Dominant Colors Aren't Just "Most Frequent Pixel"
Extracting a meaningful color palette from an image is more nuanced than simply counting which exact RGB value appears most often — a naive frequency count on raw pixel values tends to produce a palette dominated by near-duplicate shades from anti-aliasing and compression artifacts (dozens of nearly-identical off-white pixels, for instance) rather than the visually distinct colors a human would actually pick out looking at the same image.
A more useful extraction clusters similar colors together first (grouping pixels that are perceptually close, not just exactly identical), then reports the dominant color from each cluster — producing a palette that matches human color perception far better than raw frequency counting alone, which is exactly the difference between a palette that looks "right" to a designer and one that's technically accurate but visually unhelpful.
Being able to also manually pick individual pixels with a magnifier, separate from the automatic extraction, matters because automatic dominant-color detection can miss a specific accent color that's visually important but numerically rare — a small but deliberate design detail an automated frequency-based approach would understandably overlook.