DCGAN is initialized with random weights, so a random code plugged into the network would generate a very random impression. Having said that, when you may think, the network has a lot of parameters that we could tweak, along with the aim is to find a location of those parameters that makes samples created from random codes appear like the schoolin