Reference · Glossary

Diffusion model

A generative model that **starts from noise and iteratively denoises** into an image (or other signal) guided by a text prompt.

When to use

Understanding how AI image tools (Stable Diffusion, DALL·E, Midjourney-style systems) create pictures from words.

When not to

Text-only chatbots — they use LLMs, not diffusion (unless the product bundles both).

Example

Prompt "watercolor cat on a windowsill" → model removes random noise step-by-step until a coherent image emerges.