Model and representation: The effect of visual feedback on human performance in a color picker interface

User interfaces for color selection consist of a visible screen representation, an input method, and the underlying conceptual organization of the color model. We report a two-way factorial, between-subjects variable experiment that tested the effect of high and low visual feedback interfaces on speed and accuracy of color matching for RGB and HSV color models. The only significant effect was improved accuracy due to increased visual feedback. Using color groups as a within-subjects variable, we found differences in performance of both speed and accuracy. We recommend that experimental tests adopt a color test set that does not show bias toward a particular model, but is based instead on a range of colors that would be mostly likely matched in practice by people using color selection software. We recommend the Macbeth Color Checker naturals, primaries, and grays. As a follow-up study, a qualitative case analysis of the way users navigated through the color space indicates that feedback helps users with limited knowledge of the model, allowing them to refine their match to a higher degree of accuracy. Users with very little or a lot of knowledge of the color model do not appear to be aided by increased feedback. In conclusion, we suggest that, visual feedback and design of the interface may be a more important factor in improving the usability of a color selection interface than the particular color model used.