Factors influencing the choice of projection textures for displaying layered surfaces

Abstract

We report on an experiment designed to examine the effect of type and size of projection textures on the ability of human subjects to correctly estimate the surface normal on each of two terrain-like layered surfaces. Normal estimation was determined by the ability to correctly orient a probe so that it is aligned with the normal local to the surface to which it is attached. All presentations were optimized for layered surface viewing, being displayed on a high resolution stereoscope and with a rocking motion to provide motion cues. Three texture types were examined, each at both coarse and fine scales. One texture consisted of random dots, containing no consistent structural pattern. The other two texture types provided structure in two orthogonal directions, with one being a regular square grid, and the other parallel hatch marks arranged in ranks to form perceptual lines orthogonal to the hatches. The experiment demonstrates that structured patterns, indicating shape in two orthogonal directions, are superior to random dots. Using a grid on the bottom layer significantly improves perception of the bottom surface. However, between grids and hatches, there is no other statistically significant evidence that using differing texture type or scale across the two surfaces enhances the ability to estimate shape.