Color Quality in Retail and Hospitality: From CRI to TM-30 in Practice
LiteSource
|
June 30, 2026
|
Agency News
Color Quality in Retail and Hospitality: From CRI to TM-30 in Practice
A practical guide to color in retail and hospitality lighting — what CRI captures, what TM-30’s Rf, Rg, and color vector graphic add, and how to translate the metrics into accurate skin tones, food rendering, and merchandise presentation.
In an office, color rendering is a comfort consideration. In retail and hospitality, it is a business decision. A garment that reads as warm burgundy under the right light can flatten to a muddy brown under the wrong one. A cut of beef can look fresh and appetizing or grey and tired depending entirely on how the fixture handles red. A guest’s complexion in a hotel lobby or a restaurant booth can look healthy or washed out — and people notice, even when they can’t name why.
The spec sheet gives you a few numbers to work with: a Color Rendering Index (CRI) value, sometimes an R9 figure, and increasingly a full ANSI/IES TM-30 report. Each measures something real, and none of them tells the whole story on its own. This guide walks through what CRI actually captures, what TM-30 adds, and how to turn the metrics into decisions about skin tones, food, and merchandise.
What CRI Actually Measures — and What It Leaves Out
CRI is the value most specifiers reach for first, and it’s still the headline number on nearly every spec sheet. The figure you see — usually written as “CRI 80” or “CRI 90” — is technically Ra, the arithmetic average of how accurately a light source renders eight test color samples, R1 through R8, compared to a reference source at the same Correlated Color Temperature (CCT). The scale runs from 0 to 100, where 100 means the test source renders those samples identically to the reference.
Two limitations matter for retail and hospitality work.
First, Ra is an average, and averages hide weak spots. A source can post a respectable CRI 90 while rendering one or two specific colors poorly, because strong performance elsewhere pulls the average back up.
Second, and more important, the eight samples behind Ra are deliberately moderate, low-saturation pastels. They were chosen decades ago, in the fluorescent and incandescent era, and they don’t include the saturated colors that carry the most weight in a store or restaurant. The result is a well-known trap: two fixtures both rated CRI 90 can render a red dress, a steak, or a skin tone in visibly different ways, and the CRI number alone won’t warn you.
R9: The One Extended Value to Always Request
The fix that most specifiers already know is R9. CRI actually defines fifteen test samples (R1 through R15), but only the first eight feed into the Ra figure. The remaining samples — including saturated red (R9), two skin-tone references (R13 and R15), and a leaf green (R14) — are reported separately, if at all.
R9, saturated deep red, is the one to insist on. Red is everywhere in these spaces: skin tones, meat and produce, wine, wood finishes, brick, cosmetics, and a large share of textiles and brand colors all carry significant red content. Red can also be one of the harder colors for many LED spectra to render accurately, which is exactly why it gets left out of the headline number.
The consequence is stark. A fixture can advertise CRI 90 and still post an R9 near zero — or even negative, which the math allows. Under that source, red objects drift toward brown, skin looks tired, and food looks flat. Common practice is to treat R9 ≥ 50 as a minimum for any space involving people, food, or natural materials, and to look for R9 ≥ 80 or higher in color-critical environments like cosmetics counters, fresh-food display, and fine dining. Whatever the headline CRI says, request the R9 value at the CCT you’re actually specifying.
What TM-30 Adds: Rf, Rg, and the Color Vector Graphic
TM-30 — formally ANSI/IES TM-30, with the current edition published as ANSI/IES TM-30-24 — was developed to address exactly the blind spots above. Instead of eight pastel samples, it evaluates a source against 99 Color Evaluation Samples drawn from real-world objects, spanning the full hue circle across a range of saturations and lightness levels. From that larger, more representative set, it reports color rendition through several related measures and graphics rather than a single number.
Three pieces do most of the work for a specifier.
Rf, the Fidelity Index. Rf runs 0 to 100 and answers the same basic question as CRI — how closely does the source match the reference? — but across the full 99-sample set rather than eight. A higher Rf means more faithful rendering. One caution worth repeating: Rf and CRI are not interchangeable. They use different samples and a different color space, so an Rf of 90 and a CRI of 90 are not the same claim, and the two values should never be swapped.
Rg, the Gamut Index. This is where TM-30 goes beyond fidelity. Rg describes how saturated colors appear relative to the reference. It is centered on 100: a value near 100 means saturation roughly matches the reference, above 100 means the source increases saturation (colors look more vivid), and below 100 means it dulls them. Rg captures something CRI simply cannot — whether a source makes colors “pop” or fall flat.
The Color Vector Graphic. This is the diagnostic that changes how you read a source. TM-30 divides the hue circle into 16 bins and plots, for each one, an arrow against a reference circle. The direction of the arrow shows whether hues shift, and the length and direction relative to the circle show whether that hue is pushed toward more saturation (arrow bulging outward) or less (pulling inward). For the first time, you can see not just that a source renders color imperfectly, but which colors shift and in which direction. The reddish region — hue bin 1, which governs skin and warm reds — is usually the first place to look in retail and hospitality.
The Trade-Off No Single Number Shows
Here is the insight that ties the metrics together, and the reason a one-number spec will always be incomplete: fidelity and vividness are often in tension. A modest saturation boost can improve visual preference — people frequently like colors rendered slightly more vivid than the reference — but pushing too far moves the source away from a faithful match, lowers Rf, and can distort an object’s true appearance. High fidelity and a slight gamut increase can coexist comfortably; it’s the drive toward maximum vividness and maximum accuracy at the same time that pulls in opposite directions.
Neither is wrong. The right answer depends on intent. A cosmetics counter or an art-adjacent display may demand faithful rendering above all, so the customer sees the true product color. A produce aisle or an apparel display may benefit from a controlled, modest boost in saturation that makes colors read as vivid and appealing — provided it stays honest. The skill is deciding which you want, on purpose, rather than discovering it after install.
How to Use TM-30 Without Over-Specifying
The numbers can spiral into false precision, so it helps to know that TM-30 itself does not prescribe a single perfect target. Its design-intent framework (Annex E) instead organizes color goals into three intents — Preference (how pleasant and natural the rendering looks), Vividness (how saturated colors appear), and Fidelity (how close rendering is to the reference) — each with three priority levels, where Level 1 is strictest and Level 3 most flexible. As a rough anchor, the IES describes the top Preference level as the range about 90% of occupants find acceptable, stepping down to roughly 80% and 65% at the lower levels.
The practical takeaway matters more than the table: the standard deliberately does not assign intents or levels to specific applications. The right target depends on the space, the merchandise, and the viewing population, and that judgment is left to you. A high-end retail project might prioritize preference and fidelity together; the same logic lets a hospitality or gallery source be tuned for faithful, flattering color without chasing a number that doesn’t fit. If you do specify to this framework, pull the actual criteria from ANSI/IES TM-30-24 rather than relying on summary figures, and note its criteria are written for typical interior light levels (roughly 200 to 700 lux) in spaces featuring a variety of colors.
Putting It to Work: Three Targets
The metrics become useful when they map to the things these spaces are actually selling: people, food, and merchandise.
Skin Tones — Hospitality, Dining, Fitting Rooms, Lobbies
Skin reads as “healthy” largely through its red and warm undertones, which is why the red hue bin carries so much weight. The familiar fitting-room complaint — where a shopper looks drained and a sale evaporates — is often not just a light-level issue; red rendering, CCT, Duv, glare, and vertical illuminance all affect whether skin appears healthy or washed out.
The recipe is a warmer CCT (commonly 2700K to 3000K in hospitality and dining), strong red performance (R9 in the 80s or above, and high local fidelity in the red bin), and a small, deliberate positive chroma shift in red that gives complexions a warm, healthy glow without tipping into an artificial look. The Color Vector Graphic is the check here: a slight outward push in hue bin 1, not a large one.
Food Rendering — Grocery, Restaurants, Markets
Food display is one of the most red-sensitive color-rendering applications, especially for meat, berries, tomatoes, sauces, and warm bakery tones, all of which signal freshness through red and warm content. High R9 and strong red-bin performance should be treated as baseline expectations for food display and dining.
A modest gamut boost (Rg slightly above 100) can make produce and prepared food look more vivid and appetizing — this is legitimate merchandising. But there’s a line worth respecting: pushing saturation far enough to make tired meat or aging produce look fresher than it is crosses from merchandising into misrepresentation, with real customer-trust and brand-reputation implications. Keep the boost controlled and confirm it in the vector graphic. CCT is typically matched to the product and merchandising convention: warmer light flatters bakery, meat, and prepared foods, while some operators run neutral light over seafood, deli, and certain produce.
Merchandise and Brand Color — Apparel, Cosmetics, Display
For merchandise where the customer is judging the actual color of the product — apparel, cosmetics, paint, packaging — fidelity comes first. Start from a CRI 90+ baseline, then use the TM-30 report to verify which hues hold and which drift, paying attention to the bins that match the brand palette and fabric range on the floor. Cosmetics and color-matched apparel are the strictest cases: a saturation boost that flatters the display can also misrepresent the product and drive returns.
As directional starting points, a fidelity-first space holds Rf at roughly 90 or above with Rg near 100; a space that benefits from controlled vividness often runs Rg a few points above 100 (commonly cited in the low 100s) while keeping Rf high. Treat these as starting targets to confirm against the actual report and the merchandise, not as fixed rules — the right values shift with the objects on display, the light level, and the viewing population.
One more requirement spans all three: consistency. The same garment, dish, or face should read the same under every fixture in the space. That means tight color binning (a 3-step MacAdam ellipse or better is a common target) and matched CCT across the install, so the room doesn’t turn into a patchwork of slightly different whites.
CCT Still Sets the Mood
None of the above replaces a deliberate CCT decision. Color rendering metrics are always reported at a given CCT, and CCT is what sets the emotional register of the space: warm white (2700K to 3000K) for intimate hospitality, dining, and boutique retail; neutral (3500K to 4000K) for general merchandise and larger-format stores. The rendering metrics tell you how truthfully and how vividly colors appear; CCT tells you how the room feels. They are separate specification decisions, and both have to be made — and held consistent — for the space to land.
CCT alone doesn’t fully describe the white, either. Two 3000K sources can feel noticeably different depending on Duv — their position relative to the blackbody locus. A source sitting slightly above the locus reads faintly green; one below it reads slightly rosy. That difference is easy to miss on a spec sheet and very hard to miss on a face, which is why Duv matters as much as CCT in hospitality, dining, cosmetics, and fitting-room applications.
What to Request from Manufacturers
When color quality matters, ask for more than the headline number:
- Full CRI data — Ra and R9 at minimum, reported at the exact CCT you’re specifying.
- The TM-30 report — Rf, Rg, the Color Vector Graphic, and the local red-region values (Rf,h1 and Rcs,h1), again at the specified CCT and at any dimmed or warm-dim setpoints the controls will actually use.
- Chromaticity and Duv — confirm the source’s position relative to the blackbody locus, especially in hospitality, dining, cosmetics, and fitting-room applications where a green or pink tint can noticeably affect skin and material appearance.
- Design-intent fit — if the manufacturer publishes TM-30 Preference, Vividness, and Fidelity priority levels, request them; they shortcut the conversation about faithful versus flattering.
- Binning tolerance — the MacAdam step rating, so color stays consistent across a large installation.
- Configuration confirmation — verify the data is for the exact optic, lumen package, and CCT you’re specifying, tested per LM-79, not a generic family figure.
Bringing It All Together
CRI tells you the average. R9 catches the red blind spot the average hides. TM-30 shows you which colors shift and in which direction, and lets you choose — on purpose — between rendering that is faithful and rendering that is flattering. In an office, that distinction is academic. In a store, a restaurant, or a hotel, it’s the difference between merchandise that sells, food that looks fresh, and guests who look their best.