Microsoft has introduced a new generative AI model called Muse, which could significantly change approaches to game creation on the Xbox platform. Muse can generate game environments based on visual data and player actions, understanding the game’s three-dimensional world and its physics.
The model was developed through a collaboration between Microsoft Research and Ninja Theory studio, which provided access to a large amount of gameplay data. Muse was trained on seven years of Bleeding Edge gameplay data, allowing the model to process a billion pairs of images and actions. Currently, the model can generate visualizations at a resolution of 300×180 pixels.
Microsoft is demonstrating examples of using the Muse model to create real-time game visualizations. Although the frame rate is currently limited to 10 frames per second, this is just the beginning of research in this direction. The company is also exploring the use of Muse to enhance classic games and adapt them to modern hardware, opening up new opportunities for preserving gaming heritage.
Microsoft plans to share AI tools and experiments with Xbox players and developers to address real-world challenges and add new value to the creation and gameplay process. At the same time, the company emphasizes that the decision to use generative AI in games will remain with individual studios.
