From HD to UHD, more than just a leap
It has been more than a decade since the audiovisual environment was facing a major change: the move from SD to HD. Back then it was just a change in definition, we could -in perhaps too simplified terms- say. But what is going on with transition from HD to UHD is somewhat more comprehensive. Time to reflect a bit about it.
By Yeray Alfageme
Let us recall those times. The leap from SD to HD meant a fourfold increase in definition of the image being produced and received by TV viewers. Yes, just TV viewers, as back then digital content distribution platforms, either live or on-demand, were nearly non existent and very little known. And yes, YouTube was already there, but hardly anything else.
HD “simply” meant more pixels. Neither color space nor image scan type. Progressive images still coexisted with interlaced images. Neither sound nor image luminance changed. It is true, however that with regards to audio a change did take place, as distribution of surround audio was enabled either through Dolby or multichannel. As far as metadata were concerned, the signal was capable of conveying something more than just teletext; I know: I am oversimplifying.
However, the move from HD to UHD is definitely a more substantial change as not only definition -more pixels- is increased four times once again, but also are color space, luminance, frame speed and, of course, sound capabilities, all this resulting in better pixels.
At this point, I deem it convenient to draw a line between two worlds: OTT and traditional TV distribution, as they are dramatically different, not only in their ways but also in their nature, and transition from one environment into the other can -and must- be approached differently. Let us see in detail what I mean.
Traditional TV, SDI, satellite and DTT
HD-HDR images are already regarded as UHD signals. No definition leap is required for an image to be regarded as ultra high definition, keep that in mind. As laid down by the UHD Forum, different definition, luminance and chrominance are possible, and not only definition, within a UHD environment.
Moving from SDR to HDR image involves changing from 8-bit to 10-bit or even 12-bit depths based on the standard so chosen. With regards to contribution, what we could simplify as a SDI environment means switching to 3-Gbps linear signals instead of the regular 1.5 Gbps. Most equipment units are capable of handling these types of signals, but this must be borne in mind particularly with regards to compression and distribution. Maybe the kind of coder being used for compressing and sending out our signal for distribution will not be capable of handling signals having depths higher than 8 bits, which will be a problem when handling HDR signals.
A combination of these luminance and chrominance levels with a leap in definition up to 4K images, make increased bandwidth indispensable, both for contribution and distribution. Back to the SDI linear world, a 4K-HDR image requires 12 Gbps for non-compressed transfer, so hardly any 1.5 Gbps equipment supports this standard. There are some pseudo-standards such as 12G-SDI that allow conveyance of these kinds of signals on a single coaxial cable, but they are just that: pseudo-standards. Some manufacturers have adopted them, but letting our infrastructure leap rely on non-standard technology seems a hardly acceptable risk.
What is indeed official is transmission through the use of 4 3G-SDI signals for an effective 12 Gbps bandwidth. But this would require 4 coaxial cables, something for which many infrastructure setups are not ready either. Could a central control array split routing capacity in four in order to support these kinds of signals? Not in most cases.
At this juncture we switch to the world of fiber, nothing but a different medium for SDI signals which, nonetheless avoids having to use 4 cables; and to the IP world through BT.2110 and this is of real help. And the issue is that, in my opinion, switching from HD to UHD in a traditional linear TV environment practically forces the move from SDI to IP. Obviously, a UHD production environment can be implemented on SDI infrastructure, especially on fiber, but I think that this is a real constraint that has nowadays a much shorter life cycle when compared to implementation of an IP environment.
The IP world is more uncertain, though. Technology has matured and standards are finally there, but manufacturers and their technologies are not yet up for the job in some aspects. That was the prevailing opinion not long ago and doubts are increasingly dissipating, so I am quite convinced that investing at present in an IP production environment is a much better decision than just updating our SDI infrastructure to host UHD signals, even if it involves switching to fiber.
As a final world on distribution, even though coders and compression standards are fast improving and allow now increased quality on the same bandwidth it is still too early –about this I am more convinced- to replace a traditional TV distribution network -either satellite or DTT- for UHD in just one go. The few experiences recorded in both instances are pilot projects and the only existing commercial channels in UHD are via satellite as this makes it easier to control all parts of the distribution chain, coding, distribution channel and set-top box.
OTT and IPTV environments
I dare not say that the leap is any easier in the native streaming world, but I think that at least it is not as disruptive as in traditional TV, because of the fact that an OTT platform that had been so far broadcasting HD content may switch to UHD content distribution entails an update not as much as an intrinsic, essential change in model and infrastructure.
Be it an IPTV model in which the network is under control and streaming is not actually distributed through the Internet but by a telecom company owning the same; or the OTT model distributed through the Internet, quality standards under which content is distributed are already adaptable nowadays. The various platforms typically have a minimum of four and a maximum of nine quality profiles for carrying out distribution. Therefore it can be inferred that increasing the number of profiles in order to support UHD images turns out to be at least feasible.
A thing to bear in mind and, justifiably so, is distribution codec. Under H.264, the most usual codec for distribution of content in streaming, only certain profiles are capable of supporting HDR –and higher definition- content and not all decoders can successfully process these images. Furthermore, the bitrate required for conveying these signals with adequate quality is really high.
Options available nowadays are two: H.265 and AV-1. It is true that MPEG is working on VCC, the alternative that this organization aims to present versus AV-1, but it seems it will not be available until the end of the year. H.265 is obviously the natural evolution of codec H.264, already featuring numerous profiles capable of handling UHD signals in a comfortable way and with reasonable bitrates. However, this falls a bit short in the processing of metadata as well as in terms of flexibility when it comes to combining various pieces of content in varying resolutions or quality standards.
However, as AV-1 has the advantage of being much more modern, it has been specified for much more flexible capacity conditions, in addition to allowing for much higher compression rates while keeping a visual quality comparable to that offered by H.265 without requiring –as some might think- intensive CPU usage. This makes it particularly interesting for implementation in set-top boxes and low-performance devices connected. Giants such as Netflix or HBO have already set their sights on AV-1 and included it within their technical specifications as accepted distribution method. The only drawback to AV-1 is that not all hardware offers native support for coding and decoding this standard, although I think this is something that will be fixed sooner than later.
One final consideration: starting to distribute UHD content through a different codec does not mean we should change our entire header and content distribution systems or transcode all our content into the new format. The good side of streaming is that it enables combining various formats in an easy fashion provided that systems and software applications used support this, although the task is no doubt far simpler here than in traditional production environments.
In sum. In traditional production environments the leap into UHD entails -almost inevitably, if we want to go for a long-term investment- a switch to IP supported by BT.2110 as this will enable us to handle signals under any format in a more flexible way; however, this change is a substantial one and it may turn impossible in some instances. In a streaming environment the change is less marked. Nevertheless, essential decisions in regard to distribution codec to be used will make our move permanent or temporary while we wait for the industry to mature and our competitors get, in some instances, ahead of us.