DCGAN is initialized with random weights, so a random code plugged into the network would deliver a very random graphic. Having said that, when you might imagine, the network has millions of parameters that we will tweak, along with the intention is to find a environment of these parameters which makes samples created from random codes appear to be