A fixture can be the correct colour temperature, measure well on paper, and still make a face look ill. Colour temperature describes the overall cast of a white light. It says nothing whatever about how that light renders the colours it falls on, and those are separate properties that a single number cannot carry.
What the familiar index actually measures
The colour rendering index published by the International Commission on Illumination compares how a source renders a set of test colour samples against a reference of the same colour temperature. The headline figure, Ra, is the average of the first eight samples. Those eight are moderately saturated pastels.
The important omission is the ninth sample, a strong saturated red, which is not part of the average at all. A fixture can score well on Ra while rendering deep reds badly, and that is precisely the failure that shows up on skin, on lips, and on anything crimson in the costume plot. Where a fixture's data sheet publishes Ra alone, the number is not wrong, it is incomplete in the specific direction that matters most for people.
Why blue pumped white runs short of red
Most white LEDs are not white emitters. They are blue emitters with a phosphor coating that converts part of the blue into a broad yellow band. The eye adds the two together and reports white. The resulting spectrum has a tall narrow spike where the blue diode sits, a broad hump through the greens and yellows, a dip in the cyan region between them, and a tail that falls away through the deep reds.
Set a filament beside it and the contrast is structural. Thermal radiation produces a smooth continuous curve rising steadily into the red, with no spikes and no gaps, which is why tungsten renders almost everything acceptably and why it remained the reference long after it stopped being efficient.
Fixtures built from several coloured emitters rather than from phosphor conversion fill some of these gaps deliberately, adding amber, lime or deep red channels for that reason. It costs slots, cost and complexity, and it is why serious theatrical LED fixtures carry more emitters than an architectural product does.
The newer method, and why it reports two numbers
The Illuminating Engineering Society's method known as TM-30 evaluates a far larger set of samples and reports fidelity and gamut separately. Fidelity says how close rendering is to the reference. Gamut says whether the source pushes saturation up or flattens it, which is information the older single figure could not express at all: two sources with identical fidelity can make a costume look vivid or lifeless.
Both methods come from professional and standards bodies rather than from manufacturers, and both are published with their procedures. That is the reason to prefer them to any number invented for a brochure.
Metamerism, and the dress that changed colour in the wing
Two lights can produce an identical white and render one pigment identically while disagreeing completely about another. Colours matched under one source therefore separate under another, and the effect is routine rather than exotic.
The practical consequences all involve a decision made in one place and revealed in another. Costume matched under workroom fluorescents. Makeup chosen under dressing room lamps. Paint mixed under a north window. Each looks correct where it was decided and can be visibly wrong in the state it appears under. The remedy is procedural rather than technical: judge under the light the audience will see, or at least check under it before the technical rehearsal, when there is still time to change something.
Colour on colour, and the trap in gel
Rendering failures compound when a coloured filter is put in front of an already gapped source. A filter is subtractive: it removes what it does not pass. Deep red gel in front of a filament has plenty of red to work with. The same gel in front of a blue pumped white LED is asking for wavelengths that are barely present, and the result is dim, muddy and frequently a different hue than the swatch book promises.
This is one reason colour mixing fixtures exist rather than filtered white ones, and it is why a swatch book calibrated against tungsten is a poor guide to what a filter will do on a modern rig. The book is not wrong; the light behind it has changed.
Cameras do not agree with eyes
A sensor's response to the spectrum is not the eye's response, and a gap in the deep red that a live audience tolerates can appear on a recording as flat, grey skin that no white balance recovers. Any production that will also be captured should be assessed through the camera, because that assessment is a different measurement from the one made from the seats.
What we cannot verify
Rendering figures on fixture data sheets come from the companies selling the fixtures and are usually stated without the drive level, junction temperature or measurement geometry, all of which change the result; LED output shifts as a fixture warms. Values quoted for a fixture family may have been measured on one member of it. We have not measured any product and reproduce no manufacturer's figure. Where rendering matters to a production, the assessment worth trusting is the one made on the actual fixtures, at working temperature, on the actual faces and fabrics.
The short version
- Colour temperature describes the cast of a light, not how it renders colours.
- The headline rendering index averages eight pastels and omits strong red.
- Blue pumped white LEDs dip in the cyan and fall away in the deep red.
- Multi emitter theatrical fixtures add amber, lime or red to fill those gaps.
- TM-30 reports fidelity and gamut separately, which one number cannot.
- Judge costume, makeup and paint under the light the audience will see.