V-Nova’s LCEVC Technology Adopted by Brazilian Broadcast Organization
Brazil’s SBTVD Forum has been working on its next-generation broadcast/broadband solution for a while. Now, after extensive (over 5,000) and rigorous testing followed by agreement by the Brazilian Ministry of Communication, they have announced the selection of technologies that will be adopted as part of the TV3.0 Project. Among them is MPEG-5 LCEVC codec enhancement technology developed by V-Nova.
In a recent press release, V-Nova said they, along with system integrator Phase and encoding partner Harmonic responded to the Call for Proposals (CfP) in July 2020. V-Nova proposed LCEVC and Harmonic submitted Dynamic Resolution Encoding (DRE) encoding tool.
LCEVC is a processing solution that takes place before the encoder. It is basically a layering approach that transforms the incoming content into a base layer and an enhancement layer. LCEVC can process incoming 4K content to create an FHD base layer and an enhancement layer that, when combined at the receiving device, restores the 4K content. For 8K content, the base layer would be 4K. LCEVC is codec-agnostic, working with AVC, HEVC, or VVC.
Dynamic Resolution Encoding (DRE) is a tool used during the encoding process. It first analyzes the content to determine the scene’s complexity, deciding if a lower resolution version can be used while still maintaining the desired image quality. This preprocessing can save bitrate and lower delivery costs.
It is similar to VOD providers’ approach in creating the convex hull of encodes. That is, for each piece of content, multiple encodes are made at various resolutions to allow for variations in network bandwidth. The end device will automatically switch to a lower resolution encode should the network deteriorate.
A critical difference between the above, sometimes called per-title encoding, is that this requires multiple offline encodes before delivery to the CDN. With DRE, it can be done in real-time with a single-pass encoder on a scene-by-scene basis. (see more here)
An independent test lab, appointed by the SBTVD Forum and funded by the Brazilian Ministry of Communications, conducted the technical evaluation phase to evaluate the technologies that could meet the challenging targets for delivering next-generation services in Brazil. In addition to the technical evaluation, market and intellectual property aspects were also considered for the selection. LCEVC successfully complied with all the requirements of the TV 3.0 CfP.
According to V-Nova’s statement, “Following detailed technical evaluation, the combination of LCEVC with Dynamic Resolution Encoding was the only video enhancement solution selected to be used to improve a VVC encoding for broadcast and broadband delivery or H.264/AVC and H.265/HEVC encoding for the internet options.”
In our follow-up conversations, V-Nova noted that the LCEVC/DRE combination with VVC would be focused on the Over The Air (OTA) applications. In contrast, its combination with AVC and HEVC will concentrate on the Over The Top (OTT) use case. V-Nova explained that SBTVD wants to be more aggressive with OTA. They are working on new modulation and transport schemes to help make the more restricted bandwidth in OTA more useful. These will require new receivers in mobile devices or TVs.
SBTVD cannot control the client devices on the OTT side, so they wanted to be sure any rollout would work on many legacy devices. This next-generation broadcast solution is not expected to be operational until 2024.
Naturally, we also asked about the evaluation and testing of 8K content. While the CfP clearly called for the ability to deliver 8K over OTA and OTT networks in the future, the group is more focused on 4K in the near term. The 4K HDR content target is in the 5 to 10 Mbps range. This aggressive target seems to be a critical reason that LCEVC/DRE was chosen.
As for 8K, the testing appears to be more focused on validating that a solution can support 8K video so far. A lack of 8K content was one barrier to this (and an area that the 8K Association can help).
The test details have not yet been released, but part of the CfP process was to submit subjective test results of a particular tool or technology. The independent test lab then used quantitative tools to validate the claims in the CfP. Apparently, they used PSNR and SSIM metrics, but not VMAF or additional subjective testing.
Layered approaches to video distribution seem to be getting traction. While HEVC has a scalable (Layered) option, it has not gained any use we are aware of. VVC has a built-in scalable approach, and the ATSC 3.0 spec supports layered distribution. Nonetheless, LCEVC was chosen as the enhancement layer to the VVC base layer despite the existence of an organic VVC enhancement layer. Will this be the way that 8K finally comes to market?