Just like original HD, UHDTV is arriving in different flavors, in this case distinguished by frame rate and color depth rather than the basic resolution. At least the DVB Project is trying to bring order to UHD and establish a roadmap for service deployment and manufacture of fully compliant boxes, but here too there is some confusion. This is because the DVB is introducing the specification in two stages and has given approval only just now for what it calls Phase 1, defined as the standard for services being deployed in the 2014 to 2016 timeframe.
This approval for the Phase 1 specification has been given by the DVB’s Steering Board at its 77th meeting, updating the current ETSI (European Telecommunications Standards Institute) TS (Transport Stream) 101 154, its specification for the use of Video and Audio Coding in Broadcasting Applications based on the MPEG-2 Transport Stream. The latter is relevant because ETSI is the body responsible for enshrining the standard in stone for adoption throughout Europe, as well as in many other countries in Asia, Africa and Oceana.
Original HD has been deployed at three resolutions, one being 720p, which is better for fast moving sports given its progressive scan, and then 1080i, superior for slower motion video with greater picture detail because it has 1080 scanning lines rather than 720. There is also 1080p full HD, which gives the best of both worlds, but which has seen only limited deployment for linear services even now, although rather more for VoD.
At least UHD is fixed at a single resolution, 2160p, which is 2160 by 3840 pixels, four times the number of 1080p full HD. To cope with such a big increase in resolution a significant improvement in compression efficiency was essential to keep the bit rate down and this has been achieved by HEVC (High efficiency Video Coding), which has therefore been incorporated by the DVB in its Phase 1 UHD spec. The DVB has picked just a few from the wide range of options available in the HEVC Main 10 profile, selecting Level 5.1 for UHD content for resolutions up to 2160p and the less compute intensive Main profile level 4.1 for existing HD at resolutions up to 1080p.
The DVB had good news for the industry on the HEVC front in that progress tuning the coding algorithms towards optimal efficiency is progressing quicker than had been anticipated. The original hope and expectation had been that HEVC would eventually achieve about double the efficiency of H.264/MPEG-4 main profile, or halving the bit rate for a given video quality. In fact it has already exceeded that for all resolutions, while doing best of all for UHD, in line with our own original prediction.
Broadly the efficiency gain with HEVC over H.264 is in proportion to the video resolution, the reason being that at high video qualities there is more scope for compressing areas of the picture such as sky or grass within which there is little color variation. The DVB was able to cite a study for the Joint Collaborative Team on Video Coding (JCT-VC), a group of video coding experts from ITU-T, conducted in May 2014 by the BBC and University of the West of Scotland. This was a subjective study in the sense that it took mean opinion score (MOS) values to establish base cases for comparison of HEVC and H.264 at different resolutions. The finding was that for a given subjective quality, HEVC required 48% as many bits as H.264 High profile at 480p Standard Definition, 44% as many at 720p, 38% as many for 1080p and just 36% for UHD. This means HEVC is already capable of compressing UHD pictures down to little more than a third the bit rate of H.264. Put another way, UHD with HEVC will run at a bit rate just 4/3 or 33% higher than full HD using H.264 encoding.
However that is just part of the story, because tests along the same lines as the recent JCT-VC ones have also shown a strong relationship between frame rate and quality as expressed by MOS values. In particular it has emerged that frame rate needs to increase in line with screen resolution to achieve the best overall gain in quality. Even if this is not quite a linear relationship it appears that while 32 fps is fine for HD at 720p or 1080i, even 64 fps is not enough for UHD, to the extent that it becomes the weak link for quality. The DVB’s Phase 1 specifies 64 fps, but leaving scope for doubling the limit to 128 fps in the ensuing Phase 2, scheduled for release within two years for UHD deployments from 2017 onwards.
By the same token color depth also needs to be increased to achieve the full quality benefits of the higher resolutions of UHD and accordingly the DVB has specified 10 bits per pixel in Phase 1 UHD, rather than 8 bits. This increases the number of color shades the screen can display per pixel from 256 to 1024.
Along with the UHD Phase 1 the DVB has also approved technical protocols for companion screens and MPEG-DASH, which are all really part of the same overall package for what might be called next generation television. In fact the MPEG-DASH profile for delivering adaptive bitrate media over IP networks, enshrined by the DVB in its Bluebook A168, includes specific support for UHD at up to 60 fps at this stage. The DVB has also managed to cut down on the number of DASH options supported while still covering the required range of use cases, which should improve the prospects of enabling interoperability within a streaming ecosystem.
Finally the DVB has approved the specification for Companion Screens and Streams, Part 2: Content Identification and Media Synchronization. This concentrates on identification and synchronization of companion devices such as tablets or smartphones with a broadcast service delivered to either a smart TV or a display connected to a set top. This includes hooks for audio watermarking or fingerprinting as the mechanism for the companion to recognize the content in order to trigger actions such as display of secondary ads or supplementary information. It is similar in a sense to the DVB’s provision for watermarking within the latest fourth version of its CI Plus interface separating the CA from the video decoding. In both cases the DVB has enabled watermarking without specifying any particular technology or vendor.
Overall the DVB’s package highlights how UHD and DASH are still work in progress, which may make some operators as well as vendors wonder whether it is better to wait for phase 2 before plunging in. On the other hand competitive factors may compel them to run with Phase 1, given that there are already deployments of services billed as UHD or 4K that are not compliant with any published standard.