New technology for creating three-dimensional portraits - VIDEO PDF Print E-mail
Monday, 17 August 2015 17:11
A team of researchers from the Federal Polytechnic School of LaUSAnne has introduced a new technology to create three-dimensional portraits without the use of sophisticated technology - only with the camera on your phone. "We wanted to make the process quick and easy. All you need - is to remove the video and add to it a few frames from the facial expressions of the interviewee human. Everything else our algorithm will do for you" - says the senior author of the development Ikim Alexander (Alexandru Ichim). This technology will find a number of very important practical applications in the future, according to the Swiss. For example, in a rapidly growing online gaming industry, as well as in online discussions or video conferences. The authors of the new technology say that the idea of ​​creating avatars itself is elementary, but the basic algorithms of its development is not as simple as it seems at first glance. "Creation of a conventional two-dimensional image is a fairly easy task. Now we want to simplify this level and the creation of three-dimensional pictures. The main difficulties - this change in light of the object to change the angle, and blurred pictures that will appear when the user starts to do a portrait panoramic photo" - explains Alexander. The new algorithm, created by French scientists, includes many details. Specially created to Alexander and his colleagues computer program complements those inaccuracies that allows the user during the creation of the portrait on her smartphone. The algorithm takes into account not only the position of the head in space, but also facial expressions: if the first photo man smiled, then all other photographs to complement its 3D-portrait, it will have the same cheerful expression. In the near future, a new study plan to improve their technology. For example, at the moment of its most difficult task they are playing hairstyles user: a three-dimensional image to restore the position of each hair on a person's head is extremely challenging. No less difficult from the point of view of developers, to save on each panoramic image position of the teeth and ears. So far, the authors want the algorithm to focus on facial expression, before going on to detail other parts of the body. The development will be presented at SIGGRAPH in autumn 2015.