Color schemes and color blindness

Color schemes and color blindness

Post by Daniel P. B. Smi » Mon, 24 Aug 1992 19:44:30

Quote:>>You should try really hard to avoid using color in user
>>interfaces.  It should be all black and white
>Sigh. If only this were true.
>From a HCI-purist point of view, you are probavly correct. There are plenty
>of ways to produce emphasis on a decent workstation without using colour.

I understand that you don't want "angry fruit salad." I also understand
that a good designer can make his or her point in monochrome ("black and
white?" what? not black-and-amber?). And I also understand that monochrome
displays and color-deficient users are important parts of the population.
I still don't see how that justifies a proscription against color.

What I really wanted to muse about, though, was a) color-blindness, and
b) hue versus brightness.  The commonest kind of "color-blindness," by far,
is deuteranopia. Deuteranopes see all colors, but do not distinguish
between red and green. In effect they have a two-dimensional color space
instead of a three-dimensional one. Notice that their color world is not
as impoverished as many think.  They experience hue, brightness, and
saturation.  For example, they distinguish bright orange from brown (a
desaturated, dim orange).  It's just that bright green may also look like

In the case of traffic signals, for a VERY long time now (since the
twenties, maybe?) the practice has been to use a rather yellowy-orange
shade of red for the "red" light and a rather bluish green for the "green"
light. This means that most "color-blind" people see traffic signals as
being colorful, and with easily distinguishable colors.

By coincidence (well, I doubt that it's coincidence -- the people that
designed color video probably know about the Hering color theory) traditional
video is comprised of "luminance" (brightness) and "chrominance" (color),
with the color carefully factored into one axis which is basically blue/yellow
and another which is basically red/green. If you list "brightness,"
"blue-yellow color distinctions," and "red-green color distinctions," you
are in effect producing a ranking: brightness is most important in forming
a recognizable image, blue-yellow discriminations next, and red-green least.
Hence in trying to do what would now be called "image compression" in an
all-analog world, the designers of the video signal transmit luminance
with the highest bandwidth, blue-yellow next, and red-green least.

Unfortunately, early cheap designs of digital color systems (RGB signals,
two levels for each gun) have tended to produce palettes that make it difficult
to use color WITHOUT producing angry fruit salad.  In fact I'm suffering
right now because although VGA hardware lets me pick *ANY* 16 colors,
the standard Windows driver restricts me to the a miserable 16.

Well, the point (there WAS a point) is that I think we all agree that
good designs should use brightness as WELL as color to distinguish and
emphasize elements.  (Probably the defining characteristic of "angry
fruit salad" is not that the colors are tasteless or garish, but that
you have chrominance differences WITHOUT luminance differences).

This can and should be extended to say that color differences themselves
should emphasize the blue-yellow distinctions.  By doing this, it
should be possible to design displays in which people with the
common kind of color-blindness benefit from the use of color.

Now, as to HOW to do this (or is it being done now?) I haven't a clue.
Probably with a bit of research one could design a "palette-collapser"
that would reprogram a VGA card so that a normal full-color screen would
be transformed to show the normal-color-system viewer how the screen
would look to a deuteranope. (No, I don't want to get into the philosophical
issue of whether the yellow a deuteranope sees is the same yellow I see,
or the same yellow you see.  What I'm suggesting is that there's probably
some numeric modeling in the literature that could automatically show
you a screen in which you would make the same visibility distinctions
that a deuteranope makes looking at a full-color screen).

What I mean is, if you were designing for the VERY, VERY rare part of
the population with NO color vision -- the ones that truly see in
monochrome -- what would you do?  Well, you'd use a monochrome display.
Or turn down the chrominance control, if your display had one.  Similarly,
to design for color-blindness, just "turn down the red-green-axis control".

Daniel P. B. Smith


1. colors colors colors

They have also been refered to by color, with tops-blue being
Tops-10 and tops-orange being Tops-20.  Obscure refs have been
made to tops-black, which would be pre-6.04 LIR (limited interim
release) which was often seen on the early KA10's (both of mine,
serial numbers 9 and 44, were black), though I think I had some
DF10 and RC10 (and of course the BeerCooler10) gear which was

Of course, that pretty sea-breeze blue that the PDP6 came in was
pretty cool.


2. I NEED HELP getting the serial number of an IDE dirve

3. Color Blindness

4. Enterprise Template

5. Color Blindness and Computer Screens

6. What is this card??

7. To Glenn Tenney; about book chapter on color blindness

8. W2K Pro vs NT4 Server Incompatibility

9. color blindness and computers

10. Color blindness and it affects on GUIs

11. Catering for colour blindness

12. Screen Design considerations for color blindness

13. Color blindness tests NEEDED