|
port feature automatically adjusts bitrate and resolution to ensure the highest picture quality for specific network conditions . “ In situations where you ’ re concerned with cellular coverage or network congestion , a combination of blended cellu lar and satellite — as provided by our CellSat connectivity solution — provides improved reliability , enhancing picture quality .
“ In short , like strong links in a chain , the overall system matters most ,” Schneider concludes .
When HEVC / H . 265 was first introduced , it was expected to replace AVC / H . 264 as the most effi cient video compression standard ( and is still regarded as such by many ). But what if there is a new codec that can provide approximately 35 % better compression efficiency than HEVC / H . 265 ?
Meet AV1 , an open-source , next-generation and royalty-free alternative technology for delivering video over the Internet , and which was launched in late March this year by the Alliance for Open Media ( AOMedia ).
The launch of AV1 , according to David Godfrey , vice-president and general manager , Asia-Pacific , Bitmovin , marks a “ significant shift ” in the balance of power in online entertainment by challenging traditional business models , and enabling innovation in social video , mobile video , virtual reality ( VR ) and online TV — all without having to pay historical IP owners .
According to Bitmovin , AV1 provides about 35 % better compression efficiency at the same quality , when compared to HEVC / H . 265 and VP9 . Godfrey adds : “ The emergence of AV1 as a royalty-free codec is the first time that online video innovation will be possible without making payments to a pool of incumbent , traditional media and consumer electronics companies .
“ AV1 opens the playing field to Internet companies to , potentially , take more of the lead in video entertainment . It also means more predictability as companies will not be affected by changes in patent pools .”
Supported by technology titans such as Apple , Google , Facebook , Intel and Cisco , AV1 is derived primarily from the in-house codec technologies of the founding companies , including Google ( VP9 , VP10 ), Microsoft ( Windows Media ), Cisco ( Thor ) and Mozilla ( Daala ).
Emphasising the royalty-free nature of AV1 , Carl Furgusson , vice-president , portfolio strategy , MediaKind ( formerly Ericsson Media Solutions ), says : “ AV1 is royalty-free for both equipment vendors to supply and operators to use , providing anyone joining or implementing AV1 in products
|
❝ AV1 opens the playing field to Internet companies to , potentially , take more of the lead in video entertainment .❞ — David Godfrey , Vice-President and General Manager , Asia-Pacific ,
Bitmovin
grant free use of any of their patents that are relevant to AV1 , and you do not sue anyone else for patent infringement for using AV1 .”
According to Furgusson , the main uses for AV1 will most likely be low bitrate adaptive bitrate video streaming to multi-screen and mobile devices . However , because it technically supports 4:2:0 / 4:2:2 / 4:4:4 chroma sampling , AV1 can potentially be used in a wide range of professional video applications .
“ It also supports 10bit along with 8bit , so has the bit-depth to support high dynamic range ( HDR ) video ,” Furgusson continues . “ AV1 is likely to be adopted by the big streaming giants such as YouTube and Facebook , who are currently using VP9 — as well as other video codecs — and who want to achieve maximum device and customer reach for their apps .”
Crucially , he also notes that there are no fundamental technical barriers to implementing AV1 into other streaming formats . This is despite the fact that standards groups such as MPEG are yet to update specifications like MPEG- DASH to support AV1 — so that the industry has a “ clear reference ” to ensure end-to-end interoperability , says Furgusson , who adds : “ With some care in implementation , it is technically possible to carry AV1 in MPEG-2 transport streams .”
Where compression efficiency is concerned , AV1 is a natural next step , concedes Dejero ’ s Schneider . However , he is quick to add : “ AV1 provides better compression but demands significantly more computation . Right now , if you want to encode in portable , low-power , low-latency applications , AV1 is
not applicable .”
He cites the example where Facebook conducted AV1 tests , where it took them a couple of hours to encode one second of video . “ AV1 requires 2,000 to 3,000 times more computing power than HEVC / H . 265 , so it ’ s just not practical for many applications , especially low-latency and lowpower ones ,” Schneider reiterates .
Instead , he believes that AV1 will , in the interim , be a boost for the likes of over-the-top ( OTT ) service providers such as Netflix . “ The cost to encode video is going to be distributed over millions of views , so if you need piles of computing power , that ’ s fine .”
It is true that AV1 is a more computationally intensive codec than HEVC / H . 265 , acknowledges MediaKind ’ s Furgusson . “ In best performance mode , the current reference model for AV1 can take 177 days to encode one hour of 1080p60 video ,” he notes . “ With
|
the AV1 specification now completed , the next phase of work for the AOMedia is to speed up the operation of the reference model , and gain support for AV1 in popular open-source tools and developer frameworks .”
When that happens , Furgusson is convinced that AV1 and HEVC / H . 265 can deliver comparable performance results , with the former capable of achieving the 50 %+ bitrate reductions of the latter ( when compared to AVC / H . 264 ), but on a royaltyfree basis .
Broadcasters should give serious consideration to the adoption of AV1 , urges Bitmovin ’ s Godfrey . He explains : “ Adopting AV1 is important because shrinking video files means higher quality video can be sent over the same network infrastructure for a richer , more compelling user experience , and reduce delivery costs for popular content .”
With broadcasters looking to experiment with hardware-based acceleration for encoding and decoding , Godfrey predicts the investment of “ significant engineering resources ” to bring AV1 to market very quickly .
AV1 can also provide a huge boost to the 360-degree video and virtual reality ( VR ) industry , as it allows huge video files to be delivered over existing bandwidth , says Godfrey . “ This could help drive new social video experiences on platforms such as Facebook and Instagram .
“ AV1 is also a particularly interesting opportunity for emerging markets where the highest bandwidth connection can be 100Kbps . The ability to offer reliable , non-buffering — or very infrequent buffering — for HD video on devices such as mobile is very compelling .”
Agreeing with Godfrey , MediaKind ’ s Furgusson points out that , much like HEVC / H . 265 , AV1 supports tile-based encod-
❝ AV1 is a technical achievement and a step towards reducing total cost of ownership for delivery of video to a growinggobal consumer base .❞ — Carl Furgusson , Vice-President ,
Portfolio Strategy , MediaKind
|
ing , giving it support for the latest concepts in bitrate-efficent VR / 360-degree video delivery over bandwidth-restricted networks . For higher resolution video , AV1 also supports up to 128x128 pixel macro-blocking processing — a capability higher than HEVC / H . 265 ( maximum 64x64 pixel micro-blocks ), according to Furgusson .
As to which codec operators eventually decide to go for , many factors are in play , he highlights . These include : deployment timescales , delivery network , target consumer device and decode capability , as well as the volume of consumer devices and the associated royalty and licensing costs .
“ At this nascent stage , all AV1 decodes will be software-based , making it most applicable for use on PC Web browsers , or newer and more powerful tablets / phones as the decoder ,” Furgusson says . “ Work continues to improve encoding time and reduce computational requirements . Based on commercial factors , AV1 will most likely be limited to file-based processing , while encoding of live TV channels is much further away .
“ Combining all these elements make AV1 most likely to be deployable for enterprise , social media , or video-on-demand ( VoD )/ subscription video-on-demand ( SVoD )/ catch-up TV services to high volumes of PC and multiscreen devices .”
And perhaps , AV1 is not the only candidate to start a potential codec war with HEVC / H . 265 , with a joint ITU / MPEG group already formed to create another nextgeneration video compression standard called Versatile Video Codec ( VVC ) — intended to super cede HEVC / H . 265 . According to Furgusson , VVC is on track to become a standard in 2020 , and initial testing has already shown at least 40 % more efficiency than HEVC / H . 265 .
This , he adds , makes VVC a disruptive force from a bitrate-efficiency perspective , particularly for 4K / UHD and 8K , as well as 360-degree video-over-Internet delivery . “ The advent of a new MPEG codec will inevitably prompt a whole new round of debates about what the royalty situation is going to look like , and will hopefully learn from the HEVC / H . 265 situation ,” says Furgusson , who nevertheless , describes AV1 as “ a great codec waiting to happen ”.
He concludes : “ AV1 is a technical achievement and a step towards reducing total cost of ownership for delivery of video to a growing gobal consumer base . But , in the fast-moving media world , there will be a race just around the corner , between AV1 and VVC .”
|