What Is CRI vs. TLCI in LED Lights—And Why Do Cheap Bulbs Make Skin Look Pale or Green?

Walk down the lighting aisle at any hardware store—or browse budget ring lights online—and nearly every box boasts "High CRI 80+" or "90+ CRI Daylight." Yet the moment you turn on a standard household LED bulb over your desk and open a video call or mirrorless camera, your face looks waxy, pale, or faintly greenish-yellow. Why does a light source that passes an 80-point or 90-point color accuracy test make human skin look unwell? As documented in the International Commission on Illumination (CIE) Color Rendering Index specification and technical photometry analyses by UPRtek, CameraStuff, and Light Sky, the answer lies in a 50-year-old mathematical shortcut: standard headline CRI (Ra) averages only the first eight pastel test patches (R1 through R8) and completely excludes R9 (strong saturated red).

Here is how the CRI (Ra) formula works, why cheap blue-pump white LEDs deliberately sacrifice 620–680 nm red wavelengths to boost lumens-per-watt efficiency, and why cinematographers and lighting engineers rely on Extended R9, TLCI (Television Lighting Consistency Index), and IES TM-30 instead.

1. How CIE CRI (Ra) Works—And Why It Ignores R9 through R14

Introduced by the International Commission on Illumination (CIE) and refined into its modern form in 1974, the Color Rendering Index (CRI) measures how faithfully a test light source renders surface colors compared to an ideal reference illuminant of the exact same Correlated Color Temperature (CCT):

• The Reference Illuminant Split at 5,000 K: As specified in the CIE CRI method, if a lamp's correlated color temperature is under 5,000 K (such as 2700K warm household tungsten or 3200K studio halogen), its spectrum is compared against an ideal Planckian black-body radiator. If its color temperature is 5,000 K or higher (such as 5600K or 6500K daylight), it is compared against a mathematical CIE Standard Illuminant D (daylight) simulator. To even qualify for a valid comparison, the lamp's chromaticity distance from the Planckian locus must satisfy DC = Δuv < 5.4 × 10^-3 on the CIE 1960 uniform chromaticity scale.
• 14 Test Color Samples (TCS01–TCS14), Only 8 Counted in Ra: As detailed by UPRtek's photometry guide, the CIE defines 14 Munsell reference color swatches (often expanded to 15 in Asia with R15 for Asian skin tone). However, when a manufacturer prints "CRI = 80" on a box, they are reporting General CRI (Ra), which is the unweighted arithmetic mean of only the first 8 moderately saturated pastel swatches (R1 to R8):
  • TCS01 (R1): Light greyish red; TCS02 (R2): Dark greyish yellow; TCS03 (R3): Strong yellow green; TCS04 (R4): Moderate yellowish green;
  • TCS05 (R5): Light bluish green; TCS06 (R6): Light blue; TCS07 (R7): Light violet; TCS08 (R8): Light reddish purple.
• What Gets Left Out (R9–R14): The remaining six swatches—TCS09 (R9: Strong Saturated Red), TCS10 (R10: Strong Yellow), TCS11 (R11: Strong Green), TCS12 (R12: Strong Blue), TCS13 (R13: Caucasian Skin Tone / Light Yellowish Pink), and TCS14 (R14: Olive Leaf Green)—are classified as Special CRI values and are completely excluded from the headline Ra average.

2. The Physics of Cheap White LEDs: Why Blue-Pump Phosphors Starve Red Skin Tones

Why does omitting R9 cause such severe damage when evaluating modern LED bulbs? As explained in Light Sky's analysis of stage and studio CRI, a typical white LED is not natively white: it consists of a monochrome ~450 nm blue Indium Gallium Nitride (InGaN) semiconductor chip coated in a yellow cerium-doped yttrium aluminum garnet (YAG:Ce) phosphor.

• The Lumen-per-Watt Trade-Off at 620–680 nm: Human eyes are maximally sensitive to green-yellow light at 555 nm (the photopic luminous efficacy peak) and rapidly lose sensitivity in the deep red spectrum above 620 nm. If a bulb manufacturer adds deep-red nitride phosphors emitting at 630–660 nm, a large share of the LED's electrical energy goes into red photons that barely register on a lumen meter—lowering the bulb's advertised lumens-per-watt efficacy.
• Gaming the R1–R8 Pastel Average: By concentrating phosphor output in the yellow-green band (530–590 nm) and letting the spectrum cliff-dive above 610 nm, a budget household LED easily achieves CRI (Ra) = 80–83 on the eight muted pastel swatches while its R9 score sits between 0 and 20 (and cheap halophosphate fluorescent tubes actually score a negative R9).
• Why Low R9 Turns Faces Greenish-Grey: Beneath the outer melanin layer of the epidermis, human skin tone across all ethnicities is governed by the reflectance of oxygenated hemoglobin in capillary blood, which reflects strongly in the 610–680 nm red band. Under a light source with R9 = 15, there are virtually no deep-red photons available to bounce off the skin—leaving only the blue spike (450 nm) and yellow-green phosphor hump (550 nm), which renders faces ashy, pale, or jaundiced.

"In lighting design, R9 value is typically used as a benchmark to assess whether a light source can accurately reproduce red tones... An R9 value above 50 is the minimum requirement for acceptable red rendering, while an R9 value of 90 or higher indicates excellent color reproduction."

3. Why Cameras Need TLCI, TM-30, and SSI Instead of CRI

Even if an LED fixture has a high CRI (Ra), a digital camera sensor does not see light the way the human eye does. As CameraStuff's comparison of CRI, TLCI, and TM-30 explains, human vision relies on three overlapping cone cell responses (LMS) processed by the brain's chromatic adaptation engine, whereas a mirrorless camera or webcam uses discrete Red, Green, and Blue Bayer color filters over a silicon CMOS sensor:

1. TLCI (Television Lighting Consistency Index, 0–100): Developed by Alan Roberts for the European Broadcasting Union (EBU), TLCI replaces the human eye observer with a mathematical model of a standard broadcast camera sensor and evaluates 24 Macbeth ColorChecker patches. A score of TLCI 85–90 means a colorist can fix the footage with minor white-balance tweaks; a score of TLCI 95–100 means the fixture produces broadcast-ready skin tones straight out of camera with zero color-grading rescue required.
2. IES TM-30 (Rf Fidelity + Rg Gamut Across 99 Color Samples): Because LED makers learned how to tune narrow phosphor peaks to flatter the 8 specific Munsell pastel swatches in CRI, the Illuminating Engineering Society introduced TM-30, which tests 99 real-world Color Evaluation Samples (CES) selected from nature, skin, textiles, and inks. TM-30 reports both Rf (Fidelity Index, 0–100) and Rg (Gamut Index, centered on 100), revealing whether a light source desaturates reds (Rg < 95) or oversaturates them.
3. SSI (Spectral Similarity Index): Developed by the Academy of Motion Picture Arts and Sciences (AMPAS), SSI ignores human vision and fixed color swatches entirely, comparing the continuous wavelength-by-wavelength curve of an LED directly against tungsten (3200K) or CIE D55/D65 daylight so any future camera sensor will respond identically.

4. Complete Data Table: Lighting Tiers by CRI (Ra), Extended R9, TLCI, and TM-30

The dataset below—included in the companion CSV (cri-ra-vs-r9-saturated-red-tlci-and-tm30-led-lighting-skin-tone-benchmarks.csv)—compares seven real-world lighting tiers across all four photometric standards:

If your desk is lit by overhead office LEDs or mixed window light and you want flattering, natural skin tones on video calls without buying a $600 cinema COB fixture, Calibrate video call lighting in Lumet to turn your monitor into a color-accurate, temperature-matched key light with custom warmth to compensate for low-R9 room bulbs.