Monochrome Achievement

How we detect gifts where model blends with backdrop

What is Monochrome?

Monochrome achievement is awarded to gifts where the model visually blends with the backdrop due to similar colors. This creates a unique aesthetic where the model appears to merge with its background, rather than standing out.

✓ Monochrome Match

Blue model on blue backdrop — colors blend harmoniously

✗✗ No Match

Contrasting colors — model stands out from backdrop

Detection Algorithm

Model Classification

First, we classify each model as monochromatic or not. This flag is stored in the database and set in two stages:

Automatic Analysis

Script analyzes image contrast and color distribution for initial classification

Manual Verification

Our team reviews and corrects classifications manually

gift_models.monochrome = TRUE | FALSE

Color Data Storage

For each model and backdrop, we store color information in multiple formats:

Field	Example	Description
color_hex	`#4A90D9`	Original hex color from sprite/first animation frame or averaged gradient color
oklch_l	0.62	Lightness in OKLCH (0-1)
oklch_c	0.12	Chroma/saturation in OKLCH
oklch_h	245°	Hue angle in degrees

Color Space Conversion

To compare colors perceptually, we convert hex colors to OKLab color space:

// Gamma correction

sRGB → linear RGB

c <= 0.04045 ? c/12.92 : ((c+0.055)/1.055)^2.4

// Matrix transformation

linear RGB → LMS → OKLab (L, a, b)

// Extract chroma

Chroma = √(a² + b²)

Chroma measures how colorful vs. neutral/gray a color is

Color Similarity Check

Finally, we compare model and backdrop colors using three rules:

Neutral Check

C < 0.03

Gray/neutral colors (low chroma) can only match other neutrals

Euclidean Distance

ΔE < 0.162

Overall color difference in OKLab space must be small

ΔE = √((L₁-L₂)² + (a₁-a₂)² + (b₁-b₂)²)

Hue Angle

Δθ < 30°

For chromatic colors, hue directions must be similar

θ = arccos((a₁·a₂ + b₁·b₂) / (C₁·C₂))

OKLab Color Space

OKLab is a perceptually uniform color space — distances correspond to how humans perceive color differences.

Lightness

0 = black, 1 = white

Green ↔ Red

≈ -0.4 to +0.4

Blue ↔ Yellow

≈ -0.4 to +0.4

Why OKLab over RGB?

RGB distance doesn't match human perception
OKLab treats all hues equally
Better for detecting subtle color matches

Model Classification Examples

Models are classified as monochrome or not based on their visual characteristics:

Monochrome Modelmonochrome = TRUE

🐳

Whale

Single dominant hue throughout the design

Dominant color:#4A90D9

✗Contrasting Modelmonochrome = FALSE

🦜

Parrot

Multiple distinct colors in the design

Multiple colors:

Gift Examples

Even with a monochrome model, the gift only earns the achievement if backdrop color matches:

✓ Monochrome Achievement

🐳

Model:

#4A90D9

Backdrop:

#3B82F6

ΔE = 0.08 < 0.162 ✓

≈≈ Near Miss

🐳

Model:

#4A90D9

Backdrop:

#F97316

ΔE = 0.31 > 0.162 ✗

Same monochrome model, but orange backdrop creates too much contrast in OKLab space

💡

In Summary

A gift receives Monochrome achievement when its model is flagged as monochromatic AND the model's color closely matches the backdrop's color in OKLab space. This ensures the visual "blending" effect is actually perceptible.

if model.monochrome = TRUE
AND oklab_similar(model.color, backdrop.color)
then award "Monochrome"