UHD TV sales might have taken off but there's not much UHD content to watch on them. A new MPEG-5 video coding standard, LCEVC, might just address this.
Ever since regular TV broadcasting began by the BBC from my home town London on 26th August 1936, the broadcast industry and the technology ecosystem around it has been continually striving to improve it and make it look better. However, in recent years we have failed to upgrade traditional terrestrial broadcasting to ultra-high definition (UHD), with the consequence that most people who have been buying UHD TVs have never watched any actual UHD TV content on them!
The main reason UHD TVs have been a great success for TV manufacturers is because they have the opportunity to sell premium products that for the same screen size actually cost them less to manufacture than a full HD panel. This is because the display production yield for UHD is greater than full HD due to there being four times as many pixels for the same screen size, meaning the allowed percentage of non-defective pixels is much easier to meet.
Sadly, as of 2021 UHD has remained impractical for broadcasters, since current video codecs just consume too much precious spectrum. In particular, it doesn’t make economic sense to broadcast UHD, since
Another dimension is that the lack of UHD content has made it easier for new OTT services to enter the market with higher-quality content, putting further pressure on traditional broadcasters’ revenue streams.
Modern digital video codecs were designed around single-channel transmission. The industry developed high efficiency video coding (HEVC) in 2012 to reduce bandwidth requirements, but HEVC could only be used to transmit services aimed at new devices, since legacy receiving devices would not be able to decode them.
If we look back at the history of TV, we have faced a similar challenge before with the introduction of color TV in the analog world, where spectrum was even more scarce than it is in our current digital transmission world. It would have been impractical to broadcast new color channels incompatible with legacy receivers, since there would not have been enough spectrum to broadcast them along with existing black-and-white channels.
As an industry, we came up with a couple of tools to solve that problem: one was sub-sampling the color information, and the other one – what I would call the ‘genius’ solution – was that we only needed to send the extra information required for color (Chroma) alongside a backwards-compatible black-and-white (Luma) channel. This meant all viewers with legacy receivers could continue watching black-and-white channels, while customers with new TVs could watch color TV instead.
Back in 2018, MPEG recognized that on top of working on whole new codecs, which require whole new ecosystems to be built and deployed before being used, it made sense to have a low complexity enhancement codec that could be applied to any existing or future codec to allow delivery of a conventional video stream to everyone (similarly to the old black and white) along with a small amount of additional enhancement data enabling a higher-resolution enhanced picture quality for those people who have receivers compatible with the new enhancement codec.
The approach above is made even more impactful by the fact that:
An approach proposed as a starting point to address this requirement used technology from V-Nova. As a result, after much hard work by many organizations across the industry, the MPEG-5 Part 2 low complexity enhancement video codec (LCEVC) ISO/IEC 23094-2 standard was finally completed.
LCEVC also brings along an interesting “psychological” challenge to an industry that is used to making and selling new equipment with each new standard. LCEVC is “low complexity”, which means that although it can be implemented in hardware (and surely will be in the future), it can also do what no other video codec had done before: be implemented by using existing hardware blocks via a new device driver. Plus, these same blocks we are using to implement in current generation chips also exist in older chips, giving the unique opportunity to retrofit LCEVC on a large number of existing TVs and set-top boxes (STBs) via an over-the-air update, therefore enabling UHD TV to come to terrestrial broadcast TV today, without the need to wait for everybody to have purchased new equipment.
With LCEVC, a UHD channel would have a ‘base’ video of 1080p – so any device that is not LCEVC aware would still get 1080p video. LCEVC data typically adds 10-20% on top of the lower-resolution base video, so indeed the combined new channel (base plus enhancement data) is significantly smaller than a full-resolution channel broadcast without LCEVC. Most importantly, for broadcast services you don’t need to broadcast two channels at the same time.
I believe that – a bit paradoxically – being retrofittable on many existing devices and backwards compatible on the rest will foster greater sales of new UHD equipment, thanks to greater availability of high-quality UHD content. By making the delivery of UHD content commercially viable for terrestrial TV, LCEVC succeeds at what any new codec may find impossible to achieve: that is, reducing the bitrate enough so that broadcasters could afford to send whole new UHD channels on top of their legacy full HD ones.
Lastly, one of the great things about LCEVC is that it is not at all suggested to be an alternative to future codecs, since it also enhances them: using VVC or AV1 with LCEVC produces better quality than using the VVC (versatile video coding) or AV1 (AOM video 1) video coding schemes on their own, and saves a whole lot of energy for both encoding and decoding. In fact, it was already shown that the combination of LCEVC with AV1 reduces transcoding processing requirements by up to 70% and even makes AV1 delivery to mobile devices more viable, due to the fact that LCEVC-enhanced AV1 can run on many more devices and extends battery life by up to 50% vs. using AV1 on its own.
This article was originally published on EE Times.
Rick Clucas is SVP innovation and technology at V-Nova.