As most of you know, several video codecs from the MPEG are under active research and development, such as VVC, EVC, and LCEVC. We compared these codecs in the past on OTTVerse (see here), gave an overview of EVC, spoke to the architects of VVC, and conducted in-depth trials on LCEVC vs. H.264/AVC, which showed a 28% gain in compression efficiency for LCEVC with an H.264/AVC base vs. vanilla H.264/AVC.
Recently, at the 134th MPEG meeting (held online in April 2021), MPEG announced its first official results for the LCEVC codec verification tests. The tests indicated average bit rate savings produced by LCEVC when enhancing AVC to be approximately 46% for UHD and 28% for HD, and when enhancing HEVC, about 31% for UHD and 24% for HD. In addition, an overall benefit was also reported while improving VVC and EVC.
And V-Nova held a virtual press conference on May 19th with Guido Meardi, CEO of V-Nova & Co-Founder, and Simone Ferrara, SVP Technology & IP Strategy, outlining the licensing terms of V-Nova’s LCEVC Implementation. You can read all about it here.
We follow that up with an exclusive interview with the CEO and Co-Founder of V-Nova, Guido Meardi. We ask him about the future of LCEVC, open-source support, and the plans to take LCEVC to production faster than ever before.
Thank you so much for agreeing to this interview. OTTVerse is very happy to talk to you about LCEVC, and we also want to congratulate you on the pace at which you’ve gone through the MPEG certification process and even announced licensing terms for LCEVC. Great stuff from you and the rest of the V-Nova team.
I remember contacting you in Sept. 2020 to access the LCEVC encoder to run some tests and see the performance for myself. After I saw a 28% compression efficiency improvement over H.264/AVC (objectively), a 3x speed-up, and confirmed the results after a lot of subjective testing, I knew that this is something interesting that the industry should embrace and use to speed up the next-gen codec adoption process.
I want to ask you a few questions based on my experiments on LCEVC and my understanding of the licensing.
Krishna: So first off, it appears that a lot of the action (licensing, FDIS, etc.) have taken place during the pandemic. How did your team celebrate such important milestones while working from home?
Guido: We celebrated virtually and everyone received custom hoodies to mark both LCEVC and VC-6 final standard status. But I must concede that such a group effort and effective teamwork deserves a face to face gathering, which we are planning to have soon.
Krishna: HEVC and AV1 are fantastic codecs that would benefit a lot from LCEVC, but they face different challenges when it comes to large-scale adoption. Since LCEVC is an enhancement codec that depends on the base codec for its success, how do you view and approach base-codec adoption issues. For e.g., are you in talks with the AV1 (AOM) team to accelerate AV1 adoption?
Guido: It’s notable how in its early stages the interest in LCEVC was driven primarily by its ability to improve compression efficiency, but over time people are increasingly appreciating also the low-complexity nature of LCEVC and how it can accelerate the adoption of newer codecs1. By reducing the compute cost of encoding and making decoding on devices feasible without dedicated hardware (e.g., AV1 1080p decode on most current mobile devices, at much lower battery drain), LCEVC can play a big supporting role to accelerate adoption. We have various active discussions with partners and services related to that, so watch this space!
Krishna: In the licensing, you talk about cost per user. How do you define a “user,” and how does this definition vary for an ad-supported service vs. a subscription-based service? Please give us a glimpse into your licensing approach to these different business models?
Guido: V-Nova LCEVC licensing terms are different from the model traditionally used by other codecs: they are modern, and very similar to SaaS models. We designed these terms together with customers and understood that this would be a fair and proportionate model. A capped fee based on usage is fair to any size of customer base.
Now, subscription-based services are different to ad-based services. For ad-based, the metric is Monthly Active Users (MAU), whereas for subscription-based it will be number of subscribers. A suitable multiplier accounts for the difference between the different types of “user”.
Krishna: Can you give us an idea of the devices/eco-systems that you currently support via SDKs? Also, can you give us a peek into your roadmap for devices not currently supported.
Guido: Our SDK already supports a very wide range of encoding and device environments. For encoding we have libraries optimised for CPUs (both Intel and ARM), GPUs and even FPGA. For decoding we have optimised libraries to cover the most popular platforms including Android, iOS, Windows, MAC and scripted decoding for HTML5 capable browsers to deploy LCEVC avoiding the need for any plugins.
To facilitate adoption, we also have available multiple reference integrations for both encoders and decoders, including FFmpeg (with support for dozens of base encoders), ExoPlayer for Android, AVPlayer for iOS, Microsoft UWP for Windows and web players like HLS.js, Shaka Player, video.js.
We also have integrations at OS level, such as a patch for AOSP (Android Open Source Project) that makes LCEVC easily implementable in all AOSP-based systems. And, of course, we’re adding more options all the time – including professional vendor solutions – in conjunction with partners on active deployment projects.
Krishna: With LCEVC, teams have the option to tune both the base and enhancement codecs, individually and this can potentially be a double-edged sword. For e.g., tuning the rate control modes, bitrate allocation between the base and enhancement layers, choosing between different upsampling kernels for different content types, etc. So, in this regard, what’s your advice to teams interested in LCEVC? Any tutorials, best-practises, or resources that you can point to?
Guido: You are correct, though I see this as an upside that will help maintain differentiation in the implementation of codecs and therefore competition and subsequent further development and improvement.
At V-Nova, we offer dedicated support to help customers get the most out of LCEVC. General suggestions and best practices are available on our documentation site at https://docs.v-nova.com, to help anyone getting started with LCEVC.
However, whilst there are lots of controls available with LCEVC, we have worked hard to calibrate LCEVC defaults with a wide range of base encoders so that users can simply select from easy-to-use presets and get great results straight away.
Krishna: In your documentation, I saw a reference to the V-Nova Platform for producing LCEVC-conformant bitstreams. Is this a big part of your product strategy i.e., offering a cloud-based LCEVC encoding platform? Can you provide more information on this?
Guido: Our goal is to make LCEVC as easily accessible as possible. While we have plenty of reference integrations and applications ready, we also encourage anyone who is interested in trying LCEVC enhancement with their own content to head to https://platform.v-nova.com and sign up for a free trial. The platform supports both LCEVC-enhanced x264 and x265.
Anything you encode can be played back in the V-Nova demo apps which are available in all major app stores. We’re adding new functionality to our platform, though we suggest customers contact us to discuss their detailed needs.
That said, we do not intend to compete with any of the encoding vendors or streaming platform providers. The LCEVC platform and reference integrations are there to provide an easy way for people to evaluate the technology and to power proof-of-concept VOD and live workflows for our customers.
That brings us to the end of our interview. Thank you so much, Guido for taking the time to answer our questions. We wish you and the V-Nova team all the best for the future and we look forward to reporting on LCEVC deployments as they happen!
Krishna Rao Vijayanagar
I’m Dr. Krishna Rao Vijayanagar, founder of OTTVerse. I have a Ph.D. in Video Compression from the Illinois Institute of Technology, and I have worked on Video Compression (AVC, HEVC, MultiView Plus Depth), ABR streaming, and Video Analytics (QoE, Content & Audience, and Ad) for several years.
I hope to use my experience and love for video streaming to bring you information and insights into the OTT universe.