Isn't the examples not very telling if you can't see the input too? Results look good, but I don't know what the input / what kind of preprocessing what done on it.