New AI Makes 3D Models From Just a Few Photos

Imagine being able to take a few photos of a room, a product, or even a local park, and have a computer instantly turn them into a detailed 3D model. That's exactly what a new AI, called Atlas, from a company called World Labs is promising to do. It can recreate full 3D scenes using just two or three pictures, a big step up from older methods that needed many more.

This technology might sound a bit futuristic, but it's quickly becoming very practical. For businesses, especially smaller ones, this could be a game-changer. Think about real estate agents creating virtual tours of properties with minimal effort, or online retailers showcasing products in interactive 3D views without needing expensive equipment. Builders and renovators could easily visualise designs in existing spaces, making planning much simpler.

Atlas is an example of an "omni world model" – essentially, an advanced AI that can understand and represent the world in a very detailed, spatial way. Its ability to work from such limited input is what sets it apart. This means less time, less computing power, and less hassle to get a high-quality 3D representation of almost anything.

While this particular technology is still emerging from a startup, the implications are broad. It points to a future where capturing our physical world in a digital format becomes as easy as snapping a few photos. This could transform how we shop, design, learn, and even interact with maps and navigation, bringing sophisticated 3D tools to the fingertips of everyday Australians and small business owners.