Visual Preference for ClearType Technology

ClearType filtering is a sub-pixel rendering method that improves the perceived image quality of text. The method renders text at subpixel resolution and then applies a onedimensional filter to reduce color artifacts. We performed behavioral and computational experiments to analyze the effect of varying the linear filter parameters. Specifically, we systematically varied the values of a symmetric, meanpreserving, five-tap filter; such filters are described by a twodimensional parameterization. We generated ClearType versions of the same letter from a large set of these filters and asked subjects to select the version that they preferred. Subjects’ preferred renderings from a compact region in the twodimensional parameter space. Computational analyses show that the preference data are predicted by the SCIELAB metric, a spatio-chromatic metric of human visual sensitivity.