Skip to main content

Inside VVC: Q&A With Intel on the Next-Gen Video Codec

With video continually rising in percentage of total internet traffic—with estimates as high as 80 percent—and more and resolutions rising to UHD and beyond, engineers are hard at work developing new codecs to lighten the load. One of these is Versatile Video Coding, or VVC. 

The primary objective of VVC is to provide a significant improvement in compression performance over the existing High Efficiency Video Coding (HEVC) standard, but it will also aim to aid the deployment of higher-quality video services and emerging applications such as 360-degree and HDR video.

Jill Boyce, Intel Fellow and chief media architect at Intel Corporation

Jill Boyce, Intel Fellow and chief media architect at Intel Corporation (Image credit: Intel)

We reached out to Jill Boyce, Intel Fellow and chief media architect at Intel Corporation and a member of one of the core groups responsible for VVC, to learn more.

AV Technology: How are you involved in the development of VVC?

Jill Boyce: The VVC is being developed by the Joint Video Experts Team, or JVET. I serve as an associate rapporteur of ITU-T, one of the two parent bodies of the JVET group. (The other parent body is MPEG, which is under IEC.) I’m with Intel, I’m an Intel fellow, and in addition to these standards-related things, I chief media architect, which means I’m responsible for what goes into video codec implementations in Intel chips. So, we do hardware implementations of existing standards and will eventually do VVC once it’s finished. The joint teams between ITU-T and MPEG have developed many different major video coding standards over the years: HEVC, and before that AVC. I’ve been participating in these standards efforts going back to 1994.

AVT: How will VVC improve upon HEVC?

JB: The biggest thing is the goal of a 50-percent reduction in bitrate, so the goal is for the same quality to have half the bitrate. 

AVT: So this will reduce 4K down closer to the bitrate of 1080p?

JB: While 4K is a quadrupling of resolution, the bitrate ratio as you increase the resolution by 4X is nowhere near 4X. It varies depending upon your content, but 2.5x is a fairly common ratio when you quadruple your pixel rate.

When you cut the bitrate in half, 4K will still be a bit more than 1080p, because, content dependent, we have approximately 2.5 increase in the pixel rate, but you cut that down by 2. So we’re talking somewhere around a 1.25X increase in bitrate, but getting to have 4K instead of 1080p.

AVT: What are some other areas of improvement over HEVC?

JB: We just had a meeting last week, and one of the features that got adopted was better support for independent regions of pictures. So, if you have a large picture and you want to decode just a portion of it—let’s say you have a picture that’s split into four regions—in HEVC, we had limited support for that feature, but it was done in a non-normative way. But for VVC, there’s better support for this sub-picture, with better coding efficiency and more clear definition of normative support for sub-pictures. 

Another thing that’s being added is some support for basic scalability, spatial or quality scalability, and what’s called reference picture resampling. So within the same video sequence, if you wanted to switch between 1080p and 4K, then back to 1080p because your network bandwidth was changing dynamically, that feature will be supported in VVC.

AVT: What does VVC mean for the future of pro AV?

JB: I think it will facilitate 4K and 8K video. You can do 4K obviously with HEVC, but the coding efficiency with VVC will be significantly better. So not much of an increase over 1080p bitrate. So I think we’ll see a lot more 4K, and I think we’ll see 8K.

And with subpictures and reference picture resampling, it will provide a little bit more flexibility. One of the use cases talked about for 8K is that you might choose to capture in 8K but you might just pull out a section from it—maybe there’s just one area of interest that you care about. And because your total resolution was high enough, that section is still represented with a good enough resolution. So it enables that use case more.