A study on the reason why prints made with only cyan and magenta inks look like regular full-color prints

Two-tone color prints are often commercially used for low-cost color image reproduction. We sometimes can recognize the color which should not be expressed by the ink set of only two colors. Remarkable example is seen in prints made with ink set of cyan(C) and magenta(M); they look like rather regular full-color prints and missing yellow(Y) color is often recognized in the prints. Color appearances were compared in subjective evaluations between three types of two-tone prints: C+M, M+Y and Y+M. Superiority of C+M prints were reconfirmed and the reason was set as a target of our study to be clarified. We suppose that our natural color constancy works when we recognize two-tone color prints. This supposition is derived from an analogy of the case when we see objects under a colored light or through a colored filter. We attempted to explain the reason by considering the von Kreis model, which describes the chromatic adaptation to changes of lighting conditions. The von Kreis model is considered as a basis of many kinds of advanced color appearance models; the essential principle is liner gain-controls for the three components of colors. Color components were measured on selected typical points on printed images of C+M, M+Y, Y+C prints and C+M+Y print as a reference. The von Kreis coefficients were calculated by using measured results on the same typical points which show nearly black in C+M+Y prints and in each two-tone color prints. Colors of the typical points on the C+M+Y prints were estimated from two-tone color prints by a calculation using the von Kreis coefficients. Consequently, best estimation was derived from the calculated results on C+M prints. This result indicates the advantage of C+M prints for us to activate our color constancy easily. It was indicated that the same kind of compensation mechanism as is supposed under a colored light is possible for the color recognition on C+M prints: von Kries model could successfully estimate colors of C+M+Y prints from C+M prints. Thus, we conclude from this result that C+M prints tend to look like full color prints because we can easily compensate colors of C+M prints by using linier gain-control scheme at our natural vision system.