DCGAN is initialized with random weights, so a random code plugged into the network would make a very random image. Even so, when you might imagine, the network has an incredible number of parameters that we can easily tweak, and also the purpose is to locate a setting of these parameters that makes samples produced from random codes appear to be t