DCGAN is initialized with random weights, so a random code plugged in the network would generate a totally random graphic. Even so, when you might imagine, the network has countless parameters that we can easily tweak, and also the target is to find a environment of those parameters that makes samples generated from random codes appear like the tra