DCGAN is initialized with random weights, so a random code plugged to the network would generate a totally random image. On the other hand, as you may think, the network has countless parameters that we could tweak, and also the target is to find a placing of such parameters that makes samples created from random codes look like the education detai