Our colleague kixpixphotography (sorry, I can't seem to remember your actual name - it appears infrequently) has posted the results of an interesting series of tests with several white balance measurement tools (all of the "diffuser" variety. They involved a "tabletop" scene of several colored paint bottles and some other items, including what appear to be some folding test targets (with black, white, and gray portions)
He hasn't told us yet which general technique was used, but I suspect is was the "from the camera position" technique.
[This is the technique often called "reflected light". The term isn't apt, except perhaps when practiced with the Color Parrot/ColorRight diffuser. The intimation of the term is that what is measured is the light reflected from the subject, whereas in fact with most diffusers it is a far broader incidence of light - actually the incident light on the camera position. And of course that's good - the chromaticity of the light reflected from the subject is the last thing we want to measure for white balance correction purposes, unless the subject happens to be a neutral target.]
I have taken a look at the test images to attempt to assess the "accuracy" of the color balance obtained in each case. To do this, I had to make the assumption that the "white" sector of the tricolor test targets was fairly close to reflective chromatic neutrality. If in fact it was substantially off neutral, then of course all bets are off.
I read the sRGB color of that sector in each of the images (on an 8-bit RGB basis, which we realize introduces some quantization error). I then calculated the chromaticity each represented (as CIE u'v' color coordinates) and plotted the results on the du'v' chart. Here, the origin (0,0) represents "theoretically ideal" white balance color correction.
Here we see that chart:
Let me first say that I am well impressed with the performances of all the tested diffusers in this series (even the devices not intended to be white balance diffusers but pressed into service for curiosity's sake, like a plastic urinal cap).
As you know, I am not a big fan of "from the camera position" white balance measurement, owing to the lack of any comforting concept by which it should provide reliable information. Nevertheless, we continue to find that in so many cases, it produces a very useful result. I won't philosophize here in what situations bring that about, and I don't know enough about the total environment of these tests to analyze them in that regard.
Suffice it to say, all the "real diffusers" (plus the urinal cap) produced very respectable results in this particular test series (all much better than the camera's "AWB" function; we have no idea what camera was involved).
Clearly here the best "bang for the buck" comes from the urinal cap (assuming that one could get one at almost zero cost), with the runner up being the "Mennon" white balance cap (which we hear is available for $5.00).
It was also nice to learn of the Clear White diffuser, offered by Lynn Radeka, of which I was previously unaware.
Of course, that diffuser is recommended for use in the "at the subject" mode.
The Color Parrot/ColorRight diffuser is shown with the greatest error (still not very large). In fairness to our host here, I point out that this analysis is wholly dependent on the assumed chromatic neutrality of a non-pedigreed (unknown, in fact) reference object. Thus we cannot draw reliable conclusions from small differences in the "plotted" performance of the different diffusers.
That having been said, I will take a moment to give my outlook on "from the camera position" measurement.
1. Its best chance to give a desirable result is when the overall illumination on the camera position is similar to that on the subject.
2. In general, a broader acceptance pattern is more likely to successfully exploit this than a narrower one.
Of course, I have no information on the acceptance patterns of any of the "diffusers" tested except for the Color Parrot.
Thanks again for the nice round of tests results.
Re: kixpixphtography's WB tool measurememt tests 
Linear Mode
