Science

Making pictures move

Seeing images spring to life feels like the future of Instagram. Dr Jingtang Liao demonstrated his software that creates depth in 2D images. The next step is to bring this magic into the hands of smartphone users.

Adding depth to pictures. (Images: Jingtang Liao)

Once a picture contains depth information, it can be used to amplify the perception of depth by changing the depth of focus, the lighting, haziness or perspective. But how to move from a flat RGB (red, green, blue values for every pixel) picture to what Dr Jingtang Liao calls an RBGD picture that has an added value for depth in each pixel?


Liao, who graduated last week at the Faculty of Electrical Engineering, Mathematics and Computer Sciences, has developed software that makes it easy for users to annotate depth in images. Darker is closer, and further away is lighter. By choosing a matching grey tone, the user adds scribbles to the picture to add depth information. In the case of Vermeer’s Girl with the Pearl Earring, for example, the shoulder would be almost black, the neck a shade lighter, the face mid-grey, her turban lighter still and the background white. Merging the annotations with the spatial colour distribution, the software then creates a depth map in grey scale.


Darker is closer, and further away is lighter


The software will assume gradual depth transitions, except where there are sharp edges in the picture, in which case the software will assume a sudden jump in depth. Sometimes the user has to identify regions to be ignored, such as shadows and reflections in windows.


Once the RGBD map is complete, the fun can start. Depth can then be used to increase depth perception by changing saturation, haze, focus, or the perspective. Changing the point of view makes objects move more as they come nearer. So when you move the point of view horizontally, vertically or just wiggle it around, the image will spring to life as if by magic. Watch the video to get an impression of the effect.