Image Color Analyzer & Extractor

Extract dominant colors, usage percentages, exact pixel colors and full color analysis from any image — privately in your browser.

Drop an image here or paste it with Ctrl/⌘ + V Choose image Formats: JPG, PNG, WEBP, GIF, AVIF · up to 25 MB Processed locally — your image is never uploaded

No image at hand? Try a sample:

100% Magnifier

Colors Sort

Your palette

Add colors from the results or picked pixels.

About the Image Color Analyzer & Extractor

Upload, drop or paste an image. Color Atlas clusters its pixels in the perceptual CIELAB space to find dominant colors and how much of the image each one covers, lets you pick exact pixel colors, analyze a selected region, highlight where a color appears and export palettes and reports. Your image never leaves your device.

How to use it

Select an image, drag it onto the drop zone or paste it from the clipboard.
Choose how many dominant colors to extract (5 to 50) and how to sort them.
Click on the image to pick exact pixel colors; use the magnifier for precision.
Draw a rectangle in Region mode to analyze only part of the image.
Open the Detailed analysis tab for histograms, families and exact color counts.
Export the palette (PNG, SVG, CSS, JSON, TXT) or a report (PDF, CSV).

Examples

Extract a brand palette from a product photo.
Measure how much of a room photo is covered by wall color.
Find the exact color of a pixel in a screenshot.

Frequently asked questions

Is my image uploaded?

No. All processing happens locally with the Canvas API and a Web Worker. Nothing is sent to our server.

How are percentages calculated?

Every analyzed pixel is assigned to its nearest dominant color (CIELAB distance). Percentages are the share of analyzed opaque pixels in each cluster and add up to about 100%. Large images are downsampled first, which is stated in the results.

How are transparent pixels handled?

Pixels with alpha below 50% are excluded from statistics; the count of excluded pixels is shown.

What does "exact" mean in the detailed analysis?

When the analyzed image is within the exact-count limit, every distinct RGB value is counted exactly. Grouping by similarity is clearly labelled as grouped.