Diffusion Model
An AI model that generates images by learning to reverse a gradual noising process.
AI-generated
Diffusion models generate images by starting with random noise and gradually removing it, step by step, until a clean image emerges. They learn this process by training on real images: noise is added gradually, and the model learns to reverse each step. Stable Diffusion, DALL-E 2, and Midjourney all use diffusion models.
Diffusion models are the dominant approach to AI image generation. They produce higher-quality, more diverse images than previous approaches (GANs) and are more stable to train. Understanding diffusion models helps you understand how AI image generation works and why it keeps getting better.
Lilian Weng: "What are Diffusion Models?" - https://lilianweng.github.io/posts/2021-07-11-diffusion-models/
Stability AI: Stable Diffusion - https://stability.ai/