Problem: the art I'm using for the professors is terrible; the ~20 images I am using are all inconsistent and mis-match-ish.
Solution: Create a tool to munge existing art into a consistent style. Sort of like Photoshop filters, but slightly more powerful. Or to put it another way, like A Neural Algorithm of Artistic Style but somewhat less powerful. I would have gone with the neural method, except that it seems like it is, at this point, a royal pain to set up for Windows, and I'm not quite ready to install Linux & dual boot just to use this tool. Also the neural method is (at least right now) very computationally expensive, with a single attempt taking 20 - 60 minutes. So! Instead, I will try to implement an idea that has been in my CommonPlace book for a while, to take a series of basis images, and then use them to try to re-create the original image as best as possible. It's basically the Fourier transform (assuming I remember what that is at all) but with jpgs and kittens instead of boring math. So, yes, I'm ready to get started on this app. This is definitely the most efficient way to generate the 20 professor images. :)