Graphics Reference
In-Depth Information
A studio broadcast monitor has a reference brightness of 100 candelas per square
meter.
It's tempting to say that since the human eye's sensitivity to light is captured
by the candela, we could (if we wanted to do just grayscale graphics) represent
all light in terms of the candela. As mentioned in Chapter 1 and Chapter 26, this
would be a grave error. In doing so, we'd need to assign a reflectivity to each sur-
face; assuming diffuse surfaces, this would be a single number indicating what
fraction of incoming light becomes outgoing light. Suppose we have a surface
whose reflectivity is 50%. Then incoming light of a particular luminous inten-
sity would become outgoing light of half that intensity. The problem is that real
surfaces, with real light, may reflect different wavelengths differently. A surface
might, for instance, reflect the lower half of the spectrum perfectly, but absorb all
light in the upper half. If it's illuminated by two sources, one that's in the lower
half and one that's in the upper half, with equal luminous intensity, the reflected
light in the first case will have the same luminous intensity, while in the second
case it will have none at all. In other words, there are cases where this “summary
number” captures information about human perception, but masks information
about the underlying physics that brought the light to the eye. One could argue,
therefore, that luminous intensity of light should only be examined for light that
arrives at some person's eye.
Counter to this position is the fact that much of the light we encounter every
day (like that from incandescent lamps) is a mixture of many wavelengths, and
most surfaces reflect some light of every wavelength, so in practice we can use
a summary number like luminous intensity, and a summary reflectivity, and the
reflected light's luminous intensity will turn out to be the incoming intensity mul-
tiplied by the reflectivity. This summary-number approach only causes problems
in cases where the spectral distribution of energy (or of reflectivity) is peculiar.
But with the advent of LED-based interior lighting, such peculiar distributions are
becoming increasingly commonplace; many of today's “white LED flashlights”
are actually based on multiple LEDs of different frequencies, and have highly
peaked spectral distributions, for instance. This discussion is another example of
the Noncommutativity principle.
We've said that because photometric quantities represent weighted averages,
and the weighted-averaging process does not commute with various other oper-
ations (like multiplication), these photometric quantities will be of little use to
us except when applied to the light arriving at the human eye. To clarify the
statement about weighted averages, consider the following example. We take
two lists of numbers,
L
=(
1, 3, 1, 5, 6
)
and
(28.4)
R
=(
.33, .33, .33, 0, 0
)
,
(28.5)
and consider the weighted sum of each under the weights
w
=(
0.2, 0.2, 0.3, 0.3, 0.0
)
.
(28.6)
The results are 2.6 and .233, respectively.
Now consider the term-by-term product of
L
and
R
;itis
(
0.33, 1, 0.33, 0, 0
)
,
(28.7)
