Re: NTSC (Re: [ProgSoc] Microsoft's ClearType)

[Christian Kent <ckent@nospam.socs.uts.EDU.AU>]

Maybe if I pause long enough within this thread, nobody will ban it  :)

On Mon, 21 Dec 1998, Telford Tendys wrote:

> Are you telling me that an NTSC broadcast can transmit 350 vertical white
> stripes with black between them and that each stripe is individually visible

Yep, professional-quality cameras are often rated at 600 or 700 lines.
These are just conventional NTSC or PAL video cameras.  SuperVHS is rated
to at least 400 lines, and BetaCam-SP manages 600, I believe.

> on close inspection? Given that the quality of NTSC is worse than PAL and
> that PAL is pretty poor, I would have to see this with my own eyes to believe it.

Well we are talking "perfect" NTSC, which really isn't any worse than
perfect PAL.  Both are analogous to RGB, anyway.

> I could well believe that NTSC is FILMED at high resolution but how much of
> that gets through to the actual TV set is quite another matter.

What's that supposed to mean?  Ignoring the "film" terminology, I would
say that the manufacturers wouldn't bother putting a million pixels into
the camera's CCD if it didn't get used in the end.  You can expect the
entire detail to come out of the station transmitter, okay -- for the
purpose of argument I'll concede whatever happens after that.

> I still remember the days of C= 64 and ZX81 machines that did their output to
> the TV set and I remember that 40 columns of text was about the limit of what
> was readable (and that was getting fuzzy).

Well you occasionally see computer screens on the TV news, whether via a
camera (eg CAD image), or a web page printed to VT, and they always look
legible, a helluva lot better than the 40-column text you and I remember.

> Try sending NTSC signal to a PAL display monitor or a multisync monitor set
> to sync-on-green, neither of these displays will recognise the NTSC colour
> and you will see vertical stripes.

This is just the display trying to cope with NTSC it doesn't understand,
ie they're artefacts of interference.  That's why the reverse doesn't
apply, so that you =can= create these vertical stripes on NTSC if you
really wanted to.

Look, here's one more piece of logic:  NTSC was designed to be backward
compatible with B&W 525-line.  If NTSC simply makes its colours work by
drawing vertical stripes, then all the owners of B&W sets in the 50s would
have noticed a sudden change in all the greys becoming stripes.  The NTSC
colour is encoded on a subcarrier to the luminence, not the luminence
itself.

> > Sounds more like a bad generation of colour.  Test patterns always include
> > stripes that are very close together, and the quality of the image is (in
> > part) determined by how much/little colour bleeds out between the stripes.
> This is precisely because those test patterns are attempting to confuse the
> colour mechanism. They use stripes that are just a bit different from the
> colour signal to prove that the TV can correctly filter colour information out
> from intensity information. The reason that sometimes colour will bleed is
> because the TV is having difficulty with this filtering.

Ergo, the TV is at fault, not the NTSC spec.  The NTSC TV should be
decoding the test-pattern stripes in just the same way as a B&W TV would,
after all the transmitted signal is also identical to B&W test pattern's
stripes.

> > developing high-res graphics for NTSC output these days, they don't have
> > to worry about which pixel changes to what colour depending on position.
> That is exactly what I am saying!
>  Step *1*, black and white TV.
>  Step *2*, NTSC standard is written as a hack to lay colour onto B&W signal.
>  Step *3*, apple make a monochrome machine.
>  Step *4*, apple add colour by following the EXISTING hack known as NTSC.
Step 5, they reduce the hardware requirements of the colour graphics mode
by placing a dependency on the pixel colour with its position, thus
developing the world's only 2-bit colour mode  <smirk>  (pun intended)

Step 6, they develop a double hi-res mode which doubles the monochrome
resolution, or alternatively, squares the number of colours from 4 to 16.
Old hi-res mode images are completely compatible and display the same way.

Step 7, they develop a super hi-res mode, upon the same NTSC composite
connector, which does away with colour / pixel-position dependency
(depending which mode you select).

> I suspect that this is POSSIBLE but only because of the availability of
> digital filters and other similarly expensive techniques. When the standard
> was written, no one was thinking about computers nor did they care one
> iota about transmitting high contrast stripes. They wanted to transmit

Quite right -- but I put it to you that all the hi-fidelity images
transmitted today are still compliant with the spec written up in the 50s
and 60s.  A nice 15-year-old TV will display a computer image with as much
enthusiasm (actually, today's commodity TVs suck) as one made today.

> > After all what do you think they've been up to in the TV engineering
> > departments for the last 30 years, improving things?  The specs haven't
> > changed, but the implementation sure has  :-)
> I'll admit that I have never spent the money on a really top-notch TV set.
> The old 1084 C= monitors had a PAL decoder in them and you could hook them
> up to a VCR to get a TV, it generally looked pretty clean compared with

If anyone has an Amiga monitor with an RCA-input lying around, KEEP IT.
Hook it up to your VCR as a tuner, because it makes the best damn TV image
between here and the Sony MultiSync factory.

> your average lounge room set. The effect of digital filters in the modem
> scene has been awesome so I wouldn't be surprised if similar improvement
> hasn't occured in the area of TV as well, probably TVs are more price-sensitive
> than modems though.

Well that does come into it when you shop around for 100Hz sets, which are
in their third generation now and use really nice ICs to tidy up the
picture.

But regular sets today use very cheap decoder chips to get the colour
information out, and you can see how lousy the skin-tones are on some sets
by putting a Philips, Orion or GE set next to a Sharp or Toshiba.  Plus
the phosphors aren't as good anymore, but anyway.  They seem to do nothing
to the horrible crawling around the edges of graphics, either.  That gets
totally eliminated on all 100Hz sets.  Mmmmm.

CK.

--
You are subscribed to the progsoc mailing list. To unsubscribe, send a
message containing "unsubscribe" to progsoc-request@nospam.progsoc.uts.edu.au.
If you are having trouble, ask owner-progsoc@nospam.progsoc.uts.edu.au for help.

This list is archived at <http://www.progsoc.uts.edu.au/lists/progsoc/>