I was thinking about 2 hypothetical pairs of identical DeltaE errors, and I pickup an example about which I'd like to hear from you expert colortists...
Think about two near grey levels, say 40% and 50%. Now choose 2 pairs of such greys, with all 4 greys with a DeltaE of say...2.5. But with one crucial difference: the first pair of greys have errors with the same hue, while the second with an opposite hue.
Now if we happen to evaluate the quality of this 2 "calibrations" using DeltaE, we should affirm that they are of the same "quality", because trivially both exhibit an average DeltaE of 2.5
But unfortunately if we watch a greyscale steps, first time with the grey pair of the same hue, and then with the other pair containing grey of opposite hues, we would end up saying that the first calibration is better, since we can't perceive or recognise different hues in the grey steps.
The issue is that this obvious difference would not be factoried in by any average DeltaE measures commonly used here. So a sort of measure of the continuity in hues should also be taken into account...
Am I wrong? What do you think?