How can two lights look white but show colors differently?

The white point only fixes the overall balance of a light, not the shape of its spectrum. Two sources can integrate to the same white yet have very different spectral power distributions - one smooth, one spiky. A colored object reflects only the wavelengths its pigment allows, so if the light is missing those wavelengths, the color it reflects is dull or shifted, even though the light itself still looks white.

What is CRI (Color Rendering Index)?

CRI is a 0-100 score of how faithfully a light source renders colors compared with a reference source of the same color temperature (a blackbody below 5000K, daylight above). The general index Ra averages the fidelity of eight standard test colors; 100 means colors look exactly as they would under the reference, and lower numbers mean larger color shifts. Newer metrics like TM-30 add fidelity (Rf) and gamut (Rg) measures.

Why do cheap LEDs make food or skin look bad?

Many inexpensive LEDs build white from a few narrow spectral peaks with gaps in between, especially in the deep reds. Skin and food rely heavily on those red wavelengths, so under a gappy spectrum they render sallow, gray, or unappetizing even though the light looks white. A source with a smoother, fuller spectrum and a high CRI renders them naturally.

Color Rendering: Why Two "White" Lights Reveal Colors Differently

You've seen it: a shirt looks navy in the store and black in the parking lot, or a fruit bowl looks lush at home and grim under the office lights. Both lights looked white - so what changed? The answer is color rendering. A white point tells you nothing about the shape of a spectrum, and it's the shape that decides which colors a light can bring to life and which it flattens. This is the interactive guide to color rendering and the CRI.

Colorimetry · 88 4 Live Demos ~30 min read Light quality

CRI

Color Rendering Index

Ra ≤ 100

100 = perfect fidelity

reference

Judged vs a smooth spectrum

gaps

Missing wavelengths dull color

Same white, different colors

An object has no color of its own - it has a reflectance, a rule for which wavelengths it bounces back and which it absorbs. What you see is that reflectance multiplied by whatever light is falling on it. So a color can only appear if the light actually contains the wavelengths the object reflects. A perfect crimson tomato needs deep-red light to be there; if the lamp emits almost no red, the tomato has nothing red to reflect, and it looks brown - no matter how white the lamp itself appears.

And a lamp's whiteness is deceptive. The white point - its color temperature - is just the overall balance of the spectrum. Two lamps can hit the same white while having wildly different spectral shapes: one a smooth, full rainbow (sunlight, incandescent), the other a few narrow spikes with dark gaps between them (a cheap LED or fluorescent tube). They look identical on a blank wall and completely different on a bowl of fruit. That difference is color rendering.

The core idea: what you see = reflectance × light spectrum. The white point sets the balance of the light; the fullness of the spectrum sets how well real colors survive. A high color-rendering source has light at every wavelength; a low one has gaps that silently erase colors.

Two lights, one white, two renderings

Here are two light sources adjusted to the same white: a reference with a smooth, full spectrum, and a test LED-style source built from three narrow peaks. The spectra are drawn above; the same eight-color palette is rendered under each below. Watch the saturated colors - especially reds and cyans - shift and dull under the spiky source, even though its white is the same.

Interactive 01 · Reference vs test

The same palette under two "white" lights

Top: the two spectral power distributions (smooth reference in gold, spiky test in blue). Bottom: the eight test colors rendered under each - reference on the left of each pair, test on the right. The fidelity score on the right is a CRI-style average (100 = renders like the reference). A smooth reference scores ~100; the peaky source falls well short.

Turning down the quality

Color rendering isn't on/off - it's a dial. Slide a source from a smooth, full spectrum toward a gappy, peaky one and watch the fidelity fall and the palette curdle in real time. This is the difference between a premium high-CRI bulb and a bargain-bin one, made continuous.

Interactive 02 · The quality dial

From full spectrum to gappy

Drag from a smooth spectrum (high fidelity) toward a spiky one (low fidelity). The white point is held steady, so the light keeps looking white while the colors it renders drift further from true. Notice how the reds and greens are the first to suffer.

Spectrum quality 100

Where the spectrum has gaps

To see which colors a gap destroys, build the spectrum yourself. This source is three phosphor peaks - blue, green, and red - each with its own strength. Pull the red peak down and the whole warm end collapses: tomatoes, skin, and wood go gray and lifeless, because there's no red light left for them to reflect. The demo names the color that suffers most.

Interactive 03 · Build the spectrum

Remove a peak, lose a color

Three peaks make the white; adjust each one's strength. As you starve a region of the spectrum, the colors that live there stop rendering. Drop the red peak to watch warm colors die; drop green to hollow out foliage. The readout flags the worst-hit sample and the overall fidelity.

Blue peak 100%

Green peak 100%

Red peak 100%

Scoring the render: Ra and per-color fidelity

CRI turns all this into a number. It renders a set of standard test colors under the source and under a same-temperature reference, measures how far each one shifted, and reports the fidelity of each (Ri) plus their average, Ra. A single low color can drag the average down - which is why the special deep-red index R9 is watched so closely. See each color's score.

Interactive 04 · Fidelity per color

Ra is an average - look underneath it

Pick a source type and see the fidelity of each test color as a bar (100 = perfect), with the average Ra called out. A source can post a decent Ra while badly failing one or two colors - the bars reveal what the single number hides.

Source

What the numbers mean

A quick field guide to the vocabulary of light quality.

Ra (general CRI)

Average fidelity of 8 pastel test colors vs a reference. 90+ is good, 80-90 acceptable, below 80 shows visible shifts.

R9 (deep red)

A separate index for saturated red - often the weakest and most telling. Crucial for skin, food, and wood.

Reference illuminant

The yardstick: a blackbody below 5000K, daylight above. CRI measures distance from this, not from reality itself.

TM-30 Rf

A modern fidelity measure using 99 samples - a fairer average than CRI's eight, and less easily gamed.

TM-30 Rg

A gamut measure: above 100 the source boosts saturation, below 100 it dulls it. Fidelity alone can't tell you which.

Metamerism link

Poor rendering is metamerism's cousin: colors that matched under one light diverge under a gappy one.

Best practices and pitfalls

White point ≠ quality

Color temperature and CRI are independent. A 3000K bulb can render well or badly; check both numbers.

Don't trust Ra alone

Ask for R9 and, ideally, TM-30 Rf/Rg. A good Ra can still hide a red that renders like mud.

Match light to task

Galleries, retail, print proofing, and dentistry need 90+ CRI; a storeroom can live with less.

Judge color under known light

Evaluate paint, fabric, and print under a controlled high-CRI source (like D50 booths), never a random bulb.

Beware the store-vs-home gap

A product chosen under one spectrum can look wrong under another - a metameric surprise. Take a sample home.

Fuller spectrum, safer color

When in doubt, choose the source with the smoothest, most complete spectrum - it forgives every color.

"A light doesn't paint an object - it lends it wavelengths. Take away the reds and the reddest thing in the room simply has nothing left to give back. The lamp still looks white; the world just looks poorer." Editorial summary · you can only reflect what's there

The takeaway: color rendering is about the shape of a light's spectrum, not its white point. What you see is reflectance times the light, so gaps in the spectrum erase the colors that live there. CRI's Ra scores fidelity against a same-temperature reference; watch R9 and TM-30 too, because an average can hide a color that's quietly gone gray.

Test your understanding

Six questions on color rendering, spectral gaps, CRI, and the metrics beyond it. Instant feedback, no scores recorded - a wrong answer comes with a short explanation.

Quick check

Loading…

Question 1 of 6

Continue your journey

Color rendering sits at the meeting point of light sources, the standard observer, and metamerism - here's where to go next.

Physics · 28

Color Rendering: Why Two "White" Lights Reveal Colors Differently

Same white, different colors

Two lights, one white, two renderings

The same palette under two "white" lights

Turning down the quality

From full spectrum to gappy

Where the spectrum has gaps

Remove a peak, lose a color

Scoring the render: Ra and per-color fidelity

Ra is an average - look underneath it

What the numbers mean

Best practices and pitfalls

Test your understanding

Quick check

Continue your journey

Color Temperature and White Balance

Emission: Neon, LEDs, and Gas-Discharge Light

Metamerism Explained

The Standard Observer: 2° vs 10° and the Average Eye

CIE XYZ Explained

Human Color Vision: Cones, Opponent Signals, and the Brain