DCGAN is initialized with random weights, so a random code plugged into the network would deliver a very random impression. Nonetheless, as you may think, the network has a lot of parameters that we could tweak, as well as the objective is to locate a placing of these parameters which makes samp