This is only a preview of the September 1989 issue of Silicon Chip. You can view 46 of the 112 pages in the full issue, including the advertisments. For full access, purchase the issue for $10.00 or subscribe for access to the latest issues. Articles in this series:
Articles in this series:
Articles in this series:
Articles in this series:
Articles in this series:
Articles in this series:
Articles in this series:
|
THE WAY I SEE IT
By NEVILLE WILLIAMS
High definition TV: lots of
problems at the transmission end
As a follow-up to his observations in the August
issue, we give correspondent Keith Walters an
opportunity to list the transmission problems that
would arise with the introduction of high
definition, wide screen television. He says they
would not have occurred to readers unfamiliar
with professional TV production technology.
Keith Walters (K.W. for short)
assumes that our original correspondent W.G. is just such a person: quite familiar with video home
movie gear but blissfully ignorant of
the complicated equipment and procedures used every day of the week
in assembling programs and commercials from a variety of sources.
"Through kilometres of high
quality 75-ohm coax, countless BNC
connectors, Tokyo-by-Night control
panels and mixing desks, multiple
signal sources have to be synchronised and sub-carrier phases
matched to within 2°. Not simple",
he says.
Most of this equipment, according to K.W., is configured to the
TV standard in the particular country ( usually a variant of PAL or
NTSC). If another standard was
adopted, it would all have to be
discarded and production facilities
re-equipped and re-cabled at enormous cost.
The situation would have little in
common with the changeover from
monochrome to colour in the 1970s.
Most of the old monochrome equipment was nearing the end of its
useful life, anyway. It could be
replaced progressively with new
colour equipment, able to work well
in b&w mode until the changeover
was actually made.
High definition, wide screen
television, he says, calls for a "component video" approach to signal
processing, already familiar in
some applications but having its
first public exposure in TV broadcasting in the various MAC formats. The "components" referred
to are simply the luminance (Y)
signal and two chrominance signals
(B-Y and R-Y).
Until fairly recently, normal
practice has been to encode the
three signals into composite video
(luminance and a standardised colour sub-carrier) right at the source
- the TV camera, special effects
generator, &c. All such signal
source equipment can be synchronised to a master sync and subcarrier generator. Once this is
done, and allowance made for the
various lengths of connecting cable,
signals can be wiped and splitscreened, and otherwise processed,
without further ado.
The changing scene
TV video production based on
composite video was regarded as
quite satisfactory up to the late
1970s and still is for most applications. However, the emergence of
digital video processing - which
bends, folds and otherwise "mucks
about" with images - has exposed
the limitations of composite video
as a production format. At this
point, I quote directly from Keith
Walters' letter, with some abbreviation to conserve space:
"Basically, these gadgets work
by breaking up the picture into a
half-million or so pixels, and storing
them digitally in a high-speed RAM.
By varying the manner in which the
memory is accessed on playback,
the shape of the output image can
be manipulated.
"You can't do this with composite
video because stretching or shrinking would change its sub-carrier
frequency. The signal must first be
decoded back to its original components - luminance, B-Y and R-Y,
commonly referred to as 'YUV';
this, so that the effects machine can
handle the components separately
and simultaneously.
"The problem is that there is no
known process whereby composite
video can be turned back into YUV
components identical to those
originally obtained from the camera or whatever.
"This statement stands, despite
what happens in an ordinary TV
set. Usable U and V information is
recovered and (with PAL) averaged
by the delay lirte over two lines. A
4.43MHz trap extracts most of the
colour subcarrier, leaving the high
frequency luminance signals
reasonably intact. The residual
SEPTEMBER1989
93
THE WAY I SEE IT - CTD
chroma sidebands are not too intrusive because the subcarrier is
specifically chosen for minimum
visibility".
OK for receivers ... but
Overall, says K.W., the components so recovered are OK for
direct display on the screen of a TV
set, being visually quite pleasing
under optimum conditions. But a
very real problem arises in a TV
production situation if recovered
YUV components need to be reencoded on a new subcarrier. As he
says:
"It is imperative that all traces of
the original subcarrier be removed
from the Y signal, because even a
minute frequency difference between the old and new subcarriers
will cause unsightly patterning on
the reconstituted chroma signal.
" The simplest way to counter the
effect is to use a low-pass filter at
3.1MHz, the lower limit of the
chroma sidebands. But this also
limits the luminance bandwidth.
What's more, the roll-off has to be
fairly sedate to avoid ringing effects. The upshot is that practical
filters are typically flat to 2.5MHz,
3dB down at 3MHz, and rolling
quite rapidly above that.
"2.5MHz? Where have we heard
t?at _before? Yes, anti-patterning
filtermg can drag luminance bandwidth back down towards the limits
suggested earlier for VHS, Beta or
any other helical-scan recorder
that uses the 'colour-under '
system".
At this point in his letter, K.W.
digresses into an explanation of the
colour-under recording system and
the problems which show up when
an incoming composite signal has to
be taken apart for recording and
playback, then reconstituted to suit
the receiver. It is informative but
interrupts his main theme. To carry
on:
"What we really need is something that completely removes the
colour subcarrier while leaving the
high frequency luminance components intact - something that
has not been achieved to date.
"Techniques using delay lines or
94
SILICON CHIP
digital field stores ('spatial filtering') offer a partial solution. With
NTSC, a simple glass delay line can
provide good chroma/luminance
s_eparation. With PAL, two delay
Imes are required, with more complex circuitry, and it doesn't work
as well. In both cases it is at the expense of vertical resolution.
"The techniques work best on
stationary images containing lots of
vertical lines, which means that
they work better on electronically
generated test patterns than they
do on real pictures! This is not to
say that the process is impractical.
By us~ng elaborate digital filtering
techmques, quite impressive results can be obtained - at considerable cost.
" But the bottom line is that the
recovered YUV signals are still
never identical to the originals and
if they have to go through the sam~
process again, in the course of
video production, the deterioration
becomes more marked. There's a
limit to the number of generations
possible".
Avoiding compromise
Such problems could be avoided
s~ys Ke~th Walters, by keeping ali
signals m a production facility in
component form right up to the end
of the process, converting them to
composite video at the transmitter.
Component signals would be
more compatible with digital video
effects machines; NTSC " green
faces" problems would be elim-
Satellite broadcasts:
MAC or PAL?
"It's not as if satellite links can't
handle composite video. You
may need a little more power or a
slightly larger dish but it's hardly
the disaster area some champions of the MAC format would
have us believe.
"Personally, I reckon that the
adoption of the B-MAC system
by the ABC had a lot more to do
with bureaucratic desire for a fully remote controllable distribution
system than any technical consideration". (K.W.)
inated; SECAM would become a
more practical system; special effects such as chroma-key and electronic captions would benefit in all
systems; and NTSC/P AL standards
conversion would be facilitated
e_ach. becoming virtually indis~
tmghishable fr om the original
material.
These desirable benefits do not
come easily or cheaply. I quote again with some abbreviation:
"The preferred approach to
operating 'component' is to run
three cables everywhere the signal
has to go, instead of one as with
composite video. However, the
length of the three cables has to be
closely mat ched and rigidly
maintained.
" Component vision mixers have
only recently become available rend~red pra ctical by the adoption ·
of highly a ccurate and drift-free
signal processing. Slight gain errors which, with composite video
would cause negligible contrast
change, could produce noticeable
colour er rors with component
video.
" Component video r ecorders
have been available for some time
notably the Sony Betacam cam~
corders used by news crews. Using
two sepa rat e pairs of heads
mounted in parallel on the head
drum, the luminance and multiplexed chrominance signals are recorded separately on a specially
modified Beta cassette run at nine
times normal Beta speed".
Betacam & Betamax
At this point, K.W. digresses to
emphasise that Betacam has little
else in common with domestic
~etamax VCRs. Similarly, profess10nal C-format video recorders
have nothing to do with domestic
':HS-C, despite fr equent implications to the contrary in both instances. C-format uses "one-inch"
(25 .4mm) tape and a 152mmdiameter drum rotating at the field
rate. It provides for various head
options and can cope with a composite signal bandwidth of up to
6MHz. But, to continue with the
main theme:
" Getting b a ck to component
recorders, standard Betacom has a
bandwidth of about 4MHz. The new
SP (superior performance) using a
new metal tape boasts a luminance
bandwidth of over 5.5MHz - in my
experience a useful response to
over 6MHz. On a component input
monitor, the picture is incredible".
K.W. also mentions the " MII"
format, jointly developed in Japan
by Matsushita and NHK. Basically
similar to SP Beta cam, it offers
more features, having been essentially redesigned from scratch as
the basis for a full range of TV station equipment, from camcorders to
editing facilities.
"Finally, many TV stations are
looking seriously at S-VHS camcorders [fitted with 3-chip CCD TV
cameras) as a low cost format for
news gathering, particula rly in
hazardous situations. At a round
$15,000, they cost about one-sixth
of a Betacam set-up. The SIN ratio
is not as good but, if they're edited
directly onto a higher quality production format, they're adequate
for most situations.
State of the art
"But state of the art in video
recording is in the new digital video
recorders [D-1 format) made by
Sony and BTS [Broadcast Television Systems, a collaboration of
Philips and Bosch).
"D-1 machines r ec or d components digitally at an industry
standard sampling rate known as
4:2:2. They use special large
cassette tapes and involve a data
rate of 200 megabits per second.
"Three principal strategies are
used to achieve this seemingly impossible recording rate:
• The data stream is multiplexed
onto two record hea ds mounted in
parallel;
• The head drum rotates at three
times the field rate;
• A special metal particle tape is
used, formulated so as to optimise
the distinction between ones and
zeros.
"There is virtually no deterioration between as many generations
as one is likely to encounter in a
production situation. They guarantee 20; more than enough!
"Again, with Quantel ' s new
'Harry' system, about 90 seconds of
digital component video can be
stored on a combination of hard
PAL isn't dead yet!
"There's an awful lot of life left
in our present TV systems. Over
the past 10 years, the SMPTE
(Socie'ty of Motion Picture and
Television Engineers) has
published numerous articles on
compatible improvements to
current TV systems. Judging by
the accompanying photographs,
they could otter a substantial
improvement over what we have
now." (K.W.)
discs and RAM - long enough for
most commercials. It provides virtually instant access to any frame,
which appear on the monitor
screen like strips of film, making
editing a simple task for nontechnical operators. One Harry and
one digital VTR can replace a
whole multi-VTR editing suite.
To be realistic
"Component video has many
theoretical advantages but they only add up if a system is component
all the way - and that simply isn't
practical at the moment. There's
too much money tied up in composite equipment, which works well
for most present purposes.
" Where component systems are
introduced, they are usually in the
form of component "islands" ,
essentially isolated from the existing composite chain. Any video
that has to be brought in from outside the island has to be decoded
from YUV, with a corresponding
drop in quality".
In an aside, K.W. points to the
above as a reason for the lukewarm
reception by broadcasters of the
various MAC formats. They would
be ideal if everything was shot,
edited and released in component
form. In reality, present program
material comes mainly from 1-inch
tape which has to be turned back
into YUV.
As reproduced by a special MAC
receiver, the end result may still be
pretty good but in most cases, it has
to undergo further ·composite encoding/decoding before it appears
on domestic receiver screens.
The basic point K.W. is seeking to
make in all this is that HDTV is not
just another extension of present
systems. Before addressing the problems of presenting high definition
wide screen images in the home, the
TV industry must first come up with
suitable program material and that,
in turn, will involve a n a llcomponent production format.
I quote: "Despite all the complexities described above, converting to
fully component format might prove
to be a mere detail compared to going to full-scale HDTV. As I said
right at the start, I don't believe
HDTV is anywhere near ready for
general release".
Summing up
K.W. concludes by saying that he
was anxious to see HDTV in the
Japanese pavilion at Expo '88 , expecting to see real HDTV sets on
display, alongside a conventional
receiver. What he saw were thr ee
HDTV projection screens, shared
by slide projectors, viewed from a
minimum distance of 3 metres.
The cameras on show produced
noticeably noisy pictures, even in
fairly bright light, which is what
one would expect with a 30MHz
bandwidth. They're supposed to
have solved that problem with a
new type of camera tube but the
display was scarely indicative of
mature technology.
As for shooting movies on HDTV
instead of film , K.W. sees little advantage in the forseeable future. I
quote:
"The low-light performance of
the best HDTV camera today is
laughable compared with relatively
cheap and available 400 ASA film.
"Admittedly, it's easier to create
special effects on video but what is
more likely to happen is that
footage captured on film will be
processed digitally by a special
high resolution telecine scanner
[straight off the original negative).
The effects will be produced by
computer and the resulting image
transferred in negative form onto
film again. Special effects don't
have to be done in real time" .
There's more but by this time,
you've probably had your fill of
HDTV. Even the writer admits that
the letter is a bit " long-winded" but
it certainly seems that the picture
he paints is not a HDTV one.
~
SEPTEMBER198 9
95
|