Dynamic perspective cues enhance depth perception from motion parallax.

Area of Science:

• Visual Perception
• Computational Neuroscience

Background:

• Motion parallax, a key depth cue from self-movement, is typically studied with simplified stimuli.
• Previous research often used random dot textures and orthographic rendering, limiting ecological validity.

Purpose of the Study:

• To investigate depth perception from motion parallax under more naturalistic conditions.
• To compare dynamic perspective rendering against orthographic rendering for depth perception.
• To evaluate the contribution of specific dynamic perspective cues.

Main Methods:

• Used 1/f spectrum textures (random dots, Gabor micropatterns) and dynamic perspective rendering.
• Compared depth ordering performance between orthographic and perspective rendering.
• Synchronized head movements with texture motion (shearing).
• Analyzed the impact of removing individual dynamic perspective cues.

Main Results:

• Dynamic perspective rendering enhanced depth perception compared to orthographic rendering for both texture types.
• Performance with naturalistic 1/f textures was slightly lower than random dots but improved with depth-scaled micropatterns.
• Removing any of the three dynamic perspective cues (vertical displacements, speed gradients, near/far speed differences) impaired depth ordering.

Conclusions:

• Dynamic perspective rendering, with its inherent depth cues, significantly improves motion parallax-based depth perception.
• Naturalistic textures alone do not enhance performance; the rendering method is critical.
• All tested dynamic perspective cues contribute to accurate depth ordering.