Find colors that look close to a base color, with the difference measured honestly. Candidates are generated around your color in OKLCH and ranked by CIEDE2000, so "similar" means perceptually similar — you choose how many alternatives you want and how far they may drift.
Settings
Generated locally in your browser — your settings and results never leave this page.
“Similar” is a measurement, not a vibe
Two colors are similar when people struggle to tell them apart — and that is measurable. CIEDE2000 quantifies perceived difference: below 1 invisible to most, around 2 the classic tolerance threshold, 6 clearly related but distinct. This finder generates candidates around your color and admits only those within your chosen ΔE budget, each labeled with its actual score. “Similar” here is a number you set, not an algorithm’s opinion.
How the neighborhood is explored
Candidates walk a golden-angle spiral through OKLCH around the base — interleaving nudges to lightness, chroma and hue so the ring samples every direction of “different” rather than just lighter/darker. Near-duplicates are suppressed (candidates closer than ΔE 0.6 to an already-kept one are skipped), the survivors sort nearest-first, and everything is gamut-mapped so each suggestion is a real, displayable color. The walk is deterministic: identical inputs give identical results, and shared URLs reproduce them.
Situations that call for measured similarity
Rebuilding a brand color from a compressed photo or aged print (find candidates within ΔE 3 of the artifact, pick by eye). Softening a color that clashes with a neighbor without losing its identity. Generating hover and focus variants that read as “the same color responding” — ΔE 3–6 is that register. Offering alternatives when a beloved color fails contrast: often a ΔE-4 sibling passes.
Similar, shaded, or harmonized?
Three different neighborhoods, three tools: this one for perceptual nearness in all directions; Shades & Tints for the lightness axis specifically; the Harmony Generator for related-but-different scheme partners at wheel distances. Knowing which neighborhood a task lives in is most of the taste involved.
Distance, calibrated
ΔE thresholds are abstract until measured between real colors. Distances from dodger blue to five progressively-different neighbors:
| Neighbor | HEX | What it is |
|---|---|---|
| ΔE ≈ 1 | #2190fd | imperceptibly different |
| ΔE ≈ 3 | #3a8fed | hover-state distance |
| ΔE ≈ 6 | #4f8cd8 | clearly related sibling |
| ΔE ≈ 12 | #6f86b8 | same family, different member |
| ΔE ≈ 25 | #8f8f8f | no longer similar at all |
The spread slider on this page is choosing a row of this table as its outer boundary. Most practical searches live between the second and third rows — different enough to matter, close enough to substitute — which is why the default lands there.
How to use the Similar Color Finder
- Pick the base color.
- Set how many alternatives you want and how far they may drift (in ΔE units).
- Alternatives arrive sorted nearest-first, each labeled with its distance.
- Copy the list or JSON.
Frequently asked questions
How does the finder generate similar colors?
It walks a golden-angle spiral around your color in OKLCH – nudging lightness, chroma and hue in interleaved directions – then keeps candidates whose CIEDE2000 distance is within your spread, discarding near-duplicates. The result is a deliberately diverse ring of neighbors rather than ten copies of the same tweak.
What does the ΔE spread slider actually control?
The maximum allowed perceptual distance. At 2-3 you get colors most people could not pick out of a lineup; at 6 clearly-related siblings; at 15-20 loose cousins. The default 6 suits "same color, slightly different" tasks like hover states.
When would I need perceptually similar colors?
Matching a brand color that a photo or print reproduced imperfectly, generating hover/active variants that read as the same hue, softening a color that vibrates against its background, or offering near-equivalents when a color fails a contrast check.
Why are the results not just lighter and darker versions?
Because similarity has three axes: the finder varies hue and chroma as well as lightness. Pure lightness ramps are the Shades and Tints tool's job; this one explores the full neighborhood.
Are the suggested colors guaranteed to be displayable?
Yes – every candidate is gamut-mapped into sRGB before scoring, so the ΔE label describes the color you can actually copy, not a theoretical coordinate.
Can two runs give different results?
No – the candidate spiral is fully deterministic, so the same base color, count and spread always produce the identical list. That also makes shared URLs reproducible: a colleague opening your link sees exactly your results.
How do I check which of two similars is closer to a third color?
The Color Difference Calculator takes any two colors and reports ΔE2000 with a plain-language verdict – the measuring tape to this tool's search party – finder proposes, calculator adjudicates, and the two link to each other for exactly this hand-off.
