EBU Technical Review : No. 281 (Autumn 1999)
Scanning by numbers?
The development of TV has been plagued by multiple standards, particularly at the level of picture-scanning formats. The "legacy" formats of 405 and 819 lines have long since disappeared, leaving us with 625 and 525 lines – with field rates of 50 and 60 Hz (or 59.94 Hz) respectively. Can we avoid a multiplicity of scanning standards in the future world of high-definition TV? In my view, we are more likely to see a flying pig!
The CCIR Plenary Assembly in Dubrovnik in 1986 was the turning point for HDTV standards. There, the conflict was between the 1125/60 system, supported by Japan and the USA, and the 1250/50 system from Europe. The Europeans rejected a 60 Hz field rate for HDTV because, in the 50 Hz areas of the world, it would imply widespread use of standards conversion to or from 625/50 material. As standards conversion between different field rates was, and remains, a formidable problem, the Europeans selected 1250/50 as their future HDTV format. During the early 1990s, the debate about HDTV standards shifted dramatically with the prospect of digital compression and digital delivery. The North Americans decided to adopt a new system with 1080 active lines and 1920 pixels per horizontal line, rather than the Japanese 1125/60 system with 1035 active lines or the European 1250/50 system with 1152 active lines. We now faced a proliferation of image formats - with a choice of 1035, 1080 or 1152 active lines.
From the perspective of camera manufacturers, this implied three separate markets. This was regrettable as all modern cameras rely on solid-state sensors with a fixed array of pixels. It would be much better if worldwide agreement could be achieved on a common image format (CIF). The desire for a CIF had been expressed on many occasions in the ITU and elsewhere, but by 1995 there seemed to be no prospect of such an agreement. The problem was, yet again, the "not invented here" syndrome. Achieving agreement meant that proponents of two of the three image formats had to be willing to concede in favour of a competing format.
Surprisingly, such an agreement was reached by DAVIC (Digital Audio-Visual Council) in December 1996. The statutes of DAVIC specifically prohibited any discussion on commercial issues. As the main reasons for preferring any of the image formats were almost entirely commercial, DAVIC was constrained to consider only technical arguments. But none of the candidates had overwhelming technical merits – other than the fact that the 1920 x 1080 format was based on "square pixels". Given that DAVIC's policy was to avoid multiple standards, the 1920 x 1080 format was selected. Although most engineers welcomed this decision, there were some adverse reactions, particularly from a few people in Japan or Europe – one person claimed that this was a "disaster for Europe"!
Nevertheless, time is a great healer. Possibly emboldened by the consensus within DAVIC, ITU-R Study Group 11 re-examined this issue during 1997 and 1998. Although the split between 50 Hz and 60 Hz areas remained, the 1920 x 1080 image format was endorsed by ITU-R in Recommendation BT.709-3. The Technical Committee of the World Broadcasting Unions has also recommended that this format be used for HDTV programme production and exchange. Production equipment using this format, capable of operating at either 50 Hz or 60 Hz, was also exhibited at IBC 99 in Amsterdam.
Despite this outbreak of common sense, furious debates have raged over the last three years or so, mainly in the USA, about the relative advantages of interlaced and progressive scanning. At the risk of trivialising an ongoing argument, there are two opposing camps: those in favour of interlaced scanning of images with 1920 x 1080 pixels and those preferring progressive scanning of images with 1280 x 720 pixels. The first format is claimed to be "better" because each image has 2,073,600 pixels compared with 921,600 pixels for the second format: in other words, the pixel count has a ratio of 2.25. In practice, the subjective difference is not so obvious because interlaced scanning requires more lines than progressive scanning. Perhaps, by coincidence, the 720-P and 1080-I systems give almost identical performance in terms of perceived vertical resolution.
Those in favour of progressive scanning feel that decisions must not be made on the basis of "pixel counting". It seems that digital compression can be more effective with progressively-scanned TV source signals. This "coding gain" might mean that, for a given quality, lower bit-rates are possible with progressive – despite the fact that progressively-scanned TV signals require double the nominal data rate at a given picture refresh rate. This potential saving in bit-rate could be used to offer other services, such as multimedia.
From a European perspective, we can afford to watch developments on the other side of the Atlantic. However, we need to understand the issues and to evaluate the coding gain under controlled circumstances on a wide range of test material.
In the end, such arguments in favour of one scanning format or another are probably sterile. In North America, all HDTV receivers are capable of decoding 18 separate scanning formats, including interlaced and progressive formats at various picture rates. Similarly, most modern desk-top computers have CRT monitors capable of operating at different scanning formats – such as 640 x 480, 800 x 600 and 1024 x 768 pixels. Progressive scanning is the norm for computer displays – with picture refresh rates of 70 Hz or even faster.
A related issue is the inevitable trend towards flat-panel displays for TV applications. These have a fixed arrangement of pixels, but the number of active lines may well be set by the computer industry rather than by broadcasters. Broadcasting engineers have traditionally been locked into specific scanning formats – and, indeed, have passionately defended their preferred formats. Given that the computer industry has embraced multiple scanning formats, do we need to fix unique standards?
European
Broadcasting Union