Sign in

AI image generators

2022

Diffusion models such as DALL-E 2, Midjourney and Stable Diffusion turn text prompts into detailed images for millions of users.

The story

In 2022 a new kind of software let anyone type a sentence and receive a detailed, original image in seconds. OpenAI announced DALL-E 2 in April, Midjourney opened its public beta in July, and Stability AI released Stable Diffusion in August 2022. Unlike its rivals, Stable Diffusion came with its weights openly published, so anyone with a decent graphics card could run it at home, modify it and build products on top of it. Within months millions of people were producing illustrations, concept art and fake photographs.

The core method is called a diffusion model. During training, the system takes real images and gradually adds random noise until only static remains, while a neural network learns to reverse each small step. To generate, it starts from pure noise and removes it step by step, steered by a text prompt that a language model has converted into numbers. The idea was sketched in 2015 and made practical around 2020. Stable Diffusion's 'latent' version worked on a compressed representation of the image, which cut the computing cost enough to run on consumer hardware. It rested on the deep learning breakthrough, the Transformer architecture used to interpret prompts, and web-scraped image-text sets such as LAION-5B. That scraping triggered lawsuits from artists and photographers over copyright and consent.

Why it mattered

  • Generative AI reached the general public, setting the stage for ChatGPT's launch a few months later.
  • Illustrators, photographers and stock-image firms faced new competition, and lawsuits tested how copyright applies to training data.
  • Realistic synthetic images raised the stakes for misinformation and led to labeling and watermarking efforts.
  • Open model weights made it possible for communities to fine-tune and share their own image models.

Drafted by AI (Claude) from the sources above and checked against them by AI; no historian has reviewed this page. Spot a mistake? Suggest a correction. How pages are made

Community notes

Loading notes…

Learn more ranked by votes

BookMustafa Suleyman
The Coming Wave
Search Open Library
VideoSearch YouTube
how diffusion models generate images
Search YouTube