Home / Research / Color Science Beyond CRI

Color Science Beyond CRI: TM-30 Rf/Rg, CQS, GAI, TLCI & New Metrics for LED Lighting

📅 Published: 2026-05-15 · 🔄 Updated: 2026-05-15 · 📖 Scholarly Article

1. Why CRI Is No Longer Sufficient

The Color Rendering Index (CRI, CIE 13.3-1995) was designed in the 1960s for evaluating fluorescent lamps against a Planckian radiator. It uses eight pastel test color samples (TCS01–TCS08, each with moderate saturation) and computes the average color difference between test source and reference illuminant in CIE 1964 U*V*W* color space—a space known to be perceptually non-uniform. The resulting Ra (average of R1–R8) has been criticized for more than three decades for: (a) penalizing LED sources that deliberately enhance red saturation (R9 < 0 possible with high Ra), (b) allowing a single LED phosphor system to score Ra > 80 while rendering saturated reds poorly, (c) using an obsolete chromatic adaptation transform (von Kries, 1902), and (d) averaging across only eight samples, which hides spectral deficiencies. The lighting industry has responded with a suite of supplementary and replacement metrics. This article provides a comprehensive technical comparison.

2. IES TM-30-20: Fidelity (Rf) and Gamut (Rg)

2.1 The 99 Color Evaluation Samples (CES)

TM-30 uses 99 CES derived from real-world object reflectance spectra (paint, textiles, skin, flowers, food, manufactured goods). Each CES is assigned one of 16 hue-angle bins (H1–H16, each 22.5°), plus an additional 4 bins for skin tones (HSK1–HSK4). The 1,584 spectral reflectance data points (99 × 16 illuminants at 2 nm resolution) form the basis of the calculation. The color difference metric uses CAM02-UCS (CIECAM02-based uniform color space), widely regarded as the best available appearance model since 2002.

2.2 Rf (Fidelity Index, 0–100)

Rf is the average color fidelity across all 99 CES, scored non-punitively (i.e., no negative values). A difference of 1 Rf point is approximately a just-noticeable difference (JND) under optimal viewing conditions. The metric is anchored such that a perfect Planckian (or CIE daylight) reference scores 100 Rf. Typical performance:

2.3 Rg (Gamut Index, 60–140)

Rg measures the relative gamut area of the test source compared to the reference, normalized to 100 for exact reference gamut match. Values above 100 indicate increased average saturation (color enhancement); values below 100 indicate desaturation. The Rf/Rg pair is always presented together—a "good" source has both high fidelity (Rf ≥ 85) and near-neutral gamut (Rg ≈ 95–105). The color vector graphic (CVG), a polar plot of hue-angle-specific ΔC* (chroma shift) and ΔH* (hue shift), reveals at a glance in which hue bins the source deviates from the reference—an invaluable tool for lighting designers and specifiers.

SourceCCT (K)CRI RaTM-30 RfTM-30 RgR9GAI
YAG:Ce (single phosphor)50007468101−1558
YAG + SCASN (2-phosphor)40008379104+2572
YAG + LuAG + SCASN (3-phosphor)35009086100+4268
LuAG + KSF (narrow red)3500959098+6862
RGB multi-chip (tunable)40009492102+8588
D65 / NIST reference6500100100100+100100
Note on the Rf/CRI correlation: For most phosphor-converted LEDs, Rf ≈ Ra − 4 to Ra − 8. The offset is due to TM-30's use of 99 CES vs. CRI's 8, and the different color-difference formula (CAM02-UCS vs. CIE 1964). The two are not interchangeable—specifications should always state which metric is being referenced.

3. CIE 224:2017 — Colour Fidelity Index (CFI) and CQS

CIE 224:2017 introduced the Colour Fidelity Index (CFI, denoted RfCIE to distinguish it from TM-30 Rf), based on 10 moderately-saturated test samples with a non-punitive scoring range of 0–100 using CIECAM02-UCS. The CIE CQS (Color Quality Scale, Qa) is related but uses 15 test samples that are more saturated than CRI's 8, and applies a "colored" correction to avoid penalizing sources that intentionally increase color saturation. In practice, Qa is 2–8 points higher than CRI Ra for the same source, because the non-punitive scaling does not penalize saturated red enhancement as CRI does. CIE 224 has been adopted by several European national standards bodies as a preferred alternative to CRI for LED product documentation.

4. GAI (Gamut Area Index)

GAI measures the area enclosed by the eight test-color sample chromaticity coordinates (CIELAB a*b* plane) under the test source, relative to the area under D65 illumination. A GAI of 100 means the average gamut matches D65. Higher GAI values (> 100) indicate increased saturation; lower values indicate desaturation. GAI is not a fidelity metric—a source can have high GAI but poor Fidelity (e.g., a monochromatic amber + deep blue combination). The joint use of GAI and CRI (or Rf) is recommended: a source with CRI ≥ 80 and GAI ≥ 75 is considered to provide "good" color rendition for general interior lighting (NLPIP 2013, Rea & Freyssinier 2010).

5. TLCI-2012 for Broadcast and Film

The Television Lighting Consistency Index (TLCI-2012, developed by the European Broadcasting Union, EBU R 137) evaluates how a light source reproduces colors as they would appear through a broadcast camera. TLCI uses a reference camera model (SMPTE 170M colorimetry, BT.709 primaries, gamma 2.4) and 24 ColorChecker SG patches plus 6 memory colors (skin, sky, grass). Scores are reported on a Qa-like scale: TLCI-1 ≥ 85 = excellent for broadcast; 75–84 = acceptable with minor color correction; < 75 = problematic. LED luminaires with CRI 90+ (especially those with KSF red phosphor) typically achieve TLCI-1 values of 80–90, while single-phosphor YAG sources (CRI ≈ 72) score 55–70 on TLCI.

6. Practical Guide for Specifiers

ApplicationRecommended Metric(s)Minimum ValuePreferred Value
General office / retailCRI Ra or TM-30 RfRa ≥ 80 / Rf ≥ 75Ra ≥ 85 / Rf ≥ 80
High-end retail / hospitalityTM-30 Rf/RgRf ≥ 85, Rg 95–105Rf ≥ 90, Rg 98–102
Museum / galleryRf, R9, GAI, UV/IR contentRf ≥ 90, R9 ≥ 50Rf ≥ 95, R9 ≥ 70
Broadcast / film studioTLCI-2012, TM-30TLCI-1 ≥ 80TLCI-1 ≥ 90, Rf ≥ 92
Medical / surgicalRa, R9, TM-30 skin-tone binsRa ≥ 90, R9 ≥ 50Ra ≥ 95, R9 ≥ 90
Industrial / warehouseRa (legacy spec)Ra ≥ 70Ra ≥ 80

7. The Future: CIE 2025 Standard on Colour Fidelity

CIE is currently developing a new comprehensive color rendition standard (expected publication 2026–2027) that will consolidate fidelity (Rf per CAM02-UCS), gamut (Rg per TM-30 methodology), a color preference index, and hue-specific descriptors into a unified metrology system. The new standard is expected to replace CIE 13.3 (CRI) and CIE 224 (CQS) for all new product specifications after a 5-year transition period. Forward-looking specifiers are already specifying TM-30 Rf/Rg alongside legacy CRI.

8. References

  1. CIE 13.3-1995: Method of Measuring and Specifying Colour Rendering Properties of Light Sources.
  2. IES TM-30-20: IES Method for Evaluating Light Source Color Rendition.
  3. CIE 224:2017: Colour Fidelity Index (CFI) for Accurate Scientific Use.
  4. Rea, M.S. and Freyssinier, J.P. (2010). Color rendering: A tale of two metrics. Color Research & Application, 35(3), 192–202.
  5. Royer, M.P. (2021). IES TM-30-20 is approved: Progress and promise. LEUKOS, 17(1), 1–4.
  6. EBU R 137: Television Lighting Consistency Index (TLCI-2012). European Broadcasting Union.
  7. Smet, K.A.G., et al. (2016). Color quality scale and color fidelity: Are they the same? Optics Express, 24(22), 25156–25168.
  8. NLPIP Lighting Answers (2013). Color Rendering: What's New? National Lighting Product Information Program.

9. Related Articles

CRI & RA Color Rendering Index · Correlated Color Temperature Explained · Color Tolerance SDCM · LED Phosphor & Spectral Engineering · Human Centric Lighting

Sources: CIE 13.3-1995 · IES TM-30-20 · CIE 224:2017 · EBU R 137 · Royer 2021 · Rea & Freyssinier 2010 · Smet 2016
Disclaimer: This article is for educational reference. Product specifications should be verified with manufacturer-provided TM-30 or CIE photometric test reports.