Atlas Management for Augmented Reality Rendering Latency

The patent segments the eye-buffer into multiple atlases based on content importance and visibility. Instead of processing the entire full-resolution eye-buffer, the system divides it into smaller atlas regions that can be selectively rendered and transmitted, reducing overall processing latency while maintaining quality for critical regions.

The patent applies different quality levels to different regions of the eye-buffer based on their importance. High-priority regions (such as foveal regions or regions containing critical AR content) are rendered at full resolution, while less important regions use lower resolution, thereby reducing overall latency and power consumption while maintaining perceived rendering quality.

The patent extracts only the essential visual information from the full eye-buffer by identifying and isolating important content regions. These extracted regions are then encoded into compact atlases, eliminating the need to encode and decode unnecessary transparent or low-priority regions, thus reducing power consumption while preserving critical visual information.

The patent applies partial action by encoding only the necessary portions of the eye-buffer at full quality rather than the entire buffer. This selective encoding approach reduces the total computational work required for encoding and decoding operations, directly lowering power consumption while maintaining visual information integrity for important regions.