In this demo, a projector is aimed at two poster boards placed at an angle to each other.

Normally, the resulting projection should be skewed because of perspective distortion.

However, in this case a kinect camera is used to capture 3d depth information about the environment, thus giving knowledge about where the projected light hits the poster boards.

A second kinect camera is used to find the head location of the viewer relative to the kinect.

Combining this information, it is possible to pre-distort the image, so when the image reaches the viewer, it appears to be non-distorted.

In this case, the mona lisa frame appears to be rectangular, independent of the view location.

Full write-up at:

Loading more stuff…

Hmm…it looks like things are taking a while to load. Try again?

Loading videos…