Characterize Digital Camera Color

[Ruey <tmi@xxxxxxxxxxxxx>]

An interesting experiment with a digital camera would be to shoot three 
separate exposures through a sharp cutting tricolor filter set. It might 
take some testing to get balanced exposures. As a starting test I would 
try 3 stops for each of a 25-58-47B tricolor set or 29 red = +4.3 stops, 
61 green = +3.3 stops and 47B blue = +3.0 stops. for daylight. What this 
would give was a way to test digital camera color. For example the 
digital camera reviews typically say that a Sony A850 does not have as 
nice a color as say a Nikon full frame camera. I have wondered how 
subjective this judgment is. Could it be that contrast or other factors 
are confusing the color evaluation? By shooting tricolor through a 
digital camera on three separate RGB shots using filters like 25-58-47B 
or 29-61-47B you would have a way to evaluate the color yourself. First, 
you could combine the separation images in Photoshop and compare them to 
the normal digital camera shot visually. Second, you could use the 
difference combine mode to subtract one image from the other and have an 
image that was the difference between the normal digital sensor image 
and the separation RGB image of the sensor. Since the registration would 
be inherent, the images would be aligned. If the differences were 
significant this could offer a way to get nicer color for landscapes or 
still life images where movement of the subject was not an issue. Since 
the digital sensor spectrum plots I have seen - not often offered in 
detailed versions by manufacturers - seem to show significant color 
overlap I bet it might be possible to get purer, dye transfer-like 
original images this way. Has anyone tried this?

Ed Scott