Tìm hiểu về chuẩn H.265

The HEVC/H.265 standard, which offers a number of improvements over current H.264 implementations, has now been finalized. It should now be a matter of months until you begin to see devices (smartphones, graphics cards) that support H.265 decode, though whether these implementations will be in hardware, or shoddy, battery-sucking software, remains to be seen.

Why H.265 is superior to H.264

Repeated quality comparison tests have demonstrated that H.265 reduces file size by roughly 39-44% at the same quality compared to H.264. This figure can change dramatically depending on how quality control is measured. In subjective viewing tests, subjects reported HEVC’s quality was equal to or greater than H.264′s with a bitrate reduction of 51-74%. That’s substantially better than what a comparison of peak signal-to-noise (PSNR) values would predict. The subjective viewing tests were conducted with multiple subjects across a wide range of video samples, and while they aren’t the final word on the topic, these are hugely encouraging results.



The current HEVC standard describes three profiles: Main, Main 10, and Main Still Picture. Main supports 8-bit color while Main 10 implements 10-bit color. Both profiles limit chroma subsampling to 4:2:0. Extensions to the standard, expected in 2014, will allow for 4:2:2 and 4:4:4 sampling, and multiview video coding (3D).



H.265 is designed to use a more efficient means of encoding pixel data and incorporates larger blocks of pixels than H.264′s macroblocks did. It can divide a picture into tiles for more efficient parallel processing and decode slices independently for better resynchronization. Intra-prediction specifies 33 directional modes (up from eight in H.264) and offers better motion compensation processing and vector prediction.



Right now, there’s some overlap between what H.265 and various extended H.264 profiles can do. H.264 Hi10P, for example, supports 10-bit video. There’s even an H.264 profile (Hi444PP) that supports 4:4:4 chroma subsampling and 14-bit color. In these cases, the difference between H.265 and H.264 is that the former will eventually deliver the same capabilities using significantly less bandwidth.

The tradeoff is processing power. H.265 requires substantially more computational power to decode than H.264, Early chips that support H.265 have already been announced — Broadcom made a splash with its Brahma BCM7445 at CES earlier this month, but that chip is something of a beast. It’s a 28nm quad-core processor that’s capable of transcoding four 1080p30 streams or driving HEVC video at up to 4096×2160.

MPEG-2 and native content

So how long until you can buy an HEVC-capable device? That’s going to depend on a host of factors. Companies like AMD and Nvidia will likely integrate H.265 decode capabilities in fairly short order, though we may see features added as new components of the specification are developed. First-generation chips might only conform to the now-finalized standard, with support for H.265 multiview decode added once that extension is finalized.

Based on how various companies handled H.264 decode capabilities, we’ll probably see some fudging as far as device-level compatibility is concerned. Smartphones and tablets with enough CPU power to handle H.265 decode can be advertised as H.265-capable with software updates, even if they chew through battery life like a fat kid on cake when handling the task. The companies talking up H.265, like Qualcomm and Broadcom, see the new video standard as a way to differentiate devices and sell consumers on next-generation products. For the media companies actually responsible for serving up content, it’s quite a bit more complicated.

Right now, terrestrial cable is dominated by MPEG-2 broadcasts. The good news is that H.265 could finally give broadcasters reason to ditch MPEG-2 by reducing bandwidth consumption by 70-80% for the same content. That’s enough to provide the impetus for stepping up to full 1080p broadcasting and leaving 720p/1080i behind. The flip side, however, is that broadcasters have precious little reason to innovate. Most cable companies face limited competition in their target markets. Satellite companies will likely adopt H.265 first, thanks to the potential bandwidth savings.

Long-term, H.265 is going to be the standard of choice for UHDTV’s 4K and 8K resolutions (See: 8K UHDTV: How do you send a 48Gbps TV signal over terrestrial airwaves?) — but that brings up the other issue. Currently, there’s no native 4K content. HEVC’s finalization means that content owners now have a theoretical standard to map to, but they don’t have any unified way to deliver the content. Sony is planning a 4K film digital delivery service for customers who buy its first 4K televisions, and is marketing a new lineup of films as “Mastered in 4K.” These are 1080p films that were converted from 4K digital masters and will supposedly offer enhanced quality for a “near-4K experience.”

The Blu-ray Disc Association is investigating ways to support 4K in the Blu-ray Disc standard, but that’s not going to be simple. H.264 can theoretically be extended to cover this resolution, but the bandwidth problem is going to rear its head at that point. An H.264 4K Blu-ray film needs far more storage space than an H.265 version of the same content. Problem is, existing players don’t support 100-128GB BDXL discs, either.

There’s no solution that doesn’t break compatibility with the existing Blu-ray standard in some form. Who wins the debate may depend on who gets stuck bearing the cost. Updating to the H.265 standard wouldn’t require any changes to disc manufacturing but might require all-new players, while high-density discs might work in some current players, but would require a manufacturing overhaul.

Another big question for later this year will be console support. Sony’s PS2 and PS3 helped launch both the DVD and Blu-ray standards. The PS4 could theoretically do the same for 4K content, provided there’s some consensus about how 4K content is going to be delivered in the future and what standards support it.

For now, we’re cautious about making claims for the new standard. We know it does one thing well — reduce bandwidth consumption at a given quality level. That benefit could trickle down to battery life improvements when streaming video, provided the higher power consumption of decoding doesn’t offset the radio’s lower power consumption. It could make 1080p broadcasting and 4K video a reality — but not until Sony and other media giants decide how to package said content.