Abstract
Observer-induced motion parallax and binocular disparity share geometric similarities and have interact to determine perceived 3D object shape. However, it has been demonstrated that depth is often misperceived, even when both cues are available. These results are likely due to conflicts with unmodeled cues (e.g. focal blur) that are endemic to computerized displays. comparison, perceived 3D shape for physical targets in near space is veridical when binocular disparity is present and shows little improvement with the addition of motion parallax. Here we evaluate the impact of display-based cue conflicts on depth cue integration by directly comparing perceived shape for physical and virtual objects. Truncated square pyramids were rendered using Blender and 3D printed. In both physical and virtual viewing conditions we assessed perceived depth using a ratio task with 1) motion parallax, 2) binocular disparity, and 3) their combination. Virtual stimuli were viewed using a head-mounted display. Physical stimuli were presented using apparatus with precise control over position and lighting. To create motion parallax observers made lateral head movements using a chin rest mounted a horizontal motion platform. Using the method of constant stimuli, observers indicated if the width of the front face was greater or less than the distance between this surface and the pyramid base (i.e. the depth of the pyramid). We found that depth estimation accuracy was similar for virtual and physical pyramids. However, precision was higher for physical targets. Further, in both physical and virtual conditions estimates were more precise when depth was defined by binocular disparity than by motion parallax. Our analysis (using a probabilistic model) shows that a linear weighted combination model does not adequately describe performance in either physical or virtual test conditions. Our work highlights the importance of maximizing ecological validity using carefully controlled natural stimuli and tasks.