DCGAN is initialized with random weights, so a random code plugged into the network would deliver a very random impression. On the other hand, while you might imagine, the network has numerous parameters that we are able to tweak, and the goal is to locate a placing of these parameters which makes samples created from random codes seem like the coa