Path: utzoo!utgpu!watmath!onfcanim!dave
From: dave@onfcanim.UUCP (Dave Martindale)
Newsgroups: comp.graphics
Subject: Re: Film Recorder Wanted
Message-ID: <16746@onfcanim.UUCP>
Date: 18 Nov 88 21:58:27 GMT
References: <4071@psuvax1.cs.psu.edu> <325@ur-cc.UUCP>
Reply-To: dave@onfcanim.UUCP (Dave Martindale)
Organization: National Film Board / Office national du film, Montreal
Lines: 67

In article <325@ur-cc.UUCP> haake@.cvs.rochester.edu (Bill Haake) writes:
>
>Some questions have are:

>How well does the addressable resolution match what actually gets on
>the film?

The two are pretty independent.  The addressable resolution just depends
on the width of the DACs and other digital hardware chosen by the
manufacturer (and in some cases is determined by analog means that
have no specific limits).

The useful resolution depends on several things.  First, take the
diameter of a fully-illuminated spot on the CRT face (the tube's
spot size) and divide it into the diameter of the usable portion of
the tube face.  Celco claims 0.8 mil for their best recorders,
Dicomed claims 1.0, most others are probably substantially larger
(particularly if the spot size is not specified).  Typical CRT
diameters are 7 inch and 5 inch, with some smaller.  The result of
the division above will be a number in the range of several thousand;
this is the useful resolution for the image diagonal.  For a typical
image of about 4:3 aspect ratio, multiply by 0.8 to get the horizontal
width of the picture in pixels.

Ideally, the lens should have considerably higher resolution than the
CRT and the film, so it isn't the limiting factor.  In practice, there
is also a desire to use as fast (large) a lens as you can to decrease
exposure times, and a faster lens is either less sharp or more expensive.
Tradeoffs, tradeoffs.

>What is the resolution of different films both color and b/w?

Medium-speed colour negative film can hold about 30 line pairs
(that's 60 pixels) per mm at a 1.6:1 contrast ratio, going up to
about 100 lp/mm if the original's contrast ratio is 1000:1.
For a good ballpark figure, assume you need about 50 lp (100 pixels)
per mm of negative width for something that looks "really sharp".
You may be able to get away with substantially less than that for
computer-generated scenes that don't actually have enormous amounts
of background detail.  But if you want something that compares to
a photographic original, you need lots of resolution.

>How much of a difference does the size of the imaging screen make
>in the quality of the output?

See above.  The larger the screen, while keeping spot size constant,
the more useful resolution.  But it's the ratio between spot size and
screen size that is important, not the absolute value of either.

>How does an electrostatic display work and why might it be better
>than a magnetic deflection CRT?

Electrostatic deflection applies an acceleration perpendicular to the
electron beam, while magnetic deflection deflects the beam by a certain
angle (the deflection is always perpendicular to the electron's instantaneous
velocity).

With electrostatic, the amount of deflection is directly proportional
to the deflection voltage, while with magnetic deflection the angle is
proportional to the current and the total displacement to the tan of
the current.  As a result, circuitry has to be used to linearize the
deflection.  Worse yet, horizontal and vertical deflection interact
with each other, which causes the image to suffer "pincushion" distortion,
requiring further correction circuitry.

So why do people use electromagnetic?  It gives a smaller spot size,
and the deflection circuitry operates at low voltage.