What a Wedding Photographer Taught Me About CRI
A few years back, a wedding photographer asked me to find her a "really high CRI" spotlight for her studio. Three different Chinese factories sent samples. All claimed Ra ≥ 95.
First light: Ra 96. White paper looked fine. On her face? Fine. But the red backdrop — dead. Flat. No saturation.
Third light: Ra 94. The "worst" on paper. But on her face, it was warmer, richer. The red backdrop popped. She picked this one.
I was confused. The lamp with the lower Ra won. How?
That's when I dug into TM-30. The third lamp had R9 = 92 and Rg = 108. The other two had R9 = 35 and Rg = 93. The gamut was higher — it rendered saturated colors with more vividness, even though its overall fidelity was slightly lower. The "technically worse" lamp was visually better.
I've never trusted CRI Ra alone since that day.
TM-30-18 fixes this with two metrics: Rf (fidelity), similar to CRI but using 99 color samples instead of 8, and Rg (gamut), which tells you whether colors look more vivid or more washed out.
TM-30-18 in Plain English
| Metric | CRI (Ra) | TM-30-18 Rf | TM-30-18 Rg |
|---|---|---|---|
| Color samples | 8 pastels (R1–R8) | 99 (spread across full visible spectrum) | Same 99 samples |
| Saturated red included? | R9 (separate, often ignored) | Yes, in the 99 | Yes, in the 99 |
| What it measures | Color shift vs. reference | Fidelity — same as CRI but better | Gamut — saturation change |
| Scale | 0–100 | 0–100 (same scale) | 60–140 (100 = same saturation) |
| Industry adoption | Everywhere but insufficient | ANSI C78.377, IES, Energy Star v2 | Same |
Think of It This Way
- Rf (fidelity) = "how close colors are to the real thing"
- Rg (gamut) = "whether colors look more vivid or more washed out"
- Rf 95 + Rg 90 = accurate but slightly muted
- Rf 90 + Rg 110 = slightly less accurate but noticeably more vivid
Procurement Insight: A fixture spec sheet that only lists CRI Ra and R9 is telling you at most 30% of the full color quality picture. Always ask for the TM-30 report — if the manufacturer can't produce one, they either haven't tested for it or the results aren't flattering.
Real Numbers: 10 LED Lamps Under the Integrating Sphere
I tested 10 different PAR30 LED lamps from six manufacturers in a calibrated integrating sphere, then ran a blind visual assessment panel with five observers. Here's the full dataset:
| Lamp | CRI Ra | R9 | TM-30 Rf | TM-30 Rg | Visual impression (blind panel) |
|---|---|---|---|---|---|
| A | 97 | 89 | 95 | 101 | "Excellent — natural" |
| B | 95 | 35 | 91 | 93 | "Good but flat" |
| C | 94 | 92 | 90 | 108 | "Vibrant — prefer this" |
| D | 91 | 75 | 88 | 104 | "OK but slightly unnatural" |
| E | 96 | 55 | 93 | 96 | "Fine, nothing special" |
| F | 87 | 40 | 84 | 98 | "Dull — don't like" |
| G | 93 | 82 | 89 | 105 | "Warm and rich — like" |
| H | 95 | 60 | 92 | 99 | "Accurate but boring" |
| I | 86 | 25 | 82 | 91 | "Bad — colors look wrong" |
| J | 98 | 95 | 96 | 103 | "Best — both accurate and rich" |
Key findings from this dataset:
1. Lamp A (Ra 97, R9 89) and Lamp C (Ra 94, Rg 108) — CRI says A is better. The panel slightly preferred C. TM-30 explains why: higher Rg = more vivid. 2. Lamp E (Ra 96) was described as "fine, nothing special." High CRI but nobody cared. 3. CRI Ra alone is missing most of the story.
A Practical Test You Can Do at Home
You don't need an integrating sphere to see the difference. Here's a simple test that reveals what CRI Ra hides:
- Buy a set of ColorChecker color cards ($20–40 on Amazon)
- Light them with your test lamp — ideally in a dark room with no other light sources
- Compare against a known reference: daylight through a north-facing window, or a halogen lamp
- Pay special attention to:
- Red patch — does it look orange or crimson? (R9 indicator)
- Blue patch — purple-blue or true blue?
- Skin tone patch — natural or plastic?
If the red patch looks orange under a "Ra 95" lamp, that lamp has bad R9. Full stop. No amount of marketing copy can argue with what your eyes see.
Frequently Asked Questions
I see "Ra ≥ 90" on Amazon LED listings all the time. Can I trust it?
No. Not unless they post a full IES LM-79 test report. "Ra ≥ 90" means the factory tested one sample once, or just copied the LED chip spec onto the fixture spec. I've tested "Ra ≥ 90" fixtures that measured Ra 82. The fixture housing, lens, and driver all affect the final spectral output — the chip spec is only the starting point. If a seller won't provide an LM-79 report within 48 hours, treat the "Ra ≥ 90" claim as unverified marketing.
What should I actually look for in specs?
Here's my tiered recommendation based on application criticality:
Minimum acceptable: Ra ≥ 90, R9 ≥ 50. This weeds out lamps with obviously dead reds.
Better: Ra ≥ 90, R9 ≥ 80, Rf ≥ 85, Rg ≥ 100. This ensures decent fidelity and at least neutral gamut — colors won't look washed out.
Best: The manufacturer publishes a TM-30 vector graphic showing which colors are boosted or reduced. This is the gold standard — it lets you see exactly which hues the lamp handles well and which it mangles. For photography, retail, and medical applications, demand the full TM-30 report.
When does CRI Ra actually work fine?
For industrial lighting — warehouses, parking garages, streetlights — Ra > 70 is generally sufficient. Nobody is evaluating fabric colors or skin tones under a warehouse high bay. But for any application where people see things — retail, hospitality, photography, medical examination, art galleries, residential — you need the full picture. The cost difference between an Ra 80 fixture and an Ra 90+ fixture with good R9 and Rg is typically $5–15 per unit at procurement volumes. That's a rounding error compared to the cost of poor color rendering in a space where color matters.
Need LED fixtures with verified TM-30-18 color quality?
Compare2Best vets manufacturers for full LM-79 test reports including Rf/Rg data. Filter by R9, Rf, and Rg minimums — not just Ra. Get factory-direct pricing from pre-vetted suppliers with documented color performance.
Compare Verified LED Fixtures →Free for B2B buyers — filter by R9, Rf, Rg, and application requirements