DCGAN is initialized with random weights, so a random code plugged in the network would make a completely random image. However, when you may think, the network has numerous parameters that we are able to tweak, and also the goal is to locate a environment of these parameters which makes samples