A Methodology for Predicting User Curvature Perception of 3D Objects

One's perception of an object's curvature affects its perceived appeal, realism, and even distance. However, studies indicate curvature perception often differs from the object's mathematically defined curvature, and no alternative for predicting curvature perception exists. We present two pairwise-comparison studies where participants selected objects perceived to have more curvature. The results indicate some objects are perceived to have significantly more curvature than others, yielding distinct perceptual clusters. We then demonstrate that traditional curvature measures poorly predict curvature perception, and present a novel methodology with results proving it a capable indicator of how a 3D object's curvature will be perceived.