You’re close. There is such a thing as using ColorThink to round trip through a profile, but this is sort of half of that.
Instead this procedure is guessing what colors you’ll end up with if you use your profile to predict what Lab values you’ll get when you run your device values (ie RGB) through the profile … and then compare those Lab values with the actual Lab values from the measurement.
This of course is only relating to the proofing direction of a profile, how accurate your soft-proofing will be.
There is a section in the manual that talks about the function of the ColorSmart Guide:
www.colorwiki.com/wiki/ColorThink_Pro_- … arts_Guide
And here is the section just on this topic:
It is not quite so simple to get quantifiable numbers relating to the quality of the printing direction of a profile. The closest we could come up with is this procedure for visually analyzing the quality of the printing effect of a profile:
But back to your overall question: Yes, the small differences you see are there because of rounding errors in the creation of a profile, but also the smoothing function of a profile. A good profiling engine will naturally introduce some intelligent smoothing so that image gradients will have smooth transitions of color. A bad profile, and also a good profile that has to do a lot of work to yank the color around to where it is supposed to be, will likely have larger numbers in this test.
Great question. Thanks for asking it.