DCGAN is initialized with random weights, so a random code plugged into the network would crank out a very random picture. Nonetheless, when you might imagine, the network has countless parameters that we can easily tweak, and also the goal is to locate a setting of such parameters that makes samples created from random codes appear like the traini