DCGAN is initialized with random weights, so a random code plugged in the network would make a completely random picture. Nevertheless, as you may think, the network has many parameters that we could tweak, along with the purpose is to find a setting of those parameters that makes samples created from random codes appear like the education informat