From: utzoo!henry
Newsgroups: net.works
Title: more on display resolution 
Article-I.D.: utzoo.2333
Posted: Tue Aug  3 19:34:00 1982
Received: Tue Aug  3 19:34:00 1982

Actually, display terminals are a *worse* problem than phototypesetters.
People doing phototypesetting have one big advantage:  almost all their
stuff is intended to be reproduced by printing processes, which cannot
reproduce 1000 lines/inch.  (In fact, it's a bit peculiar that people
bother investing in 5000 lines/inch phototypesetters when all that extra
resolution is totally wasted on the final product.)  Trouble is, the
human eye can see 1000 lines/inch jaggies.  If you want a screen that
looks like a printed page, yea unto the very limits of the human eyeball,
you are going to need *even higher* resolution.

On the other hand, if you are a reasonable man rather than one of the
graphic-arts types who buys a 5000 lines/inch phototypesetter because
when he looks at the original (*not* the mass-production product) with
a magnifying glass he can see the dots in 1000 lines/inch output, then
you may be able to content yourself with much lower resolution.  I have
heard that the critical resolution for "smooth looking" output is somewhat
dependent on the dynamic range of light intensities, and CRTs actually
are somewhat better off than paper (i.e. need less resolution to look good).

There was a *very* interesting paper in Siggraph a year or two ago.  The
author (alas, I don't remember his name, and my Siggraphs aren't handy)
claimed that standard NTSC resolution actually approaches the limits of
the human eye, if used *right*.  You need a good monitor, plus possibly
some hardware fiddling so the scan lines touch each other (normally there
is black space between them).  You need *at least* 8 bits of gray scale,
and your software must *USE* it properly.  Oh yes, and your D-A conversion
must be calibrated properly so that that 8-bit range really does turn into
the right sort of range of light intensity.  The explanation of why this
works relies on things like halftoning effects, and I can't reproduce all
of it off the top of my head;  the key fact is that the resolution of the
human eye is not a single number, but a complex phenomenon, and by doing
things right you can do an end-run around resolution limits.  Don't flame
to me that it's impossible, read the article first.  It's in one of the
Siggraph conference proceedings, probably two years ago, titled something
like "The 8000-line display".

						Henry Spencer
						utzoo!henry