基于网格-4DGS混合神经表示的端边协作渲染与传输方法及系统
By using a mesh-4DGS hybrid neural representation method, dynamic and static elements are separated and rendering tasks are offloaded, solving the problem of insufficient computing power of terminal devices. This enables efficient free-viewpoint video rendering on devices with weak computing power, improving rendering performance and transmission efficiency.
Patent Information
- Application Number
- CN202510978526.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2025-07-16
- Publication Date
- 2026-07-17
- Estimated Expiration
- 2045-07-16
AI Technical Summary
Existing edge-to-edge collaborative rendering methods cannot be directly used for free-viewpoint videos built with 4DGS. The computing power of terminal devices cannot support real-time rendering, and traditional rasterization rendering acceleration technology cannot be applied to 4DGS.
A method based on mesh-4DGS hybrid neural representation is adopted. By separating dynamic and static elements, static elements are represented by texture mesh and dynamic elements by 4DGS. The rendering task is offloaded to the edge server. Combined with self-distillation strategy and parallel rendering pipeline, collaborative rendering between terminal device and edge server is realized.
Achieving over 120 FPS free-viewpoint video rendering on mobile virtual reality headsets and low-computing-power edge servers improves rendering performance and transmission efficiency while reducing overall latency.
Smart Images
Figure CN121099120B_ABST
Abstract
Citation Information
Patent Citations
Fast virtual viewpoint synthesis method and device based on feature texture grid and Hash coding
CN116778063A
Efficient volume video representing and rendering method and device
CN117252969A