AI Style Infrastructure

Every AI output, in your Style.
Defined once, use it everywhere.

Your style, captured as 160+ structured blocks across color, type, mood, and voice. Paste it into any AI tool.

Extract My Style — FreeNo account needed · start in seconds

Reference: a stack of rainbow plates and two mugs on a yellow floor against a teal wall

Reference: a black road bike in front of flat yellow and blue colour fields

Reference: a woman in a cobalt shirt and mustard trousers between flat red and blue walls on a yellow floor

STYLE REFERENCE — HARD CONSTRAINT (HIGHEST PRIORITY)STYLEREF.IO

Colour-Blocked Studio Set

PHOTOGRAPHY · 1:1 · SQUARE-ON · 101 STYLE ITEMS · 14 SECTIONS

#FF6BCE#00B5B8#CB997E#006EFF#000000#FFC800#61C200#FF8000#FF0000

Output Format

Creative discipline Photography
Digital
Delivery mode Still
Ratio 1:1 (Square)
Cropping Center

Mood

Bold, Clean, Confident, Playful
Minimalist
Energy Vibrant
Emotional bias Positive
Real-world accuracy Precise
Minimal narrative
Characters Allowed (detailed)

Colors

3–5 colors
Saturation Vivid
Hue contrast
Hue relationships Complementary
No gradients
Color dominance Balanced

Light & Shadow

Lighting mood style Studio-flat
Key light style Diffused
Artificial
Light source count Multiple
Shadow type Soft
Shadow density Light
Fill-light ratio Flat fill
Exposure bias Balanced
Lighting presence Explicit
Color grading style Natural skin
colour comes exclusively from physical painted set surfaces, never coloured light or gels

Composition

Flat perspective
Single focal structure
Hierarchy Strong
Symmetry Near symmetry
Balance Center-weighted
Layout axis Horizontal
Plane separation Color/value shift
Visual density Airy

Surface & material

Material simulation Paper
Surface treatment Matte
Texture Textureless painted paper and board fields
sharp natural detail reserved for skin, hair, fabric, and props
Texture usage Photographic
Transparency Opaque
Imperfection level Clean
Stylized realism
Micro-detail allowance Allowed
Micro-detail types Visible skin texture, Fabric weave realism, Natural imperfections

Camera

Full shot
Eye level
DOF Deep focus
Lens type Standard
Focal length feel Normal
Point of view Objective
Camera movement Static
Motion implication Static

Shape Language

Geometric
Primitive dominance Rectangles
Edge sharpness Razor-sharp
Silhouette clarity Recognizable
Negative space usage Dominant
Angle usage Limited

Post-processing

Tone curve Linear
Sharpening Crisp
Image treatment Full-color

Background

Background type Solid color
Environment mood Studio
Background–Subject Integrated
Background complexity Minimal
Prop styling Curated

References

De Stijl, Bauhaus, Mid-Century Modern
Memphis Design
Temporal reference Contemporary
Reference Brands Mid-century advertising
Memphis-era studio still life

Guardrails

Avoid Lens Flare, Heavy Bokeh / Blur, Film Grain / Noise, HDR over-processing, Over-saturation
Stylistic filters
Avoid Visuals coloured gel lighting
coloured rim light
light spill
blended grounds
vignetting
theatrical shadows
busy prop clutter
dark unlit gaps
subject fully camouflaged against background
Allowed Visuals matte painted flats
seamless paper
uniform saturated colour planes
clean hard seams
square-on compositions
soft attached grounding shadow
shape-led props
natural skin texture

Flat-Value Rule

Every colour field holds one uniform edge-to-edge value
no falloff, gradient, or edge brightening
Seam Composition Wall-floor joins and flat-to-flat seams act as deliberate horizontal or vertical compositional lines
Camera Alignment Square-on camera
built planes remain parallel to the sensor and frame edges
Subject Integration Wardrobe and props continue the set palette while retaining enough separation for the subject to read clearly
Naturalistic Exception Skin is the sole untreated naturalistic surface against the constructed flat-colour set

READ FROM 3 REFERENCESPASTE INTO ANY AI TOOL

Add it to any prompt.

Paste your StyleRef into the AI tool you already use, then prompt anything. Three different prompts — one style.

Walking boots with trailing laces on a yellow floor against a cobalt wall — Prompt 1A pair of walking boots side by side with the laces untied and trailing.

A border collie sitting upright in front of a teal and yellow split wall on a blue floor — Prompt 2A border collie sitting upright and facing the camera with its ears forward.

A tan swivel office chair with a green jacket, against flat blue and red walls — Prompt 3A swivel office chair turned three-quarters toward the camera with a jacket over its back.

StyleRef addedColour-Blocked Studio SetYour AI tool

Generate

Extract My Style — Free

Consistent style in generated images.

Add it to any prompt.

Paste your StyleRef into the AI tool you already use, then prompt anything. Three different prompts — one style.

StyleRef addedColour-Blocked Studio SetYour AI tool

A swivel office chair turned three-quarters toward the camera with a jacket over its back.

Generate

Extract My Style — FreeConsistent style in generated images.

01One spec, every tool

Define Style Once.
Use Everywhere.

A StyleRef is your style, written down as rules — exact hex values with proportions, named type classes, light, composition and an avoid list — marked as a hard constraint so the model reads rules instead of a mood.

You build it once, from a reference or by hand, and reuse it in every chat, every week, every tool.

Model-Optimized
Copy-Paste Portable
STYLE.md Export

Extract My Style — Free

What you paste

STYLE REFERENCE — HARD CONSTRAINT (HIGHEST PRIORITY)

This style must be applied to ALL generated content (text, image, video).
If any part of the request conflicts with this style, the style OVERRIDES the request.

Colour-Blocked Studio Set
 • Output Format: Photography; Digital; Still; 1:1 (Square); Cropping: Center.
 • Mood & Personality: Bold, Clean, Confident, Playful, Minimalist; Energy: Vibrant.
 • Colors: Solid Colors: [#ff6bce 11%, #00b5b8 11%, #CB997E 11%, #006eff 11% …]; 3–5 colors; Vivid; No gradients.
 • Light & Shadow: Studio-flat; Diffused; colour comes exclusively from physical painted set surfaces, never coloured light or gels.
 • Guardrails: Avoid gel lighting, vignetting, blended grounds, busy prop clutter …

ChatGPT
Claude
Gemini
Midjourney
Flux

160+ Style Blocks across 20 sections

The machine above filled 101 of them from three images.

01Output Format
02Mood & Personality
03Colors
04Typography
05Light & Shadow
06Voice & Language
07Spatial Structure
08Camera & Motion
09Shape Language
10Container & Boundary
11Line & Stroke System
12Surface & Material
13Post-processing
14Background & Environment
15Artistic Mediums
16UI & Web
17References
18Inspiration
19Guardrails
20Custom Style Items

What each section covers

Output FormatMedium, platform, aspect ratio, creative discipline
Mood & PersonalityEnergy, emotional bias, narrative, humor register
ColorsPalette with hex codes, roles, weights, gradients
TypographyFont family, weights, hierarchy, case, scale
Light & ShadowLighting mood, key light, atmosphere, film grading
Voice & LanguageFormality, vocabulary, sentence length, pacing
Spatial StructurePerspective, focal structure, layout axis, density
Camera & MotionShot size, angle, DOF, lens, film stock, movement
Shape LanguageShape basis, primitives, edges, silhouette
Container & BoundaryContainer shape, corners, padding, edge behavior
Line & Stroke SystemStroke usage, weight, consistency, joins
Surface & MaterialTexture, material simulation, imperfection, micro-detail
Post-processingTone curve, grain, sharpening, render style
Background & EnvironmentBackground type, environment mood, props
Artistic MediumsPhysical medium, illustrative technique
UI & WebColor theme, elevation, buttons, icons, navigation
ReferencesCultural references, temporal era, reference brands
InspirationReference image URLs
GuardrailsAvoided visuals, colors, words, conditional logic
Custom Style ItemsFree-form key-value pairs for rules that don't fit elsewhere

02The gallery

One style. Any subject.

Every row below was made from one published StyleRef. Pick a style; the subjects change, the style doesn’t. Each one is free to copy, fork or pull into your tools.

Risograph print of a monstera plant in a terracotta pot

Risograph poster: Good Times, with a pizza and an evening schedule

Risograph poster: Book Club, with a stack of books and a cat

Risograph gig poster: The Buzzkills, a guitarist mid-jump

Risograph print of a cyclist passing a bookshop and coffee sign

Showing Risograph Print, 5 images

Extract My Style — Free Browse all styles in the gallery Open Risograph Print →

03Who it’s for

Built For

Creatives who use AI professionally and need their style to stay intact — every generation, every time.

01
AI Copywriters
Lock your voice — prompt after prompt.
02
AI Product Designers
Define your UI/UX design system for AI-generated assets.
03
AI Art Directors
Consistent lighting and mood for high-end AI production.
04
AI Graphic Designers
Your visual language, replicated in every output.
05
AI Video + XR
Persistent style direction for generative video and extended reality. Your visual DNA, across every emerging medium.

Extract My Style — FreeNo account needed · start in seconds

04For agents

Your agents read it too.

Commit a STYLE.md so coding agents build in your style. Connect the MCP server so Claude or Cursor can find, apply and save styles. Or pull any StyleRef from the CLI.

Read about STYLE.md

CLI

# demo style: Colour-Blocked Studio Set — swap in your own ID
npx styleref init bmivnppb-58ff14a99580

MCP server

# Streamable HTTP — Claude, Cursor or any MCP client
https://styleref.io/api/mcp

05FAQ

Frequently Asked Questions

Still unsure? Try it — no account needed.

Extract My Style — Free

What is a StyleRef?

A StyleRef is a structured style specification — a compact, copy-pasteable block of text that defines your creative style across dimensions like tone, color, typography, mood, and constraints. Build it once, paste it into any AI tool.

Does it work with ChatGPT, Claude, and Gemini?

Yes. StyleRef generates model-optimized formats — select 'AI Tools' for ChatGPT and Gemini; 'STYLE.md' for coding agents like Cursor, Codex, Claude Code and projects; 'FLUX' for FLUX image generation; 'Midjourney' for Midjourney.

What is a style block?

Style blocks are the individual building units of a StyleRef. Each block covers a specific dimension of your style — Color Palette, Tone of Voice, Typography, Mood, Output Format. StyleRef includes 160+ blocks across 20 sections.

How is StyleRef different from a system prompt?

A system prompt is a broad instruction. A StyleRef is a structured, machine-readable style specification with enforced priority rules — designed to override default AI tone and aesthetic drift, not just guide it.

Can StyleRef extract a style from an image or document?

Yes. StyleRef includes an extraction feature that analyzes a reference — an image, copy, brand guidelines, or documents — and extracts its characteristics into structured style blocks, ready to reuse in any AI generation.