2/15/2013 7:07 AM
Ratification of the H.265 (High Efficiency Video Coding) standard by the ITU (International Telecommunications Union) late January clears the way for its adoption as the anointed successor to H.264/MPEG4.
Henceforth to be known as H.265, the questions now are how quickly will it ripple through the market, and when we will arrive at an all H.265 world?
The answer is that H.265 will have a quite different adoption profile from H.264, being slower to be taken up by makers of traditional managed set top boxes, but much quicker for consumer devices. When H.264 was at a similar stage a decade ago after its ratification in June 2003, there were in any case hardly any consumer devices for consumption of video content other than TVs and DVD players. This is the first big difference today, the existence of smartphones and tablets with a frenetic pace of product innovation and release that has vendors such as Apple eating their own lunch every six months or so. They will be the early adopters of H.265, which will be standard in tablets and smartphones by the end of 2014 if not a bit sooner.
By contrast, we will then be seeing only the first adoption of H.265 by the set tops based on chips that were announced by Broadcom at last month’s Consumer Electronics Show (CES) in Las Vegas.
The eagerness by the CE makers to adopt H.265 is driven by two markets, mobility and fixed line OTT, although with a different flavor in each case. For mobility bandwidth reduction is the primary driver as proliferating video threatens to bring cellular infrastructures to their knees. This will remain the case even though Cisco has just revised downwards its forecasts for mobile data volumes up till 2017 from the giddy numbers that had been projected.
So, although mobile data can and increasingly will be offloaded via WiFi onto broadband networks, even there the ability of H.265 to halve the bandwidth compared with H.264 will yield big cost savings. Over time, deployment of H.265 will also allow video quality to improve over mobile networks, but the emphasis in the short term at least will be on bandwidth reduction for video at existing resolutions.
In the case of fixed line OTT, it is the either way around, with quality in the driving seat. There, H.265 is seen as a way of delivering HD services over limited bandwidth, and I can see little reason why operators will not seize the opportunity. There is also an interesting IPTV angle, where the motive could be to increase the range rather than the quality, extending the existing service to consumers previously too far from the nearest exchange to obtain adequate QoS.
By halving the bandwidth required for a given quality it may be possible almost to double the distance. Even for those operators that currently deliver IPTV exclusively over Fiber to the Home, enabling multichannel HD services, H.265 may open up the possibility of reaching more customers via VDSL2 over copper, at distances up to 1 Km from the nearest fiber end point.
The other main driver for H.265 is ultra HD or 4K, and here, things get interesting. I have to admit to being among those who saw H.265 as inadequate for 4K transmission, given that it will generate 8X as many bits per second as most current 720p or 1080i HD services. But, that was before I looked more closely at the specification.
Although H.265 is touted as being about twice the efficiency of H.264, it will do far better than that for 4K. The reason is simply that at those high resolutions there will be even more scope for intra frame compression, because any area of the picture that is all a similar color can be represented in virtually the same number of bits irrespective of the pixel density.
For example, in the case of a sporting event where there is a lot of grass in the picture, that region can effectively be encoded with the same number of bits within ultra HD as standard definition, which means that the bit-rate reduction is correspondingly greater for the former. H.265 has been designed with this in mind through its support for larger blocks of pixels up to 64 x 64 than the 8x8 typically used in current H.264 codecs.
H.265 will also be able to exploit parallel processing, which will be employed in future codecs, by dividing the picture in the first instance into tiles that can be encoded independently of each other, giving further potential for efficient compression of ultra HD. For this reason I think that current estimates that H.265 will enable delivery over the Internet at bit rates between 20 Mbps and 30 Mbps, as against 45 Mbps for H.264, are too conservative. It looks now that H.265 could achieve a three or even fourfold improvement over H.264 for ultra HD, which means that 15 Mbps should certainly be achievable.
Even so, we are not going to be seeing widespread 4K services anytime soon, remembering also that it only really scores for TVs bigger than 50 inches and even then only given the right viewing distance. Although there is still debate over the relationship between screen size, distance and the resolution beyond which there is no discernible improvement in quality, it is clear that for several years ultra HD will be a niche market driven by the same cutting edge operators that currently offer 1080p60 HD.
It also looks unlikely there will be the same wholesale stampede to H.265 among set-top box makers that there was to H.264 But, in the case of tablets and smartphones, the stampede will be all the more notable.