As live Super Hi-Vision pictures are successfully sent over the internet it looks increasingly as if it will have become a practical broadcasting system by the end of the decade. Adrian Pennington reports.
The world’s first international Super Hi-Vision (SHV) transmission test over IP networks was conducted September 29 at BBC TVC by BBC Research & Development in collaboration with Japanese broadcaster NHK.
The project, spread over two days, featured a live 30-minute performance of The Charlatans recorded for digital radio station BBC 6Music, and a live 30-minute transmission back to Tokyo of Taekwondo by the British Olympic team. SHV is the ultra-high-definition video format being developed by NHK for next-generation TV broadcasting, with a resolution of 7680 x 4320 pixels, sixteen times higher than existing full HD.
It’s not the first time the two broadcasters have collaborated on SHV tests. In 2008 the first public live SHV transmission was made between London and Amsterdam (to IBC). The BBC’s main contribution has been on the encoding side, first with Dirac and subsequently in conjunction with NHK on the MPEG and ITU-T standardisation process for HEVC (High Efficiency Video Coding).
Two years ago MPEG-2 was used for compression, delivering data rates of 650Mbps. The rate has now been more than halved thanks to MPEG-4 AVC/H.264 but, though not exclusively focussed on SHV, HEVC has a target rate of 150Mbps which would be at the outer limits for compressing SHV over fibre optic broadband to the home by NHK’s target broadcasting date of 2020.
“Since the SHV frame rate is 60fps we have also started to look at generating higher frame rates,” explained John Zubrzycki, BBC R&D principal technologist. “With movement in the scene, at higher resolutions the picture will look blurred — but if you increase the frame rate you increase sharpness.”
Having tested high definition decades before it was ever a practical proposition, investigation into futuristic formats is normal for BBC R&D. “With HDTV up and running our remit is to look at what may be the next generation visual technology a decade or even 25 years hence,” he said.“It could be 4K digital cinema or 3D, both of which we are exploring, as well as SHV.”
The recent tests were shot using one of only three prototype SHV cameras sporting three (RGB) 1.25-inch 33 megapixel (8K) resolution sensors. Previously the SHV camera featured two 4K green sensors with half-pixel offset (plus 4K red and blue sensors) which achieved full resolution for vertical and horizontal luminance detail only.
Since existing lenses are not simply sharp enough to capture all the information the SHV sensors can receive, NHK has had high precision lenses custom made — each with a basic 10:1 zoom and built larger than existing lenses for the 1.25in sensor.
From the camera the video and audio signals were fed to a Fujitsu-developed SHV encoder composed of eight 1080/60p MPEG-4 AVC/H.264 encoding units. The SHV image is divided into eight overlapping tiles by a format converter and each of them fed to an encoder unit. The output streams of these eight encoders are in MPEG TS format and multiplexed together using a TS multiplexer.
“The principal that NHK are deploying to handle the data is to treat it as 16 HD SDI signals in parallel in production. A new development is that these signals are able to be muxed uncompressed into a single optical stream, carried over a single triax cable from the camera,” said Zubrzycki.
In the control gallery the feed was demultiplexed from the optical stream displayed around the studio on 4K monitors and locally stored on an array of 16 Panasonic P2 solid state recorders, each encoding and storing a sixteenth tile of the the SHV picture using AVC-Intra (100Mbps) — giving a total recorded bit rate of 1.6Gbps.
“It’s quite a challenge to manage 16 HDTV signals in parallel around a studio although in theory it can be done today with a big enough router and storage devices,” said Zubrzycki. “We added nothing fancy to the broadcast, no caption generators or vision mixing — things which will be added to a mature SHV system.
“The real challenge, and the real reason for the test, was to prove that we could transmit the signal live half way round the world and if we could do that that we could do it anywhere,” he added.
Equally important it demonstrated that SHV could be carried over the internet as a contribution signal from an event to the broadcaster’s studio, avoiding the expense of a satellite transponder and with an eye to IP as the way images will be transmitted in future.
The eight 45Mbps encoded streams combined to create a stream with a bitrate that varied between 320-380Mbps. These were IP encapsulated and packet streamed for error protection then sent over broadband IP network to Japan.
The test was routed over various research IP networks including JANET (the UK’s 40Gbps education and research network), GEANT in Europe, Internet2 in the US and Gemnet in Japan (coordinated by NTT). On reception at NHK’s Science & Technology Research Labs, the reverse operations were conducted to reproduce SHV video and audio.
“The packets were sent in defined routes through the networks since we felt that with that amount of content we needed to monitor and verify what was going on,” said Zubrzycki. “It was an effective demonstration but there was some packet loss and our investigation now is to find out what is happening to those packets. This was the real reason we did the tests – to find out how we can send SHV via IP.”
The camera’s sensitivity is the key hurdle to overcome at the acquisition end. Although the camera has half the 40kg weight and size it was in 2008 the goal has to be to make it more manoeuvrable. “For the trial we used more light than today’s normal HDTV but with similar lighting levels to HDTV at a similar stage of its development,” noted Zubrzycki.
NHK believes it can start experimental satellite broadcasting using the Ka band (21 GHz) in 2020. Before that however it has to solve the problem of attenuation caused by the amount of rainfall Japan receives in the Ka band.
“We have a plan for experiments using the Ku band (12 GHz) prior to the Ka-band experimental broadcasting,” explained Keiichi Kubota, director general of NHK R&D. “Research is also ongoing for SHV broadcasting over terrestrial broadcasting systems, although we don’t think it will be practical until after satellite broadcasting starts.”
Kubota likens the situation to that at the early stage of HD when there was no market demand when it started R&D on the format. “It is our job to create new demand, even before anyone notices the need,” he said. “The most important thing for us is to make people recognise that SHV will bring such a wonderful future. The Kyushu National Museum, Fukuoka, Japan has already introduced an SHV system. Our final goal is still SHV broadcasting to public; but before that, we have to put SHV to practical use in theatres.”