What Delta E actually measures, how it’s calculated, and what value counts as accurate enough to trust for color-critical work
Introduction
Delta E (written ΔE) is the number colorists, photographers, and monitor manufacturers reach for whenever they need to say how accurate a color is, rather than just how it looks. It shows up on monitor spec sheets, in calibration software reports, and in printing tolerances, always as a single figure that’s supposed to tell you whether a color matches its reference closely enough to trust. The catch is that Delta E isn’t one fixed thing. There are several formulas, they don’t produce comparable numbers, and a value that’s perfectly fine for one kind of work can be unacceptable for another.
What Delta E Actually Measures
Delta E measures the distance between two colors, a reference color and a sample color, as perceived by a standard human observer. The name comes from German: Delta for change, and E for Empfindung, meaning sensation. Put together, Delta E means a difference in sensation, which is a fitting name, since the entire point of the metric is to approximate how different two colors look to an actual person, not just how different their numeric values are.
A Delta E of 0 means no detectable difference at all. As the number climbs, the visible gap between the two colors grows. In practice, professional evaluations rarely deal with values above 5, since anything higher is usually an obvious, glaring mismatch rather than a fine calibration detail.
The CIELAB Color Space
Delta E is built on CIELAB (also written CIE L*a*b*), a color space developed by the International Commission on Illumination in 1976 specifically to be perceptually uniform, meaning equal numeric distances in the space correspond reasonably well to equal perceived differences in color. That property is what makes it useful as the foundation for a difference metric in the first place; RGB and CMYK values don’t have it, so measuring raw distance in those spaces doesn’t line up with how different colors actually look.
CIELAB describes any color with three values. L* is lightness, running from 0 (pure black) to 100 (pure white). a* is the red-green axis, with positive values leaning red and negative values leaning green. b* is the yellow-blue axis, positive toward yellow and negative toward blue. Every color a display or print can produce lands somewhere in this three-dimensional space, and Delta E is, at its core, a way of measuring the distance between two points in it.
How Delta E Is Calculated
The original formula, now called Delta E 1976 or CIE76, is straightforward: take the difference between the two colors’ L*, a*, and b* values, square each difference, add them together, and take the square root. It’s the same idea as calculating the straight-line distance between two points, just applied in L*a*b* space instead of ordinary geometric space.
CIE76 is easy to compute but not especially accurate to human perception, because equal distances in CIELAB space still aren’t perfectly uniform across all colors. It tends to overstate differences in some hues, like blues and greens, and understate them in others.
CIE76 vs CIE94 vs CIEDE2000
Two later formulas tried to correct for this. CIE94 adds weighting factors for saturation (chroma) and hue, making it noticeably more accurate for saturated colors than the plain Euclidean CIE76 formula. CIEDE2000 (often written ΔE00) refines this further with additional corrections for hue rotation and lightness, and is now the formula most professional display and print evaluations use, precisely because it tracks human perception more closely than either of its predecessors.
The important thing to remember is that these formulas aren’t interchangeable. A Delta E of 1 under CIE76 is not the same visual difference as a Delta E of 1 under CIEDE2000. When a monitor spec sheet or calibration report quotes a Delta E figure without naming the formula, treat the number with some skepticism, since it’s genuinely ambiguous without that context.
Delta E Value Interpretation
As a general guide to how Delta E values translate to visible difference, most colorimetry references use something close to this scale:
| Delta E Range | What It Looks Like |
|---|---|
| Less than 1.0 | Not perceptible by the human eye |
| 1.0 to 2.0 | Perceptible only through close, side-by-side comparison |
| 2.0 to 10 | Perceptible at a glance |
| 11 to 49 | Colors are more different than similar |
| 100 | Colors are exact opposites |
These bands are approximate and formula-dependent, but they’re a reasonable rule of thumb for reading a spec sheet or a calibration report without getting lost in the math.
What Counts as a Good Delta E for Monitors
For color-critical work, the widely used benchmark is Delta E of 2 or lower, which is generally accepted as accurate enough that colors will read as intended for design, photography, and video work. Delta E of 1 or lower is considered close to imperceptible in most real-world workflows and is where genuinely high-end reference monitors tend to land. Casual or gaming displays often sit around Delta E 2 to 3, which looks perfectly fine for everyday use but isn’t tight enough for work where color has to match exactly, like brand colors, skin tones, or a print deliverable.
When comparing monitors, it’s worth checking three things beyond the headline number: whether the spec quotes an average or a maximum Delta E across the tested color patches, which formula was used, and which color space the figure applies to (sRGB, Adobe RGB, and DCI-P3 coverage can all report different accuracy). A bare “Delta E under 2” claim means a lot more once those details are pinned down.
How Delta E Gets Measured
Delta E isn’t something you eyeball. It’s measured with a colorimeter or, for more demanding work, a spectrophotometer, a hardware sensor that sits against the screen and reads the actual light it emits. Calibration software displays a sequence of known color patches, the sensor measures what the screen actually produced, and the software compares that measurement against the reference values to calculate Delta E for each patch, usually reporting both an average and a worst-case maximum across the full set.
Delta E in Practice
For anyone grading footage, editing photos, or doing client-facing design work, Delta E matters most as a floor of trust: it’s what tells you that what you’re seeing on screen is close enough to reality that decisions made by eye, exposure, skin tone, brand color matching, won’t fall apart when the work moves to another screen or to print. Factory calibration on a good monitor is a reasonable starting point, but panels drift with age and use, so periodic recalibration is what actually keeps Delta E low over the long run rather than just on the day the display shipped.
Common Mistakes and Misconceptions
- Comparing Delta E numbers across formulas as if they’re the same scale. A CIE76 value of 2 and a CIEDE2000 value of 2 do not represent the same visible difference.
- Assuming factory calibration is permanent. Panels age and color accuracy drifts, so a display that shipped at Delta E 1 won’t necessarily stay there without periodic recalibration.
- Trusting a Delta E spec without knowing whether it’s an average or a maximum. A low average can still hide a few badly off patches.
- Ignoring which color space the figure applies to. A monitor can post a great Delta E in sRGB and a much worse one in a wider gamut like Adobe RGB or DCI-P3.
- Treating Delta E as the only measure of a good display. Uniformity, gamut coverage, and viewing angle all affect how trustworthy a screen is for color-critical work, even when Delta E looks good on paper.
Conclusion
Delta E gives color work something it would otherwise lack: a number instead of a guess. It won’t tell you whether a color looks good, but it will tell you whether what’s on your screen actually matches what it’s supposed to be, which is the more important question for anyone making decisions based on what they see. Know which formula is behind the number, aim for 2 or under for serious work, and treat recalibration as routine maintenance rather than a one-time setup step.