Microsoft filed 1 patent in the AI & Machine Learning category.
The application covers a system for creating photorealistic digital human avatars using AI-powered Gaussian splatting technology with prior model optimization. This approach enables real-time rendering of human digital twins for applications in virtual reality, gaming, and video conferencing.
Microsoft's single AI and machine learning patent tackles the challenge of generating lifelike digital human avatars from limited or imperfect source material. The system uses Gaussian splatting, a rendering technique that represents 3D scenes as collections of oriented 3D Gaussian primitives, but enhances it by incorporating a statistical prior model trained specifically on human appearance characteristics. This prior acts as a constraint during reconstruction, guiding the algorithm toward realistic human features even when input data is sparse or degraded. The result is fewer visual artifacts and more convincing avatars that can be rendered in real time for use in virtual environments, games, and remote communication Platformss.
All data sourced from USPTO patent filings. Google Patents may take several weeks to index recent publications. If a link is unavailable, search for the patent number at USPTO Patent Public Search.