Light-dependent mixing · 38 illuminants · experimental
Choose an Atlas reference in the table, select the lights it must match under, and calculate a new recipe. The reference identity stays fixed.
Swatches: spectrum → CIE D65 / 2° → sRGB. Their display stays at D65; the table compares every illuminant numerically.
| Illuminant | Selected | Existing recipe ΔE00 | New recipe ΔE00 |
|---|
Light-source provenance and calculation
Loading Mixing Atlas …
What do the 569 open HLC fields mean?
Open means no recipe within the HLC tolerance was found under the tested search conditions. This does not establish that the target is out of gamut.
A target may lie outside the tested palette's achievable range. Limits of four components and 100 integer model parts, or a bounded search, may also prevent a match. The cause has not yet been established.
The status applies to the tested palette, mixing model and search limits. No match found is established; unreachable with this palette still needs testing.
B/A · target, existing and new mixture
Digital model preview from reflectance spectra · CIE D65 / 2° → sRGB · no measured or photographed mixtures. The symbolic master HEX remains unchanged and may differ from this spectral preview.
The same screen conversion is applied to all three swatches. Channels outside sRGB are clipped and marked; this is independent of the open HLC search status.
Your spectral data · local import
Your base colours are calculated in a separate user palette. Existing atlas recipes and reference values are preserved. The imported file is not sent to a server or automatically stored permanently.
No user palette loaded yet.
Data validation and model limits
1–64 base colours in one declared material system. Wavelengths must increase strictly and cover 380–730 nm. Reflectance must be declared as fraction 0–1 or percent 0–100. No extrapolation. Values are linearly interpolated to 36 bands. The original import is retained in the export. Fluorescent, metallic and other effect colours with values above 1 require a different model and are rejected here.
The example contains synthetic test curves only, not measured products. Even for real single-colour spectra, the calculation initially assumes S=1. R=0 is floored to 0.000001 only inside the K/S calculation. Incomplete measurement or licence information is reported. A series name does not prove material compatibility.
A full atlas search may take several minutes depending on your device. The search uses HLC only, up to four components and 100 integer model parts. The bounded candidate pool does not prove a global optimum. An open search does not prove fundamental unreachability. Illuminant differences are assessed afterwards.
Atlas page
Select a reference to view its recipe and diagnostics. Grid swatches use the symbolic master HEX only.
| Reference | HLC | Max. ΔE00 | |Δλ*| nm |
|---|
Grid: L runs from light at the top to dark at the bottom; C from low on the left to high on the right. Dashed border: open HLC field. Hue is not assessed for neutral fields.
Current model recipe
Reflectance spectra · 380–730 nm
Additional illuminant observations
| Illuminant | ΔH° | ΔL | ΔC | ΔE00 |
|---|
These HLC/Lab values are calculated from spectra under each illuminant. They do not overwrite the target address or original master values.
Full atlas page overview
Each tile shows existing HLC matches / references. Illuminant diagnostics do not change this status.
Illuminant summary for HLC model matches
| Illuminant | Ø ΔE00 | Median | 95th percentile |
|---|
All values are model differences from the target spectrum under the same illumination. No new optimisation was performed. Differences between illuminants are not a formally calculated metamerism index.
Method, material basis and next validation steps
Reference identity is frozen in the PKL master. The existing HLC match rule is unchanged: five-unit cells with half-open boundaries, circular H and hue ignored for neutral fields. It uses the existing D50/2° kernel over 400–700 nm. The additional illuminant diagnostics use all available reflectance data over 380–730 nm, linear interpolation to 1 nm and trapezoidal integration with the CIE 1931 2° observer. The relative white point uses the same illuminant and interval; no extrapolation or chromatic adaptation.
Pigment models currently assume S=1; Sirius candidates use Dr Backes's separate K/S and drop-volume models. Sirius and other pigment models are not mixed together. Up to four components and 100 integer model parts. Raw spectra are retained. Screen colours are symbolic reference displays, not measurement evidence.
Among the HLC matches, 775 recipes use one named series, 8,084 combine different named series and 3,855 contain at least one sample with unclear series assignment. A single series alone does not establish a calibrated mixture.
ARBE λ* is an additional spectral diagnostic (here λ*V2 = 380 nm + the reflectance integral over 380–730 nm). An independent benefit from λ*-guided recipe optimisation has not yet been tested. Next digital comparison: identical base colours, HLC cells and search budgets; spectral optimisation with and without the λ* criterion. Then independent physical mixtures and reflectance measurements.
Integrity and validation results
Attribution and licensing
ATLAS references: HLC Colour Atlas XL / freieFarbe e.V.; reference system and analysis: ATLAS Clarus, Norbert Woiwod. freieFarbe · ARBE λ*. Reference values come from the supplied frozen master; no new measurements were produced.
Base samples: supplied file pigments.rs, using its product/pigment names. Provenance, licensing and measurement conditions are not fully documented for every sample. This display adds no material approval.
Sirius model: Dr Ulrich Backes, backes-art.de, mixing model for Lascaux Sirius. His message of 1 October 2026 states a switch to CC BY-SA; the exact version is not yet confirmed and is not invented here. Recipe selection, reordering, ATLAS assignment and additional model analyses are by ATLAS Clarus. Final licensing information for public redistribution remains to be clarified.
Illuminant and observer data: International Commission on Illumination (CIE), D50, D65, A, nine LED illuminants and CIE 1931 2° functions, according to metadata CC BY-SA 4.0. Spectral restriction, interpolation, integration weights and derived diagnostics are modifications. ΔE00 test data: Sharma, Wu and Dalal (2005).
Digital pilot v0.3 · 2 October 2026 · no claim of complete physical coverage or measured mixing accuracy.