Nvidia received 2 granted patents, all in AI & Machine Learning.
The patents cover systems for automatically generating live commentary by processing video frames through language model pipelines, applicable to sports, gaming, and events. Additionally, the AI & Machine Learning work includes neural network techniques that learn realistic forces for animating objects and characters in computer Graphics, enhancing physics simulation versatility.
NVIDIA's 2 AI and machine learning patents tackle distinct automation challenges in media and animation. The first system processes video frames through a language model pipeline to generate live commentary for sports, gaming, and events, replacing human commentators with a structured approach that translates visual events into natural language narration. Unlike earlier methods that relied on manual scripts or rigid templates, this approach uses a dynamic prompt construction layer that feeds visual understanding directly into large language models, enabling flexible, real-time description of on-screen action. The second patent trains neural networks on multi-aspect motion data to derive force-based control signals for animating objects and characters, addressing longstanding problems with brittle, limited-variety movement in neural animation systems. By focusing on forces rather than directly predicting positions or poses, the method produces more resilient and versatile physics simulations for computer Graphics applications.
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.