Abstract
Braunstein, Andersen, and Sauer (Psychonomic Society Meeting, 1999) reported that judgments of object depth in structure-from-motion are influenced by both the difference between the maximum and minimum projected velocities and the ratio of these velocities, with difference having the greater effect. In that study all objects were surfaces of revolution and were equal in simulated depth. The present experiments seek to generalize these results to other objects and to investigate whether the true (simulated) depth of an object is also a predictor of its perceived depth. The stimuli were computer-generated random-dot 3D scenes, each containing a ground plane, a ceiling plane, and a 7.0-cm wide and 9.3-cm high elliptical cylinder on a pole between the two planes. The cylinder rotated about its vertical axis and the entire scene translated horizontally. Velocity difference (0.7, 1.5, or 2.2 deg per sec), velocity ratio (1.25, 1.5, or 2.0) and cylinder depth (7.0, 9.3, or 11.6 cm) were varied independently. In the first experiment, the edges of the object were occluded by a virtual rectangular window 9.3 cm high and 7.0 cm wide to avoid visible contour changes. In the second experiment the edges were not occluded. Subjects viewed the displays through a collimating lens and adjusted a line on a separate monitor to match the perceived depth of the object. In both experiments, objects with greater velocity differences and with greater velocity ratios were judged to have greater depth, but the simulated depth of the object did not significantly affect its judged depth. The results indicate that the difference and ratio of the maximum and minimum velocities on an object determine judged depth in structure-from-motion displays, even when the displays vary in simulated depth. Whether or not contour changes are visible, the difference and ratio have a greater influence on the judged depth than does the simulated depth.
Supported by NIH Grant 1R01EY12437.