BBC R&D finds that AV1 produces better low-bitrate quality than HEVC, but the codec picture will get even muddier in 2020 as MPEG fast tracks VVC, MPEG-5 EVC, and LCEVC.
http://www.streamingmediaglobal.com/Articles/Editorial/Featured-Articles/VVC-and-AV1-Show-Efficiency-Gains%3b-MPEG-Announces-Third-Next-Gen-Codec-132632.aspx
Streaming standards AV1 and Versatile Video Coding (VVC) are on track to outperform HEVC as the industry adapts to a multi-codec future.
The most recent tests by BBC R&D comparing all three codecs side by side verified claims by AV1 developers Alliance for Open Media (AOM) that its codec has significantly reduced its computationally complexity.
At the same time, VVC was shown to outperform both HEVC and AV1 in terms of video quality output by up to 35% - a significant improvement on previous tests.
In results published earlier this month, BBC R&D found that VVC (a development of the Joint Video Experts Team and MPEG) performed 27% better than HEVC for HD sequences and 35% for UHD sequences.
AV1, on the other hand, performed very similarly to HEVC, with an average 2.5% loss over HD sequences, and 1.3% gains for UHD sequences (see graph at end of article).
Comparing AV1 to HEVC, BBC R&D found that AV1 could produce higher-quality decoded video than HEVC in low-bitrate scenarios, "which is highly desirable in the video coding field."
The time taken to process the videos through the codec was also tested. Increased processing times means increased complexity, and therefore more computational power is required.
BBC R&D found that the compression gains from VVC come at the cost of processing time. Encoding took around 6.5x longer than that of HEVC and decoding took 1.5x longer. AV1, on the other hand, takes about 4x as long to encode than HEVC, but is 8% quicker than HEVC to decode.
In comparison to the lab's previous test results last year, "AV1 has hugely improved these processing times. This again highlights the focus of [AOM] of producing a codec optimised for streaming."
It is worth mentioning that the test models used by BBC R&D are intended to only provide an insight into the possible quality a codec can achieve, and they are not optimised for speed. Encoding and decoding will typically happen with optimised software or on hardware, where processing times will be far quicker. BBC R&D acknowledges this but adds that its data does still give a general idea of the relative quality and complexity these codecs have to one another.
Taking these results at face value then, VVC development can be expected to deliver hefty compression gains over HEVC by the time it is finalised next year.
MPEG is targeting a 50% improvement in HEVC bandwidth with VVC by 2020. This will enable live UHD-2 encoded content (8K) to be delivered at less than 50Mbps by around 2020-2022.
Of course, this being compression schemes, life is not that simple.
Since VVC is an evolution of standards and technologies already used in HEVC and other codecs it will not come for free. There are concerns that VVC will be equally burdened by royalties and patent pool opacity as HEVC.
AV1 is also now being dogged by patent claims of certain AOM members including NTT, JVC, Orange, Toshiba, and Philips. A licence for use of AV1’s use in consumer displays (costing €0.32 per device), STBs and OTT devices (€0.11) administered by Sisvel International was issued in March.
To counteract both of these, MPEG has fast-tracked development of two additional competing codecs.
MPEG-5 EVC is being designed with a royalty-free profile, even though the basis for it is V-Nova’s Perseus codec.
A second "main" profile to MPEG-5 EVC adds a number of additional tools, each of which is capable, on an individual basis, of being either cleanly switched off or else switched over to the corresponding baseline tool.
"The main benefit of EVC will be a royalty-free baseline profile but with AV1 there’s already such a codec available and it will be interesting to see how the royalty-free baseline profile of EVC compares to AV1,” comments Christian Timmerer co-founder and CIO of Bitmovin in a blog following the 125th MPEG meeting.
Then at the 126th MPEG meeting, LCEVC was announced. Low Complexity Enhancement Video Coding is now the third video coding project within MPEG addressing requirements and needs going beyond HEVC. It will form part of the MPEG-5 suite.
Says Timmerer in another blog, "The new standard is aimed at bridging the gap between two successive generations of codecs by providing a codec-agile extension to existing video codecs that improves coding efficiency and can be readily deployed via software upgrade and with sustainable power consumption."
The coding efficiency target for LCEVC—to be at least as good as HEVC—has apparently been met, and the goal now is to achieve an overall encoding and decoding complexity lower than that of the codecs it is built on (AVC and HEVC).
With standardisation of LCEVC, MPEG-5 EVC, and VVC due in 2020—about the time AV1 as set to mature—the race is set to come to a head.
Competition has to be healthy even if three of the runners come from one stable, but weighing the merits of each might require as complex an algorithm as the technology itself.
It is not compression efficiency alone which will determine winner, but the processing complexity at both encoder and decoder, as well as the power efficiency and calculation of its business and financial aspects.
No comments:
Post a Comment