Google’s Project Genie Can Generate Photorealistic Worlds You Can Explore

Google released yesterday Project Genie, an experimental tool that lets Google AI Ultra subscribers generate playable 3D worlds and explore and remix worlds created by other users. To generate worlds, users just need to describe or upload images of what they want to create, and exploring these worlds mirrors the experience of playing a video game.

Project Genie leverages Google’s Gemini, Nano Banana Pro, and Genie 3 models. The latter is a new “world model” capable of creating dynamic worlds that can be explored for multiple minutes in real time at a 720p/24FPS quality.

“Unlike explorable experiences in static 3D snapshots, Genie 3 generates the path ahead in real time as you move and interact with the world. It simulates physics and interactions for dynamic worlds, while its breakthrough consistency enables the simulation of any real-world scenario — from robotics and modelling animation and fiction, to exploring locations and historical settings,” the company explained.

During the world sketching phase, Project Genie lets users create their 3D world and character and choose between a first-person or third-person view. At the moment, worlds can only be explored for 60 seconds, and Genie 3 also cannot always make text readable and recreate real-world locations in great detail. However, the potential is already there, and Google is already planning to introduce promptable events that change the world as users explore it.

While Project Genie is currently restricted to Google AI Ultra subscribers in the US, Google already plans to expand its access to more markets. “Building on the work we have been doing with trusted testers, we are excited to share this prototype with users of our most advanced AI to better understand how people will use world models in many areas of both AI research and generative media,” the company said yesterday.