The battle for video codec supremacy is not over and maybe never will be. Image: V-Nova
Release of the AV1 codec from the Alliance for Open Media (AOM) in March 2018 was supposed to mark the start of a heated battle with the H.265 (HEVC) codec for supremacy in the converging world of video entertainment, but for the time being they are more likely to coexist relatively peacefully in parallel universes.
That was the tentative conclusion of a panel debate held at the TV Connect show in London this month. Participants were divided over which one would end up prevailing or whether indeed there would be a single winner. There were also suggestions that another codec could emerge from a different party, with some common agreement that there would always be demand for new technology that appeared to offer substantial performance and efficiency gains without undue cost.
A key point often forgotten and made at the panel by Ian Nock, founder of Fairmile West Consulting and also head of the Interop Working Group at the Ultra HD Forum, is that AV1 and HEVC are not at the same stage of development and therefore cannot be compared directly at present. HEVC has already been deployed in trials and early services by a number of broadcasters for UHD content of some form, while AV1 has yet to become available in products. “It’s like comparing technologies of the first and second world war,” said Nock.
While that might be slight poetic licence, clearly HEVC would be expected to outperform AV1 at this stage and yet that point is deliberately ignored by proponents of the former. Indeed for this reason some codec experts believe that ultimately AV1 will prevail and become the dominant codec as it matures and leaves HEVC behind. This view was put forward at the panel by Dror Gill, CTO & VP Product at Beamr, an Israeli based developer of content adaptive encoding to enhance existing codecs.
AV1 will start to replace HEVC once it is available in hardware, according to Dror Gill, CTO of Beamr.
"It will be HEVC today but AV1 tomorrow,” said Gill. Widespread AV1 roll out will not happen until 2020 when enough devices support it in hardware, he reckoned. AV1 as a more recent development trades extra computational cost against higher encoding efficiency but the gains have yet to be achieved in practice and will require implementation in hardware over the next two years.
The case for AV1 replacing HEVC is not purely technical, given that both codecs will have successors incorporating more techniques such as content adaptive encoding in any case. The case is founded more on the commercial force behind the AOM, which includes all the technology big hitters including Google, Microsoft, Facebook, Amazon, Cisco, IBM, Netflix and most recently Apple. Their motives also provide a clue, which was growing frustration with the fragmented IP (Intellectual Landscape) of MPEG around HEVC and the associated prospect of escalating licensing fees. While several attempts have been made to fix that it could be that the damage has already been done with the codec baton passing from MPEG to AOM.
However, that is not yet certain and as also emerged from that London panel discussion the situation is not yet clear cut, with some doubt still over whether AV1 really will be free of patent claims given that its technology is not totally revolutionary and relies to some extent on earlier work. There is also the fact that although HEVC has not gained traction as quickly as its backers once hoped it still has considerable momentum. It has been adopted by many cash-strapped broadcasters who are reluctant to give it up and start again without compelling evidence that it really is going to be far superior. Such evidence is clearly lacking yet and as Guido Meardi, CEO and co-founder of London based codec technology developer V-Nova pointed out at the panel, HEVC is yielding some convincing efficiency gains already.
“For TV sets, it would be silly not to leverage HEVC,” said Meardi. “The important point is that HEVC and even H264 will stay around for a long time yet.”
Guido Meardi is CEO and co-founder of London based codec developer V-Nova.
V-Nova incidentally originally positioned its Perseus codec as an alternative to H.264 and HEVC but came to realize few broadcasters were willing to adopt a non-mainstream codec for distribution of primary services, even if they might do so for video contribution. Therefore, Perseus is presented more as a tool to supercharge other mainstream codecs, as a complement rather than alternative.
The one broadcaster/operator on the panel, Sky, took a similar line, while agreeing that it would still be using H.264 for some time yet on HD services. “With HEVC we can do more with less bandwidth,” said Jeff Web, Sky’s Principal Streaming Architect. “But most of my devices are AVC (H.264) only. “To change the whole transcoding farm, we need to reach a tipping point when it is worth doing.”
This highlights the reality that broadcasters will not be unduly influenced by the finer points of the codec debate and will only move when it really can be justified economically. It points to a near to midterm future where AV1 will gain wide adoption for streaming services offered by the big Internet players, while for a year or two yet HEVC will itself extend its footprint in the legacy broadcasting sphere.
The critical point will then come as broadcasters migrate increasingly to streaming delivery themselves at a time when AV1’s successor AV2 arrives, competing in turn perhaps with MPEG’s follow up to HEVC. The fate of any legal challenges against AOM will also play a part, but whatever happens it looks unlikely the whole video world will ever rally round a single codec as it did to an extent with H.264 around a decade ago. Even then the older MPEG 2 was still hanging on, while there were also some online alternatives, so there never has been just one absolutely dominant codec.