Why online DMC and bead colour charts disagree

Online hex and RGB tables for embroidery floss and fuse beads disagree because they are screen approximations, not official colour specifications. Different public charts can come from different scans, photographs, lighting conditions and colour-space conversions, and in our comparison two independent tables for the same brand routinely sit more than ΔE₀₀ 5 apart — a gap that is clearly visible to the eye. Comparing multiple independent sources removes the worst single-chart bias, though no screen value can replace checking real thread or beads.

Why do two hex charts for the same brand look different?

Most craft manufacturers do not publish measured digital colour values for every shade they sell. When a pattern app or reference site lists a hex code for a floss skein or a fuse bead, someone had to estimate it — by scanning a printed shade card, photographing physical samples, sampling a manufacturer PDF, or adapting an older public table.

Each of those steps can introduce its own systematic shift. Printed ink does not reflect light like mercerised cotton or glossy plastic, cameras and scanners handle white balance differently, and a slight change in exposure moves an entire chart lighter or darker at once. In our comparison of four independent public DMC tables, one chart ran roughly 9 lightness units darker on average than another across the whole palette. Even when a brand publishes a digital shade card, the RGB values in the document do not always match physical beads: for Artkal, the RGB values in the 2024 manufacturer chart PDF sit a median ΔE₀₀ of 5.1 away from published spectrophotometer readings by PerlerVault.

A second problem is that online tables are often copied from one another without saying so. Before comparing sources for Magpie Loom Colours, we checked for hidden duplicates and excluded two non-independent tables: one public Anchor table had 91 out of 107 paired colours identical to the corresponding DMC values, showing it had been copied through a conversion chart rather than compiled independently; another public DMC table was openly built as an average of several existing charts. Leaving copied tables in a comparison creates false agreement without adding new information.

How big are the differences?

To measure how far independent public tables drift apart, we converted every source colour into CIELAB and calculated pairwise differences with CIEDE2000 (ΔE₀₀) — a perceptual colour-difference scale where values below about 2 are hard to tell apart by eye and differences above 5 are clearly visible.

BrandIndependent sources comparedMedian pairwise ΔE₀₀Colours with ΔE₀₀ > 15
DMC4 public tables4.9–12.0179
Anchor2 public tables4.90
Hama4 public tables6.2–8.616
PerlerPerlerVault spectrophotometer + 1 public table (43 shared colours)4.90
ArtkalPerlerVault spectrophotometer, 2024 official PDF + 2 public tables4.6–7.013

Across all five brands, independent public estimates typically disagree by ΔE₀₀ 5 or more. Even the closest pairs hover right around that threshold, while two DMC charts can sit a median ΔE₀₀ 12.0 apart.

Which colours disagree most?

Extreme disagreements are not spread evenly across brands. Counting shades where at least two independent sources sit more than ΔE₀₀ 15 apart — a completely different colour to the eye — gives 179 colours in DMC, 16 in Hama and 13 in Artkal, compared with 0 in Anchor and 0 in Perler.

Individual examples show how misleading a single chart can be:

What does a consensus value fix — and what doesn't it?

For brands without instrument measurements (DMC, Anchor and Hama), Magpie Loom Colours builds a consensus value by converting each independent source to CIELAB and taking the per-channel median. For Perler and Artkal, we use published spectrophotometer readings from PerlerVault rather than averaging over them, and cross-check Artkal against the 2024 manufacturer PDF so large discrepancies are visible.

Taking a median across independent sources cancels out single-chart lightness bias and catches obvious outliers like dark-grey Hama Silver. To check whether that actually improved cross-brand matching, we tested 141 comparable DMC codes against two published DMC-to-Anchor conversion charts compiled by hand from physical thread. Moving both brands to consensus values raised our first-recommendation agreement from 23% to 34%, and raised agreement within our top three suggestions from 36% to 61%. Hand-compiled charts are still human judgements rather than ground truth, so 100% agreement is neither expected nor a proof of perfection, but a jump of that size shows the consensus values sit closer to real thread.

What a consensus cannot do is turn estimates into official measurements. We have not measured thread or beads ourselves with an instrument, and where sources disagree strongly — or where a shade has only one public source, such as DMC 3880–3895 — the screen swatch remains an approximation. Gloss, dye-lot variation, fabric colour and your own monitor calibration add further shifts that no hex code can capture.

How should you choose thread or beads, then?

You can read our full matching method, grade thresholds and per-brand data status on the How we match colours page, or report a wrong match if you spot a pair that disagrees with real thread or beads.