There's a specific kind of frustration that designers and developers know well. A client sends over a logo file, you zoom into it, try to eyedrop a color, and end up with three slightly different HEX codes depending on where you click and which tool you're using. Or you're building a website and trying to match the brand colors from a JPEG reference image, and every attempt feels slightly off. The image color picker exists to solve that exact problem.
Click a point on the image and the tool reads the actual pixel data from the full-resolution file, not a screenshot of it, not a compressed re-rendering. That's why the HEX code you get here for a specific point matches what you'd see if you opened the same file in Photoshop or Figma and sampled from it directly.
How the palette extraction actually works
A typical photograph contains hundreds of thousands of distinct colors. Showing all of them would be useless. What you actually want is a small set of the colors that define the visual character of the image. The tool uses a median-cut quantization algorithm to find them, working from a downsampled copy of your image for speed. Think of it like sorting a pile of thousands of colored marbles into groups by shade, then picking the most representative marble from each group. Very dark and very light pixels, plus any transparent ones, are left out of this process on purpose, so a plain white or black background doesn't crowd out the more distinctive colors in your subject.
You can extract 3, 5, 6, 8, 10 or 12 colors depending on what you need. For a brand logo or a product with a clean background, three to five colors is usually the right call. For a landscape photo or a complex illustration where you're trying to understand the full tonal range, eight to twelve gives you a more complete picture.
Picking exact pixel colors vs extracting a palette
These are two different things and they're useful in different situations. Palette extraction tells you what the dominant colors are across the whole image, minus the near-white, near-black and transparent pixels mentioned above. Pixel picking tells you the exact color value at a specific point, read from the full-resolution image, and you can pin up to 8 of these at once. If you're matching a button color to a logo, you want pixel picking, click directly on the part of the logo you're matching and you'll get the exact HEX code for that point. If you're building a color scheme inspired by a photo, palette extraction is what you want.
Taking your colors into CSS
Once you've got the palette you want, the CSS export button outputs the colors as custom properties that you can paste directly into your stylesheet, named in palette order. Something like --color-1: #2D4A8C; /* rgb(45, 74, 140) | hsl(220, 51%, 36%) */ ready to drop in and rename to something more descriptive if you want. This saves the copy-paste-reformat cycle that eats up time when you're manually transferring colors from a design tool into code. If you need the palette as a visual reference rather than code, the PNG download gives you a swatch image you can drop into a Figma file, a mood board, or a client presentation.