How SkinSpectrum Calculates a Skin's Colors
Updated July 9, 2026
Every color classification on SkinSpectrum is produced automatically from the skin's image — there is no manual tagging by default. This page explains how that works and what the Color Match Score actually measures.
Reading the image
Each skin image is loaded and resized to a small working size for speed. Transparent and near-invisible pixels are ignored so the gun's cut-out background never counts as a color.
Grouping colors perceptually
Similar pixels are grouped together and compared in a perceptual color space (not raw RGB), so colors are judged the way the human eye sees them. The dominant clusters become the skin's primary, secondary and tertiary colors, each with a percentage.
Those colors are then mapped onto SkinSpectrum's fourteen color groups — black, white, gray, red, orange, yellow, gold, green, cyan, blue, purple, pink, brown and multicolor.
Visual weight, not just area
A neutral gun body can cover a lot of pixels without being what you notice. To account for this, the analysis weights pixels by saturation — 'visual weight' — so a vivid paint isn't drowned out by a large muted area.
The Color Match Score
The Match Score from 0 to 100 combines the target color's percentage, its perceptual distance from the ideal, saturation, primary-vs-secondary rank and the confidence of the analysis. For two-color palettes it evaluates the combination, not just one color.
It is deterministic and unit-tested, so the same skin and palette always produce the same score.
Limitations
Automated analysis is very good but not infallible — some skins are legitimately ambiguous or dominated by neutral tones. Manual overrides always take precedence over the automatic result when a correction is needed.