DCGAN is initialized with random weights, so a random code plugged to the network would make a very random impression. Even so, while you might imagine, the network has countless parameters that we can tweak, and also the goal is to locate a environment of these parameters that makes samples generated from random codes look like the schooling detai