The Midjourney prompt formula (with 12 copy-ready examples)
The short answer
A reliable Midjourney prompt follows subject, action, environment, lighting, lens or medium, composition, style and parameters. Keep it under about 60 words for v7, state the light early, use --ar for aspect ratio and --stylize to control artistic interpretation, and change one variable at a time.
Why keyword soup fails
A list of forty adjectives gives the model forty conflicting instructions. It averages them into something generic: soft light, shallow depth of field, highly detailed, 8k. The result looks like stock art because it is the average of stock art.
A scene description gives the model a hierarchy. The subject is the subject. The light is a direction and a quality, not an adjective. The lens determines perspective. The parameters set the frame.
The formula in practice
Here is the structure we use across the image generation prompt collection:
- Subject: who or what, with one distinctive detail.
- Action or pose: what they are doing, stated simply.
- Environment: where, with two or three concrete details.
- Light: direction, quality, colour temperature.
- Lens or medium: 35mm, 85mm, macro, watercolour, gouache.
- Composition: framing, negative space, angle.
- Style: restrained references to a movement or production style, not a living artist.
- Parameters: --ar, --stylize, --style raw, --no.
Example: editorial portrait
``text Editorial portrait of a ceramicist in her studio, seated, looking slightly off camera. Soft north window light from the left, gentle falloff, warm grey backdrop. 85mm lens at f/2, shallow depth of field, natural skin texture, subtle grain. Calm and confident. Negative space on the right for a headline. --ar 4:5 --style raw --stylize 150 ``
Example: product hero
``text Studio product photograph of a matte black travel mug on a rough travertine plinth. Warm directional light from the upper right, soft shadow, a thin highlight on the rim. Seamless sand background, minimalist composition, generous space above for copy. 100mm macro, f/8, crisp logo-free surface. No text. --ar 4:5 --stylize 100 ``
Example: cinematic environment
``text Cinematic wide shot of an empty Lisbon street at dawn, wet asphalt reflecting warm lamps. Low mist between buildings, a single cyclist in the distance, overcast blue-hour sky with a peach horizon. Anamorphic 35mm, f/2.8, muted teal and amber grade, fine grain. --ar 21:9 --stylize 250 ``
Parameters that actually matter
| Parameter | What it does | When to use |
|---|---|---|
--ar 4:5 | Aspect ratio | Portraits and social; 16:9 for scenes |
--stylize 0-1000 | Artistic interpretation | Lower for product realism, higher for concept art |
--style raw | Less default beautification | Editorial and documentary looks |
--chaos 0-100 | Variation between grid images | Early exploration |
--no text, watermark | Negative steering | Almost always |
--seed | Reproducibility | When a composition works |
The five mistakes we see most
- Too many subjects. One clear subject beats three competing ones.
- No light direction. "Beautiful lighting" is not a direction.
- Parameters in the middle. Put them at the end so the scene reads cleanly.
- Naming living artists. It is ethically and legally risky, and the model may not reproduce the look anyway.
- Changing five things at once. You learn nothing about which change worked.
Iterating without wasting credits
When a result is close, keep the seed and change one thing: the lens, the light direction or the aspect ratio. When the composition is wrong, rewrite the subject and environment but keep the style. When the style is wrong, change the medium and stylize, not the scene.
Save every prompt with the model version, seed and parameters. A look you cannot reproduce is not an asset.
Text in images
Midjourney v7 is better at short text but still unreliable. Keep text under four words, put it in quotes, and expect to fix it in a design tool. For text-heavy graphics, Flux or a vector editor is a better fit.
A note on rights
Check Midjourney's current terms for commercial use, and avoid prompts that reproduce a trademarked character, a living person's likeness or a copyrighted artwork. Generate, then retouch and typeset in a design tool before publishing.
Browse all Midjourney prompts or try the portrait prompt and product prompt.
Copy-ready prompts featured in this guide
Copy any prompt below or open its full page with interactive variable fields and platform tabs.
Site Admin
Author & Platform Administrator
Administrator and lead editor at AIPrompt123, curating high-performance AI prompts, SEO/AEO/GEO strategies, and developer repositories.
Questions about this guide
Short answers to what readers ask most about this topic.
Keep reading
The best ChatGPT prompts for students (that actually help you learn)
Most 'study prompts' generate summaries that feel productive and teach you nothing. These twelve prompts are built around retrieval, feedback and self-testing.
How to write a system prompt that actually holds
A five-block structure for system prompts, with examples, the rules that cannot be enforced in a prompt, and how to test for drift.
Prompt engineering for beginners: zero-shot, few-shot and chain-of-thought
The techniques that matter, when to use them, and why chain-of-thought is not a universal upgrade. With copy-ready patterns.