ÐÓ°ÉÔ­´´

Television technology is at a crossroads. Everyone is agreed that the
TV set of the future will receive bigger and better pictures. The problem
is that the systems which Europe has developed to improve picture quality
are becoming outmoded before they can be put on sale.

Last month, Thomson, the French consumer electronics giant, started
selling the first domestic TV set that provides cinema-style wide-screen
and high-definition picture quality. The new set, called Space, costs around
£3000. Sales in Germany and Britain, where Thomson owns the Telefunken
and Ferguson brands, will follow later this year. But just as Thomson was
launching Space, the French Ministry of Foreign Affairs leaked a report
which warned that the only high-definition transmission technology which
is available for the set to exploit is already obsolete.

Thomson’s response – ‘How many TV engineers work at the foreign affairs
ministry?’ – conceals a pragmatic policy. Its Space set is designed to be
‘futureproof’. It contains none of the decoding circuitry for receiving
transmissions which the French ministry now fears is obsolete.

Space receives conventional terrestrial and satellite transmissions,
displays them in wide-screen format and artificially heightens definition
by doubling the number of picture lines shown on the screen. But with add-on
decoders, Space will also be able to receive wide-screen or high-definition
television (HDTV) programmes from any source in any technical format.

The wide-screen picture format is the one future development on which
the world’s electronics and broadcast industry agrees. Traditionally, the
ratio of width to depth of the picture on screens, the aspect ratio, is
4:3. The human eye, however, has a wide angle of horizontal vision. In the
1950s, the cinema stretched its screen format to give an aspect ratio of
16:9 and more to give it an edge over television.

The TV industry has agreed on a 16:9 aspect ratio for wide-screen TV,
which is the format of the new Thomson set. The 4:3 TVs will look boxy and
square in comparison with Space’s 16:9 screen.

For the best effect, the new 16:9 screens must work with transmissions
which provide a wide-screen signal of high definition – that is, using double
the number of picture lines. There is widespread disagreement on the best
way to achieve this.

In Europe, 30 electronics companies, research laboratories and broadcast
authorities have been involved since 1986 in a Eureka project under the
European technology R&D programme. The project, EU 95, is to develop
an HDTV system that is compatible with the 625-line MAC (multiplexed analogue
components) transmission system developed for satellite broadcasting.

Even the basic MAC system was designed for wide-screen broadcasting:
it can transmit 16:9 pictures which a wide-screen uses in their entirety.
Conventional 4:3 sets can show the pictures but ignore the edges.

For high-definition MAC transmission (with 1250 horizontal scanning
lines making up the picture instead of the conventional 625 for Europe)
the system takes advantage of the wider bandwidth of a satellite channel
compared with terrestrial channels. The TV station transmits a 625-line
picture, which existing sets receive as normal, and also extra detail in
digital code which the HDTV sets retrieve. So the HD-MAC system is a hybrid
analogue-digital one.

As a second string, European electronics companies and broadcasters
are developing PAL Plus. This brings the wide-screen (but not high-definition)
option to the 625-line PAL TV system which is used for terrestrial broadcasting
throughout most of Europe. Broadcasters such as Sky also use PAL from the
Astra satellite.

The developers rejected the idea of trying broadcasting extra side panels
which a wide-screen set can stitch on to the central image, fearing the
seams would show. Instead PAL Plus will rely on the transmission of ‘letterbox’
pictures. These will appear on conventional 4:3 sets as a strip picture,
with black borders above and below. PAL Plus sets will expand the letterbox
image to fill the wide screen. So PAL Plus is also a hybrid system, with
digital code buried in the black letterbox strips. Its big handicap is that
it relies on the hundreds of millions of viewers with 4:3 PAL TV sets being
prepared to watch letterbox pictures.

Until recently all the work done on HDTV was based on analogue or hybrid
technology. Adopting this would have the unfortunate effect of locking the
world into the same kind of standards muddle that exists today. The US and
Japan use the 525-line NTSC (National Television Standards Committee) system
while Europe and Australia use the 625-line PAL system. Any HDTV system
designed to be compatible with the existing TV system in the US and Japan
will inevitably be incompatible with any new system designed to work with
Europe’s sets. At the same time, any HDTV system adop ted by film studios
to replace photographic film will be incom patible with at least one of
the HDTV transmission systems.

The European Broadcasting Union describes this as a Gordian knot. ‘There
is one hope,’ says the EBU. ‘The future in this matter as in others is digital.’
If a TV signal is originated in digital code, or is converted into digital
code for transmission and recording, it can easily be converted into other
standards by mathematical processing. Digital TV cuts across all the standards
barriers.

Switching to digital could give the West a valuable advantage over Japan.
Twenty years ago, the Japanese state broadcasting station NHK began work
on Hi-Vision, an HDTV system using 1125 lines. Japan’s direct broadcasting
satellites are now transmitting Hi-Vision pictures and a few TV sets (from
Sony, Matsushita and Hitachi) went on sale in Japan in December, with a
price tag of $30 000. Hi-Vision is an analogue system, and it would be difficult
for Japan to scrap it now.

The US is best placed to adopt a digital TV standard. After a brief
flirtation with Hi-Vision, the US government decided to develop its own
HDTV technology. Initially there were around 20 competing systems, all for
analogue or hybrid systems. Now, many have dropped out and the remainder
are switching to digital technology. The Federal Communications Commission
(FCC) will test them this year, and decide in 1993 which is best for the
US to adopt as a standard.

General Instruments was first to propose an all-digital system, called
DigiCipher. The company submitted its application to the FCC in June 1990
and demonstrated a computer simulation of the system last year.

In November the Advanced Television Research Consortium – NBC, Philips,
Sarnoff Research Labs and Thomson – switched to an all-digital format. A
month later, American Telephone and Tele graph teamed up with Zenith Electronics,
the only remaining American-owned TV manufacturer, to develop an all-digital
HDTV system.

Finally, in January, General Instruments teamed up with the Massachusetts
Institute of Technology to form the American Television Alliance. The two
companies will develop two all-digital HDTV transmission systems.

William Schreiber, a retired professor from MIT, warns that the rush
to digital in the US may be suicidal. If decisions are made without enough
thought for the cost of development, efficiency and time scales, digital
technology may materialise too late to stop Japan scooping the pool with
Hi-Vision as an off-the-shelf working system.

In Europe, attention to HDTV development has centred on HD-MAC and Eureka
Project EU 95, with the much publicised participation of Bosch, Philips,
Thomson and Nokia. Another Eureka Project, EU 256, has kept a much lower
profile. The participants – which include the Italian and Spanish broadcasters
RAI and RTVE, the University of Madrid and the Italian-Spanish electronics
company, Telettra – have been developing an all-digital HDTV transmission
system since July 1988. With little publicity, EU 256 demonstrated digital
HDTV relays from the World Cup in Italy.

A project in the European research programme for advanced communication
technologies (RACE) involves the BBC in creating wide bandwidth telecommunications
highways that can carry pictures and sound as digital code. All this is
making analogue HDTV systems look like throwaway technology.

Any wide-screen HDTV signal must contain at least four times as much
information as a conventional TV signal, and best estimates are that the
technology to compress an HDTV signal into existing TV channels, at a price
domestic users will pay, will not be ready for commercial exploitation until
the end of the decade.

At the HDTV ’91 conference, held in London in December, Jean-Luc Renauld
and Gwyn Morgan of the British company, Logica Communications, drew attention
to another obstacle to HDTV, whether analogue or digital, becoming a consumer
product before the next century. The clarity of HDTV pictures becomes noticeable
only on a large screen. However, the public will not be prepared to buy
large bulky screens until they are produced as flat panels which can hang
on the wall. High-definition panels will not be ready before the year 2000.

This gives the world a breathing space to develop and agree on a digital
standard for wide-screen HDTV. Until then, existing TV systems and TV sets
such as Space which use cathode-ray tubes and artificial picture enhancement,
will do the job.

The key to HDTV television, whether digital or analogue, is data compression.
For example, simply converting a conventional 625-line TV signal into digital
code generates a data stream of 216 megabits per second (Mbit/s). For wide-screen
HDTV, it is more than 1 gigabit per second. Neither terrestrial, nor satellite
channels, can handle such large amounts of data.

For RACE, the BBC has been working on coding technology to reduce the
amount of digital code needed for a moving TV picture to the standard, 34
Mbit/s, agreed by the European Broadcasting Union and the European Telecommunications
Standards Institute. Packets of four codes, each producing 34 Mbit/s, will
be used together for HDTV.

All the new digital compression systems rely on the same basic breakthrough
in digital coding technology. This is DCT, or discrete cosine transform,
which over the past two years has radically altered the rules of digitally
coding TV pictures.

The EU 256 team has been very quiet about its work on digital coding,
both because it is politically sensitive (making EU 95 look obsolete) and
because the technology is proprietary. But patent applications filed by
RAI gives an insight into how the compression system will work and what
problems system developers have hit.

All DCT encoders work by splitting each TV picture into groups of eight
horizontal scanning lines, called ‘stripes’. Each stripe is then chopped
into blocks along the scanning lines. The chopping action is performed separately
on the luminance, or black-and-white detail information, and the two chrominance
signals which contain colour information.

Each block is then compared with the previous block. This gives a ‘difference
block’ representing how the content is changing in that area of the picture.
For example, in a landscape there will be no change so the difference block
will contain no information. But in a football match there will be many
changes, so the difference block will contain a lot of information.

By encoding only the differences between pictures, the digital bit stream
contains far fewer bits than would be needed to describe each picture in
full. But the amount of bits varies continually, with more needed for when
the football is rapidly passing from player to player than when a referee
is ticking off a player. The variations are averaged out in a buffer memory,
but this overloads when the scene is very busy. The flow of bits has to
stop and the picture starts to look grainy. At the moment, this creates
a very disturbing effect.

RAI’s answer is to put a filter ahead of the digital processor to remove
some of the fine detail from a fast action scene. The filter works dynamically,
removing just enough detail to stop the buffer from overflowing. The action
of the filter is delayed, so that the picture does not rapidly fluctuate
between sharp and soft definition. A subtle, slow change of clarity is far
less noticeable than sudden fluctuations in grain.

The question is not whether digital HDTV will be feasible, but when.
National Transcommunications (until recently the research division of Britain’s
Independent Broadcasting Authority) has demonstrated bit reduction for 625-line
pictures from 216 Mbit/s to 12 Mbit/s, without noticeable loss of quality.
General Instruments claims that DigiCipher will be able to compress an HDTV
signal of 1.2 Gbit/s to 15 Mbit/s. At these very low data rates, it will
be easy to transmit a digital HDTV programme on a satellite channel, with
analogue video bandwidth of around 10 MHz. It should even be possible to
carry HDTV on a terrestrial TV channel with bandwidth of around 6 MHz.

But much of the digital compression technology is still on the drawing
board. All the demonstrations so far have relied on computer simulations,
with a minicomputer taking up to a thousand seconds to compress and code
each second of video. There is no doubt that, with custom-designed VLSI
(very large-scale integrated) chips, the companies will eventually achieve
real-time processing of pictures for digital coding.

But with the electronics industry in recession, only a few of the largest
companies can afford to develop the chips. The irony is that most of the
companies which could make the technology work, and can afford the investment,
are Japanese. And this is one case where the Japanese, with their investment
in Hi-Vision, have every incen tive not to make the technology work.