is that colors created by following the video signal’s “recipe”
for mixing light from the red, green, and blue primaries will
result in wrong colors. Hence, the colors from consumer
monitors must be wrong! How wrong is a function of their
deviation from the standard, and I’ll look at how to measure
it below. But that isn’t the only error that plagues color accu-
racy for consumer monitors and projectors. Another error is
usually much larger when consumer TVs are purchased.
White Reference Color Temperature
It is necessary, but not sufficient, that the chromaticity
coordinates of the primaries match the SMPTE C standard
to accurately reproduce color. It is also necessary that the
relative brightness of light from the three primaries be cali-
brated to produce the standard D65 white reference color,
otherwise the contribution of light from each primary will
not be correct when creating any other colors. White is
defined to be the color represented by equal red, green, and
blue values of an RGB signal. The color produced by any
other combination of signal values depends on the initial
calibration of the relative primary light outputs for the
white signal values. But what brightness of white should be
used when calibrating the color temperature?
Grayscale Color Temperature
The answer is that all brightness levels of white should pro-
duce the same standard D65 color temperature. That means
the relative light outputs from the three primaries must
track together as the total brightness changes. As a result,
any color generated by another mix of the primaries will
also stay at the same CIE (x,y) location regardless of the
brightness of the color. Grayscale is the term used to
describe the color temperature of the reference white over
the range from dark gray to peak white. The closer the
grayscale color temperature can be held to D65, the more
accurate the colors will be at all brightness levels within the
picture. For instance, if the color temperature increases to
the more blue-white of 7500K in the middle of the brightness
range, then colors will be seen with a bluer hue than desired
when they appear at that brightness level.
So we have seen that two conditions are necessary to
achieve perfect display color accuracy. The grayscale must
maintain a perfect D65 color temperature across the entire
brightness range of the display, and the CRT phosphors
must match the SMPTE C standards. It’s just that simple.
U n f o r t u n a t e l y, we can’t seem to get any consumer products
that exactly match the SMPTE phosphor chromaticity stan-
dards, and no one has ever built a display that can be cali-
brated for perfect grayscale tracking. It is interesting to note
that direct-view CRT monitors can usually be calibrated for
significantly better grayscale tracking than CRT projectors,
but CRT projectors can have primary colors that more close-
ly match the standard because each primary color is gener-
ated by a separate CRT. This allows more specialization of
phosphor selection and possible color filtering of the light
output from the CRTs.
Now that we know that no consumer display will be per-
fect, let’s discuss what sort of technique we can use to mea-
sure color accuracy.
6. Color Measurement
Color Bars
It is helpful to have some standard video test signals that
can be used to calibrate a display, or to measure the accu-
racy of the display’s color performance. The easiest signals
to generate are the common color bars that almost every-
one has seen at one point or another. (Figure 5)
Color bar signals are generated by creating all possible
combinations of the three primary colors with each of the
RGB signals set to the same value. The resulting colors are
red, green, and blue, their complementary colors, cyan
(blue + green), yellow (green + red) and magenta (red +
blue), and white, which is always defined as equal RGB sig-
nal levels. The most common color bars are those that use
75 percent of the maximum signal value. We rather obvi-
ously refer to those as 75 percent bars, but 100 percent bars
V I D EO
Table 1 SMPTE C Color Bars
White Yellow Cyan Green Magenta Red Blue
x 0.3127 0.421 0.231 0.310 0.314 0.630 0.155
y 0.3290 0.507 0.326 0.595 0.161 0.340 0.070
Fig. 5