Immersive Media Streaming via Adaptive Tile Density

The patent applies local quality by differentiating image quality across different spatial regions. The viewport region (where the user is looking) receives high-quality rendering with full pixel density, while peripheral regions use lower quality with reduced pixel density. This resolves the contradiction by maintaining high image quality where needed while reducing overall bandwidth consumption through selective quality adjustment in different frame regions.

The patent segments the omnidirectional video frame into multiple tiles or regions, allowing independent quality control for each segment. The viewport-containing tiles are transmitted at high quality, while other tiles use lower quality settings. This segmentation approach enables targeted bandwidth allocation that improves perceived image quality without proportionally increasing total bandwidth consumption.

The patent implements preliminary action by pre-calculating and buffering video tiles for anticipated viewport positions before the user actually looks in those directions. The system predicts future head movements and prepares corresponding high-quality tiles in advance, so when the user turns their head, the required content is already available or nearly available, significantly reducing perceived download latency while maintaining adaptability to head position changes.

The patent applies dynamics by making the video delivery system adaptive to real-time head position changes. The system dynamically adjusts which tiles are prioritized for download and rendering based on current and predicted viewport positions. This dynamic adaptation allows the system to respond to frequent head movements without being constrained by fixed download schedules, reducing latency while preserving immersive experience.

The patent segments the complex 6DOF ROI calculation problem into manageable components by processing tiles independently. Instead of calculating ROI for the entire omnidirectional frame at once, the system divides the frame into multiple tiles and performs ROI calculations for each tile based on its relationship to the viewport. This segmentation reduces per-tile computational complexity while maintaining overall accuracy through coordinated processing of all tiles.

The patent applies partial action by focusing computational resources only on tiles that contain or are near the viewport region. Rather than performing exhaustive 6DOF ROI calculations for all tiles in the omnidirectional frame, the system identifies and processes only the relevant subset of tiles that contribute to the current view. This partial processing approach maintains ROI calculation accuracy for visible regions while significantly reducing overall computational complexity.