I have recently completed a test program intended to discern various aspects of the behavior and performance of the Color Parrot white balance diffuser.
An extensive report on my findings is available directly here:
http://Pumpkin.Annex.home.att.net/ar...olorParrot.pdf
Here is a very brief summary of my principal findings. Note that in the report I discuss at length the specific test procedures I followed, the rationale for them, and the significance of the results. I urge that the members look at least briefly at the report itself before asking questions of the "well, how did you determine
that?" flavor.
Chromatic neutrality
Chromatic neutrality is perhaps the most critical property of a device used for the measurement of chromaticity, such as the Color Parrot.
We have here no standardized laboratory illuminant source from which this property could be directly evaluated. Rather, we tested the chromatic response of the Color Parrot against the response of a WhiBal gray card. We have from the manufacturer of the WhiBal gray card an assertion of the tolerances on its chromatic neutrality, determined in a fashion to which I can give the benefit of the doubt.
We found the Color Parrot to have a chromatic response that differed from that of the WhiBal gray card by an amount that can be expressed in a familiar way (in terms of the CIE L*a*b* color model) as a*=-3.21, b*=-0.392 (for an assumed L* of 70, which we must know to be able to interpret the a* and b* values as a chromaticity difference).
For comparison, the manufacturer of the WhiBal gray card states that the tolerance limits for that product are a*=±0.5, b*=±0.5 (for L=73).
For comparison, chromatic neutrality tests were also made of an ExpoDisc diffuser, It was found to have a departure from the neutrality of the WhiBal gray card that was about 50% greater (worse) than for the Color Parrot.
Directivity pattern
The directivity pattern of a diffuser is of interest because (simplistically) it tells us over what range of angle does the diffuser "catch" incident light to be presented to the camera for measurement and (in a more sophisticated context) tell us the relative "weighting" to be accorded to streams of light arriving from different directions that have different chromaticity.
We found the Color Parrot to have a somewhat narrower pattern than the classical "cosine" pattern often used as a basis of comparison and sometimes considered to be desirable for measurement in a "mixed light" situation. The exact significance of this on the performance of the tool is, however, unclear.
Photographic color correction tests
Actual photographic tests were made with a Canon EOS 20D in which the taken images were given in-camera white balance correction (under the camera's "custom white balance" feature), with reference frames taken with five different techniques:
-By measurement of a WhiBal gray card.
-By measurement with the Color Parrot diffuser, with two variations:
-- Measurement at the subject, with the diffuser facing the camera position to be used for the shot
-- Measurement at the camera position, with the diffuser facing the subject
-By measurement with the ExpoDisc diffuser, with the same two variations.
Tests, featuring Carla as the model, were made in three different "illumination settings".
We found that for measurements taken "at the subject" by the Color Parrot, consistent and fairly accurate results were obtained (meaning as compared to the theoretically ideal color correction result) over all "settings".
When the Color Parrot was used "at the camera", the results were less consistent, being less accurate in two of the settings than the "at the subject" measurement but more accurate in the third.
Of course, such a relatively-modest test program, especially conducted as it was bereft of actual laboratory equipment, can only give us certain quick looks at the properties of the device under test. Still, I think the information gleaned should be of some interest and value to our community.
Carla and I are way pooped!
Linear Mode
