Guide · 9 min read

Midjourney prompt structure that actually renders

Subject, camera, lighting, and style tokens for Midjourney v6 — with examples you can copy into Discord /imagine.

Midjourney is not a paragraph generator

Midjourney parses short, high-signal phrases separated by commas or line breaks. Long narrative paragraphs dilute emphasis. Think: subject → environment → lighting → camera → style reference.

Core anatomy

Layer	Purpose	Example
Subject	Who/what	"solo mountaineer, red jacket"
Scene	Where	"alpine ridge, pre-dawn fog"
Lighting	Mood	"soft rim light, cool shadows"
Camera	Lens feel	"35mm film still, shallow DOF"
Style	Aesthetic	"editorial outdoor, muted palette"

Parameters that matter

Aspect ratio — --ar 16:9 for banners, --ar 2:3 for posters
Style raw — when you want less default beautification
No — use sparingly; describe what you want instead of long negative lists

Reference images

When using --cref or style references, keep language consistent with the reference (wardrobe, palette). Change one variable at a time when iterating.

Workflow on My Copyprompt

Open a Midjourney prompt with reference output
Copy the prompt block
Paste into Discord /imagine
Adjust --ar only if the curator note suggests a different crop

Quality checks

Before publishing your own Midjourney prompt to the community:

Include at least one output image you generated with the exact text
Add a curator note explaining which words are load-bearing
Mention version-specific quirks (v6 vs niji)

Explore further

See cinematic portrait prompts or read our Flux prompting guide for cross-model comparison.