Google has developed a groundbreaking new AI tool called Genie 2, which can generate fully playable 3D game worlds from a single image prompt. This technology enables artists and designers to quickly prototype environments, accelerating the creative process for environment design.
Genie 2 works by using a single image as input, similar to Google’s existing text-to-image AI tool Imagen 3. Users can then interact with the generated world, and the AI will adapt on the fly to create new content based on the user’s actions. This includes generating non-player characters (NPCs), physics, lighting models, reflections, and multiple animations.
While still in its early stages, Genie 2 has significant potential for training more advanced AI models and solving structural problems in embodied agent research. Google emphasizes its focus on responsible development, highlighting its commitment to improving the tool’s capabilities and consistency.
The tech giant also sees Genie 2 as a stepping stone towards creating more general AI systems that can understand and safely carry out a wide range of tasks. As such, its primary purpose is not yet clear-cut, but it holds immense potential for revolutionizing video game development and other industries.
Source: https://www.eurogamer.net/google-has-developed-an-ai-tool-that-generates-fully-playable-3d-worlds