Why Migrating to HEVC in 2018 May Be Premature
There’s been a lot of discussion in media circles about the High Efficiency Video Codec (HEVC, or H.265 to its friends). As the latest video encoding standard from the MPEG consortium, it offers a dramatic improvement in encoding efficiency over H.264/AVC – on the order of 100%. It’s the sole supported means of encoding video for Blu-ray 4K, playback is supported on new graphics chips and computing devices, and it’s got widespread industry support. Despite all that, HEVC hasn’t enjoyed the swift adoption H.264 did more than a decade ago. As it turns out, there are good reasons it hasn’t gotten traction yet.
HEVC is hard to work with.
Under the steady march of improvements to memory capacity, processor core design, and available storage, you might imagine that video encoding is a solved problem in the PC space. But it’s a moving target: to get higher quality, throw more processing power at it. And HEVC offers awesome potential, but it’s a screeching pig for processing power.
To illustrate the difference we’re talking about, I threw a common video encoding use case at my main system at home. Transcoding a Blu-ray to a smaller file size is a traditionally time-intensive process, where you sacrifice some visual quality and bitrate for the convenience of a file that’s much smaller and easy to upload to a network share, store on a home theater PC, or pop onto a flash drive to watch at a friend’s house. In this case, I used my eight core AMD Ryzen 1700 workstation to do the job. It’s not overclocked, has 16 gigs of DDR4-2400 memory, and is running the latest version of the video processing software Handbrake. The Blu-ray was The Final Member, a documentary detailing the history of the Iceland Phallological Museum, extensive interviews with its founder, and the two men committed to having their penises forever memorialized there.
Using the default Fast 1080p30 preset, which results in a 1080p video encoded at 30 frames per second at reasonable quality and with AAC audio, the x264 encoder netted a 2.1 gigabyte file at 60.8 frames per second. That means it took about half the time to encode the movie to a smaller filesize than it would to watch it. By contrast the x265 encoder managed a 1.07 gigabyte file at 31.5 frames per second, which is an amazing space savings but means it took nearly twice as long to get the video. The visuals for each are nearly indistinguishable, but if you’re a content creator, encoding H.265 will take a lot of time to render.
HEVC has competition.
HEVC’s intense computational demands aren’t the only thing holding it back. While the software is immensely capable, it’s the product of an industry consortium like the MPEG and AVC codecs before it. Many different companies and organizations invested in its development and committed technical personnel to lend expertise, but they didn’t do it for free. Everyone involved wants to get paid for their contributions, and the royalty rates are obscene. From this article:
“In order to ship a product with HEVC support, you need to acquire licenses from at least four patent pools (MPEG LA, HEVC Advance, Technicolor, and Velos Media) as well as numerous other companies, many of which do not offer standard licensing terms (instead requiring you to negotiate terms), which can potentially cost hundreds of millions of dollars (and that’s after the recent drastic cuts to HEVC royalty fees).”
Inferior or no, AVC itself is still a capable codec enjoying wide deployment. It’s enshrined in the ATSC TV broadcast standard, YouTube still recommends it for video submission, and just about every consumer electronic that handles video playback made in the last decade can handle it without fuss. It’s still good at its job, and it’s though it isn’t free it’s much less hassle.
And it’d be foolish not to mention either of the royalty-free, open source alternative video encoding schemes deep in development. Google’s VP9 is broadly competitive with HEVC itself and less computationally intensive to play back. YouTube’s been pushing it to compatible machines (particularly desktops and laptops) for at least two years. VP9’s own royalty-free successor AV1 is under development by the Alliance for Open Media, and promises up to a 40% improvement over any of its competitors. Even leaving aside its patent encumbrances, by the time it becomes practical to work with HEVC, it may no longer be a great tool for the job.