Midjourney Prompt Guide: Parameters, Examples & Best Practices 2026

Understanding Midjourney Prompt Structure

Writing effective Midjourney prompts isn't about memorizing a formula—it's about understanding how the model interprets your words and translates them into visuals. Whether you're using Midjourney V6 or the newer V7, the core principles remain the same, though the execution differs.

The Basic Prompt Format

A Midjourney prompt typically consists of three parts:

1.Subject: What you want to see (e.g., "a woman in a red dress walking through a forest")
2.Style descriptors: How you want it to look (e.g., "cinematic lighting, photorealistic, 35mm film")
3.Parameters: Technical controls that adjust the output (e.g., --ar 16:9 --stylize 750)

Natural Language vs Keyword-Based Prompting (V6 vs V7)

Midjourney V7 represents a shift toward natural language understanding. While V6 worked well with comma-separated keyword lists, V7 responds better to complete sentences:

V6-style prompt:

"professional product photography, coffee mug, wooden table, morning light, sharp focus, high detail"

V7-style prompt:

"A professional product photograph of a coffee mug on a wooden table, lit by soft morning light with sharp focus and high detail"

Essential Midjourney Parameters

Parameters give you control over the technical aspects of image generation. They're added at the end of your prompt, always starting with --.

--ar (Aspect Ratio)

The --ar parameter sets the width-to-height ratio. Common ratios:

--ar 1:1 — Square format (Instagram feed, Pinterest)
--ar 16:9 — Widescreen (YouTube, presentations)
--ar 9:16 — Vertical (Instagram Stories, TikTok)
--ar 3:2 — Classic 35mm film photography

--stylize (Artistic Interpretation)

Controls how much creative freedom Midjourney takes with your prompt. Values range from 0 to 1000, with 100 as default.

Low (0-200)

Literal interpretation, good for product photography

Medium (200-500)

Balanced artistic touch while staying true

High (500-1000)

Strong creative liberties, striking results

--chaos (Variation Control)

Introduces unpredictability into the initial image grid. Values range from 0 to 100.

Use high chaos (50-100) for exploring different interpretations, low chaos (0-25) for consistent results when you have a specific vision.

--no (Negative Prompts)

Tells Midjourney what to exclude. Example: "A clean modern office" --no clutter, people, cables

Advanced Parameters for Professional Workflows

--cref (Character Reference)

Maintains character consistency across multiple images. Add --cref [URL] to your prompt with the URL of an image containing the character you want to maintain.

--sref (Style Reference)

Applies the style of a reference image to your new generation. Perfect for matching brand visuals or recreating specific artistic styles.

--tile (Seamless Patterns)

Generates images that tile seamlessly—useful for background patterns, website textures, and fabric designs.

Midjourney Style Keywords That Work

Photography Styles

Realism keywords:

photorealistic, ultra-realistic, photographed on 35mm film, shot on 50mm lens, raw photo

Lighting keywords:

golden hour, blue hour, studio lighting, dramatic lighting, soft light, backlighting

Art Mediums

Traditional:

oil painting, watercolor, pencil sketch, charcoal drawing, ink wash

Digital:

digital art, vector art, concept art, 3D render

Common Midjourney Prompt Mistakes to Avoid

Keyword Stuffing in V7

Don't: "professional, high quality, detailed, sharp, 8k, ultra HD, photorealistic, beautiful, stunning, amazing, masterpiece..."

Do: "Professional product photography with sharp focus and high detail"

Ignoring Aspect Ratio

Match aspect ratio to your content and intended use. Landscapes need 16:9, portraits need 2:3 or 4:5.

Not Using Negative Prompts

If Midjourney keeps adding unwanted elements, explicitly exclude them with --no parameter.

10 Practical Midjourney Prompt Examples

1. Professional Product Photography

"A premium leather wallet lying on a dark wooden surface, shot from above with shallow depth of field, natural window light from the side, product photography style, high detail on texture, no background clutter" --ar 4:5 --stylize 120 --no hands, people, shadows

2. Cinematic Portrait

"A contemplative elderly man sitting alone in a dimly lit café, looking out a rain-streaked window, warm ambient light from overhead lamps, cinematic lighting with gentle shadows, photographed on 35mm film, intimate atmosphere" --ar 16:9 --stylize 400

3. Fantasy Landscape

"A floating island with ancient ruins, cascading waterfalls disappearing into clouds below, bioluminescent plants along stone paths, dawn light breaking through morning mist, fantasy art style, rich color palette" --ar 16:9 --stylize 700

4. Architectural Visualization

"A modern minimalist living room with floor-to-ceiling windows overlooking a forest, white walls, light oak furniture, indirect natural lighting, architectural photography style, wide angle perspective, clean and uncluttered" --ar 16:9 --stylize 80 --no people, clutter

5. Food Photography

"A beautifully plated gourmet dish on a white ceramic plate, rustic wooden table, overhead angle, soft natural daylight from a nearby window, shallow depth of field, food photography style, fresh and appetizing presentation" --ar 4:5 --stylize 150

Midjourney V6 vs V7: Key Differences

V6

• Prefers comma-separated keywords
• "High quality" adds detail
• Good for established prompts

V7

• Better with natural language
• Intelligent quality interpretation
• Improved character consistency

Tools to Enhance Your Midjourney Workflow

Using Prompt Extraction Tools

When you see an AI-generated image you admire, understanding its underlying prompt can accelerate your learning. Prompt extraction tools analyze images and suggest prompts that could recreate similar results.

ImageChat's Prompt Seen tool lets you upload any AI-generated image and receive a prompt breakdown. This is particularly useful for learning, deconstruction, and getting starting points for your own prompts.

Try Prompt Seen Tool →

Frequently Asked Questions

Is Midjourney free in 2026?

No, Midjourney operates on a subscription model. Plans range from basic to professional tiers, with different features and usage limits. Check Midjourney's official website for current pricing.

How do I iterate on a result?

When you generate four images, you can reroll to generate four new variations, use V1-V4 to create variations of a specific image, upscale to increase resolution, or zoom out to expand the image beyond its original borders. Iterative refinement is key to achieving your desired result.

What's the difference between --stylize and --chaos?

--stylize controls how creatively Midjourney interprets your prompt within a single image. Higher values mean more artistic freedom. --chaos controls how different the four initial images are from each other. Higher values mean more variation across the grid.

Can I use AI-generated images commercially?

This depends on your Midjourney subscription tier and local laws. Generally, paid subscribers receive commercial rights for images they generate. However, always review Midjourney's current terms of service and consider consulting a legal professional for specific commercial applications.

Related Resources

• Prompt Seen Tool — Extract prompts from AI-generated images

• Image to Prompt Guide — Learn the reverse engineering process

• Chat with Images — Conversational image analysis