Project Genie is an experimental web app that lets people create, explore, and remix interactive virtual worlds using text prompts and images, and is now rolling out to Google AI Ultra subscribers in the U.S.
Built on Google DeepMind’s Genie 3 world model, it aims to broaden access to immersive environments that evolve in real time as users move and act within them. The prototype showcases Google’s wider AGI mission by simulating dynamic physics and interactions across an open-ended range of scenarios, from robotics and animation to fictional and historical settings.
World Sketching From Prompts and Images
Project Genie’s “World Sketching” lets users describe environments and characters using text, then refine them using generated or uploaded images. Integrated Nano Banana Pro tools allow users to preview and edit visuals before entering the scene, allowing them to fine-tune worlds to better match their intended look and feel.
Creators can also define how they traverse these spaces, whether walking, riding, flying, or driving, and choose perspectives such as first-person or third-person to shape the experience.
Exploring and Remixing Interactive Worlds
Once created, each world becomes a navigable environment that responds to player movement, generating the path ahead in real time based on user actions. Camera controls also let users adjust how they view their surroundings as they travel through these interactive spaces.
Project Genie encourages experimentation through “World Remixing,” where people can build on existing prompts, browse curated worlds or even use a randomizer for inspiration, then download videos of their creations and explorations.
An Early Prototype with Clear Limits
Being an early research prototype in Google Labs, Project Genie comes with known constraints, including visuals that may not always match prompts or real-world physics, less controllable characters, and also a 60-second limit on generations.
Some Genie 3 capabilities, like promptable events that alter the world while exploring, are not yet available. Access begins with U.S.-based Google AI Ultra subscribers, with plans to expand to more regions over time.
Frequently Asked Questions (FAQ)
Q: What is Project Genie?
A: Project Genie is an experimental web app that lets users create, explore, and remix interactive virtual worlds powered by Google DeepMind’s Genie 3 world model.
Q: Who can access Project Genie?
A: Access is rolling out to Google AI Ultra subscribers in the U.S. who are 18 years or older, with expansion to additional territories planned.
Q: How do users create worlds in Project Genie?
A: Users combine text prompts with generated or uploaded images through “World Sketching,” then refine visuals and choose traversal modes and camera perspectives before entering.
Q: Can users modify or reuse existing worlds?
A: Yes, Project Genie supports “World Remixing,” allowing users to build on existing prompts, explore curated worlds, use a randomizer for inspiration, and download videos of their sessions.
Q: What are some limitations of Project Genie?
A: Known limitations include worlds that may not perfectly match prompts or physics, sometimes less controllable characters, generation durations capped at 60 seconds, and some Genie 3 features not yet included.
Emman has been writing technical and feature articles since 2010. Prior to this, he became one of the instructors at Asia Pacific College in 2008, and eventually landed a job as Business Analyst and Technical Writer at Integrated Open Source Solutions for almost 3 years.
