Q: I heard from another interviewee that China was not adopting anything from the Alliance for Open Media (like AV1) because they prefer open standards. Is that what you’re saying about China and VVC adoption?
A: China is indeed a market where things might go faster than in other parts of the world. There’s also the next-generation broadcast initiative in Brazil (TV 3.0) where VVC is in the toolbox. So there are a couple of places where I think the interest in VVC is more concrete at this point.

[GeoffreyA]

Originally Posted by ksec

Somewhat counterintuitively VVC is pretty much Chinese. Most of the contribution in H.266 came from China, either Bytedance, Tencent, Alibaba or some other Chinese companies. With some input from Qualcomm and a few other EU companies. There are still US companies representatives in the H.266 development, but in terms of actual R&D it is pretty much Chinese + Japanese + EU.

Most of the US companies went with AOM.

Interestingly, may be Thierry Fautier doesn't know. Tencent is already trialing VVC streaming. Just not widely adopted yet

Any technology with wider involvement is welcome, particularly with the rise of economic nationalism west of the Atlantic, but at the end of the day, the proof of the pudding is in the eating. In a world without AV1, VVC would have likely gained more interest.

[Z2697]

I mean, China's "three letter agencies" have AVS that's better suited for that purpose.
The companies? They are free to use whatever they want, and for some reason they like to "show off" / bluff that very well. Maybe because many of them are / have streaming platforms.
(Just see how many Made in China encoders are in the MSU test recent years)
And they have a lot of AV1 stuff as well.

(Talking about MSU, I think its reference value is declining super fast, presumably after all those chinese companies joined the show and topping off the charts. I respect the developers and efforts that put into those encoders, but it's genuinely hard to believe they really are THAT performant. But you don't have any publically available executable to verify the result.)

[modus-ms325c]

Originally Posted by GeoffreyA

In a world without AV1, VVC would have likely gained more interest.

actually VVC would still be avoided like the plague, leading to AVC still being used regardless of anything.
plus, tech companies would've scrambled to rush out a video codec that's easier to adopt while putting on a show pretending that their shiny new codec is actually much better than MPEG's offerings.

all-in-all, doesn't matter anymore. if VVC can be declared dead owing to patent situations making it difficult/impossible to even use it, let's just follow suit by calling it such, dance on its "corpse", have a ball at doing so, and call it a day.

[kurkosdr]

Originally Posted by modus-ms325c

actually VVC would still be avoided like the plague, leading to AVC still being used regardless of anything.

Most people forget that, when it comes to free-to-view web video, AVC is the exception, not the rule. We were largely using VP6 and VP7 before AVC (commonly packaged in flv files), not MPEG4 ASP, and we are now using VP9 and what became of VP10 (AV1) as the successor to AVC, not HEVC or VVC.

For one bright moment in time, the patent holders behind an ISO format (AVC) came together and collectively waived content fees for free-to-view web video, making VP8 irrelevant. But again, that was the exception to the rule. It was VP formats before that, and it's VP formats after that. Even without AOM, we'd still have gotten new VP formats from On2 Technologies, but arguably not as good.

[Z2697]

Originally Posted by kurkosdr

Most people forget that, when it comes to free-to-view web video, AVC is the exception, not the rule. We were largely using VP6 and VP7 before AVC (commonly packaged in flv files), not MPEG4 ASP, and we are now using VP9 and what became of VP10 (AV1) as the successor to AVC, not HEVC or VVC.

For one bright moment in time, the patent holders behind an ISO format (AVC) came together and collectively waived content fees for free web video, making VP8 irrelevant. But again, that was the exception to the rule. It was VP formats before that, and it's VP formats after that. Even without AOM, we'd still have gotten new VP formats from On2 Technologies, but arguably not as good.

I was just simply not born yet that time
But I think it weren't that long between the dawn of (free) online video streaming and the adpotion of AVC as the major codec.

[edison]

https://git.ffmpeg.org/gitweb/ffmpeg...7.1:/Changelog

version 7.1:
- Raw Captions with Time (RCWT) closed caption demuxer
- LC3/LC3plus decoding/encoding using external library liblc3
- ffmpeg CLI filtergraph chaining
- LC3/LC3plus demuxer and muxer
- pad_vaapi, drawbox_vaapi filters
- vf_scale supports secondary ref input and framesync options
- vf_scale2ref deprecated
- qsv_params option added for QSV encoders
- VVC decoder compatible with DVB test content
- xHE-AAC decoder
- removed DEC Alpha DSP and support code
- VVC encoding support via libvvenc
- perlin video source
- D3D12VA HEVC encoder
- Cropping metadata parsing and writing in Matroska and MP4/MOV de/muxers
- Intel QSV-accelerated VVC decoding
- MediaCodec AAC/AMR-NB/AMR-WB/MP3 decoding
- YUV colorspace negotiation for codecs and filters, obsoleting the
  YUVJ pixel format
- Vulkan H.264 encoder
- Vulkan H.265 encoder
- stream specifiers in fftools can now match by stream disposition
- LCEVC enhancement data exporting in H.26x and MP4/ISOBMFF
- LCEVC filter
- MV-HEVC decoding
- minor stream specifier syntax changes:
    - when matching by metadata (:m:<key>:<val>), the colon character
      in keys or values now has to be backslash-escaped
    - in optional maps (-map ....?) with a metadata-matching stream specifier,
      the value has to be separated from the question mark by a colon, i.e.
      -map ....:m:<key>:<val>:? (otherwise it would be ambiguous whether the
      question mark is a part of <val> or not)
    - multiple stream types in a single specifier (e.g. :s:s:0) now cause an
      error, as such a specifier makes no sense
- Mastering Display and Content Light Level metadata support in hevc_nvenc
  and av1_nvenc encoders
- libswresample now accepts custom order channel layouts as input, with some
  constrains
- FFV1 parser

Which Intel GPU support QSV VVC decoding?

[olduser217]

Originally Posted by ksec

Somewhat counterintuitively VVC is pretty much Chinese. Most of the contribution in H.266 came from China, either Bytedance, Tencent, Alibaba or some other Chinese companies. With some input from Qualcomm and a few other EU companies. There are still US companies representatives in the H.266 development, but in terms of actual R&D it is pretty much Chinese + Japanese + EU.

Most of the US companies went with AOM.

Interestingly, may be Thierry Fautier doesn't know. Tencent is already trialing VVC streaming. Just not widely adopted yet

https://www.lexisnexisip.com/wp-cont...ogy-Report.pdf

It seems like Qualcomm is still one of the largest, if not the largest contributor & patent holder related to VVC.

According to the report, VVC is probably more like an Asian dominant standard, rather than "pretty much Chinese", since both Japan & South Korea (and dont' forget Mediatek from Taiwan as well) have a lot of contributions and patents related to it as well.

[olduser217]

Originally Posted by Z2697

I mean, China's "three letter agencies" have AVS that's better suited for that purpose.
The companies? They are free to use whatever they want, and for some reason they like to "show off" / bluff that very well. Maybe because many of them are / have streaming platforms.
(Just see how many Made in China encoders are in the MSU test recent years)
And they have a lot of AV1 stuff as well.

Interestingly, AV1 is used by one of the largest video sharing website (bilibili) in China, together with AVC & HEVC.