Abstract
Motion parallax is an effective source of depth information, even for large outdoors environments. Its definition as a differential of angular velocities could be transformed to a differential of angular displacements, since time interval were the same for every stimuli seen in a natural scene. The present experiment investigated this transformed definition of motion parallax. Experimental environment was a spatial layout composed by two exocentric intervals, horizontal and depth, each one defined by two stimuli (cylindrical stakes) and with a expansion center located 15m away observer, built on a large outdoors grassy and plan field (30m × 30m). Observers (N=60), under objective instructions, verbally judge six exocentric distances (1.5, 2.4, 3.84, 6.14, 9.83, 15.73m) in random order, in two series of estimates, respective to a one-meter modulus. Three head movement apparatus were designed to control self-generated motion parallax: restrained, 15cm translational, and 35cm translational head movements. Two viewing conditions were also designed to scrutinize for binocular cue interactions: induced monocular and binocular viewing. Observers produced head movements in an 1Hz rate during judgments. Despite the fact that we found strong spatial anisotropy (undershot depth, accurate horizontal), linear regression fits indicated that large head movements improved accuracy in depth estimates. Overall differential of angular size (or displacement) of exocentric intervals were calculated for head movements conditions and plotted as a function of logarithm of perceived distance. Strong covariation between angular size variation and judged distance provided us with evidence of a possible explanation on increased accuracy in depth estimates for large head movements. This was not the case for maximum angular size which produced no differences at all for the two head movements.