D
Doug Kerr
Guest
The topic of this essay is something that probably has almost no impact on our practical concerns, but I thought it would make good filler for this initial "edition" of the Open Photography Bugle.
It is common, and entirely appropriate, to describe the reflective color of an exposure metering/white balance calibration target in terms of its coordinates in the CIE L*a*b ("CIELAB") color model. (That model was in fact introduced to describe reflective color; it has been "hijacked" to describe the color of light or the color implied for a pixel in a digital image.)
Manufacturers of WB targets, in particular, point with pride to small absolute values of a* and b* for their products as indicative of a very small departure from ideal "chrominance neutrality", and again this is entirely appropriate.
We do, however, get into subtle difficulty when we try to quantify the degree of departure from chromaticity neutrality in terms of the values of the a* and b* coordinates. For example, we might hear, "it would be desirable for a* and b* to not be outside the range ±1.5".
But that doesn't quite describe a range of chromaticity. The reason is that the a*b* plane is essentially a plane of chrominance, not chromaticity.
I won't bore you further here with a discussion of the distinction. Those who are not certain about it, and are interested, might like to read my discussion of it in my tutorial article, "Chromaticity and Chrominance in Color Definition", available here:
http://doug.kerr.home.att.net/pumpkin/index.htm#ChromatcityChrominance
In any event, for a given chromaticity, the chrominance varies with lightness. That is, two colors with the same a* and b* coordinate values, but having different values of L* (lightness), will not have the same chrominance (values of a* and b*).
For example, let's consider a hypothetical WB target whose reflective color measures 75, +2.0, +2.0 under the L*a*b model. That color is not quite chromaticity-neutral. (Its reflectance would be 48.3%, incidentally.)
Now let's look at another target of less lightness - an L* of 42.4 (reflectance of 12.8%, often considered desirable for a target for incident light metering), but the same chromaticity. Its L*a*b specification would be 42.4, +1.3, +1.3.
So if for some reason we had concluded that we needed a target whose a* and b* coordinates were not beyond, for example, ±1.5, then the second target would satisfy that, while the first one would not, despite the fact that they both had the same chromaticity - they were both just as "close to chromaticity neutral".
[This is not at all to suggest that both would be equally attractive as WB calibration targets, owing to the reflectance matter discussed elsewhere in this forum.]
Again, I don't at all claim that this is of any real importance in judging the specifications for WB calibration targets. And I don't mean in any way to demean the presentation by target manufacturers of the a* and b* "bogeys" for their products. It is probably the most practical way to describe what they have - it would be difficult for the average user to interpret a chromaticity-neutrality specification in terms of the CIE x and y coordinates, for example.
But I think it is important to realize that the a* and b* coordinates do not precisely quantitatively describe a chromaticity, nor a departure from "chromaticity neutrality".
It is common, and entirely appropriate, to describe the reflective color of an exposure metering/white balance calibration target in terms of its coordinates in the CIE L*a*b ("CIELAB") color model. (That model was in fact introduced to describe reflective color; it has been "hijacked" to describe the color of light or the color implied for a pixel in a digital image.)
Manufacturers of WB targets, in particular, point with pride to small absolute values of a* and b* for their products as indicative of a very small departure from ideal "chrominance neutrality", and again this is entirely appropriate.
We do, however, get into subtle difficulty when we try to quantify the degree of departure from chromaticity neutrality in terms of the values of the a* and b* coordinates. For example, we might hear, "it would be desirable for a* and b* to not be outside the range ±1.5".
But that doesn't quite describe a range of chromaticity. The reason is that the a*b* plane is essentially a plane of chrominance, not chromaticity.
I won't bore you further here with a discussion of the distinction. Those who are not certain about it, and are interested, might like to read my discussion of it in my tutorial article, "Chromaticity and Chrominance in Color Definition", available here:
http://doug.kerr.home.att.net/pumpkin/index.htm#ChromatcityChrominance
In any event, for a given chromaticity, the chrominance varies with lightness. That is, two colors with the same a* and b* coordinate values, but having different values of L* (lightness), will not have the same chrominance (values of a* and b*).
For example, let's consider a hypothetical WB target whose reflective color measures 75, +2.0, +2.0 under the L*a*b model. That color is not quite chromaticity-neutral. (Its reflectance would be 48.3%, incidentally.)
Now let's look at another target of less lightness - an L* of 42.4 (reflectance of 12.8%, often considered desirable for a target for incident light metering), but the same chromaticity. Its L*a*b specification would be 42.4, +1.3, +1.3.
So if for some reason we had concluded that we needed a target whose a* and b* coordinates were not beyond, for example, ±1.5, then the second target would satisfy that, while the first one would not, despite the fact that they both had the same chromaticity - they were both just as "close to chromaticity neutral".
[This is not at all to suggest that both would be equally attractive as WB calibration targets, owing to the reflectance matter discussed elsewhere in this forum.]
Again, I don't at all claim that this is of any real importance in judging the specifications for WB calibration targets. And I don't mean in any way to demean the presentation by target manufacturers of the a* and b* "bogeys" for their products. It is probably the most practical way to describe what they have - it would be difficult for the average user to interpret a chromaticity-neutrality specification in terms of the CIE x and y coordinates, for example.
But I think it is important to realize that the a* and b* coordinates do not precisely quantitatively describe a chromaticity, nor a departure from "chromaticity neutrality".