Measurement of a subjective depth percept is an interesting psychophysical problem for which a variety of techniques have been developed. Here we compare perceived depth from motion parallax (MP) using manual depth estimation (MDE) to estimates previously obtained using an inter-cue-comparison, forced-choice technique (Nawrot et al., 2011, VSS). Observers indicated the perceived magnitude of line length or depth magnitude with the distance between the thumb and index finger, which was measured with a computer-recorded linear potentiometer. Stimuli were viewed binocularly, at 36, 54 and 72 cm, through a Stereographics Z-screen that provided both stereo separation and monocular viewing of the translating motion parallax stimuli. In Condition 1, observers gave MDEs for lines lengths between 0.5 and 6 cm. In Condition 2, observers gave MDEs for perceived depth - the peak-to-trough distance - of a random-dot stimulus depicting a vertically oriented corrugated surface. Stimulus depth was specified with either binocular disparity (1 - 24 min) or motion parallax (M/PRatio: 0.04 - 0.24). Overall, MDEs were variable, with extraordinarily large standard deviations. Comparing mean MDEs to physical line length generated regression lines with slopes between 0.4 and 1.0. The mean slope was 0.75 for all viewing distances. Comparing MDEs to expected stereo depth gave an "over-estimate" at 36 cm viewing distance (slope =1.10), but an "under-estimate" of depth as viewing distance increased (slope = 0.83 for 54 cm, and 0.65 for 72 cm). Johnson (1991) reports a similar trend. MDEs for MP showed a similar change with viewing distance. Compared to the expected values from Nawrot et al. (2011), MDEs over-estimated depth at 36 cm (slope = 1.38) and 54 cm (1.06) and under-estimated depth at 72 cm (slope = 0.93). We conclude that MDE has high variability but linear trends using mean MDE values provide similar estimates to cue comparison techniques.
Meeting abstract presented at VSS 2013