Image Colour Picker
ImagesEverything runs locally in your browser — nothing is uploaded
Drop an image here, or click to choose
Or press Ctrl/⌘ + V to paste a screenshot
PNG / JPEG / WebP / GIF / BMP / AVIF · up to 30 MB each
Click anywhere on an image to read that point’s colour as HEX, RGB and HSL. Alongside single-point picking it computes the overall average colour and a palette (each swatch with its share of the image), which is far quicker than clicking sample by sample when you are after a scheme or a dominant tone. Sample radius is the one option worth understanding: 1x1 reads the single pixel, which on a compressed photo is often noise; 3x3 and 5x5 average a small block for a colour closer to what you actually see. The default is 3x3. The palette uses median cut: colour space is repeatedly halved along the widest channel, each cell’s average becomes a swatch, and the swatches are sorted by share, largest first. Unlike quantising into an 8x8x8 colour cube, this does not split near-identical colours apart and push a large dominant colour down the list. Semi-transparent pixels are excluded from every statistic — otherwise the transparent area of a cut-out PNG would count as black and drag the average down, the most common mistake in tools like this. Clicking is wired to the preview image: a click is converted to source pixels with the current radius and the result is appended to the report card. Everything runs locally; the image is never uploaded.
Features
- Click anywhere to sample a colour as HEX, RGB and HSL
- The overall average colour, a first reference for any colour scheme
- Palette by median cut, sorted by share, not by colour-cube quantising
- Choose 5, 6 or 8 dominant colours, each shown with its share
- Sample radius of 1x1, 3x3 or 5x5; 3x3 by default to dodge compression noise
- Semi-transparent pixels are excluded, so a cut-out PNG is not dragged towards black
- Exports a PNG swatch sheet and a copyable text summary
- Nothing is uploaded; it works offline
How to use
- Drop an image in, or paste a screenshot with Ctrl/⌘ + V
- Pick the sample radius and the number of dominant colours
- The average and the palette appear at once, ready to save as a PNG sheet
- Click anywhere on the preview to sample that point
FAQ
- Which sample radius should I use?
- It depends on the source. In a compressed photo the pixels inside one flat colour wobble by about ±3, so 1x1 often lands on a speck of noise; the default 3x3 averages a small block and lands closer to what your eye sees. Use 1x1 only when you need a single exact pixel, such as reading a value out of a UI screenshot.
- Why not quantise to 256 colours for the palette?
- Quantising into an 8x8x8 colour cube is the easy way, but it splits near-identical colours into separate cells, so one large dominant colour is broken into several small slices and ends up behind minor colours once sorted. Median cut halves the distribution repeatedly, so a dominant colour stays whole and its share is the real one.
- Are transparent areas counted?
- No. Pixels below half opacity are left out of both the average and the palette. That matters: in a cut-out PNG the transparent area usually stores RGB 0, so counting it would drag the average towards black, the most common mistake in tools like this.
- Is the image uploaded?
- No. Sampling and statistics run in your browser with canvas, entirely in local memory, and it works offline.
Related tools
Background Colour Remover
ImagesMake a colour transparent and export a PNG: adjustable tolerance, connected-area-only flooding, edge feathering and de-fringing.
Image Watermark
ImagesAdd a text watermark to a photo: nine positions plus tiling, with the size scaled to the image so it looks the same at any resolution.
Image Info
ImagesRead everything an image carries: format details, camera settings and where it was taken. GPS is called out separately; the report exports as text or JSON.
Image Converter
ImagesConvert between PNG, JPEG, WebP, AVIF, BMP and TIFF, with adjustable quality and size. JPEG output lays a background, so transparency never turns black.
Image Slicer
ImagesSlice an image into a grid for nine-tile posts or a tall image. Six grid shapes, per-tile padding, and names padded for posting in order.
ID Photo Crop
ImagesCrop a photo to ID sizes — 25×35 mm, 35×49 mm and US, Japanese or Schengen visa specs — at 300 or 600 DPI. Cover-cropped, so faces never stretch.