How the machines work

Member lesson

Diffusion, explained without the mathematics

Lesson 2 of 13 in How the machines work.

How AI Image Generators Work (Stable Diffusion / Dall-E) - Computerphile · Computerphile
Lesson

Take a photograph and add a small amount of noise. Do it again. Do it a thousand times and you have static. A diffusion model is trained by watching that destruction and learning to predict, at every stage, what was removed. Run that prediction backwards from pure static and you get a picture that was never photographed.

noisestep 1step 2step 3step 4the prompt pulls at everystepnothing checks it is anobject
Noise resolving into a picture over five denoising steps, with the prompt pulling at each one.
Five square prints pegged on a line. The first is pure speckled noise, and each one after it is a little clearer, until the last shows a crisp jug and lemon on a table. One thread held by a hand runs through all five prints.
Left to right, noise becomes a picture a little at a time, and the prompt pulls on every step.
Member

The rest of this is for members.

You have read the opening of this lesson. The rest of it comes with a membership.

One membership covers all four courses and Brands.

  • All lessons in the four Learn courses
  • All brand breakdowns in the Brands section
  • All quizzes, flashcards and crosswords
  • An AI teacher on every lesson, for up to 60 questions a day
  • All course material published during the membership

Price

Membership is from $12 a month on the Yearly plan, which is charged at $144 once a year. Lifetime access is $59, paid once.Membership is from ₹200 a month on the Yearly plan, which is charged at ₹2,400 + GST once a year. Lifetime access is ₹3,999 + GST, paid once.

Check what stuck

Members only

You can't take this check yet.

Members answer these 3 questions and keep the ticket for this lesson. Every lesson has one.

  1. Question 1, locked

    Which of these words means this? “Training-time only. Noise is added to a real image, step by step, until nothing is left.”

    • Captioning
    • Forward process
    • Embedding
    • Hallucination
  2. Question 2, locked

    Which of these words means this? “What happens when you generate. The model removes a little noise at a time until a picture appears.”

    • Guidance scale
    • Unit economics
    • Reverse process
    • Cross-attention
  3. Question 3, locked

    Which of these words means this? “One pass. Twenty to fifty of them make an image; more is not reliably better.”

    • Eval
    • Denoising step
    • Attention
    • RAG

Elsewhere in the library

Hear about new courses.

Leave an email and we will tell you when something new lands. We do not write often.