StyleGAN2

This is quite the improvement from v1 of this -- crazy how fluid this is, also kinda scary how computers are starting to analyse the human condition as well huh? -- we are gonna see some crazy new music videos let me tell you! :)

The style-based GAN architecture (StyleGAN) yields state-of-the-art results in data-driven unconditional generative image modeling. We expose and analyze several of its characteristic artifacts, and propose changes in both model architecture and training methods to address them. In particular, we redesign generator normalization, revisit progressive growing, and regularize the generator to encourage good conditioning in the mapping from latent vectors to images.