AnyoneLearnAI
LevelEasyStart
Brain labTry it → read → next · ~8 min

Tutorials · Chapter C (3/4) · ~8 min

Diffusion models in plain English

Try it → see it → read → next

Start from noise, denoise step by step, guided by words — that's how many image generators work.

On paths: Student

Optional on: Everyday learner

Try yourself

Playground

Diffusion in four steps

Denoise from static → image, guided by your text prompt.

Prompt: watercolor cat on a windowsill

Random noise canvas

Advance through denoising steps.

Recap

What you just did

You stepped through denoising — random static slowly becomes a coherent scene.

Teach

How it works

Training teaches the model to remove noise that doesn't fit real images. At generation time:

  1. Start with random noise
  2. Predict a slightly cleaner image
  3. Repeat many steps
  4. Prompt steers what "clean" should look like

Different from LLMs — but often bundled in the same products (chat + image).

Use it

When you'd use this

  • Marketing visuals, storyboards, concept art
  • Understanding limits of "AI art" tools

Watch out

Watch out

Text in images, hands, and logos often fail. Check rights for commercial use.