The patent describes the use of a single depth camera to capture images of an object from various angles, to generate an accurate 3D model. Unlike similar technologies, Microsoft’s technique doesn’t require the item to be rotated on a turntable — it can remove the background and anything in the way (such as a user’s hands that are holding the item) because they don’t move in sync with the object.
In addition, this method allows for capturing views of objects that would be difficult to manage when using a turntable, such as the belly of an animal figurine. It also captures the color and surface patterns of the object, making for a more realistic 3D model.
What does Microsoft want with this technology? One possibility is that the company could use it to scan items that a HoloLens user looks at, to allow them to digitally interact with their models, Minority Report-style. Let’s hope we don’t have to wait until 2054 to find out.