2 pages
Generative models that learn to denoise, enabling high-quality image and video synthesis
High-resolution image generation by diffusing in learned latent spaces