Dominant-Eye-Aware Asymmetric Foveated Rendering for Virtual Reality

Area of Science:

• Computer graphics
• Human-computer interaction
• Visual perception

Background:

• High-resolution virtual reality (VR) demands significant computational power.
• Foveated rendering reduces computational load by decreasing pixel sampling in peripheral vision.
• Existing methods inadequately utilize binocular vision, especially dominant eye theory.

Purpose of the Study:

• To introduce Dominant-Eye-aware Asymmetric Foveated Rendering (DEA-FoR) for VR.
• To enable variable and asymmetric eccentricity ranges between the eyes.
• To optimize VR rendering speed and perceptual quality by leveraging individual visual differences.

Main Methods:

• Developed DEA-FoR, allowing dynamic eccentricity settings tailored to individual users.
• Implemented asymmetric foveation configurations between the dominant and non-dominant eyes.
• Evaluated rendering speed and perceptual quality against prior methods.

Main Results:

• DEA-FoR significantly improved rendering speed compared to previous Dominant-Eye-Aware foveated rendering (DEAMP).
• Perceptual quality was maintained at improved speed levels.
• Individual visual differences and scene complexity were identified as key factors influencing optimal eccentricity settings.

Conclusions:

• DEA-FoR offers a more effective approach to foveated rendering by adapting to individual binocular vision.
• The method enhances VR performance and user experience.
• Further research into perceptual differences in binocular vision can optimize future VR technologies.