Abstract
During computer image generation, a synthetic environment is projected onto a view plane internal to the computer. This continuous, space-time image is sampled in space and time. To minimize spatial aliasing, IGs implement procedures that serve as low-pass presampling filters, in cycles/pixel. The cutoff frequency in cycles/m typically varies with image location. In a continuous space-time image representing constant-velocity motion over a flat, textured surface, the temporal frequency of a given spatiotemporal frequency equals the dot product of the 2D spatial frequency and the viewpoint velocity, in cycles/m and m/s, respectively. A given spatial frequency, in cycles/m, thus has the same temporal frequency throughout the image. In the display of a corresponding computer-generated motion sequence, some of the spatial frequencies may be temporally aliased (i.e., have a spurious, lower temporal frequency). However, in parts of the image, aliased components are likely to be eliminated by the resolution and concomitant spatial-presampling filter of the IG. Here we assessed effects of IG resolution on perceived motion quality during simulated flight over flat, textured terrain. The resolution was either 4 or 1 arcmin/pixel. In the first two experiments, observers indicated the spatial extent of poor quality motion for flights of different altitudes and speeds. In the third experiment, observers selected the maximum speed for good quality motion and the minimum speed for poor quality motion within small windowed views of the terrain. As predicted, the higher of the two IG resolutions resulted in poor quality motion (a) over more of the image for a given speed and altitude and (b) at a lower speed for a given ground distance. However, whereas poor quality motion and temporal aliasing coincided for the lower resolution, the pattern of results suggests that the passband of the human visual system was the limiting factor for the higher resolution.