Requires version 1.5.0 of the Java Runtime Environment, available from Sun Microsystems.


On the left, the available reference images are shown in the tree view. The camera parameters for the images are calculated in advance, so you do not have to input anything but the position and direction of the desired view that you want to render. To do this, simply click on the view pane (the image area to the right of the tree view), move the mouse cursor to rotate the camera and click again. At this point, the program will start calculating the visual hull of the object. This will probably take some time, and a progress bar will indicate how far along the process has come. When the calculations are complete, the image area is updated with the rendered novel view of the object. In this version, a depth map of the object is what is rendered. In the stand-alone application, it is possible to render the object using a simple (and buggy) implementation of view-dependent texture mapping which uses the information in the reference images to shade the rendered object.

If you are interested in knowing more about this project, the source code and project report are available for download.