Lesson 61 · 2 min
AI that makes pictures
Image generators start from random noise and clean it up, step by step, until it matches the words you typed.
Most image generators are diffusion models. During training they saw millions of pictures with captions, and they practised one odd skill: taking a picture covered in static and removing a little of the static at a time.
To make a new image, the model starts with pure static and removes noise over many steps, steering each step toward your prompt. A cat on a red sofa slowly appears out of the fuzz.
Results are often impressive, and they still slip. Small text, signs, the number of fingers, reflections and fine patterns are where mistakes usually hide. Two people in one scene can also blend together.
Specific prompts work best: say the subject, the setting, the style and the light. And when you share an AI image, say it is AI-made, especially if it shows real places or people.
Do it now
Write a better picture prompt
- 1. Open any chatbot or app that can make images.
- 2. Try a short prompt first, such as "a market".
- 3. Then use the fuller prompt below and compare the two results. Check the signs and faces for mistakes.
A busy morning fruit market in a Vietnamese town, seen from a low angle, soft early sunlight, watercolour style, warm colours, no text or signs in the picture.
Quick check
Keep going
Next →Why your feed shows what it showsWords from this lesson
- Image generation
- Creating pictures from a written description, using AI trained on very large collections of images and captions.
- Diffusion model
- A type of AI that makes images or video by starting from random noise and cleaning it up step by step.