Gaze-aware neural holography for enhanced foveal image quality in near-eye displays

Holographic displays offer promising capabilities for realistic three-dimensional visual experiences, yet their practical implementation is hindered by limited spatial bandwidth product (SBP) and the demand for high-quality image reconstruction across the entire field of view. Here, we propose a gaze-aware neural holography framework that leverages human visual system (HVS) characteristics to optimize holographic image quality. We introduce two complementary approaches: gaze-aware holography (GAH) based on iterative optimization and gaze-aware neural holography (GANH) based on neural networks. In GANH, we develop a gaze patch (GP) strategy to ensure real-time performance for hologram generation, enabling the neural network to adapt to spatially dependent gaze-aware loss functions. Both numerical simulations and optical experiments demonstrate that our framework significantly enhances image quality in the foveal region while maintaining acceptable performance in peripheral regions. This hardware-agnostic approach provides a promising solution for next-generation near-eye holographic displays with integrated gaze tracking capabilities.