Computer scientists at Microsoft’s research labs in Cambridge have developed “i2i”, a stereo camera for use with instant messaging technology that automatically frames and tracks its subject. This makes it seem that the user is looking at the camera, even they are actually looking at their PC screen: so video communication is more like a face-to-face conversation.
Using multiple webcams and a new algorithm, the technology creates a 3D image which can be rotated and viewed from many angles. Like 3D photography, this relies on combining two (or more) sets of camera data to produce one Cyclopean image. Unlike still photography however, the algorithm can produce the image in real time, on an ordinary laptop.
The team has also developed 3D emoticons. These can be keyed into the combined image, and appear to orbit around the users head!
Lead researcher Antonio Criminisi, who joined the research team at Microsoft from his post-doc work at Oxford, says the technology is not far off being ready for commercialisation. “Processor load is still an issue, but has come a long way in the last six months and will work respectably (at around 10 frames per second) on an unexceptional laptop now” he says.
There is, however, no guarantee that it will ever make it out of the lab and into product development. “It is what every researcher wants, to see your work make it in the real world, but we’ll wait and see,”
Once the system has worked out what is foreground and what is background, the background can be blurred, replaced, made monochrome and so on. The team is now working on a way of developing a 3D mask that could be wrapped over the users real face. This could be used to create avatars in games, for example. “Once you have reconstructed the 3D geometry of your image, there are so many cool things you can do,” says Criminisi. Click the picture above to see it in action.