Google Earth can now generate AI images out of whatever you’re looking at. Zoom to a spot in Earth on the web, hit “create image”, type what you want, and Nano Banana 2 builds it from the satellite, aerial and 3D data of that actual location. Live globally as of yesterday.
Google’s pitch leans on classrooms and property people. Their own example prompts include rendering the Pompeii ruins as the living city of 78 A.D., turning an empty Tokyo lot into a shopping district, and asking for a Statue of Liberty infographic, where Gemini pulls the historical facts and Nano Banana draws the graphic around them. There’s a “Refine Image” button after generation, and results can be saved into an Earth project. Genuinely useful for a history lesson or a pitch deck. Or for putting a lakefront cabin on your neighbor’s plot, Google suggests that one too.
Now the fine print, and there’s a fair bit of it. Web only. No app support at launch, despite Google Earth sitting on over 500 million Play Store downloads. Generation is capped at a few images per day, and here’s the odd part – paying Google AI subscribers hit that same cap, because Earth runs its own separate subscription tiers, independent of Google One. Every image carries a visible watermark plus an invisible one, and Google’s developer documentation states flatly that downloading or sharing the images outside Google Earth isn’t supported. So whatever you make lives inside Earth, stamped as AI. And the infographics inherit the usual Gemini caveat: the facts get retrieved by AI, so check them before they reach a classroom.
Fun feature, honestly. Just narrower than the launch video makes it look.
