Google Project Genie lets you create interactive worlds from a photo or prompt

Google has launched Project Genie, an enhanced version of its AI-powered world model that allows users to create interactive environments based on a photo or text prompt. Initially demonstrated with Genie 3 for a select group of testers, Project Genie is now available to paying subscribers of Google’s premium AI service. This innovation represents a step forward for the AI/ML community, showcasing how advanced AI can create dynamic and responsive simulations, albeit limited to short-term memory for detail retention.

Project Genie utilizes world sketching techniques, whereby users can input reference images or descriptive prompts to generate unique environments. It integrates with Google's advanced models like Nano Banana Pro and Gemini 3, yielding a still image representation before evolving it into a video experience. Users can interact with these environments in real time, with the system rendering at 720p and 24 frames per second. While still in its early stages, this technology highlights remarkable potential for gaming, virtual reality, and storytelling, paving the way for richer user experiences in AI-generated content.