We describe a new training methodology for generative adversarial networks. The key idea is to grow both the generator and discriminator progressively: starting from a low resolution, we add new ...
Some results have been hidden because they may be inaccessible to you