Note, when Z != 0, my program runs very slow. This is due to the fact that the other features were implemented using matrix/image operations in matlab, while this is implemented with a quadruple for loop. I tried implementing a faster version that uses matrix ops, but couldn't figure it out in time. Thus, below are images taken if you are patient enough to wait it out (about 15 secs per update).
To implement this, I calculated s',t',u',v' given the users inputted Camera X, Camera Y, and Camera Z. Therefore, I needed to do this calculation for each pixel, and also the disparity and apature calculations from before, except for every pixel instead of entire images at a time. This was implemented with a quadruple for loop, which slows down the computation considerable.