/imagine a sunset over mountains with dramatic clouds

Within 60 seconds, Midjourney generates four image variations based on your description.

The Grid and Upscaling

Your initial result is a 2x2 grid of variations. Below each grid, you'll find:

• U1-U4: Upscale individual images to higher resolution
• V1-V4: Create variations of specific images
• 🔄: Regenerate entirely new options

Settings and Parameters

• Model version (V6 is current as of 2026)
• Style settings (RAW mode for more literal interpretations)
• Variability (how different the four options are)
• Generation speed (Fast vs. Relaxed mode)

The Art of Prompting

Prompting is where mastery begins. A well-crafted prompt is worth thousands of random attempts.

Prompt Structure

Think of prompts as having components:

[Subject] + [Style/Medium] + [Environment/Setting] + [Lighting] + [Camera/Composition] + [Parameters]

Example Evolution:

Basic: "a cat"

Better: "a fluffy orange tabby cat sitting on a windowsill"

Advanced: "a fluffy orange tabby cat sitting on a Victorian windowsill, golden hour sunlight streaming through lace curtains, oil painting style, warm tones, intimate composition"

Master: "a fluffy orange tabby cat sitting on a Victorian windowsill, golden hour sunlight streaming through antique lace curtains casting intricate shadows, Johannes Vermeer style oil painting, warm amber and cream palette, intimate domestic scene, soft diffused lighting, visible brushwork, museum quality, --ar 4:5 --style raw --s 250"

Style and Medium Descriptors

Specifying artistic style dramatically affects output:

Traditional Art Styles:

• Oil painting
• Watercolor
• Charcoal sketch
• Pencil drawing
• Pastel illustration
• Ink wash painting
• Fresco
• Gouache

Digital and Modern:

• Digital art
• 3D render
• CGI
• Vector illustration
• Low poly art
• Voxel art
• Glitch art

Photography Styles:

• Portrait photography
• Street photography
• Fashion editorial
• Documentary style
• Fine art photography
• Product photography
• Architectural photography

Artist References:

Mentioning artists can guide style:

• "in the style of Studio Ghibli"
• "inspired by Monet"
• "reminiscent of Blade Runner cinematography"
• "HR Giger aesthetic"

*Note: Artist style references are a nuanced ethical area—use thoughtfully.*

Lighting Vocabulary

Lighting transforms images. Master this vocabulary:

Natural Light:

• Golden hour
• Blue hour
• Overcast diffused light
• Dappled sunlight through trees
• Harsh midday sun
• Moonlit

Artificial Light:

• Neon lights
• Candlelight
• Fluorescent
• Bioluminescent
• Volumetric light
• Rim lighting
• Backlighting

Atmospheric:

• God rays
• Dramatic chiaroscuro
• Soft ambient occlusion
• Light fog
• Atmospheric haze

Camera and Composition Terms

Shots:

• Extreme close-up
• Close-up
• Medium shot
• Full body shot
• Wide shot
• Aerial/bird's eye view
• Worm's eye view

Lens Effects:

• Shallow depth of field
• Bokeh
• Macro photography
• Fisheye distortion
• Tilt-shift miniature effect
• Long exposure motion blur

Composition:

• Rule of thirds
• Centered composition
• Symmetrical
• Leading lines
• Negative space
• Frame within frame

Essential Parameters

Parameters (added with --) give you technical control:

Aspect Ratio (--ar)

--ar 16:9 # Widescreen/landscape

--ar 9:16 # Vertical/mobile

--ar 1:1 # Square

--ar 4:5 # Portrait

--ar 3:2 # Standard photo

--ar 21:9 # Ultrawide/cinematic

Stylize (--s or --stylize)

Controls how strongly Midjourney applies its aesthetic:

--s 0 # Most literal to prompt

--s 100 # Default balance

--s 250 # More artistic interpretation

--s 750 # Maximum stylization

Lower values for precise control; higher for artistic flourishes.

Chaos (--c or --chaos)

Controls variation between the four outputs:

--c 0 # Very similar results

--c 50 # Moderate variation

--c 100 # Maximum diversity

Use high chaos for exploration, low for refinement.

Quality (--q or --quality)

Affects detail and generation time:

--q .25 # Fastest, least detail

--q .5 # Fast, reduced detail

--q 1 # Default quality

Use lower quality for rapid iteration, full quality for finals.

Weird (--w or --weird)

Introduces unexpected creative elements:

--w 0 # Normal

--w 250 # Subtle quirks

--w 1000 # Very unconventional

--w 3000 # Maximum weirdness

Great for breaking creative blocks and discovering unexpected directions.

No (--no)

Excludes elements:

--no text, watermark, signature

--no people, crowds

--no blue, red

Useful for removing unwanted elements Midjourney tends to add.

Tile (--tile)

Creates seamlessly tiling patterns:

/imagine geometric art deco pattern gold and black --tile

Perfect for textures, wallpapers, and fabric designs.

Seed (--seed)

Each generation has a unique seed number. Specifying a seed recreates conditions:

--seed 12345

Useful for:

• Consistent style across series
• Controlled experimentation (change one element, keep seed)
• Reproducing previous results

Find seeds using the envelope emoji reaction on generated images.

Advanced Techniques

Image Prompting

Use existing images as references:

/imagine [image URL] a portrait in this style but with red hair

Image weight controls influence:

--iw 0.5 # Subtle influence

--iw 1 # Default

--iw 2 # Strong influence

Multi-Prompts

Combine concepts with weighted emphasis:

/imagine space:: 3 underwater:: 1 surreal dreamscape

Higher weights = more influence. Great for hybrid concepts.

Permutations

Generate multiple variations efficiently:

/imagine a {red, blue, green} dragon in {forest, desert, ocean}

Creates 9 jobs covering all combinations.

Remix Mode

Enable Remix in settings to modify prompts while keeping image structure:

1. Generate an image
2. Click a V (variation) button
3. Modify the prompt in the popup
4. Get variations with new elements but similar composition

Describe Command

Upload an image and get prompt suggestions:

/describe [upload image]

Returns four prompt approximations—excellent for reverse-engineering styles.

Blend Command

Combine multiple images:

/blend [image1] [image2]

Creates fusion results. Add more images (up to 5) for complex blends.

Creating Consistent Characters and Styles

One of Midjourney's challenges is consistency. Here are techniques for coherent series:

Character Reference Sheets

Create reference sheets first:

/imagine character design sheet, a cyberpunk hacker woman with blue mohawk, multiple angles and expressions, white background, concept art style --ar 16:9

Then reference this in subsequent images.

Style Anchoring with Seeds

1. Generate an image with your desired style
2. Note the seed number
3. Use that seed with new prompts for consistent aesthetics

Detailed Character Descriptions

Create a consistent character prompt template:

[Action/pose], a 25-year-old woman with pale skin, long wavy auburn hair, green eyes, freckles, wearing a vintage 1940s dress, [setting], [style parameters]

Reuse this exact description across images.

The Suffix Technique

Create consistent style by appending identical style descriptors:

/imagine [unique scene] || highly detailed digital art, dramatic lighting, cinematic composition, 8k resolution, trending on artstation --ar 16:9 --s 200

The elements after || remain constant across your series.

Workflow Optimization

Rapid Iteration Process

1. Explore (--c 50, --q .5): Generate quick, varied options
2. Refine: Find promising directions, create variations
3. Upscale: Select best results for high-resolution output
4. Vary Region: Use inpainting for targeted improvements

Prompt Libraries

Maintain organized prompt templates:

Lighting Presets:

GOLDEN_HOUR = "golden hour sunlight, warm tones, long shadows"

NEON_NIGHT = "neon lights, cyberpunk, nighttime, rain-slicked streets"

STUDIO = "studio lighting, white background, professional photography"

Style Presets:

CINEMATIC = "cinematic composition, anamorphic lens, 35mm film, movie still"

FANTASY = "fantasy art, highly detailed, magical atmosphere, ethereal"

MINIMAL = "minimalist, clean lines, negative space, simple composition"