DCGAN is initialized with random weights, so a random code plugged into the network would make a totally random impression. Nonetheless, as you might imagine, the network has numerous parameters that we will tweak, and the target is to locate a placing of these parameters that makes samples created from random codes appear like the schooling detail