Abstract
Mixed reality (MR) systems often suffer from the accommodation-vergence conflict (VAC), leading to asthenopia and reduced interaction efficiency in cluttered workspaces. This study evaluates a gaze-contingent depth-of-field (DoF) rendering system implemented in an MR headset. By dynamically adjusting the focal plane based on real-time eye-tracking data, the system renders synthetic defocus blur on out-of-focus digital elements. We evaluated this approach with 24 participants performing a complex 3D object-selection task in a highly cluttered mixed-reality environment. Results showed a significant reduction in subjective visual fatigue (measured via the Simulator Sickness Questionnaire and visual comfort scales) and a 14.8% improvement in selection accuracy compared to traditional pinhole rendering. Furthermore, gaze-contingent DoF rendering lowered cognitive load, as indicated by task completion times and pupillometry metrics. These findings highlight the potential of gaze-contingent rendering to mitigate VAC and enhance user performance in professional, high-density MR workspaces.
