
Manipulating objects in photos in 3D

According to CNET , researchers from Carnegie Mellon University have developed a technology that allows you to make flat objects in the photo three-dimensional, and freely manipulate them.
The technology is based on using the base of stock 3D models of objects from everyday life (cars, furniture, clothing, dishes ...).
After a little adjustment and calibration, the system allows you to replace a flat object from a photograph with its voluminous version, which can be freely rotated in space and animated.

At the moment, there are difficulties associated with filling the base of models. In many cases, it is necessary to use a model that is close but not identical in geometry for the object being modified.
As the model database is filled, the system will allow you to quickly recognize objects and replace them with accurate 3D models.
At this stage, there are also difficulties with “pulling out” objects with uneven edges, or located on a heterogeneous background.
The technology will be unveiled at SIGGRAPH 2014, August 13, in Vancouver.
Detailed information about the project can be found on its official website , as well as in a rather voluminous and detailed description of the research (PDF).