How to build a Midjourney
master prompt template

Why Midjourney works differently from other AI tools

Unlike ChatGPT, Claude, and Gemini, Midjourney has no system prompt field, no custom instructions panel, and no persistent configuration. Every prompt starts fresh. The master prompt concept for Midjourney translates into a reusable base template - a fixed set of style descriptors and parameters you append to every image prompt to create consistent visual identity.

This is not a limitation - it is how Midjourney is designed. The prompt structure is the configuration. Once you understand how to lock style, lighting, mood, and parameters into a base template, you have the equivalent of a master prompt that produces a recognizable visual identity every time.

How to build a Midjourney base template

Step 1 - Define your visual style

Choose the medium and rendering style that defines your visual identity. Examples: "digital art," "photorealistic," "oil painting," "cinematic photography," "anime illustration." Be specific - "photorealistic close-up portrait photography" produces more consistent results than "realistic."

Step 2 - Lock your lighting

Lighting has more impact on visual consistency than almost any other descriptor. Choose one style and lock it: "golden hour lighting," "dramatic cinematic lighting," "soft diffused natural light," "studio lighting with deep shadows."

Step 3 - Set your parameters

• --ar - aspect ratio (16:9 for landscapes, 1:1 for square, 9:16 for portrait)
• --s - stylize value (0-1000, default 100. Higher = more artistic interpretation)
• --v 6.1 - always specify version for consistent output over time

Step 4 - Negative prompts

Add descriptors for what you never want: "no text, no watermarks, no blur, no distortion, no extra limbs." Including these in your base template prevents common Midjourney artifacts across all your generations.

Advanced techniques

--cref (character reference)

Similar to --sref but for maintaining character consistency. Generate a strong character image first, then use that image's URL as your character reference for all subsequent prompts featuring that character.

Style tuner

Midjourney's Style Tuner lets you generate pairs of images and pick your preferred aesthetic. The result is a unique style code you can add to all future prompts to apply that exact tuned style.

Common questions

How many descriptors should my base template have?

Between 5 and 15 is the sweet spot. Too few and output varies too much between generations. Too many and Midjourney starts ignoring or misinterpreting conflicting descriptors.

How do I maintain character consistency?

Use --cref with a reference image URL. Generate a strong character image first, then use that URL as your character reference for all subsequent prompts. See our DALL-E guide for how image generation techniques compare.