Click through three things GANs did well: crisp images, a loss learned by a judge network, and making an image in one fast pass.
For most of a decade, GANs were what people meant by image generation.
