Mock post: how to write a prompt for an AI image generator

A grid of generated images

This post exists to check how the blog looks on bluewillow.ai. Delete it once the check is done.

What a prompt consists of

A prompt has four parts, and the generator reads all of them:

  • the subject: who or what you see in the frame;
  • the setting: where it happens and what the light is like;
  • the style: photo, watercolour, 3D render;
  • the technical part: aspect ratio, camera, lens.

Two examples

A weak prompt: a cat. The generator guesses everything else on its own, so you get a different picture every time.

A strong prompt: a ginger cat on a windowsill, morning light, shallow depth of field, 35mm photo. Here the generator has the subject, the light and the optics, so the results stay close to each other.

Where to go next

Try the same prompt with different styles and keep the ones that work. The LimeWire image generator runs on the same models.