About

Human color perception operates within a narrow band of the electromagnetic spectrum (~380nm to 700nm). Distinguishing between two colors separated by fewer than 3 - 5 digital units (out of 256 per channel) approaches the threshold of most displays and most eyes. This game exploits that limit. You are shown a target color defined by four channels: R (red), G (green), B (blue), and A (alpha/opacity). Your task is to reproduce that color using sliders. Scoring uses weighted Euclidean distance in RGBA space. A perfect match scores 100. A guess at maximum distance scores 0. Typical untrained players average 55 - 70 on medium difficulty. Professionals in print production or UI design often exceed 85.

The alpha channel is the hardest to judge. Without a reference checkerboard pattern, transparency differences below 0.1 are nearly invisible on uniform backgrounds. This tool renders both target and guess over a standard checkerboard so alpha shifts become detectable. Note: results depend on your monitor calibration. An uncalibrated display can shift perceived color by 10 - 20 units per channel, which limits achievable accuracy regardless of skill.

Formulas

Accuracy is computed as the normalized Euclidean distance between the target color vector and the guessed color vector in four-dimensional RGBA space.

d = √(R_t − R_g)² + (G_t − G_g)² + (B_t − B_g)² + (A_t ⋅ 255 − A_g ⋅ 255)²

d_max = √255² + 255² + 255² + 255² = 510

Score = max(0, 100 − dd_max × 100)

Where R_t, G_t, B_t, A_t are the target channel values, and R_g, G_g, B_g, A_g are the guessed values. Alpha is scaled to 0 - 255 to equalize its weight with RGB channels. A score of 100 means a pixel-perfect match. A score below 50 indicates the guess deviated by more than half the color space diagonal.

Streak bonus applies when consecutive round scores exceed 85:

Bonus = streak × 5

Reference Data

Difficulty	Alpha Range	Avg. Score (Untrained)	Avg. Score (Trained)	Max Distance	Rounds
Easy	A = 1.0 (fixed)	65 - 75	88 - 95	441.7	10
Medium	0.4 ≤ A ≤ 1.0	50 - 65	80 - 90	469.6	10
Hard	0.0 ≤ A ≤ 1.0	40 - 55	72 - 85	510.0	10
Color Channel Reference
Channel	Range	Bit Depth	Human JND	Weight in Score	Notes
Red (R)	0 - 255	8-bit	~3 - 5 units	1.0	Most sensitive in warm tones
Green (G)	0 - 255	8-bit	~2 - 4 units	1.0	Human eye most sensitive here
Blue (B)	0 - 255	8-bit	~4 - 7 units	1.0	Hardest to perceive shifts
Alpha (A)	0.00 - 1.00	Continuous	~0.05 - 0.10	1.0 (×255)	Requires checkerboard to detect
Notable Color Pairs (Hard to Distinguish)
Pair	Color A	Color B	Distance	Difficulty	Why
Navy vs. Dark Blue	rgba(0,0,128,1)	rgba(0,0,139,1)	11.0	Extreme	Low luminance blue range
Salmon vs. Coral	rgba(250,128,114,1)	rgba(255,127,80,1)	34.5	Hard	Similar warm hue, slight orange shift
Grey 50% vs. 55%	rgba(128,128,128,1)	rgba(140,140,140,1)	20.8	Hard	Neutral greys lack hue cues

Frequently Asked Questions

RGB channels produce visible hue and brightness changes that map to distinct cone responses in the retina. Alpha (opacity) affects only the blending ratio between the foreground color and the background. On a uniform white background, reducing alpha from 1.0 to 0.9 on a mid-tone color shifts perceived lightness by roughly 5-10 units, which is near the just-noticeable difference threshold. The checkerboard pattern in this tool makes alpha shifts more detectable by introducing high-contrast edges that visibly change with transparency.

An uncalibrated consumer monitor can deviate by 10-20 units per channel from sRGB standard. This means even a perfect guess on your screen might correspond to a different RGBA value on a calibrated display. If your monitor has a blue/warm tint, your blue channel guesses will be systematically biased. For best results, use a display calibrated to sRGB (D65 white point, gamma 2.2). The game compares numeric values, not perceptual appearance, so calibration directly impacts fairness.

On Easy difficulty (no alpha), untrained players average 65-75 out of 100. Trained designers or artists typically score 88-95. On Hard difficulty with full alpha range, scores above 80 are considered expert-level. A score of 100 requires matching all four channels exactly, which is statistically unlikely on a random target. Consistent scores above 85 across 10 rounds indicate strong color perception skills.

CIEDE2000 operates in CIELAB color space, which accounts for non-linear human perception. It is more scientifically accurate for measuring perceived difference. However, this game operates in sRGB/RGBA space because the input mechanism (sliders for R, G, B, A) is channel-based. Using Euclidean distance in the same space as the controls ensures the scoring is transparent and directly interpretable: each unit of slider movement has equal scoring weight. Implementing CIEDE2000 would penalize blue-channel errors less than green, which would confuse players who think in RGB terms.

Difficulty only changes the alpha channel range. Easy locks alpha at 1.0 (fully opaque), removing one variable entirely. Medium restricts alpha to 0.4-1.0, where transparency is visible but not extreme. Hard uses the full 0.0-1.0 range, where near-zero alpha values make the color almost invisible against the checkerboard. All difficulty levels use 10 rounds per game. The scoring formula remains identical across all difficulties.

Players with red-green color vision deficiency (protanopia or deuteranopia, affecting ~8% of males) will find it harder to distinguish reds from greens, reducing scores on targets in those hue ranges. The game does not currently offer a colorblind mode. However, the numeric RGBA values are shown after each guess, which allows learning even if perception is limited. Players with tritanopia (blue-yellow deficiency) will struggle with the blue channel specifically.