A Style-Based Generator Architecture for Generative Adversarial Networks

12 Dec 2018Tero Karras • Samuli Laine • Timo Aila

We propose an alternative generator architecture for generative adversarial networks, borrowing from style transfer literature. The new architecture leads to an automatically learned, unsupervised separation of high-level attributes (e.g., pose and identity when trained on human faces) and stochastic variation in the generated images (e.g., freckles, hair), and it enables intuitive, scale-specific control of the synthesis. The new generator improves the state-of-the-art in terms of traditional distribution quality metrics, leads to demonstrably better interpolation properties, and also better disentangles the latent factors of variation.

Full paper

Evaluation


Task Dataset Model Metric name Metric value Global rank Compare
Image Generation CelebA-HQ 1024x1024 StyleGAN FID 5.06 # 1
Image Generation FFHQ StyleGAN FID 4.40 # 1