Quasi-3D: reducing convergence effort improves visual comfort of head-mounted stereoscopic displays

Abstract

The diffusion of virtual reality urges to solve the problem of vergence-accommodation conflict arising when viewing stereoscopic displays, which causes visual stress. We addressed this issue with an approach based on reducing ocular convergence effort. In virtual environments, vergence can be controlled by manipulating the binocular separation of the virtual cameras. Using this technique, we implemented two quasi-3D conditions characterized by binocular image separations intermediate between 3D (stereoscopic) and 2D (monoscopic). In a first experiment, focused on perceptual aspects, ten participants performed a visuo-manual pursuit task while wearing a head-mounted display (HMD) in head-constrained (non-immersive) condition for an overall exposure time of ~ 7 min. Passing from 3D to quasi-3D and 2D conditions, progressively resulted in a decrease of vergence eye movements—both mean convergence angle (static vergence) and vergence excursion (dynamic vergence)—and an increase of hand pursuit spatial error, with the target perceived further from the observer and larger. Decreased static and dynamic vergence predicted decreases in asthenopia trial-wise. In a second experiment, focused on tolerance aspects, fourteen participants performed a detection task in near-vision while wearing an HMD in head-free (immersive) condition for an overall exposure time of ~ 20 min. Passing from 3D to quasi-3D and 2D conditions, there was a general decrease of both subjective and objective visual stress indicators (ocular convergence discomfort ratings, cyber-sickness symptoms and skin conductance level). Decreased static and dynamic vergence predicted the decrease in these indicators. Remarkably, skin conductance level predicted all subjective symptoms, both trial-wise and session-wise, suggesting that it could become an objective replacement of visual stress self-reports. We conclude that relieving convergence effort by reducing binocular image separation in virtual environments can be a simple and effective way to decrease visual stress caused by stereoscopic HMDs. The negative side-effect—worsening of spatial vision—arguably would become unnoticed or compensated over time. This initial proof-of-concept study should be extended by future large-scale studies testing additional environments, tasks, displays, users, and exposure times.