As the name suggests Gans help generate anything and everything. As I take notes of my learning of gans here on my github , let me start with a meme here.
The concept of Gans is based on a fight between the Generator network and the discriminator network. The generator network as the name suggests learns to generate images based on a random noise , while the discriminator tries to understand whether an image is real or fake. A point comes when it becomes difficult for the discriminator to understand whether an image is a real one or a fake one. It is at this point that the training can be stopped.