Abstract
The three fundamental attributes of color, i.e., brightness, hue, and saturation, have not found their singly corresponding physical correlates. Changes along one physical dimension also interfere with other color attributes. One example is the Helmholtz–Kohlrausch (H-K) effect, where an increase in saturation with constant hue and luminance also increases brightness as one manifestation of the luminance versus brightness discrepancy. Our recent work has collected Ralph Evans’s zero-grayness luminance to equalize brightness and thus incorporate the H-K effect. Furthermore, MacAdam’s optimal colors (as an interpolation handle) were found to be highly correlated with those equally bright colors. In this work, by navigating along the optimal color surface, the interaction between saturation and hue was further investigated. The stimulus set included five Munsell principal hues R (red) / Y (yellow) / G (green) / B (blue) / P (purple), which had dominant wavelength of 629, 576, 512, 485, 560 nm (the complementary), respectively, at different excitation purities. Observers (N = 14) were asked to first perform partition scaling for saturation given the neutral and one max-saturation hue as anchors. 50% midpoints and further 25% & 75% levels of saturation were collected. In the second part, the saturation scales across max-saturation hues were normalized relative to the max-saturation red via magnitude estimation by the same observers. The results suggest that 1) saturation is approximately linearly related to excitation purity with increasing slopes for Y < R/P < B < G; 2) when saturation is equated, the hue contribution in the H-K effect, quantified by B/L ratio, is ordered as Y/G < B < P/R; 3) the CIECAM02 and CIELAB saturation metrics either did not correlate with the saturation scales linearly or was not isotropic in hue. This work provides more solid scaffolds to develop independent color scales for each attribute.