>
> > They can’t be “identical” and go on to have “slightly more red
> sensitivity.”
> Thanks for that. I hate to always be the one.
Simple explanation, actually. But you know that "simple" ain't suitable,
therefore here is a typical Schnozz answer.
MOST sensors use a three color sensor array consisting of Red, Green and
Blue sensitive detectors. In MOST cases, the peak sensitivity is at the
primary colors at or near 650nm (Red), 540nm (Green), and 450nm (Blue). As
these are the primary colors, all other colors in the visible spectrum can
be mixed from them. The IR cut filter kicks in just past 650nm which is a
good thing because all detectors are equally sensitive to IR at 850nm. If
you didn't cut off the IR wavelengths bad things happen. (Leica M8).
In the case of the Kodak sensors found in the E-1, E-300, E-500 and E-400,
all of the green detectors have a peak sensitivity at approximately 540nm
and they are identical in how sensitive they are at that wavelength. What is
different, however, is that half of the green detectors are slightly more
sensitive to the longer wavelengths. In some documents, these particular
detectors are referred to as "Orange" pixels, but that is a misnomer because
the peak wavelength sensitivity is still pure green, but the anti-red
filtering isn't quite as strong.
As a point of comparison, the typical human eye's cones are sensitive to
three different wavelengths of light too, but definitely do not match the
primary colors we are familiar with.The cones are sensitive to short (blue),
middle (green) and long (red) wavelengths, but they peak at 425nm, 525nm and
570nm. The color "red" is not directly seen by the human eye, but is
extrapolated based on the absence of green.
I know some of you must be thinking "what about film?" I'm glad you asked.
Kodachrome's peak color sensitivities are 425nm, 550nm and 650nm with a very
hard IR cut kicking in almost immediately with the sensitivity essentially
nulled by 700nm. Fujicolor Reala has FOUR sensitivity layers with the
standard color overlaps between the layers for the Red, Green and Blue
layers, but the fourth layer is not sensitive past 575nm which is where the
red sensitivity layer starts. This fourth layer is the Cyan layer.
What does all of this have to do with practical application? Well, I'm glad
you asked that question too!!! How does a film, digital or the human eye
see the color purple? And what exactly is the color purple? We all know
that with paints, if you mix the primary colors of blue and red together you
get purple, right? That is correct, but does that actually MAKE the color
purple or is it fooling the eye into seeing purple? It doesn't actually
form the color purple, but it blends to make the eye think it is the color
purple. However, there are pigments which truly are purple with a
reflectance in the 400nm area. If you filtered out red light, they would
still reflect purple. An African Violet is a bit different. An African
Violet has a peak reflectance of 450nm-475nm which is blue and another peak
reflectance starting at 675nm and extending far into the IR band. An African
Violet isn't actually purple, but is blue and red. Unfortunately for most
sensors, the red portion is actually far into the IR cut wavelengths so most
digital cameras only see blue. If the real color of a pigment or flower is
purple with a peak reflectance around 400nm, not all is lost because the Red
detectors in a digital camera have a second minor peak in the deep purple
range (just beyond blue) so a true purple (instead of a mixture of blue and
red) is actually visible to most cameras.
Which brings up a point. What about the Macbeth ColorChecker or an IT-8
target? Can you determine color response of a film or sensor with it? The
answer is yes and no. It totally depends on the pigments or dyes used in the
target itself. If they are printed with cyan, magenta and yellow dyes then
the camera is ONLY seeing a mixture of cyan, magenta and yellow. That purple
square in the target? It's NOT purple. It's actually magenta and cyan. You
cannot create colors in printing or painting, just blend them. A blended
paint or ink does not lose its native pigmentation--it just blends the color
dots much smaller than a half-tone printed page. So, the reality is, we're
judging how our sensors see blended colors, not true spectral response which
would only be determined through refracted sunlight. (prism or DVD/CD).
Fortunately, however, most colors in the natural world are blended colors
anyway.
The sources of the information above include:
1. Vision and Art, The Biology of Seeing, by Dr. Margaret Livingstone.
2. Kodak specification sheets for the KAF5101CE, KAF8300 sensors and
Kodachrome.
3. Fujifilm Reala specification sheets.
4. Miscellaneous sources to support the above theories.
AG
--
_________________________________________________________________
Options: http://lists.thomasclausen.net/mailman/listinfo/olympus
Archives: http://lists.thomasclausen.net/mailman/private/olympus/
Themed Olympus Photo Exhibition: http://www.tope.nl/
|