How to Write Better Midjourney Prompts: The Complete Guide
Midjourney is one of the most powerful AI image generators available — but the gap between a weak prompt and a strong one is enormous. The same concept, described differently, produces wildly different results. Understanding how Midjourney reads prompts is the single biggest lever for improving your output quality.
How Midjourney Reads Your Prompt
Midjourney treats your prompt as a set of weighted concepts rather than a sentence. Each word and phrase carries influence proportional to how semantically dense and unambiguous it is. Common filler words ("the", "a", "of") carry almost no weight. Specific, uncommon, visually concrete terms carry significant weight.
The implication is that how you say something matters as much as what you say. "A beautiful woman in a forest" gives the model too many degrees of freedom. "Young woman::2 emerald forest::2 dappled sunlight::1.5, hyperrealistic, cinematic, 8K" gives it clear instructions about what to prioritise.
The Core Structure of a Strong Midjourney Prompt
Subject
Start with the main subject — be specific about what it is, who it is, and what they are doing. "A scientist" is weak. "Female scientist in her 40s, examining a glowing specimen under a microscope, focused expression" is strong.
Environment and setting
Where is the scene? What time of day, what weather, what era? "A city at night" is weak. "Neon-lit Tokyo alley, rain-slicked cobblestones, midnight, 1980s cyberpunk aesthetic" is strong.
Lighting
Lighting has an outsized effect on Midjourney output. Be specific: "dramatic side lighting", "golden hour backlighting", "candlelight", "overcast diffused light", "studio three-point lighting". This single addition transforms mediocre results into publication-quality images.
Style and medium
Tell Midjourney what kind of image you want: "photorealistic", "oil painting", "watercolour", "pencil sketch", "concept art", "illustration", "3D render", "digital art in the style of Studio Ghibli". Without this, Midjourney makes its own stylistic choices, which may not be what you wanted.
Technical parameters
--ar 16:9— aspect ratio (landscape)--ar 9:16— aspect ratio (portrait, ideal for social media)--ar 1:1— square--v 7— use the latest Midjourney version--q 2— higher quality rendering (slower)--s 750— stylisation level (higher = more artistic interpretation)
Word Weighting: The Most Underused Technique
Midjourney's :: syntax lets you assign explicit weight to specific concepts. Adding ::2 after a phrase doubles its influence. ::0.5 halves it. This is how experienced Midjourney users control which elements dominate the composition.
Example:
Without weighting: a woman in a red dress at sunset on a clifftop
With weighting: woman::2 red dress::1.8 clifftop::2 golden sunset::2 ocean background::1 —ar 2:3 —v 7
The weighted version consistently produces images where the woman, dress, and clifftop are the compositional focus rather than the model choosing arbitrarily.
Common Prompt Mistakes
Over-describing relationships
"A woman who is standing next to a tree that has red leaves" is weaker than "woman::2 red autumn tree::2, standing beside, forest path". Remove the connective tissue and let the weights carry the meaning.
Contradictory instructions
Asking for "minimalist but highly detailed" or "dark and light simultaneously" produces unpredictable results. Be specific about which direction you want when there's a genuine tension.
Ignoring negative prompts
The --no parameter removes concepts. If you keep getting unwanted elements (extra limbs, text in the image, watermarks, ugly faces), add --no text, watermark, extra fingers, blurry. This is often more effective than trying to describe what you DO want more specifically.
Using Prompt Power Pro to Automate This
Writing well-weighted Midjourney prompts manually is time-consuming until you've internalised the methodology. Prompt Power Pro is a Chrome extension that applies word weighting automatically — you describe what you want in plain language and it rewrites the prompt with weighted syntax, the appropriate aspect ratio parameters, and structured output ready to paste into Midjourney.
The extension uses your own OpenAI key (BYOK), so your prompts go directly to GPT-4o — no third party processing. One-time £7.99.
→ Get Prompt Power Pro — £7.99Frequently Asked Questions
What makes a Midjourney prompt good?
Specificity, weighted concepts, clear style direction, and lighting information. Strong Midjourney prompts are visually concrete (not abstract), tell the model what to prioritise through weighting, specify the medium or style, and include lighting details. Aspect ratio parameters ensure the output matches your intended use.
How does word weighting work in Midjourney?
Adding ::2 after a word or phrase in your Midjourney prompt doubles its influence on the output. ::0.5 halves it. This lets you explicitly control which elements dominate the composition rather than letting the model decide. Example: mountain::2 sunset::1.5 solitary figure::1.
What aspect ratio should I use in Midjourney?
For social media: --ar 9:16 (portrait) for Instagram Stories and TikTok, --ar 1:1 for feed posts. For widescreen and landscapes: --ar 16:9. For cinematic film stills: --ar 2:1 or --ar 21:9. Always specify aspect ratio — the default square often isn't what you need.
What is the difference between Midjourney v6 and v7?
Midjourney v7 (released 2026) produces significantly more coherent and photorealistic outputs than v6, with better text rendering, improved anatomy, and more natural lighting. Use --v 7 to access the latest version. Earlier versions may produce results that feel more painterly or stylised — some users prefer them for artistic work.