ClassroomThe classroom
Back to the lesson libraryMECHANISM · 21 MIN
M22.6 CONNECT THE MECHANISM

How an image can emerge from noise

Add static to a photo a thousand times and it becomes pure snow. Teach a network to undo one step, and it can turn fresh snow into a picture nobody took.

LESSON OVERVIEW21 min lesson

Lesson overview

Add static to a photo a thousand times and it becomes pure snow. Teach a network to undo one step, and it can turn fresh snow into a picture nobody took.

What you’ll explore

  • A diffusion-style image generator learns a denoising-related task and uses a chosen sampling procedure to turn noise into a structured sample, often conditioned on text.
GO TO THE SOURCE

Original explanations, connected to the research.

Denoising Diffusion Probabilistic Models (Ho, Jain & Abbeel, 2020)Deep Unsupervised Learning using Nonequilibrium Thermodynamics (Sohl-Dickstein et al., 2015)Score-Based Generative Modeling through Stochastic Differential Equations (Song et al., 2020)Scalable Diffusion Models with Transformers (Peebles & Xie, 2022)U-Net: Convolutional Networks for Biomedical Image Segmentation (Ronneberger, Fischer & Brox, 2015)
Suggest a correction

A precise note can make an explanation better.

Choose the scene and describe what needs attention. Download a feedback file to share through a channel you already use. This page does not send feedback or connect you with a reviewer.

The file includes this note, the scene title, and lesson metadata. Your saved progress and quiz responses are excluded. Download before leaving or reloading to keep your note.