Fly-though representations of the Colosseum, the Trevi fountain and the Croatian city of Dubrovnik was made by the scientists from the University of Washington. They have used advanced photo analysis and modeling techniques to do it.
They used a specially-devised algorithm to detect and arrange photos showing various angles of the same building. The same algorithm then analyzed these images – taking into account the viewpoint of the photographer – to produce composite 3D models.
Rome’s most famous landmarks were generated in less than 24 hours from 150,000 photos, while 250,000 images were scanned to make 3D models of Venice.
Although the Croatian city was recreated from just 4,600 images, the researchers say that the complex modeling techniques could be used to map huge areas in detail.
The researchers say that they now plan to work on using still images to create photo-realistic 3D tours, rather than just sketchy models.
They hope to preserve threatened urban areas – like the sinking island-city of Venice – for future generations, and say that the technology could also be used to create hyper-realistic video games and GPS systems.