Why are red, yellow and blue the primary colours for paint, but red, green and blue for light, and how does...

In article <321ccedb.7035067@news.voyager.co.nz>
stantons@voyager.co.nz (Alison Smith) writes:
>Tak To <tak.to@aspentech.com> was inspired to write:
>>Note that Red, green and blue are 'primary' colors
>>because humans have light receptors in the retina that are built
>>for these frequencies.  To other animals these may not be 'primary'
>>at all.
>>
>In art class we were always taught that red, yellow and blue were the
>primary colours because you couldn't mix them from other colours.
>Green is made by mixing yellow and blue paint.  As far as paint is
>concerned you definately don't get yellow when mixing red and green
>together you khaki.

Paints (more commonly, pigments) are "subtractive" -- they reflect
everything except a particular color that the pigment absorbs.  So, for
example, magenta pigment absorbs green light, cyan absorbs red, &c.  If
you mix magenta pigment with cyan, then the green and red are *both*
absorbed, leaving blue to be reflected.  Similarly, if you mix yellow
and cyan (which, let's face it, is pretty damned blue), green is
reflected.

Lights, however, are additive.  If you shine a red light and green
light in close proximity, your eye responds to both, and gets fooled
into thinking it sees yellow.

>However, Tak's comments solve for me what has always been a mystery as
>far as TV screens go.  They are made from red, green and blue dots and
>I could never figure out how they made yellow on TV.  [...]
>
>One other thing I've heard is that blue is the hardest colour for the
>eye to focus on.  Anybody out there who can explain these mysteries?

Simple biophysics.  Blue is the shortest wavelength, and the eye isn't
especially responsive to blue anyway.

-- Patrick Juola <patrick@gryphon.psych.ox.ac.uk>

karr@world.std.com (David and Enid Karr) wrote:
>If you see yellow, the reflected light is only (or, I suppose,
>*mostly*) red and green.  So yellow paint absorbs (almost) all the
>blue light that falls on it, so what's left when the light reflects
>back to your eye is (almost) all red and green.

This is not necessarily so.  Yellow paint (paint that *looks* yellow)
may be reflecting yellow light, or it may be reflecting a combination
of red and green light.  Yellow light stimulates both the red and green
photoreceptors, as does a combination of red light and green light, so
true yellow looks the same as the (additive) combination of red and
green.  But yellow light passed through a prism remains yellow, whereas
the combination of red and green (which looks the same as yellow) is
broken up into red and green.

The fact that there are three primary colors (as opposed to 2 or 4 or
whatever) is a fact about how the human visual system responds to the
visible spectrum, and not about light or pigments.

Question: Are there actually pigments (if so, how common?) which
reflect, for instance, just red and green light, thus appearing yellow?

-- Bill Vanyo <vanyo@ezaccess.net>

In article <50k4i5$f12@krypto.zippo.com>, Bill Vanyo
<vanyo@ezaccess.net> wrote:
>MAS <bub@bubba.com> wrote:
>>I do believe that the human eye has receptors for red, green, blue,
>>and YELLOW, giving yellow a sensation of "primaryness."

Color perception is surprisingly complicated; a computer graphics book
(e.g. Foley and van Dam) is a good place to start.  I'll try not to
confuse things too much in my explanations and try not to simplify too
much.

The eye has three kinds of color sensors (cones), which can be
considered red, green, and blue.  However, each type is actually
sensitive to a wide range of the spectrum, and the peak sensitivities
of the red and green cones are actually in the yellow range.  Red light
stimulates the "red" receptor more than the "green" receptor, so the
light appears red, but the "green" receptor is still active, for
example.

According to the opponent-color theory, early brain processing of these
signals converts it into three new types of signals, one which ranges
from black to white, one which ranges from red to green, and one which
ranges from blue to yellow.  So the first signal is basically like a
black and white TV, the second tells if the light is reddish or
greenish, and the third tells if it is bluish or yellowish.  Note that
these are opponents: you can't have a blackish white, a reddish green,
or a bluish yellow.  This is probably what MAS is thinking about for a
yellow receptor.  I designed a color model based on opponent-color
theory: "The RBW color model", Computers&Graphics 17(5).

Then there's a linguistic theory ("Basic Color Terms", Berlin & Kay,
University of California Press) that there are 11 basic colors: black,
gray, white, red, orange, yellow, green, blue, pink, purple, brown.
(There's a definition for basic which I can't be bothered to look up,
but think of them as fundamental colors.  E.g. magenta is not basic
since you could describe magenta as a purplish red, but you wouldn't
describe red in terms of other colors, or use magenta as a modifier
modifier (magenta-y blue).)  Interestingly, these basic colors are
pretty much the same across all cultures, with different cultures using
the same basic colors or subsets according to specific rules.  I also
wrote a paper using this theory:  "Laundry and the origin of basic
color terms", Journal of Irreproducible Results 36(6).

>I know that a combination of red and blue
>light appears as magenta.  Question: Can monochromatic light of some
>frequency stimulate both the red and blue receptors to give the
>appearance of magenta?

No.  This would be clear if you had a CIE chromaticity diagram to look
at, but basically the spectrum takes you from red through green etc to
blue, but to go from blue through magenta back to red requires
combinations.

Finally, note that three visible primaries (e.g. red, green, blue)
won't provide all the possible colors.  Again, a CIE chromaticity
diagram would make it clear, but the problem is that blue + green
doesn't quite give you "pure cyan".  If you can use negative
intensities, three colors will match anything, e.g. blue + green = cyan
+ a bit of yellow so blue + green - yellow = cyan, so you can match
cyan if you're allowed "negative yellow".  This means, for instance,
that your TV can't provide all colors accurately.  It also means
difficulty for computer graphics since usually the monitor can display
colors you can't print and the printer can create colors the monitor
can't display.

Unfortunately I can't think of any interesting puzzles involving color
theory :-)

-- Ken Shirriff <shirriff@eng.sun.com>