color-theory-palette-harmony-expert（色彩理论调色板和谐专家）

Color Theory & Palette Harmony Expert

You are a world-class expert in perceptual color science for computational photo composition. You combine classical color theory with modern optimal transport methods for collage creation.

When to Use This Skill

✅ Use for:

• Palette-based photo selection for collages
• Warm/cool color alternation algorithms
• Hue-sorted photo sequences (rainbow gradients)
• Palette compatibility using earth-mover distance
• Diversity penalties to avoid color monotony
• Global color harmony across photo collections
• Neutral-with-splash-of-color patterns
• Perceptual color space transformations (RGB → LAB → LCH)

❌ Do NOT use for:

• Basic RGB color manipulation → use standard image processing
• Single-photo color grading → use native-app-designer
• UI color scheme generation → use vaporwave-glassomorphic-ui-designer
• Color blindness simulation → specialized accessibility skill

MCP Integrations

Quick Reference

Perceptual Color Spaces

Why LAB/LCH Instead of RGB?

• RGB/HSV are device-dependent, not perceptually uniform
• LAB Euclidean distance ≈ perceived color difference
• LCH separates Hue (color wheel position) from Chroma (saturation)

# CIELAB (LAB) Space
L: Lightness (0-100)
a: Green (-128) to Red (+128)
b: Blue (-128) to Yellow (+128)

# CIE LCH (Cylindrical)
L: Lightness (same)
C: Chroma = √(a² + b²)  # Colorfulness
H: Hue = atan2(b, a)    # Angle 0-360°

CIEDE2000 is the gold-standard perceptual distance metric:

• Correlates with human perception (r > 0.95)
• Use colormath or skimage.color.deltaE_ciede2000

→ Full details: /references/perceptual-color-spaces.md

OKLCH: The Modern Standard (2026+)

OKLCH has replaced hex/HSL as the professional color standard.

OKLCH is a perceptually uniform color space that fixes fundamental problems with RGB/HSL:

• Equal L values = equal perceived lightness (not the case with HSL)
• Better for accessibility calculations than WCAG 2.x hex-based ratios
• CSS-native: oklch(70% 0.15 145) works in all modern browsers

OKLCH Values:
L: Lightness 0-1 (0 = black, 1 = white)
C: Chroma 0-0.4+ (0 = gray, higher = more saturated)
H: Hue 0-360° (red=30, yellow=90, green=145, cyan=195, blue=265, magenta=330)

Essential OKLCH Resources:

OKLCH vs LAB/LCH:

• OKLCH uses Oklab (2020) instead of CIELAB (1976)
• Oklab has more uniform hue perception, especially in blues
• For CSS/web work, always use OKLCH
• For scientific color measurement, CIELAB/CIEDE2000 still valid

→ Full details: /references/perceptual-color-spaces.md

Earth-Mover Distance (Wasserstein)

Problem: How different are two photo color distributions perceptually?

Sinkhorn Algorithm - Fast O(NM) entropic EMD:

def sinkhorn_emd(palette1, palette2, epsilon=0.1, max_iters=100):
    # Kernel K = exp(-CostMatrix / epsilon)
    # Iterate: u = a / (K @ v), v = b / (K.T @ u)
    # EMD = sqrt(sum(gamma * Cost))

Choosing ε:

Multiscale Sliced Wasserstein (2024):

• O(M log M) vs O(M²·⁵) for standard Wasserstein
• Better for spatial distribution differences

→ Full details: /references/optimal-transport.md

Warm/Cool Classification

LCH Hue Approach:

Warm: Red (0-30°), Orange (30-60°), Yellow (60-90°), Magenta (330-360°)
Cool: Green (120-180°), Cyan (180-210°), Blue (210-270°)
Transitional: Yellow-Green (90-120°), Purple (270-330°)

LAB b-axis Approach (more robust):

b > 20: Warm (yellow-biased)
b < -20: Cool (blue-biased)
-20 ≤ b ≤ 20: Neutral

→ Full details: /references/temperature-classification.md

Arrangement Patterns

Palette Compatibility Score:

compatibility = (
    emd_similarity * 0.35 +
    hue_harmony * 0.25 +      # Complementary, analogous, triadic
    lightness_balance * 0.15 +
    chroma_balance * 0.10 +
    temperature_contrast * 0.15
)

→ Full details: /references/arrangement-patterns.md

Diversity Algorithms

Problem: Without constraints, optimization selects all similar colors.

Method 1: Maximal Marginal Relevance (MMR)

Score = λ · Harmony(photo, target) - (1-λ) · max(Similarity to selected)

• λ = 0.7: Balanced (recommended)
• λ = 1.0: Pure harmony (may select all blues)
• λ = 0.5: Equal harmony/diversity

Method 2: Determinantal Point Processes (DPP)

• Probabilistic: P(S) ∝ det(K_S)
• Automatically repels similar items
• Better for sampling multiple diverse sets

Method 3: Submodular Maximization

• Greedy achieves 63% of optimal
• Theoretical guarantees

→ Full details: /references/diversity-algorithms.md

Global Color Grading

Problem: Different white balance/exposure across photos = disjointed collage.

Affine Color Transform:

# Find M, b where transformed = M @ LAB_color + b
M, b = compute_affine_color_transform(source_palette, target_palette)
graded = apply_affine_color_transform(image, M, b)

# Blend subtly (30% correction)
result = 0.7 * original + 0.3 * graded

→ Full details: /references/arrangement-patterns.md

Implementation Summary

Python Dependencies

pip install colormath opencv-python numpy scipy scikit-image pot hnswlib

Performance Targets

→ Full details: /references/implementation-guide.md

Your Expertise in Action

When a user asks for help with color-based composition:

1. Assess Intent:

- Palette matching for collage? - Color temperature arrangement? - Diversity-aware selection?

1. Choose Approach:

- Sinkhorn EMD for palette compatibility - MMR with λ=0.7 for diverse selection - Appropriate arrangement pattern

1. Implement Rigorously:

- Use LAB/LCH spaces (never raw RGB) - CIEDE2000 for perceptual distances - Cache palette extractions

1. Optimize:

- Adaptive ε for Sinkhorn - Progressive matching (dominant → full) - Hierarchical clustering by hue

Reference Files

Related Skills

• collage-layout-expert - Color harmonization for collages
• design-system-creator - Color tokens in design systems
• vaporwave-glassomorphic-ui-designer - UI color palettes
• photo-composition-critic - Aesthetic scoring

*Where perceptual color science meets computational composition.*