Abstract
We investgated the effects of terrain-texture resolution (spatial frequency bandwidth) on the perceived speed of simulated self-motion. To do so, we used a set of random processes to create a “base” texture with features that varied in size, orientation, and intensity. This texture was then subjected to two low-pass filters with different cutoff frequencies. When the resulting high- and low-resolution textures were mapped to flat terrains using standard bilinear-magnification and mipmap-trilinear-minification filters, the spatial patterns on the near terrains differed while those on the distant terrains were virtually identical. To assess speed perception, we used a 2IFC method of constant stimuli in which one temporal interval contained a 1-s standard motion sequence and the other contained a 1-s test motion sequence. Three standard speeds were presented in separate blocks. For each standard speed, the relative test speeds ranged from −0.10 to 0.10, in steps of 0.05 log units. Given the 60-Hz update rate of the image generator, over half of the spatial frequencies in the high-resolution, near terrain were temporally aliased at the highest standard speed.
We found that when the textures in the standard and test sequences were the same, the speed discrimination threshold was lower for the high-resolution texture than for the low-resolution texture. When the standard and test textures differed, the two psychometric functions were displaced by approximately ±0.07 log units, respectively: Simulated motion over the high-resolution texture appeared faster than simulated motion of the same speed over the low-resolution texture. The magnitude of this effect varied little with standard speed. In contrast to previous research (Lindholm et al., 1993), only 2 of the 6 observers showed an increase in perceived speed clearly attributable to temporal aliasing.