NHK Describes Olympian Data
January 6,2013, Storage Visions Conference, Las Vegas—Yoshiaki Shishikui, head of the advanced TV systems research division of NHK described the equipment, workflows, and other requirements to handle the live broadcast of the 2012 Summer Olympics in ultra high definition. The experience demonstrated the potential for UHD as well as the need for more development work before the 4K systems are ready for prime time.
The first UHD cameras had 33M pixels in a 8K by 4 K array. The system is a dual green, so there are 2 green pixels for every red and blue. The pixel count is determined by the needed viewing angle and the resolution. As screens get larger, they require more pixels to maintain the illusion of smoothness. The new format is designed for a viewing angle of 100 degrees, compared to the 30 degrees for HD. As a result, the viewing distance for UHD is about 0.75X the horizontal dimension of the screen compared to the 3X for HD.
The broadcasts used 22.2 channel full 3-D sound for a new audio-video experience that rivaled being there. The 22.2 channel sound is comprised of 9 top channels, 10 middle, and 3 bottom, plus 2 sub-woofers. The cameras are designed to go up to a 120 Hz frame rate, resulting in a data rate of 144 Gb/s. the cameras weighed about 20 kg and include an integrated AVC/H.264 codec.
For live viewing, the sites included 3 in the UK, 3 in Japan, and Washington D.C. They covered a number of venues and events during the Olympics. Production facilities included trucks with editing stations and 28 monitors. The cameras had remote focus pull, since the operators could only see a 2-D image. Most of the coverage was with wide angle lenses that matched the 100 degree field of view. The opening ceremony used only 3 cameras.
The cameras and associated processing equipment produced UHD, slow motion, and telestrator images. all video feeds to the switches and then to the studios for further processing. Audio came though 22 mikes at the site plus other audio sources and fed an audio mixer, which output to a recorder and streamed to the video feed. The production flow combined the audio and video streams with further edits at the BBC TCO studio and then to the playback and transmission equipment.
The transmission infrastructure took the raw feeds from a fiber connection and fed them to the TV center where it was compressed. The streams then split to an IP network and unicast structure where the two signals then went to broadcast facilities. Post production operations took UHD clips and down converted them to HD for edits while maintaining the full data. After editing, the files were reconverted back to UHD video and audio. The whole process took about 8 hours, from midnight to 8 A.M. the next day. The edit system is configured to hold 256 TB of data.
Audio post had a equally complex flow. Two feeds, one for Japan and one for the rest of the world were mixed live and the highlights were recorded. Transmission started from a master to an encoder to an IP stream. The data were sent to the IP network where the receivers converted the IP streams to a TV format, where another decoder converted the files for playback.
The live broadcasts went to various screens. One in Tokyo was 520 inches, Washington was an 85-inch display and the others were intermediate sizes. The viewers reported mostly positive feedback, with high immersion being the most stated attribute. Some noted some blurring on fast moves like a camera pan. Overall, the production and playout functions are like a daily broadcast.
The roadmap towards UHD broadcast are ’20 for direct satellite in the 21 GHz band, followed by terrestrial broadcast sometime later. The standards for studio and video are complete, while the audio is still being developed. Other areas needing work include compression, transmission, and others. The roadmap for equipment has similar constraints as broadcast.
