A Quality-Centered Analysis of Eye Tracking Data in Foveated Rendering

Area of Science:

• Computer Graphics
• Human-Computer Interaction
• Virtual Reality

Background:

• Foveated rendering optimizes image synthesis based on user gaze to enhance performance.
• High rendering performance is critical for virtual reality (VR) to ensure immersion and reduce nausea.
• Eye-tracking data analysis is essential for evaluating advanced rendering techniques like foveated rendering.

Purpose of the Study:

• To analyze eye-tracking data from a user study on foveated rendering for head-mounted displays (HMDs).
• To evaluate the precision of the eye-tracking system and user accuracy in fixation tasks.
• To investigate the relationship between fixation modes, accuracy, and perceived visual quality.

Main Methods:

• Collected eye-tracking data during user studies involving fixation tasks with static and moving targets in VR.
• Estimated eye-tracker precision and analyzed participant accuracy in focusing targets.
• Assessed eccentricity-dependent quality ratings and compared them with user-reported quality.

Main Results:

• Quantified eye-tracker precision and user fixation accuracy across different visual conditions.
• Identified specific fixation modes that correlate with higher or lower accuracy.
• Established a significant link between fixation accuracy, employed fixation modes, and subjective quality ratings.

Conclusions:

• The study demonstrates a clear correlation between how accurately users fixate on targets and their perceived visual quality in VR.
• Fixation modes significantly influence both accuracy and subjective experience, offering insights for optimizing foveated rendering.
• Results provide valuable data for refining foveated rendering algorithms to improve VR user experience.