LED Spectrum Mixer

Switching to Custom sources keeps your current mix. Add a source from the example library; it starts at zero until you increase its weight. Remove a source to remove its contribution. Each source has one weight control.

Save A before experimenting. Restoring or swapping A restores the complete source selection and mode. Share links and CSV exports also retain that selection. Five-channel loads the original 4000 K starter; saved A is kept.

These are relative spectral weights. A single source keeps the same colour and normalised spectrum when its weight changes. Combine sources to change the shape, or save A and use Shared scale to inspect changes in relative output. The weights are not calibrated dimmer settings, watts or strip lengths.

Yes. CCT describes the white point, while colour rendering depends on the full spectrum. Use “White vs RGB” to compare two example mixes with approximately the same white point. Their Ra and R9 values are calculated from their different spectra.

This tool is about these example sources. It does not establish the performance of a particular RGB or white LED product.

For near-white light, positive Duv lies above the blackbody locus, generally toward green; negative Duv lies below it, generally toward pink. CCT alone cannot describe this difference. Duv is a geometric colour metric, not a universal preference score.

Mixing two white sources does not normally keep the mix on the blackbody locus. The balanced 4000 K example therefore includes a small green contribution. Every participating channel remains visible.

“Shape · %” normalises each curve to its own maximum. Proportional changes preserve the shape. Save A, increase the weights, then select “Shared scale” to see both curves against one common relative scale.

These weights have no calibrated relationship to watts, lumens, current or dimmer percentages. Absolute brightness cannot be inferred from this tool.

All channels off has no defined colour. A strongly coloured mix does not have a useful white-light CCT in this tool. CRI also has a reference-chromaticity applicability limit: we withhold Ra and R9 when the CIE 1960 uv distance to the selected reference exceeds 0.0054, or the reference temperature is outside the implemented range.

“—” means not applicable or unavailable. It does not mean a score of zero. Valid negative colour-rendering indices are shown as negative numbers.

Example data, not LEDYI product measurements. Warm white uses CIE LED-B1 and cool white uses CIE LED-B5. __SOURCE_METHOD__ Each channel has equal unit integrated spectral area before its weight is applied.

Calculations use a 1 nm grid and the CIE 1931 2° observer. CCT and Duv use Robertson 1968. CRI uses the CIE 1995 test-colour method with a Planckian reference below 5000 K and a CIE daylight reference at or above 5000 K. Ra is the average of R1–R8; R9 is computed independently. No TM-30, room illuminance, electrical or driver model is included.

The displayed spectra cover 380–780 nm. Source whites have that same supplied range and are zero outside it. The computation grid extends to 830 nm; the exact calculation domains and independent comparison are recorded in the verification notes.

CIE reference LED spectra · CIE observer data · Independent CRI reference

Numerical verification pending.

The R1–R14 colour keys are calculated from each sample’s spectral reflectance under D65, retaining lightness when converted to sRGB. They are display approximations, held fixed as the LED mix changes. CIE test-colour sample data

CIE 2018, Relative spectral power distributions of illuminants representing typical LED lamps, 1nm spacing, DOI 10.25039/CIE.DS.dhcw57sd, CC BY-SA 4.0. Changes: LED-B1/B5 extracted, zero-padded and area-normalised; these derived datasets retain that licence. Additional scientific tables: Colour Developers, Colour Science for Python 0.4.6, BSD-3-Clause; original CIE provenance retained.

The design follows a narrow task: understand a mix, change it, compare it and take the result away. The forum posts below are individual user experiences, not a representative survey or an optical standard.

Tool references: Yuji’s spectrum / R9 comparison informed the side-by-side task; ams OSRAM ColorCalculator informed explicit channel inputs; PNNL / IES Spectral Calculator informed data provenance and exports. Their measurements, code and product claims have not been copied into this tool.