Lighting
A light changes the colors of the objects it falls on, and no two lights change them alike. This page takes a test light and a reference light and measures how far objects move in WLab between the two lights, for many objects and many observers at once.
An observing condition (hereafter "condition") is an observer under a light, and each condition has its own Wpt normalization, which puts its light's white at the same place: what is left is how the light renders the objects. Take an object and an observer. The WLab difference (ΔEWLab) between the object under the test light and under the reference light, each through the observer's own normalization for that light, is how far the object moves between the two lights. The mean of that difference over every object and every observer is this page's main measurement metric, Illuminant Rendering Equivalency (IRE). It is 0 when the two lights render every object alike for every observer, and the larger it is, the more the two lights differ in how they render objects; it leaves out the color of the lights themselves. Beside it, the page gives the test light's correlated color temperature and its standard color rendering indices.
1. Test light
2. Reference light
The light the test light is compared with: a single light you pick, or one found from the test light. Found from it: the Planckian (blackbody) or daylight light with the lowest IRE, the one that renders the objects most like the test light does (section 4, the default); the one with the lowest IRE among lights of your own; or the reference the color rendering indices make at the test light's correlated color temperature (CCT, section 5).
3. How far objects move between the two lights
The IRE of the test light against the reference light, and how its WLab differences spread: one for each object and each observer.
Selecting objects and observers for the IRE
The objects are the materials the lights are compared on, and the observers those who see them; the more varied the objects' spectra, the more a difference between lights shows. They apply to the IRE wherever it appears (sections 3 to 6); the color rendering indices of section 5 use their own samples and observer.
Objects
A CGATS measurement file (with spectral fields such as SPECTRAL_NM380, as spectrophotometers and chart software write them) or a table: wavelength in the first column and one object per column, or one object per row with the wavelengths across the header row; values from 0 to 1 or in percent, covering at least 420 to 680 nm. A chart you measured yourself can be used this way. The file is read in this page and goes nowhere.
Observers
Objects that move most
The 15 objects with the largest mean ΔEWLab over the observers. The shares split each one's differences (their squares, summed over the observers) into lightness, chroma (toward or away from neutral) and hue (around neutral), along the object's own hue under the reference light; the three add to 100%.
Every object under the reference light in the aw–bw plane of WLab, and for the 15 that move most, an arrow to where the test light puts them, for the observer chosen. Hover over a ring for the object.
Observer by observer
Each row is one observer's share of the IRE: the WLab differences of the objects between the two lights for that observer.
4. Closest Planckian and daylight references
The IRE of the test light against every Planckian light from 1750 to 12000 K and every CIE daylight from 4000 to 14000 K, in steps of 250 K. The lowest point of each line is the reference of that kind that renders the objects most like the test light does.
Two temperatures, two questions. The temperature of the lowest IRE is about the similarity of material color rendering: the Planckian or daylight light under which objects, each observer through its own normalization, keep their material colors closest to those the test light gives them. It depends on the whole spectrum, and not on the color of the light, which each normalization takes away. The CCT is about the closeness of the light's white point to the Planckian locus in a chromaticity diagram: the temperature of the Planckian light whose white is nearest, from the light's chromaticity alone (here in the CIE 1960 u, v diagram), whatever spectrum makes that white. Two lights with the same white point have one CCT and can still render material colors differently, so the lowest IRE can lie away from the CCT, marked by the dashed line.
5. Color rendering indices
The standard measures of color rendering also compare the test light with a reference, one made from the test light's CCT: the temperature of the Planckian light whose color is closest to the test light's, found from the light's chromaticity alone (the CIE 1931 observer in the CIE 1960 u, v diagram). Duv is the light's distance above (+, toward green) or below (−, toward magenta) the line of Planckian colors. Each index compares the colors of its own samples under the two lights, for one observer, and is 100 when the test light renders them as the reference does:
- the color rendering index Ra (CIE 13.3): 14 Munsell samples and the CIE 1931 observer, with a Planckian reference below 5000 K and daylight from 5000 K up; Ra is the mean of the special indices R1 to R8, each 100 minus 4.6 times the sample's color difference (in the U*V*W* space of CIE 13.3, after an adaptation to the reference's white);
- the color fidelity index Rf (CIE 224:2017, the same as that of ANSI/IES TM-30): the CIE's 99 color evaluation samples, the CIE 1964 observer and color differences in CAM02-UCS, with a Planckian reference below 4000 K, daylight above 5000 K and a mix of the two between;
- the gamut index Rg (TM-30): the area the 99 samples' mean colors in 16 hue bins span under the test light, as a percentage of the area under the reference; above 100 the colors are more saturated on average, below 100 less.
Unlike the IRE, each uses one observer, a fixed set of samples and a reference set by the CCT; the IRE against the same references is given with them.
The special color rendering indices R1 to R14
Each test color sample's index, 100 minus 4.6 times its color difference between the two lights; Ra is the mean of the first eight. R9 to R14 (in gray), strong red, yellow, green and blue, a skin tone and a leaf green, are not part of Ra.
16 hue bins of TM-30
The 99 samples grouped by their hue under the reference. For each bin: its local color fidelity, and the shift of its mean color along the angle bisecting the bin (chroma, as a share of the reference's distance from neutral; + more saturated) and across it (hue, in radians; + toward a larger angle). In the graphic, each bin's mean color under the reference is put on the dashed circle, and the arrow goes to where the test light moves it. Checked against colour-science 0.4.7 for the CIE fluorescent and LED lights: Rf and Rg to 0.002, the bins' shifts to 0.0001.
6. Many test lights at once
Sections 1 to 5 look at one test light. This section takes a set of test lights, chosen below, and does for each of them what sections 4 and 5 do for the test light of section 1. It searches the Planckian lights from 1750 to 12000 K and the CIE daylights from 4000 to 14000 K, every 250 K, for the one with the lowest IRE against the test light: the same search that gives section 2 its default reference. And it gives each test light's CCT and color rendering indices. The test light and reference chosen in sections 1 and 2 play no part here.
The table gives, for each test light, its CCT; then the Planckian light with the lowest IRE and that IRE, and the daylight with the lowest IRE and that IRE, the smaller of the two in italics; and last its Ra, Rf and Rg. The first two plots show each test light's IRE against every Planckian (left) and every daylight (right) reference, colored by the kind of light, with the lowest point of each line marked. The last two put each test light's lowest IRE against its Rf and against its Ra, to set the IRE beside the color rendering indices.
The lowest IRE is the smaller of each light's closest Planckian and closest daylight IRE. Hover over a point for the light.