DCGAN is initialized with random weights, so a random code plugged to the network would create a very random graphic. Nonetheless, as you may think, the network has countless parameters that we can easily tweak, along with the target is to find a setting of those parameters which makes samples generated from random codes appear like the coaching de