Midjourney Complete Guide: From First Image to Pro Results
Everything you need to know about Midjourney v6 — account setup, prompting strategies, style parameters, aspect ratios, and advanced techniques.
Midjourney produces the most aesthetically impressive AI images available. But it has a real learning curve — vague prompts produce mediocre results, and the parameter system rewards the time you invest in learning it. This guide gets you from setup to professional-quality outputs as quickly as possible.
Getting Access to Midjourney
Midjourney is available at midjourney.com. The web interface (available to all subscribers) is now the recommended starting point. You can also use it through Discord if you prefer the community experience.
Plans start at $10/month (Basic, ~200 images), $30/month (Standard, unlimited relaxed + 15 fast GPU hours), and $60/month (Pro, stealth mode + more fast hours). There is no free tier.
Tip: The Standard plan at $30/month is the right starting point for most users. Basic runs out quickly if you're experimenting, and Pro's main benefit (stealth mode) isn't needed until you're doing professional work.
Your First Image Generation
In the Midjourney web interface, type your prompt in the Create bar. A good starting prompt structure: [subject] [style] [mood/atmosphere] [technical details].
Example: 'A lone lighthouse on rocky coastline, dramatic storm clouds, golden hour light breaking through, cinematic photography, 8K --ar 16:9'
Midjourney will generate four variations. You can upscale any of them (U1-U4) for a higher-resolution version, or create variations (V1-V4) to explore alternatives from a specific image.
Essential Parameters
--ar (aspect ratio): Controls the output dimensions. Common values: --ar 1:1 (square), --ar 16:9 (landscape), --ar 9:16 (portrait/mobile), --ar 3:2 (photo standard)
--v (version): Specifies the model version. Default is v6.1. For style variety, experiment with --niji 6 for anime/illustration styles.
--stylize (or --s): Controls how opinionated Midjourney is with aesthetics. Range: 0-1000. Default is 100. Higher values = more artistic interpretation, lower values = closer to the literal prompt.
--chaos (or --c): Controls variation between the four generations. Range: 0-100. Higher = more unexpected results. Good for exploration.
Tip: --stylize 50 is great when you want precise prompt adherence. --stylize 250-500 when you want Midjourney to apply its aesthetic judgment more freely.
Prompting for Photorealistic Images
The best photorealistic prompts include: a camera system (Sony A7R V, Canon EOS R5), a lens (85mm portrait lens, 24mm wide angle), lighting conditions (golden hour, overcast diffuse light, dramatic side lighting), and post-processing style (Lightroom edit, film grain, high contrast).
Example: 'Portrait of a middle-aged woman in a bookshop, natural window light, soft shadows, Sony A7R V, 85mm f/1.8, shallow depth of field, warm color grade --ar 4:5'
For photorealistic still life or product shots, add: 'studio lighting, white background, product photography, commercial quality, 8K resolution'
Art Style References
Reference specific art movements, photographers, or aesthetics to give Midjourney direction. Effective style references include: 'in the style of Edward Hopper', 'Wes Anderson color palette', 'brutalist architecture photography', 'impressionist oil painting', 'retro 1970s magazine illustration'.
For specific vibes, use descriptive aesthetic keywords: 'cottagecore', 'dark academia', 'solarpunk', 'vaporwave', 'cyberpunk noir'.
Add '/describe' in the Midjourney Discord to analyze an image you like and get its approximate prompt — very useful for reverse engineering styles you want to replicate.
Tip: The /describe command is one of Midjourney's most underused features. Upload any image and it will generate four prompt suggestions. Use these as starting points and modify.
Advanced: Consistent Characters and Style
For consistent characters across multiple images: use --cref [image URL] to reference a character image. Use --cw to control how closely character features are preserved (0-100).
For consistent style: use --sref [style reference image URL] to apply the style of a reference image to your prompt.
For consistent multi-image projects: create a style guide image first (e.g., a color palette and mood board), then reference it across all subsequent generations to maintain visual coherence.
Key Takeaways
- Detailed prompts with style, lighting, and technical specs produce far better results than short prompts
- --ar, --stylize, and --v are the three parameters you'll use most
- The /describe command helps you reverse-engineer prompts from images you like
- --cref and --sref enable consistent characters and styles across multiple images
- Iteration is essential — generate many variations and refine the ones that are close