NAB 2018: Bitmovin Talks Codec Comparison Test Findings
At NAB, Jan Ozer met with about a dozen companies with stakes in HEVC and/or AV1. This is another in a series of video interviews he conducted with them.
Jan Ozer: Jan Ozer here back with you from the Bitmovin booth at NAB. I'm here with Christian Feldmann.
Christian Feldmann:Hi. Nice to meet you.
At NAB, Jan Ozer met with about a dozen companies with stakes in HEVC and/or AV1. This is another in a series of video interviews he conducted with them.
Jan Ozer: What do you do with Bitmovin, Christian?
Christian Feldmann: I'm working on encoders, so I'm in the encoding team. I'm working mostly on all of the encoders that we support, ranging from AVC up to we're investigating new technologies, AV1, of course, and the standardization, which is currently going on at MPEG, so the whole range.
Jan Ozer: This is the NAB of AV1. I understand you've done some testing. What are your major findings?
Christian Feldmann: Basically, we did an evaluation with the University of Klagenfurt. We created a DASH data set. So H.264 is in there of course, HEVC, but also VP9, and AV1, using a snapshot from December from the AV1 reference software. We created a data set using various different settings, various different sequences, and for that specific test with a fixed bit rate ladder, we found that AV1 was able to outperform VP9, but also HEVC by quite a margin.
Jan Ozer: By about what percentage of this did you see?
Christian Feldmann: We could see gains of AV1 compared to HEVC of about 17% on average BD rate [Bjøntegaard-Delta bitrate].
Jan Ozer: So BD rate means at 17% lower bit rate it produced the same quality?
Christian Feldmann: Correct. PSNR basically is the test.
Jan Ozer: And what were your findings regarding encoding time? And how did you make it a fair fight? Because HEVC, obviously, can encode forever if you use those settings. How do you get it roughly equivalent on the encoding time side?
Christian Feldmann: In this test, the encoding times were not identical. Basically, so the AV1 reference encoder we used is not optimized for speed at all. It's an experimental encoder, basically, meant for experiments. It's not optimized for speed currently. We tried to estimate at what speed AV1 would operate once optimized. Really a shot in the dark. It's very hard to compare all of these settings. All of the settings are, of course, in the description of the data set and the associated papers. All of the codings, which codings we used, which settings we used, that's all in the paper.
Jan Ozer: Where can people find the papers?
Christian Feldmann:You can find the paper on our homepage, there's a link to the data set, which is publicly available and of course to the associated paper with all of the details.
Jan Ozer: So that's bitmovin.com?
Christian Feldmann:That's bitmovin.com, and then there's a blog page, and there's a link to the data set and to the paper.
Jan Ozer: What's your sense of when people will start using AV1?
Christian Feldmann: Well the standardization is just finished, so the final bitstream is now available. It will probably take some time until we can see fast and efficient encodings which can actually be used at a reasonable CPU cost, so that can take some time. But considering that the standard was just released, that's not unusual, I would say. Probably within 2018, within this year, we will see some usable encoders and some more reasonable outputs there.
Jan Ozer: What did you learn from your real-time testing? You guys did some real-time streams, last year here. Was that correct? You produced real time streams on the encode side and then decode via Firefox, is that right?
Christian Feldmann: The decoding demo's also probably available on our homepage. With the Firefox nightly, you can stream using DASH. I think the live encoding demo we did, here at the NAB, that's true. Basically, we used our horizontally scalable encoding approach, where we chunk up the video and just distribute it in the cloud, so we were able to use that in order to achieve real time encoding using the rather slow reference software encoder.
Jan Ozer: Okay. Well listen, Christian, I appreciate you spending time and have a great show.
Christian Feldmann: Thank you.
Related Articles
Mux CEO Jon Dahl talks about the company's QoE flagship QoE product and Mux Video, a new API to video hosting and streaming.
17 Apr 2018
NGcodec CEO Oliver Gunasekara talks about how his company uses field programmable gate arrays (FPGA) available on standard cloud computers to accelerate high-quality encoding of HEVC and VP9 video, and what this might mean for AV1.
17 Apr 2018
V-Nova co-founder and CEO Guido Meardi describes how his company has taken the PERSEUS codec launched three years ago and produced a finished product playable in all browsers and mobile devices, and many STB and OTT platforms.
17 Apr 2018
Streaming Media's Jan Ozer interviews Capella Systems' Ikuyo Yamada in the Capella booth at NAB and discusses HEVC, AV1, and Dolby Vision.
17 Apr 2018
Jan Ozer spoke with Greg Heil, CEO and co-founder of encoding.com about the results of the Encoding.com Global Media Format Report (2017), particularly as it relates to the actual usage of HEVC and VP9 by Encoding.com customers.
17 Apr 2018
Streaming Media's Jan Ozer interviews Beamr VP of strategy Tom Vaughan in the Beamr booth at NAB 2018 about HEVC and AV1
17 Apr 2018
Streaming Media's Jan Ozer interviews Alliance for Open Media executive director Gabe Frost at NAB 2018, talking about AV1's launch, its future, and how it will likely co-exist with HEVC
16 Apr 2018
Mozilla researchers Nathan Egge and Michael Bebenita discuss Daala's role in the AV1 codec, and how a codec can be developed without stepping on third-party IP. They also discuss early performance numbers and when AV1 decode will be available in Firefox.
16 Apr 2018
Streaming Media's Jan Ozer interviews Bitmovin CEO Stefan Lederer in the Bitmovin booth at NAB 2018 to talk about the company's $30 million funding, AV1, and its new machine learning-based encoding.
16 Apr 2018
Companies and Suppliers Mentioned