David Tinapple

15-463: Computational Photography
INSTRUCTOR: Alexei (Alyosha) Efros
Computer Science Department
Carnegie Mellon University

Project 5

For this project I photographed a room in a way that provided a single point perspective image, with one vanashing point in the center of the image. From this image 5 textures were retreived, of the two side walls, the floor, the ceiling, and the far facing wall. These textures were extracted by hand using photoshop (not very computational, I know, but quick and accurate).

Using the Max/MSP/Jitter visual programming environment, I created a simple box room with roughly the same relative dimensions as the room in the photograph. I used the extracted textures to map back onto the faces of the virtual room. I adjusted the dimensions of the geometry and the properties of the virtual camera by hand, until they looked right.

The reason for using Max was that I wanted to track the position of my head (my viewpoint) in real-time and use that data to position the virtual viewpoint. Using live video input with a camera behind the screen, facing me (the viewer), I manually seed the tracking system with a starting point, using the point between my eyes as a reference. The XY position of my head is mirrored in the XY position of the virtual camera. As I move in real space, my virtual viewpoint matches.

Its fun, and gives a very strange sensation.

Also, see the quicktime movie demonstrating the work.

-David