Microsoft's Kinect SDK has allowed apps to perform 3D mapping for a while, but the Kinect is hardly ubiquitous or portable—let alone something many people already stuff into their pockets. Smartphones, on the other hand, are everywhere, and a project from Microsoft Research can turn them into pocketable 3D scanners. Microsoft calls the process MobileFusion. Check it out:
Neat, right? The idea is pretty great: use your camera to take video of an object, which an app on the phone will use to build a model. The app is constantly polling the video, comparing each frame to the frame before it to build stereoscopic images.
The full research paper is available from Oxford University (PDF) and it goes into a lot of detail. Using nothing but an iPhone 6, the research team was able to capture 320x240 video at 25 Hz and convert that video into 3D meshes with textures in real-time. This allowed them to get immediate results, and also to see how complete the mesh that the phone creates is. No additional post-processing is performed on the mesh—what comes out is purely from reading video frames in pairs, which provides depth information.
Peter Ondruska, a Ph.D. candidate at Oxford, worked on the project while he was an intern at Microsoft Research. He told Microsoft Research that the system isn't hard to use. "What this system effectively allows us to do is to take something similar to a picture, but it's a full 3D object."
Shahram Izadi, one of the principal engineers who worked on the project, told Linn the technology has plenty of uses. He suggested users might make 3D scans of landmarks they see on vacation, or of items they might want to sell online. "This is really about the accessibility and ubiquity of 3D scanning," Izadi said.
The research team hopes to support a wide range of platforms, specifically mentioning Android, iOS, and Windows Phone. Microsoft Research hopes to release this to the general public someday, but it hasn't made definite plans.