How to Use Midjourney: Beginner to Advanced Guide 2025
Midjourney produces the most stunning AI-generated images available today. This comprehensive guide takes you from first-time user to advanced prompter, with practical examples and techniques.
Getting Started with Midjourney
What is Midjourney?
Midjourney is an AI image generation tool that creates images from text descriptions (prompts). It's known for exceptional aesthetic quality—images that look artistic and intentional rather than "AI-generated."
Setting Up Your Account
Step 1: Join Discord Midjourney operates through Discord. If you don't have Discord:
- Go to discord.com
- Create an account
- Download the app (recommended) or use the web version
Step 2: Join Midjourney
- Go to midjourney.com
- Click "Join the Beta" or "Sign In"
- This will take you to the Midjourney Discord server
Step 3: Subscribe Midjourney requires a paid subscription (no free tier currently):
| Plan | Price | Fast Hours | Features |
|---|---|---|---|
| Basic | $10/mo | 3.3 hrs | ~200 images |
| Standard | $30/mo | 15 hrs | Unlimited relaxed mode |
| Pro | $60/mo | 30 hrs | Stealth mode, more fast hours |
| Mega | $120/mo | 60 hrs | Heavy users |
Recommendation: Start with Basic to learn, upgrade to Standard for serious use.
Step 4: Start Creating
Use the /imagine command followed by your prompt:
/imagine prompt: a sunset over mountains
Where to Generate Images
Option 1: Midjourney Discord Server
- Use #newbies channels or create threads
- Results are public
Option 2: Direct Messages
- DM the Midjourney Bot
- More private
- Access via Discord DM
Option 3: Web Interface (Alpha/Beta)
- midjourney.com/imagine
- Rolling out to users
- More intuitive interface
Basic Prompting
Your First Prompt
Start simple:
/imagine prompt: a red apple on a wooden table
Midjourney will generate 4 image variations. You can then:
- U1-U4: Upscale that image (create full resolution)
- V1-V4: Create variations of that image
- 🔄: Rerun the entire prompt
Prompt Structure
Basic format:
/imagine prompt: [subject] [details] [style] [parameters]
Example:
/imagine prompt: a majestic lion, golden fur, dramatic lighting, wildlife photography --ar 16:9
Adding Details
The more specific you are, the better results:
Vague:
/imagine prompt: a house
Specific:
/imagine prompt: a cozy cottage with a thatched roof, surrounded by wildflowers, morning mist, watercolor style, peaceful atmosphere
Style Keywords That Work
Photography styles:
- photorealistic
- professional photography
- portrait photography
- macro photography
- aerial photography
- 35mm film
- cinematic still
Art styles:
- oil painting
- watercolor
- digital art
- concept art
- illustration
- anime
- Studio Ghibli style
- Pixar style
Mood and atmosphere:
- dramatic lighting
- golden hour
- moody
- ethereal
- vibrant
- muted colors
- dark and mysterious
Quality keywords:
- highly detailed
- 8k
- masterpiece
- trending on ArtStation
- award-winning
Essential Parameters
Parameters modify how Midjourney generates images. Add them at the end of your prompt with --.
Aspect Ratio (--ar)
Change image dimensions:
--ar 16:9 (widescreen, landscapes)
--ar 9:16 (vertical, mobile, portraits)
--ar 1:1 (square, social media)
--ar 4:3 (classic photo)
--ar 2:3 (portrait photography)
--ar 3:2 (landscape photography)
Example:
/imagine prompt: vast desert landscape at sunset --ar 21:9
Version (--v)
Specify Midjourney version:
--v 6 (latest, most capable)
--v 5.2 (previous stable version)
--v 5.1 (older)
--niji 6 (anime/illustration focused)
Example:
/imagine prompt: anime girl with blue hair, school uniform --niji 6
Stylize (--s)
Controls artistic interpretation (0-1000):
--s 0 (strict prompt adherence)
--s 100 (default)
--s 500 (more artistic)
--s 1000 (maximum artistic freedom)
Low stylize: More literal, less artistic High stylize: More aesthetic, less literal
Example:
/imagine prompt: a mountain landscape --s 750
Chaos (--c)
Controls variation between the 4 images (0-100):
--c 0 (similar images)
--c 50 (moderate variety)
--c 100 (maximum variety)
Example:
/imagine prompt: abstract patterns --c 80
Negative Prompts (--no)
Exclude specific elements:
--no text
--no people
--no watermark
--no blur
Example:
/imagine prompt: serene beach scene --no people --no text
Quality (--q)
Adjust generation quality/time:
--q .25 (faster, lower quality)
--q .5 (balanced)
--q 1 (default, full quality)
Tile (--tile)
Create seamless patterns:
/imagine prompt: floral pattern, botanical illustration --tile
Seed (--seed)
Use a specific seed for reproducibility:
/imagine prompt: landscape --seed 12345
To find an image's seed: React with ✉️ to get details.
Intermediate Techniques
Image Prompting
Use images as input:
Method 1: URL
/imagine prompt: [image URL] modern interpretation
Method 2: Upload + Reference
- Upload image to Discord
- Copy image URL
- Use in prompt
Blending Images
Combine multiple images:
/blend [image1] [image2]
Or with the /imagine command:
/imagine prompt: [imageURL1] [imageURL2] fusion, combined elements
Image Weight (--iw)
Control image vs text prompt influence:
/imagine prompt: [imageURL] cyberpunk style --iw 0.5 (more text influence)
/imagine prompt: [imageURL] cyberpunk style --iw 2 (more image influence)
Range: 0.5 to 2
Character Reference (--cref)
Maintain consistent characters across images:
/imagine prompt: [character description] walking in park --cref [URL of character image]
Character weight (--cw):
--cw 0 (face only)
--cw 100 (full character, default)
Style Reference (--sref)
Maintain consistent style:
/imagine prompt: a cat in forest --sref [URL of style reference]
Style weight (--sw):
--sw 100 (default)
--sw 1000 (maximum style adherence)
Multi-Prompts
Combine multiple concepts with :::
/imagine prompt: sunset::2 mountain::1 river::.5
Numbers indicate relative weight.
Example:
/imagine prompt: cyberpunk city::2 nature elements::1 peaceful atmosphere::0.5
Permutations
Generate multiple variations efficiently:
/imagine prompt: a {red, blue, green} car
This creates three separate jobs.
Example:
/imagine prompt: portrait of a woman, {oil painting, watercolor, digital art} style
Advanced Techniques
Prompt Engineering for Specific Results
For Photorealistic Portraits:
/imagine prompt: professional headshot of [description], natural lighting, shallow depth of field, shot on Canon 5D, 85mm lens, f/1.8, studio lighting, high resolution --ar 2:3 --s 100
For Fantasy Art:
/imagine prompt: [subject], fantasy art, epic composition, dramatic lighting, highly detailed, cinematic, by Greg Rutkowski and Alphonse Mucha --ar 16:9 --s 750
For Product Photography:
/imagine prompt: [product], professional product photography, studio lighting, white background, commercial photography, clean, minimalist --ar 1:1 --s 50
For Architectural Visualization:
/imagine prompt: modern house design, architect magazine cover, architectural photography, dramatic sky, golden hour, professional exterior shot --ar 16:9
For Logo/Icon Design:
/imagine prompt: minimalist logo for [brand], simple, clean lines, vector style, single color, professional, iconic --s 50
Advanced Parameter Combinations
Maximum Control:
/imagine prompt: [detailed prompt] --ar 16:9 --v 6 --s 100 --c 0 --no blur, noise, artifacts
Maximum Creativity:
/imagine prompt: [loose concept] --s 1000 --c 100 --weird 1000
Consistent Style Across Images:
/imagine prompt: [subject] --sref [style URL] --sw 100 --s 50
The Weird Parameter (--weird)
Add unusual, unexpected elements (0-3000):
/imagine prompt: a forest --weird 500
/imagine prompt: portrait --weird 1500
Higher values = more surreal results
Stop Parameter (--stop)
Stop generation partway for softer/abstract results:
/imagine prompt: landscape --stop 50 (50% completion)
Midjourney Styles and Aesthetics
Photography Styles
/imagine prompt: [subject], shot on Hasselblad, medium format, film photography
/imagine prompt: [subject], shot on Leica, street photography, black and white
/imagine prompt: [subject], drone photography, DJI Mavic, aerial view
/imagine prompt: [subject], iPhone photography, natural, candid
Artist Style References
/imagine prompt: [subject] in the style of [artist name]
Photography:
- Annie Leibovitz (portraits)
- Ansel Adams (landscapes)
- Steve McCurry (documentary)
- Peter Lindbergh (fashion)
Digital Art/Concept:
- Greg Rutkowski
- Craig Mullins
- Wlop
- Artgerm
Classic Art:
- Monet (impressionism)
- Van Gogh (post-impressionism)
- Rembrandt (lighting)
- Alphonse Mucha (art nouveau)
Time Period Aesthetics
/imagine prompt: [subject], 1920s aesthetic, art deco, vintage
/imagine prompt: [subject], 1980s aesthetic, synthwave, neon
/imagine prompt: [subject], Victorian era, steampunk, ornate
/imagine prompt: [subject], futuristic, 2050, sci-fi
Color Palettes
/imagine prompt: [subject], warm color palette, orange and red tones
/imagine prompt: [subject], muted colors, pastel palette
/imagine prompt: [subject], monochromatic, blue tones
/imagine prompt: [subject], complementary colors, vibrant
Common Use Cases
1. Professional Headshots
/imagine prompt: professional corporate headshot of a [age] year old [description], warm smile, confident expression, business attire, studio lighting, clean background, shot on Canon 5D Mark IV, 85mm portrait lens --ar 2:3 --s 50
2. Book Covers
/imagine prompt: book cover for a [genre] novel titled "[Title]", [mood/atmosphere], [key visual elements], cinematic, dramatic lighting, professional design --ar 2:3 --s 200
3. Social Media Content
/imagine prompt: Instagram-worthy [subject], aesthetic, trendy, [mood], vibrant colors, perfect for social media --ar 1:1
4. Marketing Materials
/imagine prompt: [product/service] advertisement, modern marketing, clean design, professional, [brand colors], minimalist --ar 16:9
5. Game Assets
/imagine prompt: [game element] for video game, concept art, game asset, detailed, stylized, [art style], RPG fantasy --s 100
6. Interior Design
/imagine prompt: [room type] interior design, modern [style], [colors], natural lighting, architectural digest, professional interior photography --ar 16:9
7. Fashion Design
/imagine prompt: fashion design, [garment type], [style], haute couture, fashion photography, runway, editorial --ar 2:3
Troubleshooting Common Issues
"This doesn't look like what I wanted"
Solutions:
- Be more specific in your prompt
- Use lower --stylize values (--s 50-100)
- Use --no to exclude unwanted elements
- Try multiple generations and iterate
Distorted Faces/Hands
Solutions:
- Specify "detailed hands" or "beautiful face"
- Use lower --chaos values
- Try portrait-specific prompts
- Use --no distorted, deformed, ugly
Image Too Busy/Cluttered
Solutions:
- Add "minimalist," "simple," "clean"
- Use "negative space"
- Reduce the number of elements in prompt
- Use --no clutter, busy
Colors Not Right
Solutions:
- Specify colors explicitly: "red dress," "blue sky"
- Use color palette terms
- Reference specific color systems
- Add color mood: "warm tones," "cool palette"
Wrong Style/Mood
Solutions:
- Add style keywords at the end
- Use artist/photographer references
- Specify lighting: "dramatic," "soft," "golden hour"
- Include medium: "oil painting," "photography," "digital art"
Best Practices
Do's
✅ Start with clear subject ✅ Add style keywords ✅ Use appropriate aspect ratios ✅ Iterate and refine ✅ Save prompts that work ✅ Use --no to remove unwanted elements ✅ Experiment with parameters
Don'ts
❌ Use very long prompts (keep it focused) ❌ Use conflicting styles ❌ Expect exact prompt matching ❌ Include copyrighted characters directly ❌ Ignore aspect ratio for composition
Frequently Asked Questions
Q: Is there a free version of Midjourney? A: No, Midjourney requires a paid subscription starting at $10/month.
Q: Can I use Midjourney images commercially? A: Paid subscribers own their images and can use them commercially. Check the terms for specifics.
Q: Why are some prompts blocked? A: Midjourney has content policies against certain content. Avoid explicit, violent, or problematic content.
Q: How do I get consistent characters? A: Use --cref with a reference image and --cw to control character consistency.
Q: Why don't hands/faces look right? A: AI still struggles with these. Use face-specific prompts and --no distorted to improve results.
Q: How do I improve image quality? A: Use upscaling (U buttons), specify "highly detailed," "8k," and use version 6.
Q: Can I edit generated images? A: Midjourney has limited editing. Use external tools like Photoshop or the Vary Region feature.
Conclusion
Midjourney mastery comes from understanding two things: descriptive prompting and parameter control. Start simple, learn what works, and gradually add complexity.
Key takeaways:
- Be specific but not overwhelming
- Use parameters strategically
- Iterate and refine
- Build a library of working prompts
- Experiment with styles and artists
The best Midjourney users aren't just technicians—they're creative directors who know how to communicate their vision. Practice regularly, study what works, and develop your own prompt style.
Happy creating!