DCGAN is initialized with random weights, so a random code plugged to the network would generate a totally random graphic. Even so, when you might imagine, the network has a lot of parameters that we are able to tweak, plus the objective is to locate a location of these parameters that makes samples created from random codes look like the instructi