CRI vs. TM-30: Which Metric Matters for Color Accuracy?

CRI vs. TM-30: Which Metric Matters for Color Accuracy?

Color rendering is one of the few lighting specs that directly affects what people see and how spaces perform. Most spec sheets still headline CRI (Color Rendering Index), but newer standards push hard toward TM-30 as a richer way to describe color. The practical question for specifiers, contractors, and serious DIY users is simple:

When do you rely on CRI, and when do you insist on TM-30?

This guide answers that from a field-spec perspective: how each metric works, where they fail, and how to use them together for better decisions in warehouses, shops, retail, and color-critical areas.

---

1. CRI in Practice: What It Really Tells You (and What It Hides)

1.1 Quick definition

CRI (Color Rendering Index) is a legacy metric defined in CIE 13.3 that compares a test source to an ideal reference (Planckian or daylight) using up to eight pastel color samples (R1–R8). The familiar “CRI 80” or “CRI 90” value is the average of those eight.

Key points:

• Scale: 0–100 (100 = identical to reference for those samples).
• Reference: smooth-spectrum incandescent/Planckian or daylight.
• Samples: mostly low-saturation pastel colors.

According to the CIE color rendering publication, CRI was never designed for narrow-band LED spectra; it predates solid-state lighting by decades.

1.2 Why CRI is still unavoidable

Even though it is old, CRI still shows up everywhere:

• Energy programs & baselines: Many codes and utility programs still specify minimum CRI values (often Ra ≥ 80) as a shorthand quality gate.
• Standards & specifications: Documents like DOE FEMP solid-state lighting guidance still present CRI as a primary color quality indicator in summary tables, especially for basic commercial/industrial applications.
• Procurement language: Most master specs and bid forms ask for “CRI ≥ 80” or “CRI ≥ 90” with no mention of TM-30.

So for compliance and procurement, you cannot ignore CRI. But relying on it alone is risky.

1.3 Hidden trap: CRI ignores R9 and spectrum details

A common misconception is:

“If a luminaire has CRI 90, color will look great.”

In real projects that assumption fails often.

CRI’s main issues for LEDs:

• R9 (strong red) is not part of the Ra average. A fixture can advertise CRI 90 while having R9 ≈ 0, which produces flat skin tones, dull meat and food, and lifeless warm materials. The same CIE document cited above notes this as a known limitation for LED evaluation.
• Pastel-only samples. Because Ra is built on muted colors, a spectrum that mistreats saturated pigments can still score high.
• Reference spectrum bias. As summarized in a critical review of CIE color rendering metrics, LED spectra with sharp peaks can achieve high Ra even when they shift hues noticeably compared to real daylight.

From field experience, R9 is the single most common “surprise” in submittals: the spec sheet claims CRI 90, but the LM-79 report or TM-30 data reveals R9 < 0. Skin, textiles, and food all look worse than under a decent CRI 80 source with R9 ≥ 30.

Practical rule:

• Never accept CRI alone for color-sensitive areas. Demand R9 and, where possible, TM-30 and full spectral data.

---

2. TM-30: What It Adds Beyond CRI

2.1 Overview of TM-30

TM-30 is a method from the Illuminating Engineering Society that evaluates color rendition using 99 test color samples and provides multiple metrics:

• Rf (fidelity index): Similar to CRI but based on 99 colors rather than 8 pastels; describes overall similarity to a reference spectrum.
• Rg (gamut index): Describes average saturation change. Rg > 100 = increased saturation, Rg < 100 = reduced saturation.
• Color vector graphic: A 2D plot that shows where in hue space colors shift (e.g., reds slightly more saturated, blues slightly muted).

The TM-30 technical memorandum and FAQ note that Rf correlates more strongly with perceived fidelity than CRI Ra, especially for LED sources, because the larger sample set better represents real-world object colors.

2.2 Why Rf is a better fidelity indicator than CRI

Controlled experiments summarized in the same TM-30 FAQ show:

• Higher correlation between Rf and perceived color accuracy under LED lighting than between Ra and perception.
• CRI may overestimate fidelity for spiky LED spectra, where Rf reveals the mismatch more clearly.

From a specifier’s perspective:

• Rf ≥ 80 tracks well with “basic commercial quality” suitable for offices and warehouses.
• Rf ≥ 90 aligns with most occupants judging colors as “natural” or “accurate” in retail and inspection areas.

These are not regulatory limits, but they are consistent with what is seen in LM-79 reports and field mockups.

2.3 Rg: The missing dimension—saturation

One of the biggest advances in TM-30 is Rg and the associated vector graphic. As the IES notes in the TM-30 documentation, Rg tells you how a source changes the average saturation of colors compared to the reference:

• Rg ≈ 100: Neutral saturation.
• Rg > 100: Colors are more saturated on average (“punchy” or “vivid”).
• Rg < 100: Colors are desaturated (“washed out”).

Research on preference and gamut, such as the work discussed in Davis & Ohno’s color quality studies, shows that many people actually prefer slight saturation boosts, with preference often peaking around Rg 105–115 rather than at strictly neutral Rg = 100.

This leads directly to application-specific tuning instead of one-size-fits-all CRI:

• Retail displays: modest saturation boost (Rg 105–110) often makes merchandise “pop” without looking artificial.
• Museums or conservation: Rg close to 100 to avoid distorting artwork appearance.
• Industrial inspection: depends on the task; sometimes modest saturation increases improve defect visibility.

2.4 Why TM-30 is additive—not a replacement

A frequent oversimplification is “TM-30 replaces CRI.” In reality, energy codes, rebate rules, and legacy specs still lean heavily on CRI, while TM-30 adds nuance CRI never had. The TM-30 FAQ explicitly frames TM-30 as a more comprehensive method, not a regulatory replacement.

For now, the pragmatic stance is:

• Treat CRI as the minimum threshold for compliance and prequalification.
• Use TM-30 as the design tool that lets you tailor color rendition to the space.

---

3. Head-to-Head: CRI vs. TM-30 for Real Projects

3.1 Comparison table

Use this as a quick reference when reviewing spec sheets and LM-79 reports.

Aspect	CRI (Ra)	TM-30 (Rf, Rg, vectors)
Origin	CIE 13.3, pre-LED era	IES TM-30, designed for modern SSL
Color samples	8 (Ra), plus optional R9–R15	99 test color samples
Fidelity metric	Ra (0–100)	Rf (0–100)
Saturation metric	None (implicit)	Rg (≈ 60–140)
Direction of hue shifts	Not described	Explicit in color vector graphic
Strengths	Simple, broadly recognized; still used in codes and programs	More accurate fidelity for LEDs; reveals saturation and hue behavior
Weaknesses	Ignores R9 in headline value; can be fooled by spiky spectra	More complex; thresholds not yet universal in regulations
Typical use	Baseline requirement (e.g., Ra ≥ 80)	Fine-tuning color quality by application

3.2 Myth to debunk: “More test colors means perfect prediction”

Another misconception is that TM-30’s 99 colors make it foolproof:

“If Rf and Rg look good, I don’t need to worry about anything else.”

Research on metamerism and complex materials, such as the work summarized in advanced color rendition studies, shows this is not the case. Even with 99 samples, TM-30 cannot fully capture how special pigments, multi-layer coatings, or textured surfaces behave. Certain automotive finishes, interference coatings, and specialty textiles can still render unpredictably under high-Rf “good” sources.

Expert Warning (Pro Tip)

Relying solely on Rf/Rg is a common trap. Our field experience—and museum lighting literature such as gallery lighting evaluations—shows that real-world mockups routinely overturn paper decisions. Gloss, texture, and background context change appearance in ways no single metric predicts.

For any project where appearance truly matters, always:

1. Shortlist products using CRI, R9, and TM-30 (Rf/Rg, vectors).
2. Install 1–3 sample fixtures over representative surfaces.
3. Observe for at least 48–72 hours with real tasks and distances.

This small effort often reveals subtle hue shifts or flatness that metrics alone failed to warn about.

3.3 Case study: Three fixtures with the same CRI

Consider a typical high-bay retrofit for a manufacturing quality-control area:

• All three candidate luminaires: CRI 90 on the spec sheet.

The deeper data tells a completely different story:

Fixture	CRI (Ra)	R9	TM-30 Rf	TM-30 Rg	Observed effect in mockup
A	90	5	88	95	Reds and warm neutrals look washed out; fine scratches hard to see
B	90	60	92	98	Neutral, accurate colors; good skin tone and defect visibility
C	90	30	89	110	Strong color pop, but some reds look too “hot”; overshoots for inspection

All three “pass” a basic CRI 90 requirement, but only Fixture B supports the color-critical task effectively.

This mirrors patterns seen in projects where high-CRI high bays are specified for QA lines: insisting on R9 ≥ 50 and Rf ≥ 90 with Rg between ~95–105 avoids most surprises. For deeper guidance in industrial contexts, see how similar thresholds are applied in high-CRI manufacturing plant recommendations and in practice-oriented guides like our article on when to specify CRI 90+ in manufacturing.

---

4. Practical Thresholds by Application: CRI, R9, and TM-30

Instead of chasing “the highest numbers,” focus on fit for purpose. The ranges below reflect a blend of TM-30 research, standards guidance, and real project outcomes; they are practical targets, not hard code limits.

4.1 Warehouses, distribution centers, and general industrial

Typical goal: Safe, efficient throughput with recognizable colors, but not strict color matching.

Recommended minimums:

• CRI (Ra): ≥ 80
• R9: Prefer ≥ 0; ≥ 10 is noticeably better for safety markings and skin tone.
• TM-30: Rf ≥ 80, Rg ≈ 95–105

This aligns with “good commercial” color quality discussed in TM-30 application guidance and with the performance levels often referenced in broad efficiency documents like DOE FEMP’s SSL overview.

In these spaces, putting budget into efficacy, layout, and controls often yields more value than chasing CRI 90, as long as you avoid clearly deficient spectra (e.g., CRI < 80 or very low R9).

For layout implications—getting uniform brightness so color cues are consistent—pair this article with our guide on achieving lighting uniformity in a warehouse.

4.2 Shops, garages, and DIY workspaces

Goal: Clear visibility for tools, wiring colors, vehicles, and project work. Color quality noticeably affects both safety and satisfaction.

Suggested range:

• CRI (Ra): ≥ 80 (≥ 90 preferred if budget permits)
• R9: ≥ 20 for general work; ≥ 50 if you care about finishes, paints, and detailing.
• TM-30: Rf ≥ 85, Rg ≈ 98–105

Many garage owners and small shops aim for daylight-like appearance. For a deeper dive on how color quality interacts with CCT in these spaces, see our CCT-focused guide 4000K vs. 5000K for your garage workshop.

4.3 Retail, showrooms, and customer-facing spaces

Goal: Make merchandise attractive while preserving believable color.

Practical targets:

• CRI (Ra): ≥ 90
• R9: ≥ 50 (critical for skin tones, warm materials, and food).
• TM-30: Rf ≥ 90, Rg ≈ 100–110.

Preference research summarized in gamut and preference studies shows that moderate gamut expansion (Rg ~105–115) tends to be liked more than strictly neutral gamut (Rg = 100). That matches retail experience: a slight bump in saturation makes displays feel more vivid without obviously faking colors.

Our article How High CRI Lighting Boosts Retail Sales Performance explores how CRI 90+ with robust R9 improves perceived quality and dwell time, but TM-30 lets you tune how much saturation you introduce by looking at Rg and the vector graphic.

4.4 Manufacturing QA, laboratories, and color-critical inspection

Goal: Minimize color errors and maintain consistent appearance across lines and over time.

Recommended thresholds:

• CRI (Ra): ≥ 90
• R9: ≥ 50, often higher.
• TM-30: Rf ≥ 90, Rg ≈ 95–105; vector graphic should be compact, with minimal directional bias.

Here, TM-30 is especially valuable because it exposes directional shifts (e.g., reds slightly more saturated, greens pulled toward yellow) that can cause acceptance/rejection inconsistencies in QA. Practical guidance for these environments is discussed in TM-30 application notes and reflected in industry practice, as described in our piece on high-CRI UFO high bays for factory quality control.

4.5 Museums, galleries, and conservation

Goal: Present artwork and artifacts as intended, with minimal spectral bias.

Typical targets from museum-oriented literature such as critical reviews of color rendition metrics for museums:

• CRI (Ra): High, but not decisive alone.
• TM-30: Rf often ≥ 92–95; Rg close to 100 is preferred to avoid distorting saturation.
• Special attention to blue and red channels to prevent hue shifts in sensitive pigments.

In these spaces, mockups under actual artworks or reproductions are indispensable. Metrics guide the shortlist; visual evaluation under final conditions decides.

---

5. A Practical Framework: How to Specify CRI and TM-30 Together

5.1 Documentation checklist for color-sensitive projects

When color matters, treat the LM-79 report as your performance “scorecard,” and insist on the following data bundle (LM-79 is described in the ENERGY STAR summary of LM-79 as the standard method for measuring electrical and photometric performance):

1. LM-79 report with:
   • Total lumens, efficacy, CCT, CRI (Ra), and power factor.
   • Spectral power distribution (SPD) chart or CSV.
2. CRI breakdown: Ra plus R1–R15, especially R9.
3. TM-30 report: Rf, Rg, and the color vector graphic.
4. .ies file (LM-63 format) for layout in tools such as AGi32.

The combination of LM-79 + CRI breakdown + TM-30 + IES file allows you to:

• Verify brightness and distribution in your layout software.
• Confirm that Rf and Rg match the application targets above.
• Spot spectral oddities in the SPD that may affect unique materials.

5.2 Heuristic targets by application type

Here is a condensed decision asset you can paste into your project spec.

Application	CRI (Ra)	R9	TM-30 Rf	TM-30 Rg	Notes
Warehouse / general industrial	≥ 80	≥ 0 (≥ 10 preferred)	≥ 80	95–105	Prioritize lm/W and layout once these are met
Garage / DIY workshop	≥ 80 (≥ 90 ideal)	≥ 20 (≥ 50 for finishing)	≥ 85	98–105	Daylight-like feel with good visibility for projects
Retail / showroom	≥ 90	≥ 50	≥ 90	100–110	Slight saturation boost often preferred
QA / lab / inspection	≥ 90	≥ 50	≥ 90	95–105	Minimize directional hue shifts; check vector graphic
Museum / gallery	High (≥ 90+)	High but secondary to TM-30	≥ 92–95	≈ 100	Visual mockups are mandatory

These ranges synthesize findings from TM-30 research and preference studies such as gamut vs fidelity analyses, combined with observed field performance across multiple retrofit projects.

5.3 Trade-offs: Color quality vs efficacy and cost

Optimizing spectra for high Rf and tailored Rg usually costs lumens. Analyses of LED design trade-offs, such as those outlined in studies on efficacy vs color quality, show that pushing from CRI 80 to CRI 90 typically reduces efficacy by about 5–12%, depending on CCT and phosphor mix.

In practice, our own comparisons of high-bay and shop fixtures show similar patterns:

• A CRI 80, warehouse-grade high bay might deliver 150–170 lm/W.
• A CRI 90, high-fidelity variant at the same CCT often lands closer to 135–155 lm/W.

That trade-off is usually acceptable in color-critical zones (QA, retail, detailing bays) but harder to justify in bulk storage aisles. TM-30 helps you spend those lumens intelligently by ensuring that the extra power goes toward better fidelity and appropriately tuned gamut, not just a higher Ra headline.

---

6. How to Read a TM-30 Report in 3 Minutes

When you receive a TM-30 file (typically PDF), this is the fastest way to understand what it means for your project.

1. Check CCT and Duv first. Confirm CCT matches your design intent (e.g., 4000K, 5000K) and Duv is close to 0 (too far positive or negative can look greenish or pinkish).
2. Look at Rf.
   • If Rf < 80: reject for most interior tasks.
   • 80–89: acceptable for back-of-house, warehouses, offices.
   • ≥ 90: candidates for appearance-sensitive spaces.
3. Look at Rg.
   • 95–105: generally safe, neutral behavior.
   • 105–115: more vivid; evaluate with mockups for retail/branding.
   • < 95: likely to look flat unless there is a very specific reason.
4. Scan the color vector graphic.
   • Compact, nearly circular near the reference line: good, balanced behavior.
   • Pronounced bulges in one sector: expect saturated shift in those hues (e.g., reds pushed, blues muted).
5. Cross-check with R9 and SPD. If Rf looks good but R9 is low or SPD shows sharp gaps, be cautious with applications involving warm materials and skin.

This 3-minute pass tells you how the light will behave, not just “good or bad.”

---

7. Common Mistakes When Using CRI and TM-30

7.1 Overinterpreting small numerical differences

Designers often agonize over Rf 88 vs 90 or CRI 82 vs 84. The TM-30 FAQ points out that there are no universally accepted, fine-grained acceptability thresholds; small numerical differences are frequently imperceptible in practice.

Use bands, not single-value targets: for example, “Rf 85–90” rather than “Rf 87 exactly.”

7.2 Ignoring TM-30 at the project level

Another mistake is treating TM-30 as something only “flagship” projects use. In reality, simple heuristics—like Rf ≥ 80, Rg ≈ 98–102 for offices, or Rf ≥ 90, Rg ≥ 100 with decent R9 for retail—are easy for any specifier or contractor to apply.

These heuristics, discussed in practice-oriented TM-30 guidance such as the IES application notes, allow smaller projects to avoid poor spectra without heavy analysis.

7.3 Assuming lab metrics hold perfectly in the field

Datasheets are based on controlled test conditions. Real spaces are not.

A review of LED color stability and binning practices, such as that in technical reports on SSL binning, highlights several factors that shift effective Rf/Rg in the field:

• Mixing multiple luminaire types or CCTs.
• Driver variations and dimming behavior.
• Manufacturing tolerances and binning drift.
• Aging and lumen depreciation over time.

This is why the earlier mockup procedure is essential: install real fixtures, at real heights, over real materials, and confirm that appearance at 20–30 ft viewing distances matches expectations.

---

8. Key Takeaways: Which Metric Matters—and When

For lighting designers, contractors, facility managers, and advanced DIYers, the path forward is not “CRI or TM-30.” It is using each metric for what it does best.

• Use CRI (Ra) as the legacy compliance gate. Meet or exceed code and utility requirements (often Ra ≥ 80) so projects qualify for programs and pass basic quality checks.
• Always check R9 where people or warm materials are involved. CRI without R9 is incomplete.
• Use TM-30 Rf to judge overall fidelity—especially for LED products—because it tracks human perception more reliably than Ra.
• Use TM-30 Rg and the color vector graphic to tune appearance. Decide whether you want neutral, vivid, or restrained saturation, and confirm that hue shifts align with the space’s purpose.
• Expect trade-offs. Higher fidelity and tuned gamut usually cost 5–12% in efficacy; spend that “budget” where it delivers business value (retail, QA, client-facing areas).
• Validate with mockups. For any space where color accuracy affects revenue, safety, or brand, lab metrics are the starting point—not the final decision.

If you adopt this mindset—CRI for compliance, TM-30 for design, and mockups for validation—you will consistently deliver lighting that not only meets standards but also performs in the real spaces where people work and see.

---

Disclaimer

This article is for informational purposes only and does not constitute engineering, safety, or design certification advice. Always consult applicable standards, local codes, and qualified professionals when specifying or installing lighting for commercial or industrial projects.

---

Sources

• CIE – Method of Measuring and Specifying Colour Rendering Properties of Light Sources
• IES – TM-30-18 Technical Memorandum and FAQ
• IES – TM-30 FAQ (Committee Report)
• Davis & Ohno – Color Quality and Preference Studies
• Gamut and Fidelity Preference Analysis
• Critical Review of Color Rendition Metrics for Museum Lighting
• CIE – Limitations of Color Rendering Metrics for LED Spectra
• Studies on Efficacy vs Color Quality in LED Design
• ENERGY STAR – LM-79-08 Summary
• DOE FEMP – Solid-State Lighting Solutions