Proprietary vs. Open-Source Streaming Codecs
What are some of the key advantages of proprietary versus open-source codecs? There are many factors that come into play that are primarily based on finding the right balance of customer needs in various use cases.
“Why are people choosing your codecs as opposed to open-source for either h.264 or SVT-AV1?” Jan Ozer, Principal, Streaming Learning Center, Contributing Editor Streaming Media, asks Thomas Davies, Distinguished Engineer, Visionular.
“We have a unique situation and we're very lucky that there are two very good open-source implementations with generous licensing that you can look at and that's raised the whole bar for the field of paid-for codex,” Davies says. He emphasizes that Visionular’s customers are ultimately looking for the “holy trinity” of speed, quality, and bitrate. Different customers will have different thresholds of trade-offs between those elements in order to reach their production goals. “Often you are trying to get to a new resolution, a new application, and you can't do it at the bit rates that you want to, at the density that you want to,” he says. “So we see ourselves as getting a big premium over the open-source solutions in those dimensions…but which dimension you get the premium in depends on how you want to tune it – how you want to trade off speed versus quality versus bit rate.”
Davies says that Visionular is ideal for customers who seek to highly optimize the production setup for their specific requirements. He notes that customers range from very high-end cases looking to encode at 1 FPS or slower to more mid-range requirements which entail “3 to 5 FPS, suitably parallelized across a server, necessarily,” he says. “And then at the other end, you move RTC use cases where you are trying to actually provide not just real-time, but much faster than real-time because you'd like to use a fraction of a CPU, not a whole CPU in many RTC applications. So, you have to spot a very wide range of operating points and hit the sweet spot in each one, depending on how the customer wants to balance those things.”
David Ronca, Director, Video Encoding, Facebook says that SVT-AV1 is a very good implementation of AV1. “The fundamental level is production-grade,” he says. “But there's a lot of capabilities and features that are not there.” He notes that the purpose has never been to create just a single codec that everybody can use. “The intent is really to seed the market with a very good baseline,” he says. “I think that Visionular and others are showing that there is a real value in closed-source.” Still, he says, while SVT-AV1 is a great baseline for a production encoder, there are many aspects remaining that can be expanded upon. “It’s really up to the ‘Visionulars’ of the world to take that baseline and turn it into a tier 1 production-grade,” he says.
Learn more about codecs at Streaming Media West 2022.
Watch full-session videos from Streaming Media Connect 2022.