Why RGB distance doesn't match perceived difference
The simplest way to measure how "different" two colors are is to treat each as a point in 3D RGB space and compute the straight-line (Euclidean) distance between them. That number is easy to calculate but doesn't correspond well to what a human eye actually perceives, because human color vision isn't equally sensitive across all regions of RGB space — small RGB changes in some hues are barely noticeable, while equally small changes in others are quite visible. Two color pairs with identical RGB distance can look nearly identical in one case and clearly distinct in another, which is exactly the gap Delta-E was designed to close.
What Delta-E actually measures
Delta-E (ΔE) is calculated in the CIE Lab color space rather than RGB. Lab was specifically designed so that equal numeric distances correspond much more closely to equal perceived differences — its three axes represent lightness (L), a red-green axis (a), and a blue-yellow axis (b), derived from RGB through an intermediate conversion to CIE XYZ, a model built directly from experiments on human color perception rather than from how screens produce light. This tool uses the CIE76 formula, the original and simplest Delta-E definition: the straight-line distance between two points in Lab space, analogous to the RGB distance calculation but performed in a space where that distance actually tracks perception reasonably well.
Reading the ΔE number
The scale used here follows the commonly cited CIE76 interpretation bands: under 1 is imperceptible even under close side-by-side comparison; 1–2 is perceptible only on very close inspection; 2–10 is noticeable at a normal glance; 10–49 covers clearly distinct colors that still share some visual relationship; and 50 or higher indicates colors that are essentially opposite or unrelated. These bands are approximate guidelines rather than hard physical thresholds — perceptibility also depends on viewing conditions, display calibration, and the specific hues involved — but they give a useful rule of thumb for interpreting an otherwise abstract number.
Newer Delta-E formulas exist, and why this tool doesn't use them
CIE76 was refined twice since its introduction — CIE94, and then CIEDE2000, both adding correction terms to account for ways human perception deviates from Lab space's assumptions, particularly around saturated colors, where CIE76 is known to overstate perceived difference. CIEDE2000 is more accurate but considerably more complex to compute, involving several weighting functions and a hue-rotation term. CIE76 remains a reasonable choice for a general-purpose comparison tool because it's simpler, well understood, and close enough for most practical purposes — distinguishing "basically the same," "somewhat different," and "clearly different" colors doesn't usually require CIEDE2000's extra precision.
Similarity percentage: a simpler, complementary metric
Alongside Delta-E, this tool also reports a plain RGB-distance-based similarity percentage — a simpler, less perceptually accurate but more intuitive number, scaled so that identical colors read as 100% and maximally distant colors (black versus white) read as roughly 0%. It's included because Delta-E's absolute scale isn't always intuitive at a glance, whereas a percentage is immediately readable even without knowing the CIE76 interpretation bands. The two numbers usually agree in direction — a higher Delta-E corresponds to a lower similarity percentage — but they can diverge somewhat for specific hue pairs, since one is perceptually weighted and the other isn't.
Practical uses for a color difference calculator
Checking whether a slightly-adjusted brand color still reads as "the same" color as the original, comparing a rendered screenshot's colors against a design spec to catch color-management or compression drift, or deciding whether two similar-looking colors in a palette are different enough to serve distinct purposes are all situations where a numeric, perceptually-grounded difference measure is more useful than eyeballing two swatches side by side — especially when the comparison needs to be repeated across many color pairs, which is exactly what this tool's batch mode is for.