||[Feb. 1st, 2004|12:18 pm]
I have been hoping to perform a test of my camera's actual sensitivity to infrared light- trying to determine what is the actual response to various frequencies of light. I have been concerned that using the Hoya R72 filter, since a small amount of red is passed through the filter I might be fooling myself into believing what I was observing was actually infrared. So, I performed a test: I obtained a bag of prisms and some diffraction gratings from Edmunds Scientific.
Actual experimental setup: I placed a diffraction grating slide in the path of a slice of light coming from the sun through our front window, and projected the refracted rainbow onto two pieces of white paper. The result was a pretty rainbow pattern as follows:
I then took two images less than a minute apart, the bottom image is unfiltered, 1/8 sec exposure at f2.8, iso 100, using the Olympus C-4000 camera. The top image is 16 sec exposure, f2.8, iso 100. Both images are contrast and color corrected.
I'm very intrigued by the results. The light that is being picked up by the camera is clearly not "visible" light. However, I'm also bothered by the fact that the diffraction grating I'm using (500 grooves/mm, 13500 grooves/in, mounted slide) produces not just one rainbow, but a series of repeating rainbows (due to inteference pattern generated from the slits) so I can't be 100% positive that what I'm seeing through the camera is the real deal. Nonetheless, it's a good start, and it will provide the rudimentary platform to test my new camera, when I buy it.
I'm also surprised, at some level, that the camera is generating different "colors" for the various frequencies- bluer for longer wavelengths, and redder for shorter, in the "infrared" region. It explains why the images I take are always very purple-magenta befor I desaturate. But, I would expect that the ability to differentiate the two ranges of wavelengths as shown would mean that I could generate some degree of color tonality since I would expect that different materials would have different reflectivities to the two different wavelength regions. It would seem, also, that the absence of significant green color in the filtered image would indicate that the Bayer filter pattern on the CCD in the camera blocks infrared best on the green pixels, meaning that the pigment used for each color has different transmittance characteristics in the infrared. What does it mean? I can't be sure.
It's also interesting that the camera desn't seem to correctly reproduce the violet region of the spectrum: it's a little too blue at the end.
Off to Columbus, Ohio.