Originally posted by PDL ... except that the one you found was for HIE, not EIR and several search hits on EIR, which I checked, were all not accessible anymore. That included
https://125px.com/docs/unsorted/kodak2/ti2323.pdf. I didn't know that the site allowed browsing, so with the help of your remark I found it at
https://125px.com/docs/unsorted/kodak2/ti2323.pdf, must have been moved after indexing. Looking at the diagram on page 7, the spectral sensitivity of the IR sensitive cyan layer extends to 900nm, but never drops very far in the visible spectrum - similar to a monochrome IR film, just with another sensitive peak at around 530nm, where HIR drops. All of the colors layers have blue sensitivity. That is indeed not exactly possible to replicate with a color sensor which mixing channels. Kodak writes: "As indicated in Figure 2, all three layers are inherently sensitive to blue radiation. To limit the exposure of each layer of color infrared film to only its intended spectral region, a yellow filter (minus blue), such as a KODAK WRATTEN Gelatin Filter No. 12 (or equivalent), is always used over the camera lens. With the yellow filter in place, the layers act as though they are sensitive only to green, red, and infrared (all blue radiation is absorbed by the filter)."
Color filters in front of sensors have overlapping sensitivities in order to achieve a good balance between luminosity and color sensitivity. (OT: Just by chance, one of my first hits while searching for sensor curves is a scientific
article by DxOlabs - published in the proceedings of Electronic Imaging 2009 and is made available as an electronic reprint (preprint) with permission of SPIE - discussing this very topic.). A few measurements, which compare a unmodified cameras with 'full-spectrum' converted ones, can be found on
JMC Scientific Consulting Ltd | Cosmetic science consulting, see e.g.
Camera sensor sensitivity measurement – UV to IR | JMC Scientific Consulting Ltd for a Canon 5DSR or
Project Mirrorless ? Part 2, the camera conversion | JMC Scientific Consulting Ltd for a Sony sensor. So camera sensors do extend to 800nm, but lose quite a bit of channel separation. Tweaking a filter to work with the built-in color matrix
in camera must result in a technically fairly rough approximation, tuned for the look to be similar for 'typical' subjects. In fact, given the dissimilarities, it's amazing how close they got. With a different filter and a custom color profile (fairly skewed), one could get a lot closer to the actual EIR spectral response, but it would always require external raw image processing.
I'm feeling an itch to have my rarely used K-5 converted, but I already have enough toys which I rarely find the time to play with ...