Abstract
To recover absolute depth from binocular disparity, horizontal disparities must be scaled by an estimate of fixation distance. Conventional 3d digital displays present stimuli at one focal distance and this could result in distortions of perceived depth at simulated distances other than the actual distance to the display. To investigate this we independently varied distance specified by the eyes' vergence and by focal distance using a novel display that allowed us to interleave stimulus presentations at different focal planes within a block of trials. Observers viewed sparse random-dot stereograms depicting two planes forming a hinge (“open book”) and adjusted the dihedral angle until it appeared to be 90 deg. The vertical size of the stimulus was always small (2 deg) to minimize the contribution of distance information from vertical disparity. Depth constancy (with changes in vergence-specified distance) was assessed by calculating, for each stimulus configuration, the distance at which the patterns of disparities set by the observers would have corresponded to a dihedral angle of 90 deg. When focal distance was fixed, depth constancy was poor (around 40%). However, when the focal distance varied appropriately with vergence-specified distance, depth constancy increased (in some subjects to as much as 65%). Depth constancy was highest in a baseline condition in which the stimulus was a real hinge with all depth cues available, but even then constancy rarely exceeded 80%. These results show that focus cues to distance can have a significant effect on the percept of depth in 3d displays and point to the need in vision research to develop means of circumventing or minimizing cue conflicts created by conventional displays.