# Diffusion Model

> It starts with random dots. Step by step, it cleans them up into a picture.

**Field:** Artificial Intelligence  
**Category:** Machine Learning  
**Also known as:** Denoising Diffusion Model

## Think of it like

Think of it like a fuzzy picture. It gets clear, one step at a time.

## Definition

A diffusion model makes images by removing noise from them, step by step. In training, it takes real images, adds random noise to them, and learns to undo that noise. To make a new image, it starts from pure noise. Then it cleans it up over many steps until a picture appears. A text prompt can steer which picture it makes.

## Example

Lucy types 'a red fox reading a book'. A diffusion model starts with static and, over 50 steps, turns it into that picture.

## Why it matters

Most of today's image and video makers are built on diffusion models. They took over from GANs.

## Common mistake

Expecting a clean picture after a few steps. Quality comes from many small steps, so fewer steps is faster but rougher.

## Three principles

- **Learns to remove noise.** From noisy real images.
- **Starts from static.** Pure random noise.
- **Cleans step by step.** Until a picture appears.

## Related terms

- [Generative AI](https://learn.benibix.com/term/ai/generative-ai)
- [Generative Adversarial Network (GAN)](https://learn.benibix.com/term/ai/generative-adversarial-network)
- [Neural Network](https://learn.benibix.com/term/ai/neural-network)
- [Deep Learning](https://learn.benibix.com/term/ai/deep-learning)
- [Text-to-Image](https://learn.benibix.com/term/ai/text-to-image)
- [Stable Diffusion](https://learn.benibix.com/term/ai/stable-diffusion)
- [ControlNet](https://learn.benibix.com/term/ai/controlnet)

## Told apart from

- [GAN vs Diffusion Model](https://learn.benibix.com/compare/gan-vs-diffusion-model): People often think every AI image maker works the same way. These two do not. A GAN makes a picture in one go, and a second network judges it. A diffusion model starts from noise and cleans it up over many steps. Most image makers today use diffusion.

## Taught in

- [GANs and Diffusion Models](https://learn.benibix.com/learn/ai-machine-learning/gans-and-diffusion-models): Machine Learning, 16/18
- [Midjourney and AI Image Tools](https://learn.benibix.com/learn/ai-ai-tools/midjourney-and-ai-image-tools): AI Apps and Tools, 9/10

![Five squares in a row. The first is pure noise. Each next one has less noise, and the last one shows a clear picture of a house.](https://learn.benibix.com/diagrams/nn-diffusion.svg)

---

Source: https://learn.benibix.com/term/ai/diffusion-model
Author: Behnia Zarei (a developer in Vienna, Austria, who builds and teaches data, AI and business intelligence at BENIBIX)
Publisher: BENIBIX Learn (https://learn.benibix.com/)
Last updated: 2026-11-17
Licence: CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). Quote freely with attribution.
