Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!mnetor!seismo!cmcl2!rna!dan
From: dan@rna.UUCP (Dan Ts'o)
Newsgroups: comp.graphics
Subject: Re: Frequency distributions of red, green, and blue.
Message-ID: <572@rna.UUCP>
Date: Wed, 19-Nov-86 20:56:47 EST
Article-I.D.: rna.572
Posted: Wed Nov 19 20:56:47 1986
Date-Received: Thu, 20-Nov-86 01:27:01 EST
References: <647@husc6.HARVARD.EDU>
Distribution: net
Organization: Rockefeller Neurobiology
Lines: 61


>From clyde!rutgers!seismo!husc6!endor!greg Fri Nov 14 09:58:22 EST 1986
>Organization: Harvard
>
>>Does anyone have know (or have references to) the exact frequency profiles of
>>red, green, and blue phosphorous on a color monitor, red, green, and blue
>>film, and the red, green, and blue receptors in a person's eye?
>>Greg
>>
>You've fallen into a trap common to computer technologists and
>engineers: you've assumed that the eye and brain work like a computer.
>The cones in the eye's retina respond differentially to frequency
>of the incoming radiation.  There are not three well-defined types
>of receptors which one would call red, green, and blue.

	Hmm... A bit of disinformation here...

	There, of course, ARE three well-define cone receptors as well as the
rods. The spectral sensitivity of the cones DO peak in the red, green and blue.
However, they are broadly tuned such that, e.g., a monochromatic green light
will also excite (to a lesser extent) the red cones. This is as it should be,
since three sharply tuned receptors would show nothing for light outside their
bands, forcing the necessity for many more than three receptors.

	Yes, the beginnings of color vision is computed early on differentially.
The concept is called color opponency. There are two major color opponent
systems known: red/green and blue/yellow. There may also be a green/blue
opponeny system. Yellow is thought to be derived from the summed inputs from
red and green cones.

	However, color perception is much more complicated than that. One
important property which indicates this is called color constancy. Color
constancy is a high level computation performed somewhat more globally (in
visual space.) It is manifest, e.g. in the relative invariance of the perceived
color of a group of object, irrespective of the spectral content of the
illumination. For example, things seem to have nearly the same color under a
wide range of illuminations: dawn, dusk, midday, even under highly colored
lighting. The point is, color perception is much more complicated than just
detecting the wavelengths of light arriving at the retina.

	Back to the original question. The exact output of a color monitor
is, of course, dependent on the phosphors that a monitor uses, among other
things. There are quite a few phosphors available. The bottom line is that
you need to find out what monitor you have and what options are on it (the
most often distinction is between "standard" and long persistence phosphors)
and then contact the manufacturer for the curves. Those companies well-based
in the U.S. are usually more capable of answering the question. For example,
I have the output curves for my Tektronics 690SR and a Conrac monitor. (If
you're just curious, I could perhaps describe them to you by email.)

	As far as the spectral sensitivity of the human cones, there are
numerous references and methodologies. The most classical studies just
measured via microspectroscopy, the absorption curves of the cone pigments.
Then there are psychophysical and physiological measurements. Most of the
results from these methods coinside, though not completely.

	Perhaps the best place to start is a standard reference text on the
retina (any decent science or medical library should have one). An example
is, I believe, Retina by Rodieck.

	BTW, since when were computers able to "see" ? :-)