Build detailed prompts for AI image generators. Choose a subject, style, lighting, camera angle and mood, and copy prompts formatted for ChatGPT, Midjourney and Stable Diffusion.
Loading the tool…
What makes a good image prompt
- Subject: who or what, with key details.
- Setting: where and when.
- Style or medium: photo, watercolour, 3D render, anime.
- Lighting: golden hour, studio, neon, moody.
- Composition: close-up, wide shot, low angle, lens.
- Mood and colours.
- What to leave out.
Example
Midjourney: a red fox curled up asleep in a snowy pine forest at dawn, photorealistic photo, soft golden hour light, shallow depth of field, 85mm lens, calm and peaceful, soft blues and warm orange colour palette –ar 16:9 –no people, text
Prompt formats by tool
| Tool | Best format |
|---|---|
| ChatGPT, Gemini, Copilot | Full descriptive sentences |
| Midjourney | Short comma-separated phrases, plus –ar for aspect ratio and –no to exclude things |
| Stable Diffusion, Flux, Leonardo | Comma-separated keywords, plus a separate negative prompt |
Most image tools won’t create pictures of real people without consent or copy trademarked characters, so describe original subjects.
Frequently asked questions
How do I write a good AI image prompt?
Describe the subject, setting, style, lighting, composition and mood. Specific details give better results than long lists of adjectives.
What does --ar mean in Midjourney?
It sets the aspect ratio, such as –ar 16:9 for widescreen or –ar 9:16 for vertical video.
What is a negative prompt?
A list of things you don’t want, such as blurry, watermark or text, used by Stable Diffusion and similar tools.
Can I use these prompts commercially?
The prompts are yours. Whether you can use the images depends on the terms of the image tool you use.
Last updated: