Reference

How we match paint colors across brands

Paint Color HQ has 26,000+ paint colors from 13 brands and over a million pre-computed cross-brand matches. This page documents how those matches are calculated, what the closeness labels mean, and where the methodology has limits.

Short version: every color in our database has a hex code, RGB triplet, and (when available) an LRV value. To find cross-brand matches we compute the CIEDE2000 Delta E score between two colors — a perceptual color-difference standard published by the International Commission on Illumination in 2001 (CIE Publication 142) and widely used in paint, fabric, and ink manufacturing for quality control. Lower scores indicate closer digital colors. They do not guarantee that two paints will match on a finished wall or work for touch-ups. Compare physical samples in the intended finish and lighting.

The matching pipeline

Comparing one color with the catalog requires at most one comparison per candidate color. Comparing every color with every other color is much larger, so we pre-compute cross-brand candidates for the color and match pages.

1. RGB pre-filter. For each source color, candidates are narrowed to colors whose RGB channels fall within ±30 of the source. This Euclidean filter is cheap and runs on indexed columns; it reduces the candidate set before Delta E scoring, so the stored shortlist is not a physical paint test.
2. LAB conversion. The pre-filtered candidates are converted from sRGB to CIE LAB color space — the perceptual color space CIEDE2000 operates on. LAB is designed so that equal distances correspond (approximately) to equal perceived differences, unlike RGB where two visually distinct colors can be numerically close.
3. CIEDE2000 calculation. The 2001 CIE formula (Luo, Cui, Rigg) computes Delta E using LAB lightness, chroma, and hue with chroma-dependent weighting. We use the standard implementation with kL=kC=kH=1, which is the default for surface coatings (paint, fabric, ink) per ISO 11664-6:2008.
4. Ranking and storage. For each source color we keep a limited ranked shortlist of cross-brand matches, ordered by Delta E ascending. These pre-computed matches are what the color detail and match listing pages read at runtime, so the cross-brand pages serve from the cache rather than re-computing.

How we label digital similarity

The plain-language verdict labels you see on color and match pages come from this table. These are shortlist labels for digital values, not certified paint tolerances.

Delta E	Verdict	How to use the digital result
< 2	Nearly identical	The difference is hard to spot, even side by side. Still a digital result, so confirm with a physical sample.
2 to < 5	Very similar	Close, but you may notice a difference side by side. Compare hue and lightness before sampling.
≥ 5	Visible difference	The colors are clearly different. Treat it as an alternative to evaluate, not a substitute.

Data sources

The 26,000+ colors in our database come from each manufacturer's published color chip data: name, color number, hex code, and digital color values. Our importers also calculate lightness estimates; those estimates are not manufacturer-measured LRV. We do not measure paint chips ourselves — check the manufacturer reference for current specifications. Unless explicitly sourced as manufacturer LRV, treat catalog lightness values as estimates. Color names and product codes are trademarks of their respective brands.

Undertone and color-family classifications are derived from each color's LAB position relative to family centroids. The classifier sometimes diverges from a human eye on borderline colors — a famously-warm gray with LAB a* and b* close to neutral may classify as neutral rather than gray. We acknowledge this and use the brand's own family label (when set) before falling back to the classifier output.

The matches table is recomputed whenever the underlying color set changes — new brand imports, color additions, or LAB recalibrations. Match records carry no expiration date because color chip data is essentially immutable; if a brand discontinues a color we mark the row but keep the match results for historical lookup.

Known limitations

CIEDE2000 measures perceived color difference under standardized viewing conditions. Real walls don't live in standardized viewing conditions, so a Delta E match always carries caveats:

Sheen and finish. A flat and a satin in the same hex code will not look identical on a wall. Sheen affects light reflection and apparent value — we recommend matching sheen across rooms before comparing colors.
Base coat. Paint over a tinted base reads differently than paint over a pure white primer, especially for thin or transparent pigments. Most differences disappear by the second coat, but the first coat can be misleading.
Lighting. A color sampled under 2700K warm bulbs looks meaningfully different from the same color under 4000K daylight bulbs. Two colors that look nearly identical under one light source can show a visible difference under another. Always sample in your room's actual lighting, at multiple times of day.
Manufacturer batch variation. Published hex codes are nominal — actual mixed paint has tolerances. Even a nearly identical digital result is a shortlist entry, not a guarantee, because the same tolerances apply when a brand mixes its own batches.
Mineral pigments and complex formulations. Brands like Farrow & Ball use mineral pigments and proprietary base structures that produce a finish character (chalky depth, color shift through the day) that can't be captured by a single hex code. A nearly identical digital match between an F&B color and a standard latex color can still look different on a wall.

The recommendation across all of these limitations is the same: always sample physically before committing to a full room, at minimum 4×4 inches, on the actual wall, viewed at multiple times of day.

Independence

Paint Color HQ is independent of every paint brand listed on the site. No brand pays for placement, ranking, or favorable mention. Color matches are derived from published data and calculated mathematically — they don't reflect editorial or commercial judgment.

The site is run by Philip Cameron, a homeowner-builder who hit the cross-brand matching problem during a personal renovation. Read more about the origin in the about page.