Item

Abstract

3d displays on digital media are often perceived as different from the portrayed object or scene, even when the display creates the “correct” 2d retinal images. In principle, there are at least three depth cues created by digital displays that could contribute to such distortions: 1) inappropriate focus cues, 2) pixelization, and 3) inappropriate motion parallax during head movements. We measured the contribution of these inappropriate screen cues to perceived slant by varying independently the slant specified by the computer graphics algorithm (“computed slant”) and the physical slant of the CRT on which the stimuli were presented (“screen slant”). Planes with different computed and screen slants were presented (tilt = 0 deg) and observers indicated the amount of perceived slant. Precise spatial calibration ensured that retinal-image shapes, texture gradients, and disparity gradients were determined by only the computed slant. Observers were unaware of the screen slant from trial to trial. Across different experiments, we examined the influence of display type (monocular vs. binocular), screen distance (30–200 cm), head stabilization (bite bar, chin rest, and free), amount of slant, and conflict between computed and screen slant. Screen slant had a significant effect on perceived slant in a wide variety of conditions. The effect was larger in monocular than in binocular viewing conditions, at short distances, with head unstabilized, and at large screen slants. We used regression analyses to determine the effective weight given inappropriate screen cues across the various conditions. These results show that inappropriate screen cues can have a significant effect on 3d percepts and that the size of the effect depends strongly on viewing condition.