Graphics Reference
In-Depth Information
should vary from the background not just in chromaticity, but in brightness (per-
ceived intensity) as well—especially for colors containing blue, since relatively
few cones are sensitive to blue. Thus, the edge between two equal-brightness col-
ored areas that differ only in the amount of blue will be perceived as fuzzy.
Blue and black differ very little in brightness, so this is a particularly bad
combination. Similarly, yellow on white is hard to distinguish.
The eye cannot distinguish the colors of very small objects, so color coding
should not be applied to small objects. Judging the color of objects subtending less
than 20 to 40 minutes of arc is error-prone [BCfPRD61, Hae76]; at a typical view-
ing distance of 24 inches, objects 0.1 inch (i.e., many pixels) tall subtend about
this angle. The color of a single pixel on a modern monitor is almost impossible
to discern.
The color of a region can affect our perception of its size. Cleveland and
McGill discovered that a red square is perceived as larger than a green square
of equal size [CM83]. This could well cause the viewer to attach more importance
to a red object than to a green one of similar size.
If you stare at a large area of saturated color and then look away, afterimages
appear. This effect is disconcerting and distracting, so the use of large areas of
saturated colors is unwise.
For a number of reasons, red objects appear closer than do blue ones; there-
fore, simultaneously using blue to represent foreground objects and red to repre-
sent background ones is unwise. The opposite coding is fine (although the use of
saturated red text on a saturated blue background is particularly annoying to many
people).
With all these perils and pitfalls of color usage, is it surprising that one of our
first rules was that you should apply color conservatively?
The workings of the human eye and the human visual system, and they way in
which they combine to provide the perception of color, cover many fields; the
application of color for aesthetic or persuasive or communicative goals occupies
others. For more information on the human eye, presented in a manner particularly
suited to computer graphics researchers, see Glassner's book [Gla94]; other use-
ful references are [BS81, Boy79, Gre97, Hun05, Jud75, WS82] and [Poya]. For
more background on artistic and aesthetic issues in the use of color for graphics,
see [Fro84, Mar82, Mei88, Mur85, MSK04]. For more information on calibration
and cross-calibration of displays, see [Cow83, SCB88, Con12, Int03].
Exercise 28.1:
You want to interpolate between two similar colors. They're repre-
sented in RGB space. You've heard that YIQ space is more naturally related to the
human eye, so you convert to YIQ, interpolate there, and convert back. Explain
why you get exactly the same result as if you'd interpolated in RGB. For which
other color-description systems in this chapter will this turn out to be true, and
why?
Exercise 28.2:
The chapter claims that if you interpolate colors using what-
ever color triples they're represented by,
and
if the colors are nearby, then it
