Color Hue Perception Test

Drag the color chips until each row is a smooth, even gradient. The chips at both ends are locked to show you the direction. When you're done, hit Check — the lower your error score, the sharper your eye for hue.

Ends are locked. Drag the middle chips into order. On a keyboard: focus a chip and use ← →, then Enter to check.

What does the error score mean?

Every chip in a row is the same brightness and the same saturation — only the hue changes. That means you can't lean on light and dark to sort them; you have to read the hue itself. Your score adds up how far each chip's color sits from the chips beside it, so a perfectly even gradient scores zero. Swap two nearly identical neighbors and the score barely moves; drop a color far from where it belongs and it climbs fast.

Scores depend on your screen and your lighting as much as your eyes, so the most useful comparison is against your own past runs — or against a friend on the exact same board. Start on Easy to learn the feel, then work up to Hard, where the hue steps get tiny.

The task: sorting by hue alone

What you are doing here is an arrangement task — you are handed chips that step around a slice of the colour wheel in the wrong order and asked to put them back in sequence. The format has a long history in colour-vision research, where the best-known example is the Farnsworth–Munsell 100 Hue Test.

Every chip is built at the same saturation and the same lightness, so only the hue moves. That is the whole trick: it stops you falling back on light-versus-dark, a much easier judgement than hue. Each row starts at a random base hue and walks a fixed arc — Easy is one row of nine chips across 165 degrees, Medium two rows of eleven across 140, Hard four rows of thirteen across just 90. Difficulty is therefore not "more chips" so much as the same journey chopped finer. The end chips are locked, the interior is shuffled from a seeded generator, and the seed is written into the page's URL — so a link you copy rebuilds the identical board for whoever opens it, which is the only fair way to compare with someone.

How the error score is built

Each chip's colour is converted from sRGB into CIELAB, a space arranged so that distance between two colours roughly tracks how different they look, and the gap between two chips is the straight-line distance between them there. For every movable chip the code adds the distance to the chip on its left and the distance to the chip on its right, totals that across all rows, subtracts the same total measured on the correct arrangement, and rounds — with a floor of zero.

So the score is not a count of misplaced chips. It is how much longer your path through colour space is than the shortest one available. Swap two neighbours whose hues are nearly identical and the total barely moves, because the detour you added was tiny; drop a chip several places from home and it climbs fast, because you have made the ramp double back on itself twice. Zero means your ordering is exactly as smooth as the intended one. The strictly positional count is reported separately as "chips out of place", so it is quite normal to see a few flagged there and still post a low error score. The grade is the error score divided by the number of movable chips, which stops Hard's extra rows dragging it down by themselves:

0: Flawless — as smooth as the intended ramp.
under 5 per chip: Sharp eye.
under 12 per chip: Solid.
under 24 per chip: Getting there.
24 or more: Keep training.

Your lowest score per difficulty is kept in your browser's own storage under sellkit_huetest_best, and stays on your device.

Things that move your score that are not your eyes

The chips are specified in sRGB and then handed to your display, which is where the certainty ends. Panel type matters most: on a cheap TN laptop screen colour shifts visibly with viewing angle, so the chips at the ends of a wide row are seen at a different angle from the ones in the middle, and tilting the lid can move the "right" answer. Brightness matters too — turned down, the small hue steps on Hard compress toward each other.

Then there are the filters. Night-shift and blue-light modes rotate the whole ramp toward orange and squash the blue end — precisely the distortion that makes a hue sequence hard to read — so turn them off. Ambient light does the other half of the damage: a warm lamp or a bright window behind you shifts how your eyes are adapted. And unless your display has been profiled, the colours in front of you are only approximately the ones the code asked for. The comparison worth making is your own score on your own screen over time, or a friend on the identical seeded board. The reaction time test next door has the same problem wearing different clothes — it measures your monitor's refresh rate and your mouse alongside your reflexes — so read that number the same careful way.

This is not a colour-blindness screen

It cannot tell you whether you have a colour vision deficiency and it is not built to. A poor run does not indicate one and a good run does not rule one out — the likelier explanations for a bad score are a hard board, a dim room, a warm-tinted screen, or rushing.

Clinical testing is a different exercise. Screening is commonly done with pseudoisochromatic plates — the Ishihara plates, printed, viewed under controlled lighting and administered by a professional — and where an arrangement test is used clinically it uses physical calibrated caps under a standard light source. None of those conditions exist in a browser tab. If you have a real reason to know, because colour is giving you trouble at work or a job requires it, see an optometrist or ophthalmologist. This page is a game.

Frequently asked questions

How does the color hue test work?

Each row is a slice of the color wheel where only the hue changes — brightness and saturation stay the same, so you can't cheat with light and dark. The two end chips are locked. Drag the middle chips until the colors step evenly from one end to the other, then hit Check. Your score adds up how far each chip's colors are from its neighbors — lower means a smoother, more accurate gradient.

What's a good score?

Zero is a flawless sort. Most people land a little above zero, especially on Hard, where the hue steps are tiny. Try Easy first, then work up. Scores depend on your screen and lighting too, so compare yourself over time rather than chasing a single number.

Is this a real color-blindness test? Is anything uploaded?

No on both counts. It is a playful perception game, not a medical or clinical test, and a poor round here means nothing on its own — if you are actually worried about your colour vision, an optometrist has calibrated equipment and you have a laptop screen. And nothing about the test is uploaded: the chips are generated in your browser with a bit of maths, the comparison happens there too, and your score is written only to this browser's own storage. There is no account and no score server. Like every page here it does carry Google ads, which set their own cookies — that part is covered in our privacy policy.