PromptOrix

Midjourney produces some of the most aesthetically stunning AI-generated images available. It also has a learning curve that trips up most newcomers: its prompt syntax is different from natural language tools, its parameters change the output dramatically, and knowing which descriptors actually work versus which ones look good in a prompt but do nothing is knowledge that takes trial and error to build.

This guide shortcircuits that process. You get 80+ tested prompts across every major category, a parameter cheatsheet, an art style reference table, and the specific language patterns that consistently produce exceptional output in Midjourney v6.

How Midjourney Prompting Differs from Other Tools

Midjourney does not read paragraphs. It reads a weighted list of descriptors. The key differences from tools like DALL-E:

• Use commas, not sentences. “A warrior, ornate armour, standing in ruins, dramatic lighting, cinematic” outperforms the sentence equivalent.
• Parameters go at the end after –. Aspect ratio, version, stylize level, and quality all go after your descriptors separated by –.
• Front-loading matters. Midjourney weights early tokens more heavily. Put your subject and the single most important quality first.
• :: assigns weight. warrior::2 ruins::1 makes the warrior twice as prominent as the ruins in the composition.
• Negative prompts use –no. --no text, watermarks, blur excludes those elements without a separate field.

Parameter Cheatsheet

Parameter	What It Does	Common Values
–v 6	Use Midjourney version 6 (current best)	–v 6
–ar	Aspect ratio	–ar 16:9, –ar 1:1, –ar 2:3, –ar 9:16
–s or –stylize	How strongly MJ applies its aesthetic. Higher = more artistic	–s 0 (literal) to –s 1000 (very stylised); default 100
–q or –quality	Render quality and detail level	–q .5 (fast), –q 1 (default), –q 2 (max detail)
–chaos	Variation in the 4-image grid	0–100; higher = more diverse results
–no	Negative prompt — what to exclude	–no text, watermark, blur, ugly
–seed	Fixes randomness for reproducible results	Any integer
–iw	Image prompt weight when using a reference image	0.5–2; higher = closer to reference
–tile	Creates seamless repeating pattern	–tile

Photorealistic Prompts

Street documentary

candid street photograph, elderly man selling newspapers 
at a rain-soaked corner in 1970s New York, steam rising from grates, 
yellow cabs blurred in background, Kodak Tri-X film, 
Garry Winogrand style, decisive moment, 
high contrast black and white --ar 3:2 --v 6 --s 200

Golden hour landscape

Icelandic volcanic landscape at magic hour, obsidian lava fields, 
lone snow-capped volcano glowing amber against deep purple stormclouds, 
thin river of molten light on the horizon, 
ultra-wide shot, Hasselblad X2D, long exposure, 
photorealistic, award-winning landscape photography --ar 16:9 --v 6 --q 2

Underwater world

sunlit coral reef 15 metres below surface, 
bioluminescent jellyfish drifting through shafts of turquoise light, 
school of parrotfish passing a brain coral, 
Nikon D850 underwater housing, fisheye lens, 
National Geographic quality, hyperrealistic --ar 16:9 --v 6 --s 300

Atmospheric interior

small Italian coffee bar at 6am, elderly barista 
pulling an espresso for the first customer of the day, 
marble counter worn smooth by decades of use, 
dawn light through half-shuttered windows, 
cigarette smoke, grain, warmth, 
Magnum photographer aesthetic, 35mm film --ar 3:2 --v 6 --s 150

Portrait Prompts

High-fashion editorial

high fashion editorial portrait, woman with severe cheekbones, 
dramatic sculptural headpiece in white ceramic and gold, 
black couture gown, stark white studio background, 
Annie Leibovitz for Vogue, three-point studio lighting, 
8K, razor sharp --ar 2:3 --v 6 --s 400 --q 2

Character study

environmental portrait of a 70-year-old ship captain, 
weathered tan skin, sharp blue eyes, white beard, 
standing at a harbour at dusk, 
trawlers in background, worn navy peacoat, 
Rembrandt lighting, genuine expression, 
Steve McCurry style, rich tones --ar 2:3 --v 6 --s 250

Ethereal studio portrait

ethereal studio portrait, woman, translucent skin, 
surrounded by slowly drifting white flower petals, 
long exposure motion blur on petals, sharp face, 
soft diffused key light, pale mint and white palette, 
fine art photography, Tim Walker aesthetic --ar 2:3 --v 6 --s 500

Fantasy and Concept Art

Epic battlefield

epic fantasy battlefield at dawn, 
armies of thousands clashing on a fog-covered plain, 
a lone armoured knight on a white horse watching from a ridge above, 
dramatic volumetric light breaking through stormclouds, 
Greg Rutkowski, cinematic composition, 
oil painting quality, highly detailed, 8K --ar 21:9 --v 6 --q 2

Ancient city ruins

abandoned ancient city swallowed by jungle, 
stone temples covered in vines and moss, 
shafts of golden light through forest canopy, 
a lone explorer with a lantern at the entrance to a vast dark doorway, 
atmospheric perspective, matte painting quality, 
Tomb Raider concept art aesthetic --ar 16:9 --v 6 --s 300

Magical creature

a vast ancient dragon perched on a storm-lashed cliff at night, 
scales like obsidian reflecting lightning, 
wingspan filling the sky, 
bioluminescent markings along its spine, 
epic scale, impossibly detailed, 
cinematic wide shot, Charlie Bowater meets Weta Workshop --ar 21:9 --v 6 --q 2

Anime and Illustration

Ghibli-style scene

Studio Ghibli style background art, 
summer afternoon, small harbour town on a Mediterranean cliff, 
whitewashed buildings with terracotta roofs, 
laundry drying in warm wind, 
turquoise sea below, painterly hand-drawn quality, 
soft golden light, nostalgic and magical --ar 16:9 --v 6 --s 600

Anime protagonist

anime key visual, female protagonist, 
silver hair with twin tails, red eyes, school uniform with tactical gear overlay, 
standing on a rooftop at night, neon city below, 
dynamic pose, wind in hair, 
Makoto Shinkai lighting style, 
hyperdetailed illustration, 4K --ar 2:3 --v 6 --s 500

Cozy slice-of-life

cozy anime illustration, 
small second-hand bookshop interior on a rainy afternoon, 
warm lamp light, a tabby cat sleeping on a pile of books, 
rain on the window, cups of tea steaming, 
Kyoto Animation aesthetic, 
inviting and peaceful, highly detailed background --ar 4:3 --v 6 --s 400

Architecture and Interior

Brutalist monument

aerial view of a monumental brutalist government building, 
raw exposed concrete, geometric repetition, 
vast empty plaza in front, 
overcast sky creating flat even light, 
architectural photography, 
Le Corbusier meets Soviet Constructivism, 
black and white, extreme detail --ar 16:9 --v 6 --s 150

Futuristic living space

interior of a luxury apartment in 2060, 
floor-to-ceiling smart glass walls overlooking a vertical forest city, 
biomorphic furniture in warm whites and natural wood, 
a living moss wall, diffuse ambient lighting, 
Zaha Hadid Studio meets Kengo Kuma, 
architectural visualization quality --ar 16:9 --v 6 --q 2

Landscape and Nature

Autumn forest path

narrow path through an ancient beech forest in peak autumn, 
cathedral-like canopy of gold and amber above, 
fallen leaves carpeting the ground, 
morning mist in the distance, 
single shaft of sunlight on the path, 
Eliot Porter nature photography style, 
medium format quality, rich colours --ar 2:3 --v 6 --s 200

Minimalist snow scene

minimalist winter landscape, 
single bare black tree at the centre of a vast white snowfield, 
pale grey sky, no horizon visible, 
absolute silence implied, 
negative space as the dominant element, 
Japanese ink painting sensibility meets fine art photography, 
monochrome --ar 1:1 --v 6 --s 600 --no colour

Art Style Reference Prompts

Style Reference	Prompt Addition	Effect
Impressionism	Monet style, loose brushwork, dappled light, plein air painting	Soft, textured, light-focused
Art Nouveau	Alphonse Mucha style, flowing organic lines, decorative borders, botanical motifs	Ornate, elegant, poster-like
Ukiyo-e	Japanese woodblock print, Hokusai style, flat colour, bold outlines, traditional	Flat graphic, cultural depth
Baroque	Caravaggio chiaroscuro, dramatic shadows, religious grandeur, oil painting texture	High drama, extreme contrast
Cyberpunk	neon-lit rain-soaked city, teal and magenta palette, dystopian, Syd Mead concept art	Urban, futuristic, moody
Surrealism	Salvador Dali surrealism, dreamlike, impossible physics, melting objects, fine detail	Unsettling, inventive
Watercolour	delicate watercolour illustration, translucent washes, white paper showing through, botanical art	Soft, delicate, artisanal
Pop Art	Andy Warhol pop art, bold flat colours, halftone dots, silkscreen aesthetic	Bold, graphic, commercial

Product and Commercial

Luxury watch advertisement

luxury watch advertisement, 
close-up of a stainless steel dress watch on a black velvet surface, 
single spotlight creating a reflection below, 
hands set to 10:10, 
Rolex or Patek Philippe aesthetic, 
commercial photography quality, extreme detail on dial texture, 
black background --ar 1:1 --v 6 --q 2 --s 100

Skincare brand visual

minimalist beauty brand campaign image, 
frosted glass serum bottle surrounded by fresh botanicals, 
dewy water droplets on the glass, 
soft north window light, 
white and sage green palette, 
Aesop brand aesthetic, product photography --ar 4:5 --v 6 --s 200

Common Midjourney Mistakes

Using natural language sentences

Midjourney is not ChatGPT. “A photograph of a beautiful sunrise over mountains taken with a professional camera in warm light” produces weaker results than “sunrise, mountain range, golden hour, long shadows, Hasselblad, photorealistic, –ar 16:9 –v 6.” Comma-separated descriptors outperform connected sentences every time.

Ignoring –stylize

The default stylize value is 100. Low values (0-50) produce more literal, less aesthetically polished results. High values (500-1000) introduce Midjourney’s aesthetic taste aggressively. For most creative work, 200-400 is the sweet spot. Always try your prompt at two or three stylize levels before settling on an output.

Not using –no for quality exclusion

The –no parameter is your quality filter. --no text, watermark, blur, low quality, extra limbs, bad anatomy should be added to almost every serious prompt. Most users skip it and then manually regenerate images that failed because of predictable quality issues.

Frequently Asked Questions

What is the difference between Midjourney and DALL-E for prompting?

Midjourney uses comma-separated descriptors with parameters and produces more consistently artistic, aesthetically polished output. DALL-E handles natural language sentences better and excels at photorealism, accurate text rendering, and following precise spatial instructions. For stunning art, Midjourney leads. For precise, instruction-following generation, DALL-E is often more reliable. See our best AI image prompts guide for a full platform comparison.

Do these prompts work on Midjourney v5 as well as v6?

Yes, though v6 produces significantly better photorealism and handles longer, more detailed prompts more accurately than v5. Change –v 6 to –v 5.2 for older-style output if you prefer v5’s aesthetic. For image generation specifically using ChatGPT’s DALL-E integration, see our ChatGPT image prompts guide.

How do I get consistent characters across multiple images in Midjourney?

Use the –cref (character reference) parameter with an image URL of your character: --cref [image URL] --cw 100. The –cw value controls how closely the character features are followed (0-100). This feature is v6-specific and significantly more reliable than trying to describe the same character consistently in text.

Keep Building

The principles behind great Midjourney prompts — specificity, style anchors, lighting detail — are the same principles that make any AI image prompt work. Our complete AI image prompts guide covers these across every platform. For Stable Diffusion-specific syntax, negative prompting, and checkpoint selection, the Stable Diffusion prompts guide goes deep on everything SD-specific. And for prompts across every use case beyond images, the ChatGPT prompt mega-list is the reference to keep open.

More AI image prompt libraries at Promptorix — AI prompts for everything.