Midjourney V7 Complete Tutorial 2025: Master Parameters, Prompts & Advanced Techniques

Executive Summary

What's New in Midjourney V7

Release Timeline

• Launch: April 3, 2025
• Default Model: June 17, 2025 (V7 became default)
• Access: Available to all Midjourney users

Major Improvements Over V6

Image Quality:

• Text Rendering: Stunning precision (finally handles text correctly)
• Bodies & Hands: Coherent, realistic anatomy (previous weak point)
• Textures: Richer, more detailed surfaces
• Objects: Better coherence and accuracy

Technical:

• Faster generation in draft mode
• Better prompt following
• Improved parameter control
• More consistent outputs

When to Use V7 vs Earlier Versions:

• V7: Default for all new projects (best quality)
• V6: Only if you need specific V6 aesthetics
• V5/earlier: Not recommended (outdated)

Getting Started: Basic Midjourney V7 Workflow

Step 1: Access Midjourney

Via Discord:

1. Join Midjourney Discord server
2. Navigate to any #general or newbie channel
3. Type /imagine to start

Via Web (midjourney.com):

1. Log in with Discord account
2. Click "Create" button
3. Enter prompt in input field

Step 2: Your First V7 Prompt

Basic Structure:

/imagine [description] [parameters]

Example:

/imagine a cozy coffee shop interior, warm lighting, plants, Scandinavian design --ar 16:9 --v 7

Result: Midjourney generates 4 variations in ~60 seconds (standard mode) or ~6 seconds (draft mode).

Step 3: Refine Your Image

Options After Generation:

• U1-U4: Upscale images 1-4
• V1-V4: Create variations of images 1-4
• 🔄: Reroll (generate 4 new variations)
• Vary (Strong/Subtle): Create variations with more/less deviation

Pro Tip: Use draft mode (--mode draft) for initial iterations, then switch to standard mode for final upscale.

Essential Parameters: Your V7 Toolkit

1. --s (Stylize) - Control Artistic Interpretation

Range: 0-1000 (default: 100)

What It Does: Controls how much Midjourney adds its own artistic interpretation vs following your prompt literally.

Examples:

Low Stylize (--s 20):

/imagine a red apple on a white table --s 20

• Result: Literal, photographic, minimal artistic flourish

Medium Stylize (--s 100) [Default]:

/imagine a red apple on a white table --s 100

• Result: Balanced—realistic but aesthetically pleasing

High Stylize (--s 800):

/imagine a red apple on a white table --s 800

• Result: Highly artistic, abstract elements, dramatic lighting

When to Use:

• Low (0-50): Product photography, technical diagrams, literal interpretations
• Medium (50-200): General use, balanced creativity
• High (500-1000): Artistic projects, concept art, surreal imagery

2. --chaos - Control Variation Diversity

Range: 0-100 (default: 0)

What It Does: Determines how different the 4 generated images are from each other.

Examples:

Low Chaos (--chaos 0):

/imagine mountain landscape at sunset --chaos 0

• Result: 4 similar images with minor variations

Medium Chaos (--chaos 50):

/imagine mountain landscape at sunset --chaos 50

• Result: 4 distinctly different interpretations of the prompt

High Chaos (--chaos 100):

/imagine mountain landscape at sunset --chaos 100

• Result: 4 wildly different images, unpredictable variations

When to Use:

• Low (0-20): Consistency needed, minor refinements
• Medium (30-60): Exploring options, creative direction unclear
• High (70-100): Brainstorming, need radical diversity

3. --ar (Aspect Ratio) - Image Dimensions

Common Ratios:

• --ar 1:1: Square (1024×1024)
• --ar 3:2: Portrait (1536×1024)
• --ar 2:3: Vertical portrait (1024×1536)
• --ar 16:9: Landscape/widescreen (1792×1024)
• --ar 9:16: Vertical video (1024×1792)

Custom Ratios: Any ratio from 1:2 to 2:1

Examples:

Square for Instagram Post:

/imagine modern minimalist logo design --ar 1:1

Landscape for Desktop Wallpaper:

/imagine sci-fi cityscape, neon lights, cyberpunk --ar 16:9

Vertical for Mobile/Stories:

/imagine fashion model, elegant dress, studio lighting --ar 9:16

Pro Tip: Aspect ratio affects composition—Midjourney optimizes layouts for each ratio.

4. --seed - Reproducible Results

Range: 0-4294967295

What It Does: Sets a random seed for consistent base patterns across generations.

How to Use:

Step 1: Generate Image:

/imagine serene lake with mountains

Step 2: Get Seed (react with ✉️ emoji, Midjourney DMs you the seed):

Seed: 123456789

Step 3: Reuse Seed for Consistency:

/imagine serene lake with mountains, add a cabin --seed 123456789

Result: New image maintains similar composition/style to original.

When to Use:

• Creating consistent series (e.g., character designs)
• A/B testing prompt variations with same base
• Maintaining style across multiple images

Pro Tip: Seed only ensures consistency with same prompt and parameters. Change prompt significantly and seed effect diminishes.

5. --mode draft - 10x Faster Iteration

What It Does: Generates low-resolution images 10x faster at half the GPU cost.

Syntax:

/imagine [prompt] --mode draft

When to Use:

• Rapid prototyping (iterate quickly on composition)
• Testing prompt variations before final render
• Budget-conscious iteration (saves credits)

When to Switch to Standard:

• Final upscaling (always use standard mode)
• High-quality deliverables needed
• Fine details matter

Workflow:

1. Draft mode: Iterate on composition (--mode draft)
2. Standard mode: Final upscale (remove --mode draft)

6. --oref and --ow - Reference Image Control

New in V7: Omni-Reference system for precise reference image control.

Syntax:

/imagine [prompt] --oref [image URL] --ow [0-1000]

--oref: Input reference image URL

--ow (Omni-Weight): Controls reference influence (0-1000, default 100)

• 0: Ignore reference entirely
• 100: Balanced influence (default)
• 500: Strong reference adherence
• 1000: Maximum reference matching

Example:

With Reference Image:

/imagine futuristic sports car --oref https://example.com/car-reference.jpg --ow 300

Result: Car design heavily inspired by reference image styling.

When to Use:

• Matching specific aesthetics (reference photo/artwork)
• Maintaining brand consistency
• Recreating artistic styles

Pro Tip: Upload reference to Discord first, right-click → Copy Link to get URL.

7. --quality (or -q) - Render Quality vs Speed

Values: 0.25, 0.5, 1, 2 (default: 1)

What It Does: Balances render time vs quality (higher = more detail, slower).

Examples:

Fast Preview (--q 0.25):

/imagine urban street scene --q 0.25

• 4x faster, less detail (good for quick tests)

Balanced (--q 1) [Default]:

/imagine urban street scene --q 1

• Standard quality/speed

Maximum Quality (--q 2):

/imagine urban street scene --q 2

• 2x slower, maximum detail (rarely needed)

Recommendation: Stick with default (--q 1) unless specific need.

8. --no (Negative Prompt) - Exclude Elements

Syntax:

/imagine [prompt] --no [unwanted elements]

Examples:

Remove People:

/imagine bustling city square --no people

Remove Colors:

/imagine vibrant sunset landscape --no purple, pink

Remove Objects:

/imagine office interior --no computers, screens

Pro Tip: Be specific. "No clutter" is vague; "no cables, papers, objects on desk" is clear.

9. --tile - Seamless Patterns

What It Does: Creates seamless, tileable patterns (repeat infinitely).

Syntax:

/imagine [prompt] --tile

Example:

/imagine floral wallpaper pattern, vintage style --tile --ar 1:1

Use Cases:

• Wallpaper/background designs
• Fabric patterns
• Game textures
• Website backgrounds

10. Raw Mode - Photorealism & Control

Syntax:

/imagine [prompt] --style raw

What It Does: Removes Midjourney's default aesthetic, creates more photorealistic/literal outputs.

Example:

Without Raw:

/imagine portrait of a businessman

• Result: Artistic, stylized, "Midjourney aesthetic"

With Raw:

/imagine portrait of a businessman --style raw

• Result: Photographic, realistic, minimal artistic interpretation

When to Use:

• Photorealistic images
• Product photography
• Portraits requiring realism
• When default style is too "artistic"

Advanced Prompting Techniques

A. Prompt Structure Best Practices

Effective Prompt Formula:

[Subject] [Action/Pose] [Details] [Style/Mood] [Lighting] [Parameters]

Example:

/imagine 
an elderly Japanese craftsman 
carefully carving intricate patterns into wood
in a sunlit traditional workshop, wood shavings on floor
warm, nostalgic, documentary photography style
soft morning light through shoji screens
--ar 3:2 --s 150

Why It Works: Structured prompts give Midjourney clear hierarchy of importance.

B. Weight Modifiers (::)

Syntax: Use :: to emphasize or de-emphasize words.

Examples:

Equal Weight (Default):

/imagine hot chocolate

• Result: Likely beverage

Emphasize "hot":

/imagine hot::2 chocolate

• Result: Something melted or heated

Emphasize "chocolate":

/imagine hot chocolate::2

• Result: Chocolate-focused (maybe cake or dessert)

Multiple Concepts:

/imagine sunset::3 ocean::2 sailboat

• Sunset is primary focus (3x), ocean secondary (2x), sailboat background (1x)

C. Permutations - Batch Generation

Syntax: Use {} to generate multiple variations in one command.

Example:

/imagine a {red, blue, green} sports car --ar 16:9

Result: Generates 3 separate jobs:

1. A red sports car
2. A blue sports car
3. A green sports car

Advanced:

/imagine {modern, rustic, minimalist} {kitchen, bedroom, office} interior --ar 3:2

Result: 9 jobs (3 styles × 3 rooms)

When to Use: Testing variations systematically, batch content creation.

D. Remix Mode - Iterative Refinement

Enable: /settings → Toggle "Remix Mode"

What It Does: When creating variations (V1-V4), allows editing prompt.

Workflow:

Step 1: Generate Initial Image:

/imagine mountain landscape at sunset

Step 2: Click V1 (Variation)

Step 3: Edit Prompt (Remix mode active):

mountain landscape at sunset, add a cabin with smoke from chimney

Result: Variation of image 1, but with added cabin.

When to Use: Iterative refinement, adding elements to existing compositions.

Practical Workflows: From Concept to Final Image

Workflow 1: Concept Exploration → Final Render

Goal: Create a professional book cover illustration.

Step 1: Draft Mode Brainstorming (5 mins):

/imagine fantasy book cover, dragon and warrior, epic --mode draft --chaos 50 --ar 2:3

• Generate 3-4 batches, identify best direction

Step 2: Refine Composition (10 mins):

/imagine [refined prompt based on best draft], fantasy book cover, dragon breathing fire, warrior in foreground, epic battle scene --ar 2:3 --s 200

• Remove draft mode, add details

Step 3: Upscale & Finalize (5 mins):

• Select best image, click U1-U4 to upscale
• Use "Vary (Subtle)" if minor tweaks needed

Total Time: 20 minutes (vs hours of manual illustration)

Workflow 2: Character Consistency Series

Goal: Create 5 images of the same character in different poses.

Step 1: Generate Seed Image:

/imagine a young female elf ranger, green cloak, detailed fantasy portrait --ar 1:1 --s 150

Step 2: Extract Seed (React with ✉️):

Seed: 987654321

Step 3: Generate Series with Same Seed:

/imagine a young female elf ranger, green cloak, aiming a bow --seed 987654321 --ar 1:1 --s 150

/imagine a young female elf ranger, green cloak, sitting by campfire --seed 987654321 --ar 1:1 --s 150

[etc.]

Result: Consistent character across images (note: not perfect consistency, but much better than random).

Workflow 3: Style Matching with --oref

Goal: Create product images matching specific brand aesthetic.

Step 1: Prepare Reference:

• Upload brand reference image to Discord
• Copy image URL

Step 2: Generate with Reference:

/imagine [product description] --oref [brand-reference-URL] --ow 400 --ar 1:1

Step 3: Adjust --ow If Needed:

• Too different? Increase --ow (e.g., 600)
• Too similar? Decrease --ow (e.g., 200)

Common Mistakes & How to Avoid Them

Mistake 1: Overly Complex Prompts

Bad:

/imagine a beautiful sunset with orange and pink hues over a calm ocean with gentle waves and a sailboat in the distance with white sails and a lone seagull flying overhead with clouds that look like cotton candy and the sun partially hidden by the horizon creating a golden glow

Problem: Too many details overwhelm Midjourney, often ignored or conflicting.

Fix (Prioritize Key Elements):

/imagine sunset over calm ocean, sailboat with white sails, seagull flying, golden glow, cotton candy clouds --ar 16:9 --s 200

Rule: 20-30 words is sweet spot. Use parameters instead of lengthy descriptions.

Mistake 2: Ignoring Aspect Ratio

Bad:

/imagine vertical portrait of a person [generates 1:1 by default]

Problem: Square default cuts off vertical composition.

Fix:

/imagine vertical portrait of a person --ar 2:3

Mistake 3: Not Using Draft Mode for Iteration

Bad: Using standard mode for every test iteration (wastes credits and time).

Fix: Use --mode draft for first 5-10 iterations, then switch to standard for final render.

Mistake 4: Forgetting --no for Unwanted Elements

Bad:

/imagine serene lake landscape [often includes people]

Fix:

/imagine serene lake landscape --no people, buildings, boats

Mistake 5: Using Default --s for Everything

Problem: Default stylize (--s 100) may be too artistic for product photos or too literal for concept art.

Fix: Adjust per use case:

• Product photos: --s 20-50
• Concept art: --s 400-700
• General use: --s 100-200

Parameter Cheat Sheet (Quick Reference)

Pro Tip: Parameters can be combined freely (except conflicting ones like --mode draft + --q 2).

Tips from Midjourney Experts

Tip 1: Less is More

"Most beginners over-describe. Trust Midjourney's training—simple prompts often work better." - @midjourney_pro

Tip 2: Study the Gallery

Browse Midjourney community gallery, note prompts of images you like, reverse-engineer successful patterns.

Tip 3: Build a Prompt Library

Save successful prompts with notes about use cases. Build your personal style over time.

Tip 4: Iterate Systematically

Change ONE parameter at a time when refining. Changing multiple makes it hard to know what worked.

Tip 5: Use Descriptive Styles

Instead of "beautiful" or "amazing", use specific terms: "cinematic", "minimalist", "brutalist", "impressionist", "documentary photography".

Conclusion

Midjourney V7 is a powerful tool, but mastery requires understanding its parameter system. Focus on these priorities:

Week 1: Master --ar, --s, --chaos (covers 80% of use cases) Week 2: Learn --seed and --oref for consistency Week 3: Experiment with advanced techniques (weights, permutations, remix)

The 80/20 Rule: Good prompts + --ar + --s will solve 80% of your needs.

Tutorial Updated: 2025-10-14 | V7 Release: April 2025 | Verdict: Most capable AI image generator in 2025