Google's Project Genie adds Street View integration to generate explorable 3D worlds from real locations

Google announced at I/O 2026 that Project Genie, its experimental world-building tool, now integrates Google Street View imagery to create interactive 3D environments based on real locations across the United States.

How it works

Project Genie uses Google's Nana Banana Pro models to generate small, explorable worlds at 20-24 frames per second. According to Google, users can now tap a Maps pin to select any US location with Street View coverage, choose an optional style like "Desert Sands" or "Stone Age," and describe a character to place in the world.

The system generates hundreds of referenced images to create navigable scenes controlled with standard WASD video game controls. Sessions are limited to 60 seconds of exploration.

Technical constraints

Project Genie remains available only to users on Google's AI Ultra subscription tier. The company cites the tool's high processing demands as the reason for this limitation.

Accuracy is explicitly not guaranteed. Google states it plans to improve detail sharpness and accuracy in future updates, acknowledging current limitations in the generated environments.

Availability

The Street View integration is currently limited to US locations only. Google says it plans to expand to additional regions "over time" but provided no specific timeline.

Project Genie originally launched in early 2026 with the ability to generate only fictional places. The tool offers both first-person and third-person viewpoints.

What this means

Project Genie represents Google's exploration of real-time 3D world generation using AI, competing with similar research efforts in interactive environment synthesis. The Street View integration suggests a path toward mixing real-world data with generative AI for location-based applications. However, the 60-second session limit, US-only availability, and acknowledged accuracy issues indicate this remains an early research prototype rather than a production-ready feature. The requirement for an AI Ultra subscription also signals significant computational costs that currently make broad deployment impractical.