DCGAN is initialized with random weights, so a random code plugged to the network would produce a very random impression. Nevertheless, while you may think, the network has a lot of parameters that we are able to tweak, plus the objective is to locate a environment of those parameters which makes samples produced from random codes appear like the i