Models & Architectures
Diffusion Model
The AI architecture behind Stable Diffusion, DALL-E, Sora 2, and most modern image/video generators.
Also known as: diffusion models
A diffusion model is the AI architecture behind most modern image and video generators — Stable Diffusion, DALL-E 3, Sora 2, Flux Pro, Imagen, Midjourney (partially). The intuition: take a clean image, gradually add random noise until it is pure static, then train a model to reverse the process. To generate, start with random noise and ask the trained model to "denoise" it toward the prompt. Diffusion models have become the standard because they produce sharper, more coherent images than older GAN-based approaches and scale beautifully to video (Sora 2, Veo 3, Hunyuan Video).
