Art Quizzes & Games

Hue Sorting Game

Arrange rows of scrambled color tiles into smooth gradients, then see an error score and a chart of exactly which parts of the hue circle your eye finds hardest.

0:00

Sort each row into a smooth gradient between its two pinned end caps (marked •). Drag a tile, or tap one and then tap where it belongs. Keyboard: Tab to a tile, Space to pick up, ← → to move, Space to drop.

Neighboring tiles differ by about ΔE₀₀ 3.

A game for training your eye, not a medical color vision test. Screen brightness, color calibration and room light all change the result.

Runs entirely in your browser. Free, no sign-up.
Preview embedded tool

Works in an HTML block on WordPress, Wix, Squarespace or a custom website. Change the height if your layout needs more space.

How to use the Hue Sorting Game

  1. Choose a Level. Levels 1–5 split the whole hue circle into four rows and add tiles while lowering the color intensity. Level 6 gives you three short, narrow bands of hue that are hard in a different way.
  2. Sort each row. The two tiles marked with a dot at each end of a row are pinned. Drag the tiles between them into a smooth run from one cap to the other. On a phone, tap a tile and then tap the spot where it belongs. With a keyboard, Tab to a tile, press Space to pick it up, move it with ← and →, and press Space again to drop it.
  3. Press Check my order. The board locks, and you get a total error score (0 is perfect), the number of tiles in their exact spot, your time and your best on that level. Small white dots mark every tile that is out of place.
  4. Read the chart. Bars show the error for each tile in hue order, with the tile’s own color underneath, and the sentence below names the hue range where most of your points came from.
  5. Compare and replay. Show correct order flips between your arrangement and the answer. Copy result includes a link to the identical puzzle, and Download result card saves a PNG of your rows and chart.

How the score works

The scoring follows the rule of the Farnsworth-Munsell 100 Hue test, which Dean Farnsworth published in the 1940s and which still uses 85 colored caps in four trays, each tray anchored by fixed caps. Each tile is scored by its neighbors, not by its position:

error = |left neighbor − tile| + |right neighbor − tile| − 2

where the numbers are the tiles’ true positions. A tile sitting between its true neighbors scores 1 + 1 − 2 = 0.

Take a row with end caps 0 and 6. Swap two neighbors, so the row reads 0 | 1 2 4 3 5 | 6, and four tiles each score 1: one swap costs 4 points. Now take 0 | 2 3 4 5 1 | 6, where a single tile has been dragged from the front to the back. No tile is in its exact spot, yet the error is only 11, because the run 2-3-4-5 is still smooth. That is the point of neighbor scoring: it measures how smooth the gradient you built is, which is what your eye was actually judging.

Why equal hue steps aren’t equal

Ask software for colors every 10 degrees around a hue wheel and the steps will not look even. Measured with CIEDE2000 (ΔE₀₀, the standard formula for how different two colors look), a 10° step at lightness 70 and chroma 30 is about 3.1 among the pinks but about 4.6 among the periwinkle blues — roughly 50% larger.

So the game does not space tiles by angle. It walks the hue circle in tiny increments, adds up ΔE₀₀ as it goes, and places tiles at equal perceptual distances. Every neighboring pair on a level is, by the formula, equally hard to tell apart. When your errors cluster in one region anyway, that tells you something about your eyes or your screen rather than about uneven spacing.

How hard each level is

Level Movable tiles Lightness L* Chroma C* Step between neighbors (ΔE₀₀)
1 · Warm-up 24 70 36 5.4
2 · Easy 36 70 34 3.6
3 · Medium 52 70 30 2.4
4 · Hard 68 68 24 1.6
5 · Expert 84 68 18 1.1
6 · Narrow bands 30 70 34 1.1–1.6

A ΔE₀₀ of about 1 is roughly the smallest difference most people can see with two patches side by side. Lowering chroma shrinks the whole circle too: at chroma 18 one lap of the hue circle measures about 100 ΔE₀₀ units, against about 150 at chroma 36. That is why muted colors are so much harder to sort, and why painters who work in grayed, tonal palettes need finer hue discrimination than people who work with tube-bright color.

Tips from practice

Place the easy tiles first. Find the tiles that obviously belong next to each end cap and put them there. The middle of a row is where the subtle decisions live, and it is easier once the ends are settled.

Judge each tile against both neighbors. Looking at one tile alone tells you little. Ask whether it sits halfway between the tile on its left and the tile on its right; if it leans toward one side, it probably belongs one step over.

Turn off night modes and auto-brightness. Night Shift, Night Light and similar warm-screen filters shift every color toward orange and compress the blues. Room light falling on the screen does the same in a gentler way.

Train your weak region. If the chart keeps pointing at the same hues, play Level 6 repeatedly; its narrow bands concentrate the practice. The Color Perception Test checks value and saturation as well as hue.

Rest between rounds. After a few minutes of staring at one color family, your eyes adapt and fine differences fade. A short break restores them more than trying harder does.

FAQ

Is this the same as the Farnsworth-Munsell 100 Hue test?

No. It borrows the layout and the scoring rule, but it uses colors generated on your screen rather than physical Munsell caps viewed under a standard daylight lamp. Scores are therefore not comparable with clinical results, and a high score here says nothing medical about your vision.

Why is the board gray instead of white or black?

A neutral mid-gray surround shifts every tile equally, if at all. A white or colored background would change how the nearest tiles look through simultaneous contrast and make some hues harder to judge than others for reasons that have nothing to do with your eyes.

Can I challenge a friend with exactly the same puzzle?

Yes. The page address holds the level and a short seed, and Copy result adds that link under your score. Anyone who opens it gets the identical colors in the identical scrambled order, so you can compare error scores fairly.

What counts as a good score?

There is no official benchmark for this game, so use your own best on each level as the baseline. As a guide, a single pair of swapped neighbors costs 4 points. Zero on the first three levels is very achievable with care; on Expert, small errors are normal.

Does the timer change my score?

No. Time is recorded separately and only breaks ties between equal error scores in your personal best. Untick Timer if the clock makes you rush; accuracy is the point of the exercise.