Estimating depth magnitude for flat, forced and reverse perspectives

Abstract

OBJECTIVE: Assess perceived depth magnitude in stimuli with linear perspective cues that are congruent or incongruent with physical geometry. Further, determine whether perceived depth magnitude for bistable stimuli depends on percept (veridical versus illusory). METHODS: Observers viewed three scaled-size versions (small, medium, large) of two 3D stimuli (forced-perspective, reverse-perspective) and one 2D (flat) stimulus. Stimuli contained painted cues: either congruent (proper) or incongruent (reverse and flat) with physical geometry. Reverse-perspectives were bistable (veridical or illusory); forced-perspectives and flat stimuli only afforded one percept. Observers’ task: estimate perceived depth magnitude by physically adjusting distance between two rods. Results were compared to baseline depth estimations for two 3D stimuli (convex and concave) without painted cues. RESULTS: Physical depth was overestimated for all sizes and types of stimuli, including baseline stimuli. However, overestimation decreased as stimulus size increased. For proper-perspectives and reverse-perspectives, estimated depths were roughly: 3.8 times the actual depth for small stimulus; 2.6 times for medium; 1.8 times for large. Surprisingly, estimated depths for the three flat stimuli, despite containing no actual depth, were roughly equal to those for the 3D stimuli of corresponding size. Depths had a trend – but not statistically significant – to appear larger for forced than for reverse perspectives stimuli. Further, estimates did not differ significantly for the reverse-perspective stimulus when perceived as veridical versus illusory. Depths for the non-painted baseline stimuli were about 1.5 times their actual depth. CONCLUSIONS: The large overestimation for smaller versus larger stimuli may be due to the smaller binocular disparity for the smaller stimuli, causing observers to rely more on linear perspective cues than disparity. The large depth estimates for the flat stimuli, combined with the small perceived depth differences for the three 3-D conditions point to a significantly larger role of painted perspective cues as compared to binocular disparity.

Meeting abstract presented at VSS 2013