TN21 ΡΡ The Current State of High Definition Television Charles Poynton [now Poynton@inforamp.net] Sun Microsystems, Inc. 89/03/09 11:33 SCOPE Equipment for High Definition television (HDTV) is now commercially available, and HDTV is expected to be of great importance in industrial, scientific, and medical applications as well as in entertainment. This document outlines the background of HDTV, describes the basic parameters of the 1125-line system and their derivations, lists available commercial equipment, outlines potential application areas, reviews the standardization of HDTV to this point, and briefly describes standardization issues currently under discussion. INTRODUCTION Nomenclature A video system is usually denoted by its total line count and its frame rate (in hertz) separated by a solidus. For example, 525/59.94 represents the scanning parameters of the U.S. broadcast television system. Television system nomenclature refers to the total number of lines in a raster. Computer graphics nomenclature is generally concerned with the number of picture lines (lines per picture height, L/PH), which is about 8% less than the total lines in the raster in order to accommodate vertical blanking interval overhead. For example, a 1280 by 1024 system may have 1066 total lines. HDTV is defined as having approximately double the number of lines of current broadcast television at approximately the same frame rate. Hence the line rate is approximately doubled (from about 15 kHz to about 34 kHz). Luminance bandwidth is increased such that horizontal resolution is also approximately doubled. Resolution HDTV has about double the (linear) resolution of 525-line television, or about 5.5 times its luminance spatial resolution. Computer graphics is not generally concerned with the resolution of the picture on the display screen, but refers to "resolution" as the number of visible pixels stored in the frame buffer. The number of line pairs (cycles) actually resolved on the face of the display screen may be substantially less than the number of pixel pairs stored for each scan line, due to electrical and electro-optical filtering effects. Psychophysics The fundamental development work for HDTV was done at at the NHK (Japan Broadcasting Corporation), after extensive psychophysical and perceptual research led by Dr. Takashi Fujio. HDTV is capable of generating pictures substantially brighter, sharper, and of better colourimetry than 35 mm motion picture film. Many psycho-physical studies have shown that human viewers to position themselves relative to a scene such that the smallest detail of interest in the scene subtends an angle of about one minute (1/60 degrees) of arc, which is the limit of angular discrimination for normal vision. For the 481 visible lines of 525-line television the corresponding viewing distance is about 7 times picture height, and the horizontal viewing angle is about 11 degrees. For the 1035 visible lines of 1125-line HDTV, the corresponding viewing distance is 3.3 times screen height and the viewing angle is about 28 degrees. Psychophysical research has shown that a viewer's involvement in a motion picture is increased when the picture is presented with a wide aspect ratio. The aspect ratio 16:9 has been chosen for HDTV. This value, about 1.78:1, is nearly as wide as the most common theatrical film format of 1.85:1. The viewer of HDTV thus does not normally perceive increased "definition" (resolution) for the same size picture, but rather moves closer to the screen. Thus HDTV should more properly be called "wide screen television", and some argue that this designation would also be more appropriate to consumer marketing and product differentiation than "HDTV". Recent NHK research has revealed that high-quality stereo sound impacts the psychophysical response of the viewer to the picture (in particular, the viewer's eye-tracking response). Quality The picture quality of HDTV is superior to that of 35 mm film. The limit to resolution of motion picture film is not the static resolution of the film, but judder and weave in the projector. [35 mm motion picture film is conveyed vertically through the projector, and has an aspect ratio of 1.85:1, so the projected film area is only about 21 mm by 11 mm, compared to 36 mm by 24 mm for slide or print film.] Also, the colourimetry obtainable with the colour separation filters and CRT phosphors of a video system is greatly superior to that possible with the photochemical processes of a colour film system. The quality of various proposed transmission systems varies widely, and in some cases has yet to be demonstrated. The 1125/60 System This section outines the basic parameters of the 1125/60 system. More detailed information about the detailed parameters are available in an associated document. Broadcasters have proposed 1250/50 and 1050/59.94 systems in Europe and the U.S. respectively, based on "compatibility" with local broadcast standards. No commercial equipment, and very little experimental equipment, exists for these standards. Such systems will be outlined in a later section of this document. Basic Parameters of 1125/60 HDTV The scanning parameters of 1125/60 are chosen to be closely related to 525-line and 625-line systems: The total line counts 1125, 625, and 525 are odd multiples (45, 25, and 21 respectively) of 25. Vertical blanking for each system is chosen to be exactly 8% (2/25) of the total line count, thus the active line counts 1035, 575, and 483 are the odd multiples (45, 25, and 21 respectively) of 23. The target luminance bandwidth of HDTV is generally agreed to be 30 MHz, about five or six times the bandwidth of current broadcast television, although not all currently-available HDTV equipment can meet this bandwidth, and most of the proposed transmission systems cannot meet this bandwidth. SMPTE 240M Parameters The basic parameters of the 1125-line HDTV system are contained in the ANSI/SMPTE 240M standard recently-adopted in the U.S., and are contained in Annex II to the international CCIR Report 801-2. The basic parameters are as follows: Number of scanning lines: 1125 Number of active lines: 1035 Field rate: 60.00 Hz Interlace: 2:1 Aspect ratio: 16:9 Samples per active line: 1920 for luminance (Y), 960 for colour difference (Cb, Cr) ANSI/SMPTE 240M specifies RGB or YUV components, with well- characterized colourimetry and transfer functions. Luminance bandwidth is specified as 30 MHz. The signal has zero setup (pedestal), and includes a tri-level sync signal. 1125/60 Equipment Commercial hardware operating with the 1125-line system is widely available from Japanese manufacturers. Equipment which is commercially available includes: - video monitors (Sony, Hitachi, Ikegami, Barco); - video projectors (Sony, Hitachi, Ikegami); - cameras (Sony, Ikegami, Hitachi, BTS); - videotape recorder (Sony); - telecine [film-to-video] (Rank-Cintel); - framestores (Sony, Asaca/Shiba-Soku, Toko); - paintbox (Quantel); - down-converter (Sony, Ikegami) - switchers (Grass Valley Group, Sony); - test equipment (Tektronix, Magni); and - blue-screen matte (Ultimatte). 1125-line equipment which has been demonstrated, but is not necessarily commercially-available, includes: - MUSE broadcast encoder, decoder (NHK); - MUSE optical videodisc player (Sanyo, JVC); - MUSE videocassette recorder (Hitachi); - YUV videocassette recorder (Sony); - YUV videodisc player (Sony); - digital HDTV videotape recorders (Hitachi [648 Mb/s] and Sony [1.188 Gb/s]); - laser telecine (Sony, NHK) - laser film recorder (NHK); and - large-screen video projector (General Electric, Eidophor, Barco). Although current-generation 1125-line equipment is universally 2:1 interlaced, there is general agreement that the industry will tend towards progressive (non-interlaced) systems for production and display. Interlace may or may not be retained for transmission. HDTV APPLICATIONS Commercial/Industrial/Scientific It is generally thought that the initial application of HDTV will be in industrial, medical, and scientific applications where pickup, recording, and distribution of moving images is important, but where 525-line resolution is insufficient. It is also likely that HDTV technology will contribute to printing and publishing applications, even though the images in that application are stills. For example, Quantel is promoting the HDTV Paintbox as a product for printing and publishing. It is also quite likely that the 1125-line HDTV format will become an output format for computer graphics equipment. There is a strong trend in computer graphics towards higher resolution, but no preferred format. The HDTV format (about 1 k by 2 k picture samples) satisfies the need for higher resolution and high colour accuracy, and offers the opportunity to exchange, record, and distribute images among various application areas. Also, 1125- line video monitors and projectors will rapidly benefit from the economies of scale of manufacture of monitors in large quantities for other applications, and this format is therefore a good choice. Film Production HDTV is currently viable for production of material to be released on film. HDTV is equivalent (and in many respects superior) to 35 mm motion picture film. Its acceptance as a production medium awaits the wider availability of HDTV production facilities, and more knowledge of HDTV production techniques on the part of the film production community. There are advantages in producing entertainment material in HDTV, even if the end product is to be down-converted to 525-line or 625-line television. For example, it has been demonstrated that image compositing using Ultimatte can be done very effectively in HDTV. There are currently four commercial HDTV production studio facilities: Rebo Associates, Zbig Video, and 1125 Productions in New York, and Captain Video in Paris, France. A facility in Los Angeles is being planned. Consumer Broadcasting of HDTV is probably five to ten years away, except in Japan. Broadcasting requires spectrum allocation, which is subject to domestic and international political concerns. It is certain that any consumer HDTV receiver equipment will include up- conversion capability, to display 525-line or 625-line signals with improved quality. IDTV describes receiver techniques to improve the quality of standard broadcast signals. A receiver is considered IDTV if it employs frame-rate doubling to eliminate inter-line twitter, although additional techniques such as noise reduction may also be employed. EDTV describes a 525-line or 625-line broadcast television signal with altered or augmented signal content which makes possible higher quality at consumer receivers. Distribution of high-quality material for consumers could take place using HDTV either through cable systems (in the manner of Home Box Office), or on consumer HDTV videocassette (for sale or rental), prior to use of HDTV for broadcast in either North America or Europe. This approach to consumer HDTV may arise due to both the technical difficulty of HDTV broadcast (because of its large spectrum/bandwidth requirement), and the difficulties that the traditional broadcasting networks are likely to face in adopting HDTV. Japan seems to be absolutely committed to HDTV broadcast in the near future. It is certain that the primary origination medium for consumer HDTV in any form will initially be 35 mm motion picture film, due to the vast amount of existing material in that medium. Standards Since it is now evident that there will be no single international standard, discussions in the standards communities have now concentrated on three different areas: production, distribution, and transmission. Production is the shooting and assembling of program material, distribution is the exchange among program producers, and transmission is to the consumer. History Various Japanese manufacturers exhibited HDTV equipment at the World's Fair at Tsukuba in 1985. That equipment, and some of the equipment currently in experimental use around the world, has a picture aspect ratio of 5:3 and the same 59.94 Hz field rate as 525-line NTSC television. Picture aspect ratio was changed 16:9, and field rate was changed to exactly 60.00 Hz in order to facilitate international agreement on standards, resulting in unanimous agreement to present a set of basic parameters to the CCIR Plenary session in June of 1986. Adoption of these changes by the Japanese represented a major concession to the Europeans: tooling for the manufacture of CRT display tubes with 5:3 aspect ratio was already complete at a number of companies, and the field rate change required re-engineering of equipment. The proposal was not accepted by the CCIR at that time, due to lack of agreement from the European members. The Europeans stated at the time that "serious" technical problems existed in the down-conversion of 60 Hz 1125-line HDTV to 50 Hz 625-line PAL, but both Sony and NHK developed and demonstrated extremely sophisticated down-converters prior to the 1986 CCIR Plenary Session. Experts viewing the 50 Hz output from these converters perceived no motion artifacts. Many knowledgeable individuals believe that the European governments impeded the adoption of 1125-line HDTV in an attempt to protect their domestic studio and consumer equipment manufacturers. Certainly no serious technical proposals for an alternative HDTV system have been presented by the Europeans, and only a very small quantity of experimental equipment has been built in Europe. The Europeans (and the Australians) have a political interest in not adopting HDTV at this time, due to their recent adoption of multiplexed analog component (MAC) encoding for direct broadcast from satellite (DBS) systems in these countries. Receiver manufacturers now include MAC decoders in their new receivers, but consumers must install set-top converters for old receivers in order to receive MAC. The European broadcasting community would find it embarrassing to require consumers to purchase a new converter for another new standard, just a few years after the introduction (with much fanfare) of MAC. MAC is therefore currently being promoted in Europe as being capable of upgrade for HDTV (ED-MAC, for extended definition ), but there are few serious technical proposals indicating how ths can be achieved, and no converters currently being delivered that can accommodate the signal formats being proposed. Production Standards The ANSI/SMPTE 240M standard for an 1125-line Production system was adopted in the U.S. in February 1989. Disclaimers on this document carefully delineate its applicability to production use only. The standard essentially represents agreement on the detailed parameters of the 1125/60 system. The standard applies to the analog signal. Discussions on the digital representation of 1125-line HDTV are currently underway. Distribution Standards The de facto international television program distribution standard is, surprisingly, 35 mm film. In North America, film is transferred to video using a "3-2 pulldown" scheme which scans alternately three then two video fields from successive film frames. The film is run 0.1% slow to result in the 59.94 Hz field rate. In Europe, film is run 4% fast with 2-2 pulldown to result in 50 Hz frame rate. Discussions of distribution standards are in an early stage, but there is general agreement that film "friendliness" will be important. Transmission Standards A transmission standard is likely to mandate some form of framestore in the receiver to minimize transmission bandwidth, and to provide for standard-television backward compatibility. All proposed transmission standards involve the reduction of transmission bandwidth by exploiting the spatio-temporal properties of the human visual system, as first characterized by Drs William and Karen Glenn. Fundamentally, spatial detail is transmitted at low frame rate, and the information transmitted at the high frame rate necessary to portray motion has low spatial detail. Japan The Japan Broadcasting Corporation (NHK) in Japan is expected to commence HDTV broadcasting in 1992, using direct broadcasting from satellite (DBS) with the MUSE (MUltiple Sub-Nyquist Encoded) system. MUSE is not compatible with any current broadcast system, and has a bandwidth of about 8.4 MHz. North America VHF and UHF spectrum is controlled in the U.S. by the FCC. Many "proponents" of HDTV in the U.S. have informed the FCC to consider their systems for adoption as a U.S. standard. U.S. networks have proposed systems based on 1050-line, 59.94 Hz rasters, with a line rate of exactly twice that of the NTSC system, and about 966 L/PH. The claim is made that such systems are "compatible" with NTSC. Cable and satellite operators are unrestricted in their choice of transmission standards. It is quite conceivable that Japanese manufacturers could introduce consumer hardware. In any case, the vast majority of program material for HDTV will come from existing theatrical films from Hollywood. Europe The standardization process in Europe is substantially different from the standardization process in North America. Most broadcasting organizations are state-owned. Standards are agreed by the European Broadcasting Union, which is a union of the broadcasters. These meetings are closed; manufacturers (and other interested parties) attend only when invited. Systems based on 1250-lines and 50 Hz, with about 1152 L/PH, have been proposed by the Eureka-95 project in Europe. 1125/60 Parameters Under Discussion Although the combination of 1125 total lines, 60 Hz field rate, and 2:1 interlace produces a line rate of exactly 33.75 kHz, and therefore a line time of about 29.63 us, neither the total number of samples per line nor the sampling frequency is specified in ANSI/SMPTE 240M or CCIR Rep. 801-2. Appendix II of CCIR Rep. 801 specifies 1920 luminance samples per "active" line, following the terminology of CCIR Recommendation 601-1, but the term "active" is not defined in either document. In CCIR 601-1, it is implicit that some number of leading and trailing "active" samples are at blanking (or black) level, and some additional "active" samples are taken up by transition samples from blanking to video and video to blanking. Although the field rate of HDTV is exactly 60 Hz (emphasized by the "60.00" in the document), there is a movement in North America to accommodate operation at the NTSC field rate of 59.94 Hz, to maximize compatibility with existing NTSC equipment. Some current HDTV equipment is configurable for operation at either rate. The Square Pixel Issue Detailed parameters which have not yet been agreed upon are the sampling rate for digital HDTV systems, the number of "active" (non-blanking) digital samples per line, and the number of samples per picture width. These parameters are mutually interrelated, and should be related by simple integer ratios to the internationally-standardized digital sampling parameters for 525-line and 625-line television systems (in particular, the 13.5 MHz sampling rate and the count of 720 "active" samples per line), as documented in CCIR Recommendation 601-1. Poynton Vector has made a proposal for a sampling frequency of 70.875 MHz (2100 samples per total line) which achieves a sample aspect ratio which is exactly unity. One difficulty in the current discussion of possible parameter values for standardization is that nearly all manufacturers of HDTV equipment (and all manufacturers of cameras and VTRs) are Japanese. Thus, North American and European contributions are made from a position lacking in experience. Another problem is that all discussion is taking place within the traditional television constituency of broadcasters and broadcast studio equipment manufacturers. Users in non-broadcast areas, potentially the largest users of HDTV in the two- to six-year time frame, are not represented at all. Thus the SMPTE is likely to recommend parameter values for standardization which are appropriate for broadcast applications, and not necessarily appropriate for other applications. For example, the television constituency can maximize compatibility with existing NTSC equipment by choosing the NTSC frame rate of 59.94 Hz, but this rate would place a burden on non-broadcast users and European users, who would both prefer exactly 60 Hz. However, unless the concerns of non-broadcasters are expressed within the SMPTE, the broadcast-orientation will prevail, and by the time the impact of these issues is felt by the non-broadcast users it will be too late to change.