Midjourney V7 represents a meaningful step forward from V6 — better understanding of mood and energy, improved prompt coherence, and a personalization system that actually adapts to your aesthetic preferences over time. If you have been on the fence about Midjourney, or if you tried V6 and found it inconsistent, V7 is worth another look. This guide covers everything you need to go from zero to generating professional-quality images.
Getting Started: Access and Interface
Midjourney runs through Discord, though the web interface at midjourney.com offers a more approachable experience for beginners who do not want to navigate Discord servers. Either way, the core workflow is the same: you enter prompts in a message box prefixed with /imagine, and the bot generates a grid of four images for you to refine or upscale.
The first step is choosing your plan. The free tier gives you a limited number of generations — enough to test the interface but not enough to learn seriously. A Basic plan at $10/month gives you enough generation time for regular use. The Standard plan at $30/month includes unlimited relaxed mode generations and fast generation time for priority processing.
Understanding Prompts: The V7 Shift
The biggest practical change in V7 compared to V6 is how prompts are interpreted. In V6, longer prompts with more visual detail generally produced better results. In V7, the model responds equally to mood, energy, and atmosphere as it does to literal visual descriptions. A shorter prompt that establishes a clear mood often outperforms a long catalog of visual elements.
Structure your prompts in three parts: the core subject and its relationship to space, the artistic style or medium, and the lighting or atmosphere. For example, a weathered lighthouse on a rocky cliff at dusk, film photography, volumetric fog — this establishes subject, style, and mood in a compact form that V7 can interpret with less ambiguity than a longer descriptive paragraph.
The Parameters Worth Knowing
Parameters go at the end of your prompt, separated by double dashes. The most important ones for V7:
Aspect ratio (–ar) controls dimensions. Default is 1:1 (square). Use –ar 16:9 for desktop wallpapers, –ar 9:16 for mobile or Instagram Stories, –ar 3:2 for photography-style framing, –ar 2:3 for portrait orientation.
Stylize (–s) controls how much artistic interpretation Midjourney applies. Default is 100. Start at –s 200 for an editorial look with moderate artistic interpretation, or –s 50 for more documentary-style literal rendering.
Chaos (–c) controls variation across the four-image grid. Keep it between 5 and 30 for controlled, consistent outputs. Above 30, results become genuinely unpredictable — useful for brainstorming but risky for targeted generation.
Raw (–raw) reduces Midjourney default stylization, producing more cinematic, photorealistic results. This is particularly useful when you want the AI to interpret your prompt more literally rather than applying its default aesthetic beautification.
V7 Reference Tools: The Consistency Game-Changers
V7 introduced or significantly improved three reference tools that solve the hardest problem in AI image generation: consistent style and character appearance across multiple images.
Style references (–sref) let you upload an image whose visual aesthetic — color palette, texture, mood — Midjourney will apply to your generation without copying the subject. This is the tool to use for brand-consistent imagery: upload your brand asset, reference it in every generation, and get output that maintains your visual identity.
Character references (–cref) anchor a character face and body across different scenes. Generate a character you like, get its URL, and use –cref [URL] in subsequent prompts to generate that same character in new contexts. Use –cw (character weight) to control how much is copied — 100 copies face and outfit, lower values copy only the face.
Personalization (–p) is V7 most impactful underused feature. It requires a setup investment of about ten minutes rating 200 image pairs, after which the model builds an aesthetic profile that subtly tunes every generation toward your taste. Once set up, use –p [your personal code] to apply your profile explicitly, or leave it on by default and let it run in the background.
The Iterative Workflow That Actually Works
Professional Midjourney users rarely get what they want in a single generation. The reliable workflow is: generate broadly first with a simple prompt, evaluate which quadrant or result is closest to your vision, use Vary Region to refine specific areas, then upscale the best candidate. Do not try to perfect everything in one shot — build toward your target image through successive generations.
Start broad with low chaos values to establish a baseline, then increase chaos to explore variation in directions that show promise. Keep track of seed numbers for results you want to build on — the same seed with a modified prompt produces consistent compositional variation that you can compare systematically.
Common Beginner Mistakes to Avoid
Do not write paragraph-long prompts. In V7, direction matters more than detail. Focus on mood, subject, and style rather than cataloging every visual element. Do not ignore aspect ratio — a prompt without –ar defaults to square, which looks wrong on any platform designed for landscape or portrait content. Do experiment with style references early: uploading a reference image you find visually compelling and asking Midjourney to work in that aesthetic is the fastest way to learn what the tool can do and to develop an intuition for what parameters and approaches work for your specific goals.