Midjourney Style Modifiers: The Complete 2026 Reference

If you have ever typed a careful Midjourney prompt and gotten back something technically correct but completely wrong in feel, the gap is almost always Midjourney style modifiers — the words that tell the model not just what to draw but how it should look. This reference collects 200+ tested v7 modifiers, organized by category, with the small number of mixing rules that actually matter. Use it as a working palette, not a wishlist.

Style modifiers are the difference between "a woman in a city at night" and a frame that reads as Tokyo neon-noir shot on tungsten film. The subject barely changes. The modifier stack changes everything.

This guide is current for Midjourney v7, which became the default model on June 17, 2025 and remains the workhorse for most prompt libraries. It also notes where behavior shifts in the newer v8 alpha so your saved prompts keep working as the platform moves.

What are Midjourney style modifiers and how do they work?

Midjourney style modifiers are descriptive text phrases — like "cinematic lighting," "35mm film," or "art nouveau" — that you add to a prompt to control the aesthetic, medium, mood, and era of the generated image rather than its subject. They work because Midjourney's model has learned strong visual associations for these phrases from training data. The most reliable approach is to combine two or three modifiers from different categories (a medium, a lighting cue, and an optional mood or era) rather than stacking many similar words.

Mechanically, modifiers compete for influence with two other forces: your subject description and Midjourney's built-in house aesthetic. That third force is controlled by the --stylize parameter, whose default value is 100 on a scale from 0 to 1000. When you stack too many modifiers, you are asking the model to average competing visual signals, and the result is muddy. When you pick a clean, complementary set, the model has a clear target and renders it crisply.

Think of a prompt as having three layers:

1. Subject — what is in the frame ("a lighthouse on a cliff").
2. Modifiers — how it should look ("cinestill 800t, blue hour, anamorphic lens flare").
3. Parameters — technical controls (--ar 16:9, --style raw, --s 150).

This guide is mostly about layer two, with enough of layer three to keep the modifiers behaving.

How many style modifiers should you use per prompt?

Two to three. That is the single most important rule in this entire reference, so it goes first.

Past three modifiers, you stop steering and start blending. "Cinematic + dramatic + atmospheric + moody + epic" is five different ways of saying "make it feel important," and the model averages them into a generic, contrast-heavy mush that looks like every other AI image. A tight three-modifier stack — one medium, one lighting cue, one mood — gives the model a coordinate to render toward.

A clean stack looks like this:

• One medium / format anchor — e.g. 35mm film, oil on canvas, octane render
• One lighting cue — e.g. golden hour, chiaroscuro, neon noir
• One optional aesthetic, era, or mood — e.g. anamorphic lens flare, 1970s warm grain, melancholic

Avoid stacking from the same category. If you already wrote "cinematic," you do not also need "filmic," "movie still," and "dramatic lighting." Pick the strongest single word and spend your other two slots on a different dimension.

/imagine a fox sleeping in a meadow, 35mm film, golden hour, melancholic --ar 3:2 --style raw

That is three modifiers doing three different jobs. It will produce a consistent, intentional look. Add two more synonyms and it gets worse, not better.

The nine categories of style modifiers

Everything below is grouped into nine working categories. Pull from different categories when you build a stack; do not pull three from the same one.

The sections that follow expand each category into copy-pasteable phrases with notes on what each one actually does.

Which photography style modifiers are most reliable?

Photography modifiers are the most dependable in Midjourney because the training data is enormous and the associations are tight. Group them into four buckets: film stocks, cameras and formats, photographic lighting, and genres.

Film stocks

Film-stock names are some of the strongest modifiers in the whole system because each stock has a recognizable color science.

• 35mm film — generic film grain and color
• kodak portra 400 — warm, flattering skin tones, soft grain
• kodak gold 200 — saturated, nostalgic, slightly warm
• ilford hp5 — high-contrast black and white
• ilford delta 3200 — grainy, moody black and white
• fuji superia — cool greens, everyday-snapshot feel
• cinestill 800t — tungsten-balanced, red halation glow around lights
• polaroid — square format, faded, instant-film softness
• expired film — unpredictable color shifts and light leaks
• tri-x 400 — classic black-and-white reportage

Cameras and formats

• leica q3 — sharp, German-optics rendering
• hasselblad medium format — clean, high-resolution, commercial
• contax t3 — compact, punchy contrast
• mamiya 7 — medium-format film look with smooth tonality
• holga — toy-camera vignetting and softness
• disposable camera — flash-forward, casual, slightly blown out
• iphone photography — flat, modern, computational look
• drone aerial photography — top-down or high-angle perspective
• 4x5 large format — extreme detail, shallow plane of focus
• pinhole camera — soft, dreamy, long-exposure feel

Photographic lighting

Lighting is where most beginners under-spend their modifier budget. A single precise lighting word often does more than a paragraph of subject description.

• golden hour — warm, low-angle sun
• blue hour — twilight, cool tones
• harsh midday sun — high contrast, hard shadows
• candlelight — warm, intimate, flickering
• neon noir — saturated city lights on wet streets
• studio softbox — clean, even, commercial
• ring light — flat, frontal fashion light
• rim light — bright edge separating subject from background
• backlit — silhouette or glowing rim
• chiaroscuro — extreme light-to-shadow contrast
• diffused overcast light — soft, shadowless
• harsh flash — paparazzi, raw, direct
• mixed warm/cool light sources — color contrast within the frame

Photography genres

• fashion editorial photography
• street photography
• documentary photography
• environmental portrait
• candid lifestyle photography
• product hero photography
• food photography (overhead flatlay)
• food photography (moody side-lit)
• wedding photojournalism
• sports action photography

For any of these to read as a true photograph rather than a Midjourney painting, add --style raw and consider lowering --stylize. The raw style disables the model's automatic beautification so it follows your prompt more literally. More on that below.

Which cinema modifiers create a film look?

Cinema modifiers borrow the visual grammar of movies: aspect ratios, lenses, cinematographer signatures, and genre lighting. They are more abstract than film stocks, so pair them with a concrete lighting cue.

Cinematographer and director styles

• cinematic — broad, slightly contrasty, filmic grade
• shot on alexa — digital-cinema color and latitude
• anamorphic lens — 2.35:1 widescreen feel, oval bokeh
• anamorphic lens flare — horizontal blue streaks across the frame
• david fincher style — cool palette, precise, controlled framing
• wes anderson symmetry — dead-centered composition, pastel palette
• denis villeneuve atmospheric — vast scale, monochromatic, brooding
• roger deakins natural light — motivated, environment-sourced lighting
• emmanuel lubezki — natural light, immersive, flowing
• bradford young — warm, low-key, soulful

Cinematic lighting and mood

• cinematic lighting — general motivated lighting
• moody atmospheric
• film noir — high-contrast B&W, venetian-blind shadows
• neo-noir — color noir, neon-soaked
• horror lighting — low angle, top-lit, uneasy
• nordic minimalism — cool, sparse, restrained
• A24 cinematic aesthetic — intimate, naturalistic, slightly desaturated
• 70s cinematic — warm grain, soft focus
• 80s VHS aesthetic — saturated, scan lines, chroma bleed
• 90s indie film — handheld, raw, available light

Camera and lens (cinema)

• handheld camera shake
• steadicam smooth tracking
• dolly push-in
• crane shot
• crash zoom
• shallow depth of field
• wide establishing shot

Cinema modifiers pair beautifully with photography lighting. anamorphic lens flare + neon noir is a complete sci-fi-city look in two words.

Which illustration and painterly modifiers should you use?

Illustration is the broadest category and the one where artist references carry the most weight. The rule here is to always anchor a style with a medium so the model knows whether you want ink, watercolor, or vector.

Animation and anime styles

• studio ghibli — soft, pastoral, watercolor backgrounds
• makoto shinkai — luminous skies, melancholic light
• jean giraud / moebius — clean line art, surreal color
• mike mignola — heavy black ink, gothic shadow shapes
• pastel anime
• chibi style — cute, super-deformed proportions
• lineless illustration — soft, gradient-based forms
• ligne claire — Tintin-style clean, uniform lines
• ukiyo-e — Japanese woodblock print

Painterly and fine art

• oil on canvas
• acrylic painting
• watercolor and gouache
• alla prima oil sketch
• impressionist brushwork
• john singer sargent — confident, wet-into-wet portraiture
• vermeer — soft window light, quiet interiors
• caravaggio — dramatic single-source light
• rembrandt — warm gold light, deep shadow
• art nouveau (alphonse mucha) — decorative, flowing line, floral framing
• art deco — geometric, gold leaf, symmetry
• bauhaus — minimal, geometric, primary colors

Editorial and graphic illustration

• editorial illustration — conceptual, magazine-style
• christoph niemann — witty, minimal
• saul bass — mid-century minimalist poster
• malika favre — bold flat color, negative space
• comic book illustration
• jack kirby — kinetic action, dot shading
• frank miller — heavy ink, high-contrast noir
• manga style
• seinen manga — mature, detailed
• risograph print — limited palette, registration offset
• screen-printed texture
• letterpress print

Midjourney v7 became smarter at understanding style, which means recognizable artist signatures like Mucha and Sargent now render with more fidelity than they did in v6. Use that, but pair the name with your own subject so the output is inspired rather than imitative.

Which 3D and digital modifiers work best?

3D and digital modifiers fall into render engines, stylized digital aesthetics, and game/pixel looks. Render-engine names are surprisingly literal — the model associates each with a recognizable lighting and material quality.

Render engines

• octane render — glossy, high-contrast, product-viz lighting
• unreal engine 5 — real-time game cinematic look
• blender cycles — clean, physically based
• redshift render
• vray
• cinema 4d
• arnold render
• isometric 3d
• voxel art

Stylized digital aesthetics

• low poly stylized
• glitch art
• cyberpunk aesthetic
• vaporwave — pink and cyan, geometric, retro-future
• synthwave — neon grids, sunset gradients
• retro futurism
• y2k aesthetic
• pixar 3d animation
• disney 3d animation
• spider-verse animation style — comic halftone over 3D motion

Game and pixel

• 16-bit pixel art
• 32-bit pixel art
• 8-bit retro
• modern pixel art
• low poly geometric

A diorama formula that almost always works: octane render + isometric 3d + pastel palette. One render engine, one perspective, one color direction. Three categories, no overlap.

Which era and period modifiers add cohesion?

Era modifiers are powerful because a single decade implies a whole package of color, fashion, type, and texture. The danger is mixing eras: that produces incoherent output fast.

Modern subcultures and movements

These contemporary aesthetic labels also behave like era modifiers:

• bohemian
• dark academia
• cottagecore
• minimalist scandinavian
• maximalist eclectic

Within a single era, layering is fine — 1920s + art deco + flapper reinforces one period. Across eras, you get chaos: 1920s + 1980s + cyberpunk will not resolve into a coherent frame.

Which material, mood, and color modifiers fine-tune a look?

These three categories are your finishing touches. They rarely carry an image alone but they polish the stack.

Materials and textures

velvet, silk, linen, denim, leather (worn, patina), marble (polished), raw concrete, weathered wood, corrugated metal, frosted glass, crystal, iridescent material, holographic foil, chrome / metallic, liquid mercury, smoke / vapor, water droplets, dust particles, watercolor paper texture, cold-press paper texture.

V7 improved texture rendering, so material modifiers like velvet and marble have a stronger, more believable effect than in earlier versions.

Mood and atmosphere

dreamy, ethereal, melancholic, contemplative, foreboding, nostalgic, intimate, vast / epic, claustrophobic, serene, chaotic / kinetic, minimalist, maximalist, gritty / raw, polished / commercial.

Color palettes

• monochromatic blue
• duotone (specify two colors)
• pastel palette
• earth tones
• jewel tones
• neon palette
• muted desaturated
• warm gold + cool blue contrast
• analogous warm (reds + oranges + yellows)
• complementary (e.g. orange + teal)

A mood word plus a color palette is often the cheapest way to rescue a flat image. melancholic + muted desaturated reads completely differently from serene + pastel palette, even with an identical subject.

What's the difference between a style modifier and a style reference (--sref)?

This is the most consequential distinction in 2026 Midjourney prompting, so it deserves its own section.

A style modifier is text. You type cinematic lighting and the model interprets the words. It is flexible, fast, and portable, but it is also fuzzy — "cinematic" means something slightly different in every render.

A style reference is --sref followed by either an image URL or a saved style code. It points the model at a concrete aesthetic and reproduces it. Per the official Midjourney style reference docs, v7's system is "much smarter at understanding the style of an image" and far less prone to subject leakage than v6. You can also use --sref random to discover new style codes.

The strength of a reference is controlled by --sw (style weight). The default is --sw 100 on a range of 0 to 1000; --sw 0 cancels the reference entirely, low values (0-50) let the prompt dominate, and high values push the reference hard. Note that in v7, style weight has more impact with numeric style codes than with reference images.

The pro move is to combine them: use text modifiers for the broad look and a --sref code to lock the exact palette and rendering style across a series. That is how studios keep a campaign visually consistent across dozens of generations.

/imagine a cyberpunk street vendor, neon noir, cinestill 800t --sref 1234567890 --sw 200 --ar 16:9

If you maintain a library of these, store the working --sref codes alongside your text templates so you never lose a look you liked. A prompt manager with reusable {{variables}} makes this trivial — see how the Prompt Architects Midjourney workflow handles saved presets and variables.

How do the --stylize and --raw parameters change your modifiers?

Modifiers do not act alone. Two parameters constantly shape how they land.

--stylize (or --s) controls how heavily Midjourney applies its own house aesthetic. The default is 100 and the range is 0 to 1000. Low values keep the render literal and obedient to your prompt; high values produce more artistic, polished, "very Midjourney" images that may drift from your exact text. If your photography modifiers keep coming out looking like illustrations, your stylize is too high.

--style raw disables the automatic beautification. The parameter list documents raw mode as removing default styling for stronger prompt adherence. It is essential for photorealism, product shots, and any case where the painterly default is fighting you.

A practical settings cheat sheet:

For comparison, the v8 alpha encourages even heavier stylization, with the developers suggesting users lean into personalization and push --stylize 1000 when relying on the style systems. If you migrate prompts forward, expect to re-tune these numbers.

Which modifier pairings consistently produce great results?

These combinations have predictable, repeatable output. They are good defaults to keep in your library.

Notice that every pairing draws from different categories — a medium plus a lighting cue plus, occasionally, a color direction. None of them stack three synonyms.

What are the most common style-modifier mistakes?

Most disappointing Midjourney output traces back to one of these five mistakes.

1. Stacking five or more modifiers. "Cinematic + dramatic + atmospheric + moody + epic" averages to mush. Pick two or three.
2. Combining conflicting aesthetics. "Studio ghibli + david fincher" satisfies neither — soft pastoral watercolor and cold precise thriller are opposite poles.
3. Forgetting --style raw on photo prompts. Without it, Midjourney layers a painterly aesthetic over your photograph and kills realism.
4. Stacking same-era references that conflict. 1920s + art deco + flapper is coherent; 1920s + 1980s + cyberpunk is not.
5. Using an artist name as the only anchor. Pair it with a concrete medium (oil on canvas, 35mm film) and your own subject for stronger, more original results.

How should you build a personal modifier library?

Treating modifiers as a one-off typing exercise is the slow path. The professionals build a reusable system.

• Keep a tested-combinations list. Every time a pairing produces what you wanted, write it down with the parameters that worked. This becomes your private cheat sheet.
• Save winning templates with placeholders. A template like {{subject}}, {{film_stock}}, {{lighting}}, {{mood}} --style raw --ar 3:2 lets you swap one variable and regenerate instantly. The Prompt Architects image-enhancement workflow ships starter templates like this.
• Store your --sref codes next to your text. A look you can reproduce is worth ten you cannot.
• Use --seed plus the same modifiers for character or series consistency across multiple generations.
• A/B test one modifier at a time. Keep the prompt fixed, swap a single modifier, and compare. This is how you learn what each phrase truly contributes — and how you avoid carrying dead weight in your stacks.

If you frequently reverse-engineer looks you admire, pair this reference with a reverse-prompting workflow so you can extract the modifier stack from an existing image and rebuild it in your own style. And if you want a faster on-ramp to polished aesthetics, our roundup of aesthetic Midjourney prompts collects ready-made stacks you can adapt.

What changed in v7, and what's coming in v8?

Knowing what shifted between versions keeps your library from silently degrading.

In v7 (default since June 2025), the headline changes that affect modifiers are:

• --style raw is more aggressive at suppressing the house aesthetic. For photorealism, drop --s to 100-150.
• Texture rendering improved, so material modifiers (velvet, marble, frosted glass) have a stronger, more convincing effect.
• Some artist references shifted weight — Mucha and Sargent produce more recognizable output than they did in v6.
• Style references got smarter, with reduced subject leakage and a new --sref random option.

Looking ahead, Midjourney released a v8 alpha on March 17, 2026 with stronger personalization and a clear bias toward longer, more specific prompts and heavier reliance on style systems. The team explicitly encourages cranking --stylize and leaning on personalization profiles, srefs, and moodboards. Critically for anyone maintaining a library: Midjourney warns that old --sref style codes from earlier versions "may not produce the same styles anymore" after a major retrain. Re-test your saved codes and your most-used modifier stacks whenever a new model ships.

The portable lesson across every version is the same: modifiers are a moving target. Build your library so it is easy to re-test, keep your winning combinations documented, and never assume a phrase that worked last year still carries the same weight.

Frequently asked questions

How many style modifiers should I use per Midjourney prompt? Two to three maximum. Beyond three, the model averages between conflicting signals and produces muddy output. Pick modifiers that combine productively (e.g. 35mm film + golden hour) rather than stacking similar ones (cinematic + dramatic + atmospheric).

What is the difference between a style modifier and a style reference (--sref)? A style modifier is a text phrase like cinematic lighting typed into your prompt. A style reference (--sref code or URL) points Midjourney to an actual image or saved style code whose aesthetic you want to match. Per Midjourney's docs, --sref is more precise and repeatable; text modifiers are more flexible. Combining both gives the tightest control.

Why do some style modifiers stop working after a Midjourney update? Midjourney retrains its underlying model with each major version, so a phrase that worked in v5 can produce different output in v7 or v8. Midjourney explicitly warns that old --sref style codes "may not produce the same styles anymore" after the v7 retrain. Re-test your modifier library whenever a major version ships.

What does the --stylize parameter do and what is its default? --stylize (or --s) controls how strongly Midjourney applies its own house aesthetic on top of your prompt. The default is 100 and the range is 0 to 1000. Lower values stick closer to your literal prompt; higher values produce more polished, artistic, "Midjourney-looking" images that may drift from your text.

Should I name specific artists in Midjourney prompts? Yes, artist-style references still work as aesthetic anchors in v7 (alphonse mucha, roger deakins, wes anderson). Use them as a style direction, not exact replication, and pair the name with a concrete medium (oil on canvas, 35mm film) and your own subject so the output is on-brand rather than a copy.

Do these modifiers work in Niji 6, v6, and the v8 alpha? Most transfer. Niji 6 carries a heavier anime default; v6 renders slightly more saturated and painterly than v7 with the same words; and the v8 alpha rewards longer, more specific prompts and heavier use of style references and personalization. Test on the exact version you ship on, because modifier weights drift between models.

What is --style raw and when should I use it? --style raw disables Midjourney's automatic "beautification," so the model follows your prompt more literally. Use it for photorealism, product shots, and any time the default painterly aesthetic is fighting your intent. Combine raw mode with a lower --stylize value (around 100-150) for the most accurate, true-to-prompt photographic results.

By Nafiul Hasan — Founder of Prompt Architects, where we build prompt-enhancement tooling tested across ChatGPT, Claude, Gemini, Midjourney, Veo 3, and Kling. Last updated: June 10, 2026.