DCGAN is initialized with random weights, so a random code plugged in the network would create a very random picture. Even so, when you might imagine, the network has millions of parameters that we will tweak, and the goal is to locate a location of these parameters which makes samples generated from random codes appear like the coaching facts.Elev