DCGAN is initialized with random weights, so a random code plugged in the network would deliver a totally random image. However, as you might imagine, the network has numerous parameters that we will tweak, along with the purpose is to locate a location of those parameters which makes samples created from random codes seem like the training infor