Midjourney Prompt Generator
Build a correctly formatted Midjourney prompt — subject, style, lighting, and camera, plus –ar, –v, –stylize, –chaos, and –no parameters.
Midjourney prompts respond specifically to parameter syntax — aspect ratio, stylization, version, and quality flags — that don't do anything in a plain descriptive sentence. This tool builds a subject description plus the actual --parameters, so the result is genuinely ready to paste into Midjourney, not just descriptive text you'd still need to add flags to yourself.
How to use it
Be as specific about lighting, composition, and medium as you would describing a photo to someone.
These become actual --ar, --stylize, and --v flags appended to the prompt.
The output is ready to paste as-is, flags included.
Tips for better results
- Use higher stylization values for abstract or artistic subjects, not product shots. High stylization can drift away from literal accuracy, which matters more for some subjects than others.
- Match aspect ratio to where the image will actually be used. A square crop of a 16:9 generation loses composition you specifically prompted for.
- Keep the subject description concrete over abstract adjectives. A specific physical scene outperforms a vague mood phrase for consistent results.
Example output
“A weathered lighthouse keeper studying a storm through a rain-streaked window, cinematic lighting, oil painting style --ar 16:9 --stylize 250 --v 6”
TL;DR
Midjourney reads two very different things at once: a natural-language description of the image, and a string of dash-parameters (–ar, –v, –stylize, –chaos, –no) that control technical output settings the description alone can’t. Mixing the two badly — burying a parameter mid-sentence, or forgetting one entirely — is the most common reason a Midjourney prompt doesn’t do what you expected.
What each parameter actually changes
- –ar (aspect ratio) sets the canvas shape. 16:9 and 9:16 suit video-adjacent or mobile-first images; 1:1 suits social posts and icons.
- –v (version) picks the underlying model. v6.1 is the current general-purpose default; niji 6 is tuned specifically for anime and illustrated styles.
- –stylize controls how much Midjourney’s own aesthetic bias is applied on top of your description. Low values track your wording more literally; high values lean into more dramatic, “Midjourney-looking” compositions even if that drifts from a literal reading of the prompt.
- –chaos controls variation between the four grid options Midjourney generates. Low chaos gives four similar takes on the same idea; high chaos gives four quite different interpretations, useful for early exploration before you’ve settled on a direction.
- –no is a negative-prompt shorthand — listing things to keep out of frame (text, watermarks, extra fingers) rather than describing what to include.
Subject: “a lighthouse on a cliff at dawn,” Style: oil painting, Lighting: golden hour, Camera: wide-angle, Aspect ratio 16:9, v6.1, Stylize 250, Chaos 0, No: text, watermark. This generator turns that into:
“a lighthouse on a cliff at dawn, oil painting, golden hour lighting, wide-angle shot –ar 16:9 –v 6.1 –stylize 250 –no text, watermark”
Every descriptor is grouped before the parameters, and every parameter is correctly dash-prefixed and ordered — the exact structure Midjourney’s own documentation recommends.
Who this is for
Anyone using Midjourney who’s tired of guessing at parameter syntax, or who keeps forgetting whether it’s --ar or --aspect. It’s equally useful for beginners learning the syntax and for regular users who want a faster way to assemble a prompt than typing parameters from memory each time.
FAQ
What’s the difference between –stylize and –chaos?
Stylize controls how much Midjourney’s own aesthetic preference overrides your literal wording. Chaos controls how different the four grid results are from each other. You can raise both, either, or neither.
Do I need to use niji 6 for anime-style images?
It’s not required, but niji 6 is specifically tuned for anime and illustrated styles and generally gives more consistent results than describing “anime style” under the general v6.1 model.
Why does my –no list sometimes get ignored?
–no works best with concrete nouns (text, watermark, extra limbs) rather than abstract concepts. Very long or vague –no lists tend to have a weaker effect than two or three specific exclusions.
Can I use this prompt in the Midjourney web app as well as Discord?
Yes, the prompt and parameter syntax are identical in both.
Does this tool save my prompts?
No. Everything runs in your browser and nothing is stored or sent anywhere.
