Diffusion Models
Generative models that create data by gradually denoising random noise.
Diffusion models learn to reverse a process that slowly adds noise to data. Starting from pure noise, the model denoises step by step until a coherent image, video, or audio clip emerges. Operating in a compressed latent space (latent diffusion) makes this fast enough for everyday use.
They are the backbone of modern image and video generators, including Stable Diffusion, FLUX, and DALL·E.