Creating a Recipe for Codec Success to Please All Palates
Last month, I hosted a Bitmovin event called Battle of the Codecs that compared various characteristics of H.264, VP9, HEVC, AV1, and VVC to identify each codec’s deployment opportunities. Preparing for this event caused me to think through the factors that convince video publishers to adopt new codecs, which I present herein.
H.264 was the last codec home run and might be the last ever. It succeeded in the broadcast space because it was a standard, was commercially reasonable, and provided better encoding efficiency than MPEG-2. H.264 dominated streaming because it outperformed VP6 and was commercially reasonable and because Adobe and Apple deployed H.264 decode, providing near playback ubiquity. The fact that H.264 was a standard eliminated the fear, uncertainty, and doubt (FUD) factor that plagues non-standards-based codecs and encouraged playback ubiquity. It also ensured a ready supply of low-cost silicon for hardware implementations. Twelve years later, H.264 is still the predominant video codec.
In contrast, VP8 performed similarly to H.264, and so had few takers, other than YouTube. VP9 offered substantially better encoding efficiency, but never got a foothold in the broadcast space and has a definite FUD factor, although this has diminished over time. Adoption has largely been limited to high-volume, longtail use cases in deployments by companies like YouTube, Facebook, and Netflix.
HEVC offers better encoding efficiency than H.264 but has been hampered by commercial terms that are widely perceived as unreasonable. Nonetheless, because HEVC is a standard, adoption has been significant in hardware markets in which royalties can be passed on to the consumer, like mobile devices, over-the-top (OTT) devices, smart TVs, and set-top boxes. As a standard, HEVC has been incorporated into most high dynamic range (HDR) technologies, locking out VP9 and VC1, at least for now. As a result, HEVC has been widely deployed by companies distributing to the living room, particularly 4K and HDR videos. HEVC is also widely deployed in markets like security, contribution, and on-camera encoding, which almost always are dominated by standards-based technologies.
In streaming, however, HEVC has very little traction and still isn’t playable in browsers representing close to 90 percent of market share on desktop PCs (Chrome, Firefox, Internet Explorer, and Opera). More than 18 months after Apple added HEVC playback to iOS/Mac/ Apple TV devices, it appears that only a handful of companies have leveraged this opportunity to improve QoE or reduce bandwidth costs.
What have we learned? First, encoding efficiency and encode/decode performance are table stakes; if the technology doesn’t outperform existing technologies, it’s got no chance. Second, producers will eagerly deploy a new codec when it provides access to a new set of devices and customers. Third, other than companies like Netflix, Facebook, or YouTube, most streaming producers won’t adopt a new codec to harvest bandwidth savings or improve QoE. Fourth, many factors can derail a codec’s market acceptance, including unreasonable commercial terms and IP-related FUD.
What does this mean for AV1? At this point, although encode times are dropping and decode requirements are seemingly benign, compression efficiency is still in doubt. Certainly, the companies in the Alliance for Open Media will ensure a ready supply of hardware for computer, mobile, and broadcast applications, as well as browser support and an impressive list of early content adopters. This and the flip-flopping importance of the streaming and broadcasting markets could propel AV1 further than VP9 could ever have hoped to go. However, some potential content adopters are concerned about potential IP-related issues, and at this point, it’s unclear if AV1 will create new markets not served by HEVC or its successor, VVC. Another possible outcome is that AV1 is primarily used by companies in the alliance, like VP9 is today.
What about VVC, the standards-based successor to HEVC, which some early tests show as having significantly greater compression efficiency than AV1? Because it will be a standard, VVC will dominate broadcast markets and secondary markets like security, on-camera encoding, and contribution. However, absent some dramatic change in royalty policy, it won’t succeed in traditional streaming markets, as is the case with HEVC today.
What does this mean for you? Absent some Tom Clancy-like event that destroys the world’s installed base of computers, set-top boxes, smart TVs, and mobile and OTT devices (an electromagnetic pulse, perhaps?), no single codec will ever replace H.264; we’re in a multi-codec world. And, as you already know, the mere fact that a codec is available for deployment doesn’t mean you should deploy it. You’ll likely find codecs that enable entry into new markets very alluring; those that allow you to serve an existing market more efficiently, not so much.
[This article appears in the November/December 2018 issue of Streaming Media Magazine as "The Recipe for Codec Success."]
Ahead of Streaming Media West, a meeting of codec experts offers new developments in leading-edge codecs, as well as field reports from companies already using them.
Is AV1 all that people expect it to be? How much better would HEVC be doing with a fair royalty policy? Look to these charts for the answers to tomorrow's codec questions.
FFmpeg 4.0 gives many video engineers their first chance to test the new AV1 codec against H.264, HEVC, and VP9. The results? In our tests, quality was impressive, but glacially slow encoding times make AV1 a non-starter for most publishers until hardware acceleration becomes available.
Companies and Suppliers Mentioned