Document the steps you use to calculate color difference by saving both original opaque #RRGGBB inputs, the Delta E 2000 output, and the tolerance and viewing context for the decision. This turns a one-time screen comparison into a traceable note: another person can see exactly which values were compared, which calculation path produced the number, and how the result was interpreted.
The documentation should preserve two related layers of the task. The calculation layer contains the entered colors, conversion assumptions, and returned Delta E 2000 value. The decision layer contains the application, chosen tolerance, viewing or production conditions, and resulting action. Keeping them together answers the question directly while avoiding the mistake of presenting a formula-based number as a universal pass or fail.

Separate the calculation from the decision
Start by naming the question the comparison answers. “Are these two design-token colors close enough for this interface revision?” is different from “Does this screen color predict that two printed swatches match?” The first can be supported by entered sRGB coordinates; the second requires physical measurement. The calculator does not sample a real object, infer a printer profile, or certify a match, so the note should not imply any of those outcomes.
Use a simple record with two parts:
- Calculation evidence: the exact two HEX values, accepted color representation, conversion and formula settings, Delta E 2000 result, and any D50 CIELAB coordinates you need to retain.
- Decision evidence: the workflow, tolerance source, viewing or production context, limitations, and the action taken.
That structure makes the entry useful for design-token revisions, repeatable QA notes, or comparisons of sampled image colors. It also keeps a mathematically close digital pair separate from a production claim that would require calibrated equipment.
Prepare the colors without changing the comparison
The Color Difference Calculator accepts exactly two opaque six-digit sRGB HEX values in #RRGGBB form. Include the leading number sign and all six digits. Because the input space and opacity are fixed, prepare the color values before opening the calculator rather than expecting it to interpret shorthand, transparency, another gamut, or a named CSS color for you.
| Color form or condition | How to prepare it | Why preparation matters |
|---|---|---|
| Exactly #RRGGBB | Enter both values unchanged as opaque sRGB colors. | This matches the calculator’s only accepted input format. |
| Three-digit #RGB shorthand | Expand it to six digits in your source workflow first, and record both the source and expanded values. | The calculator rejects shorthand rather than applying a hidden parsing rule. |
| Alpha-bearing #RRGGBBAA | If transparency matters, composite both colors over the same known background before comparing, then enter the opaque results. Retain the source alpha separately. | A translucent foreground is not one standalone opaque sRGB color. |
| Named CSS color or Display P3 value | Resolve it to a defined six-digit sRGB value in a separate color-management workflow and document that source. | This makes the comparison space explicit; the calculator is not an ICC profile converter. |
| Empty, malformed, or extra-digit input | Correct the value before comparison. The calculator returns an explicit error and no result. | A calculation should never proceed from a guessed or reinterpreted color. |
Keep the distinction between the source and the comparison value. If a foreground has transparency, the source color and the opaque composited value are different facts; note both, along with the background. If a color came from an image sample, the calculator can compare the resulting six-digit values, but image picking and sampling happen outside it. This preserves the original application context rather than silently replacing it.
Run the CIEDE2000 comparison and capture its output
- Open the Color Difference Calculator and enter the first color as an exact opaque #RRGGBB value.
- Enter the second color using the same six-digit, leading-number-sign format.
- Select Compare colors so the tool converts both entries to D50 CIELAB and calculates CIEDE2000.
- Copy the numeric Delta E 2000 result exactly as displayed; do not replace it with a visual estimate or an unlabeled “difference” value.
- If the audit trail needs the intermediate values, copy each color’s D50 L*, a*, b* coordinates shown by the tool. Treat them as displayed values: Lab coordinates are rounded for display, while the unrounded numbers continue into the difference formula.
The documented conversion path normalizes each sRGB channel, linearizes it, converts through D65 XYZ, adapts D65 to D50 with the Bradford matrix, and converts the result to CIELAB. It then applies CIEDE2000 with the common default parametric weights kL = kC = kH = 1. This is not a straight Euclidean distance between the three displayed Lab numbers. The method follows the W3C CSS Color 4 sample conversion path.
Everything runs locally in the browser, and the entered color values are not uploaded. That supports a concise handling note, but it is not a substitute for the source record: your documentation should still contain the exact inputs and output needed to repeat the comparison.
Create a documentation entry others can reproduce
A copy-ready note is more reliable than a screenshot alone. A screenshot may omit the purpose, tolerance, or conditions that give the number meaning; the written entry should make those fields explicit.
| Documentation field | What to record |
|---|---|
| Purpose and application | The decision being supported and where the colors are used, such as a digital token revision or sampled-image comparison. Keep digital and physical claims separate. |
| Input A and Input B | The exact two #RRGGBB values entered, including their original order. |
| Source and preparation | Where each color came from, any alpha value, the compositing background, and any conversion performed before entry. |
| Calculation method | Opaque sRGB converted to D50 CIELAB, followed by CIEDE2000 with kL = kC = kH = 1. |
| Reported output | The numeric Delta E 2000 value as displayed, plus the D50 Lab coordinates if they are needed for the audit trail. |
| Tolerance and action | The workflow-specific limit and its source, or a statement that no universal threshold was applied. Record the decision made from the result. |
| Context and limitations | Relevant viewing or production conditions and any reason the result cannot prove a physical match. |
| Review trail | Your own reviewer, date, design-file version, production revision, or other details needed to identify the comparison later. |
Suggested wording: Compared opaque sRGB colors [HEX A] and [HEX B] with the Color Difference Calculator. The calculator converted both to D50 CIELAB and returned Delta E 2000 = [reported result] using kL = kC = kH = 1. Decision tolerance: [workflow-specific limit and source]. Context: [display or production conditions]. This is a comparison of the entered coordinates, not a physical measurement or universal acceptance ruling.
Fill the placeholders rather than inventing values. In particular, do not add a generic pass/fail number. If your organization has an approved tolerance, name that rule and its scope; if it does not, write that no universal threshold was applied. For a token revision, record the old and new token identifiers. For sampled colors, identify the sampling source and note whether compositing or conversion occurred before the comparison.
Interpret the Delta E 2000 value in context
A larger Delta E 2000 value means a greater difference under the formula. It does not automatically mean a larger visible error on every display or in every lighting condition. CIEDE2000 includes adjustments for lightness, chroma, and hue, including behavior near neutral colors, hue angles across the zero-degree boundary, and the blue-region rotation term. Those features make it more nuanced than a simple three-coordinate distance, but they do not create one universal tolerance.
Two useful checks are built into the result. Identical colors return zero, and comparing A with B gives the same result as comparing B with A under the same default conditions, apart from floating-point display precision. These checks help catch inconsistent entries or transcription errors. They do not tell you whether a nonzero difference is acceptable for paint, textiles, printing, brand review, photography, or interface design; those workflows can use different instruments and acceptance limits.
Record the conditions that could affect interpretation: display calibration and brightness, ambient light, monitor gamut and profiles, surface texture, surrounding colors, and observer vision. A screen swatch is only a visual reference. Browser rendering can change appearance, so the numeric result and source colors provide the reproducible evidence, while the preview supplies context. If the decision concerns a physical sample, use calibrated measurement equipment and the standard, tolerances, illuminant, and observer conditions required by that process.
Check the record for gaps and errors
Before signing the note, read it as if you were the reviewer who did not run the calculation.
- Both entered values are exactly six-digit #RRGGBB strings with leading number signs.
- Each value represents an opaque sRGB color, and any alpha compositing background and source color are recorded.
- The Delta E 2000 number is copied from the calculator rather than estimated.
- Any retained Lab coordinates are identified as D50 values and treated as rounded display values.
- The note states the default kL = kC = kH = 1 weighting and the sRGB-to-D50 conversion path.
- The tolerance, viewing or production context, decision, and limitations are present.
- The language does not claim that a screen calculation proves a print, paint, textile, or other physical match.
For an order sanity check, run the same pair again with the two fields reversed. Under the default conditions, the result should remain the same apart from floating-point display precision. If it does not, check for a mistyped or substituted HEX value before drawing a conclusion.
The implementation has also been checked against eight sign-sensitive reference pairs from the Sharma, Wu and Dalal supplementary CIEDE2000 data. That verification supports the formula implementation, but it does not replace a process-specific tolerance or physical measurement. With those distinctions in the record, the note documents both what was calculated and what the result can legitimately support.
Related reading: How to Document Steps When Using Color Palette Generator.
Related reading: How to Fix a Wrong CIEDE2000 Color Difference Result.