Google DeepMind is merging its Street View imagery with Project Genie, a generative AI system that converts simple text prompts into interactive, photorealistic environments. The result is something that sounds like it belongs in a sci-fi film: type a description, and a navigable 3D-ish world materializes in real time, complete with controllable characters and dynamic conditions.
The tool is built on Genie 3, DeepMind’s latest world model, and it targets applications across robotics training, gaming, and virtual exploration.
How Project Genie actually works
Genie 3 takes a text prompt, or even a single image, and generates an interactive environment without requiring traditional 3D assets. No hand-modeled buildings, no texture libraries, no months of level design. The AI handles all of it.
Users can create worlds, select how they want to explore them, and generate movement paths in real time. Prompts that reference “Google Street View” or “GoPro” perspectives are particularly effective at producing convincing urban landscapes.
