Chronology Current Month Current Thread Current Date
[Year List] [Month List (current year)] [Date Index] [Thread Index] [Thread Prev] [Thread Next] [Date Prev] [Date Next]

Re: a link to some interesting illusions



Larry Woolf wrote:

The question addresses the common misconception that "all of the
colors are in the rainbow."

This may not be a misconception among physics teachers, but it is
among many students. Magenta is easily used as an example of a color
that consists of a mixture of short and long wavelengths - a
non-spectral color.

There are indeed lots of misconceptions about color
and the perception of color.

It is 100% true that "magenta is a non-spectral color"
but the point is lost on many students, unless you
provide a lot of foundation and preparation.

Part of the problem is that many students have a
terribly impoverished vocabulary for describing
colors. If you tell them the subtractive primaries
are "cyan, magenta, and yellow" the typical reaction
is
-- what is cyan?
-- what is magenta?

Some people know the difference between cyan and
blue, but nobody is born knowing it, and lots of
people don't know it.

Similarly a lot of people use the word "purple"
to cover a whole range of colors including
magenta and violet. Then you've got a big
problem, because violet _is_ present in the
spectrum. So in the absence of some fairly
serious color training, a student could semi-
reasonably think that violet->purple->magenta
is a spectral color. That's wrong, but it'll
take a lot of explaining to explain why it's
wrong.

===============

There's actually some biophysics in this.

Physicists use "violet" to describe the
short-wavelength end of the spectrum. You
would think this would be "bluer than blue"
and as unlike red as anything could possibly
be. So why do people equate purple with
violet and purple with magenta, when the
latter has a lot of red in it?

The answer apparently is that the filter
used by the 'red-sensitive' cells in our
eyes is not a mathematically-perfect low-pass
filter. Look at the little "foot" on the
rho curve in the curve
http://www.photo.net/photo/edscott/eyespect.jpg
and check out
http://www.photo.net/photo/edscott/vis00010.htm
to put it in context.


===============================

Another point:

There is a spectral yellow color that can be
pretty well but not exactly represented by a
mixture of red and green lights.

More significantly, there is a spectral cyan
color that can be only rather poorly approximated
by a mixture of blue and green lights.

To make the same point with fancier terminology
and a little more generality: there are a whole
lot of colors (including many spectral colors)
that are "out of gamut" for an RGB monitor.

There is another large set of colors that
are out-of-gamut for an CMYK printer. Check
out e.g.
http://www.popularphotography.com/pdfs/colormanagp2.pdf

to see how bad it is.

I found that using:
http://www.google.com/search?q=cmyk+rgb+6-color+gamut+cie

The physics here is that there is no way to
make a light that excites the green receptor
without also exciting the red and/or blue
receptors.

==================

So far we've just discussed point-by-point colors.
There are actually nonlocal effects, whereby the
light coming from one part of the image affects
the perception of color in another part of the
image. This makes sense from a teleological point
of view (discounting the illuminant is very sensible)
but it sure makes things complicated. Physics is
only part of the story; our vision system is not
a camera or a spectrophotometer.