Conveying shape with texture: An experimental investigation of the impact of texture type on shape categorization judgments

As visualization researchers, we are interested in gaining a better understanding of how to effectively use texture to facilitate shape perception. If we could design the ideal texture pattern to apply to an arbitrary smoothly curving surface in order to enable its 3D shape to be most accurately and effectively perceived, what would the characteristics of that texture pattern be? In this paper we describe the results of a comprehensive controlled observer experiment intended to yield insight into that question. In previous studies, using a surface attitude probe adjustment task, we had found that judgments of shape under conditions of anisotropic texturing were most accurate when the directions(s) of anisotropy were aligned with one or both of the principal directions of curvature over the surface, as opposed to being aligned with an arbitrary constant direction over the surface, or with a direction that varied over the surface in a manner unrelated to the surface geometry. However, many open questions remained. Here we report the results of a new study comparing the relative accuracy of observers' judgments of shape type (elliptical, cylindrical, hyperbolic or flat) and shape orientation (convex, concave, both, or neither) for local views of boundary masked quadric surface patches under six different principal direction texture pattern conditions plus two control texture conditions (an isotropic pattern and a non-principal direction oriented anisotropic pattern), under both perspective and orthographic projection conditions and from both head-on and oblique viewpoints. Our results confirm the hypothesis that accurate shape perception is facilitated to a statistically significantly greater extent by some principal direction texture patterns than by others. Specifically, we found that, for both views, under conditions of perspective projection, participants more often correctly identified the shape category and the shape orientation when the surface was textured with the pattern that contained oriented energy along both the first and second principal directions only than in the case of any other texture condition. Patterns containing markings following only one of the principal directions, or markings oriented obliquely to the principal directions, or containing information along other directions in addition to the principal directions yielded poorer performance overall. In examining the effects of projection type and view direction, we found that observers retained the ability to make correct shape category judgments under conditions of orthographic viewing under many texture conditions, as long as the surface patches were viewed from an oblique vantage point rather than head-on. However, our observers were unable to reliably disambiguate convex from concave surface orientations in the absence of perspective projection, regardless of texture type.