DCGAN is initialized with random weights, so a random code plugged in the network would crank out a very random graphic. Having said that, while you may think, the network has a lot of parameters that we could tweak, along with the objective is to locate a environment of such parameters that makes samples generated from random codes appear like the