When analyzing a brand banner, UI screenshot, or product photograph, knowing which colors exist is only half the equation—you also need to know how much visual weight (% of surface area) each color carries.
Utiliome's Dominant Color Extractor runs an 8-iteration deterministic K-Means++ clustering pipeline directly inside your browser:
- Alpha-Aware Pixel Filtering: Transparent background pixels (
alpha < 128in PNG, WebP, or SVG cutouts) are stripped before clustering so transparent space is never miscounted as black (#000000) or white (#FFFFFF). - K-Means++ Spread Initialization: Seeds
kinitial centroids by maximizing squared Euclidean distance (d² = ΔR² + ΔG² + ΔB²) between starting seeds, guaranteeing deterministic, repeatable results every time you drop the same file. - Centroid Convergence & Role Classification: Ranks converged clusters by pixel share (
%) to identify the Primary Dominant surface, Secondary Tone, and highest-chroma Accent Highlight.