A team of researchers from Carnegie Mellon University has created experimental software which is capable of reasonably extrapolating a 3D shape from a 2D image.

With the creation of the experimental software underscoring a noteworthy push by the researchers into the third dimension, the technique used by the researchers essentially enables 3D object manipulation in 2D photos. The technique compares the 2D objects in a regular photo with 3D models that are freely available online.

As such, the software created by the Carnegie Mellon University researchers - led by associate researcher of robotics Yasser Sheikh - basically makes use of freely available online stock 3D models of everyday objects, including furniture, cookware, clothing, appliances, and cars.

However, since 3D models of objects may not exactly match the objects in a 2D photo, the technique created by the researchers semi-automatically aligns the model to the shape of the object in the photo. Thereafter, the technique extrapolates lighting and what the concealed portions of the object might look like.

Explaining the technique, the research's lead author Natasha Kholgade - a PhD student at Robotics Institute - said that "In the real world, we're used to handling objects -- lifting them, turning them around or knocking them over;" and elaborated: "We've created an environment that gives you that same freedom when editing a photo. Instead of simply editing 'what we see' in the photograph, our goal is to manipulate 'what we know' about the scene behind the photograph."