There exist a set of ``non-orthodox'' processes which don't simply apply a process to the left and right components of a StereoImage, but rather use the left and right components together, often to create a single image which uses information from both components combined in an non-standard, often ingenious way.
For example, Paul Cronin from the Physical Optics department of the School of Physics, University of Sydney, has developed an algorithm which allows video phone users to appear to be looking directly at their party. The problem is typically that the video camera in such scenarios is on the side of the screen, but the conversing party is displayed on the screen. When the user looks at their party, the camera records their image from the side, and so to their party the user appears to be looking to one side.
The algorithm takes a webcam on either side of the screen and returns a single image, which is the image of the user looking directly at the screen, reconstructed from the two composite images. This eliminates the problem described above, and would allow both conversing parties to appear to be looking directly at one another.
This is not a traditional piece of processing, as would be implemented in the Stereocam system, yet it still fits into the framework nicely. It would be desirable to have at least one of these kinds of non-orthodox processes, in order to showcase the flexibility and genericity of the Stereocam design and framework.