Graphics Reference
In-Depth Information
1.0
Scotopic
Photopic
0.8
0.6
0.4
0.2
0.
350
400
450
500
Wavelength (nm)
550
600
650
700
750
Figure 28.7: The luminous efficiency in photopic vision has a peak at 555 nm; the peak for
scotopic vision is closer to 520 nm.
average-light-level adjustment for the rods, which are the primary receptors in use
for seeing things in the dark.
28.4.1.2 Brightness
Our discussion so far has addressed the issue of how light of different wavelengths
is perceived. There's a separate issue: how light of different intensities (but con-
stant spectral distribution) is perceived. In other words, if we have a diffuser—a
piece of frosted glass, for example—and it can be lit from behind by 100 identical
lamps, and we turn on one, or ten, or all 100 so that the diffuser appears to be
a variable light source, how will our eyes and brains characterize the change in
brightness? Given the wide range of intensities that we encounter in daily life, it's
hardly surprising that the response can be modeled as logarithmic: The change
from one lamp to ten is perceived as being the same “brightness increase” as the
change from ten lamps to 100. (Here we are using
brightness
in a purely percep-
tual sense, not as something physical to be measured, but as a description of a
sensation.) That is to say, this model says that the perceptual strength associated
to seeing a light of luminous intensity
I
is
S
=
k
log(
I
)
.
(28.8)
In support of the idea of a logarithmic model of our sensitivity to luminance,
we can display two lights of the same luminous intensity and then adjust one until
it becomes just noticeably different from the other. By doing this over and over
at different starting sensitivities, we find that the
just noticeable difference
(or
JND
) is about 1% (i.e., 1.01
I
is noticeably different from
I
) for a wide range
of intensities. In very dark and very bright environments, the number increases
substantially, but for a range that includes the intensity ranges of virtually all of
today's displays, it is about 1%. So if we adjusted the intensity repeatedly by 1%,
we might expect to say that the brightness had increased by several “steps,” and
that
k
steps of increase would be achieved by multiplying by
(
1.01
)
k
. (This rea-
soning makes the assumption that each JND seems to the viewer to be of the same
“size,” however.) This implies that the response is proportional to the logarithm of
the intensity.
