Tint in camera raw versus cc index in color meters

[coa_lund]

I am right now using a Canon 400D to check the calibration of a Minolta Colormeter II (I mostly use film).
The color temperature is easy, because everyone agrees what is a Kelvin degree. I just read off the value in
camera raw (for a shot against a gray card or light source after having adjusted the WB the that particular source) and compare with the Minolta readings. In this way I have found that my Canon camera and the color meter agree within about 1 dekamired which I think is just fine. The calibration along the green/magenta axis is more problematic, however, since the tint values in camera raw and the cc index (in dekamireds) from Minolta do not correspond directly to each other. Does anybody know how the tint index in camera raw is defined and how one converts it to cc index?

I should say that from light sources close to black-body emitters such as tungsten bulbs, the tint or cc values are always very low as they should, and I have done some simple tests with a cc 30 magenta filter with the color meter which is passes.

I would be grateful for any help.

Carl

[digitaldog]

The color temperature is easy, because everyone agrees what is a Kelvin degree.
Carl

Actually not so. Kelvin defines a range of colors (a pretty big range). See: http://www.ppmag.com/reviews/200512_rodneycm.pdf

[coa_lund]

I of course also know that the photographic color temperature and the T parameter in Planck's radiation law (1900) are not the same thing, and that the photographic color temperature is related the the relative intensities of the blue/red channels. To make sense, a black-body radiator T should have photographic color temperature T K, but any light source can be assign a photographic color temperature. (The space of spectral shapes is basically infinite-dimensional whereas photographic color spaces relates to the way the human vision analyzes light and is basically two-dimensional, i.e. a two-parameter thing.) I assume that different measuring devices such as color meters, photographic sensors, etc, use the same algoritm for computing the photographic color temperature for a given light source, and this was what I pointed out in my post.

However, the purpose of my post was not so much to give an in-depth discussion of color spaces but to provide tips on how one can check the calibration of color meters. Color meters were surprisingly expensive in the past, but now you can get one for low prices second hand and they are still useful if you shoot film. However, a color meter which is not calibrated is useless. A few workshops in the U.S. can do recalibration, but this is expensive and quite awkward if you live in Europe and have to send the device to the U.S. This is why I came up with the idea to use a digital reflex camera and compare measurements to check the calibration. As I mention in my post, my Minolta color meter and my Canon 400D give very consistent values for the color temperatures (they agree typically to within about 1 dekamired for a broad range of different light sources.) Wth a 400D I can only fetch the color info via camera raw. However, since I do not know how the tint scale in camera raw is defined I cannot use the info along the magenta/green axis for calibration purpose.

If I can help with this I plan to post how such a calibration check can be done in practice.

Carl

[digitaldog]

The K values are based on a theoretical object (the black body radiator), any values you get are correlated and as such, if you get two items that match or mismatch, all it tells you is they don’t correlate. IOW, your two items that are within 1 value don’t mean much anymore than two items that are farther off. You are dealing with a range of color not tightly defined.

You want exacting numeric values, you need spectral data. And then one could argue that one Spectrophotometer versus another and how broad band they measurements are could play a role. But the point is, you’ll have actual numbers that are meaningful.