Recommended CRI Values for Art Gallery Lighting

艺术画廊照明的推荐显色指数值

Recommended Values by Application

Specification Comparison

Why Cri Matters

Let's put numbers to this. A typical halogen track light used in galleries for decades delivers Ra 100 and R9 100—it's the gold standard. But halogens waste 90% of their energy as heat. Modern LED replacements often claim "CRI 90," but if you look at the fine print, that's Ra 90 with R9 maybe 30. What does that mean for a Rothko painting with deep crimson fields? The reds appear muddy, almost brownish, because the LED lacks spectral power in the 620–700 nm range. I've seen curators reject entire LED installations because a single Vermeer looked like a cheap print.

Here's the thing: CRI is an average of eight color samples (R1–R8). R9 isn't even in the average. So a fixture can score Ra 92 with R9 of 10—and that's legal per CIE 13.3. For art, you need to specify R9 separately. The CIE 177:2007 standard on museum lighting explicitly recommends R9 ≥ 80 for sensitive works. I'd argue that's the floor, not the target.

What about TM-30? It's a newer metric from IES that uses 99 color samples instead of 14. Rf (fidelity) correlates well with CRI, but Rg (gamut) tells you if colors are oversaturated or washed out. For a gallery, you want Rg between 100 and 105—slightly enriched without being cartoonish. I've measured fixtures with Rg 108 that made a Monet haystack look radioactive. Not good.

Application Scenarios

Scenario 1: The Old Masters Gallery (oil on canvas, 17th–19th century). These paintings have thick impasto and deep shadows. You need Ra ≥ 95 with R9 ≥ 90 to render the carmine reds and ultramarine blues accurately. Use a 3000K LED track head with a 30° beam, delivering 500 lux at the canvas per EN 12464-1:2021 §5.26. I recommend a fixture with a secondary optic that cuts UV to < 0.1 µW/lm—otherwise the varnish will yellow in five years.

Scenario 2: Contemporary Art with Fluorescent Pigments. Some modern artists use DayGlo paints that peak at 540 nm. Standard CRI doesn't capture this well. You need a full-spectrum source with Ra ≥ 98 and R9 ≥ 95, plus a spectral power distribution that hits 540 nm without a dip. I've seen installations where a 4000K LED with Ra 92 made the fluorescent orange look dull gray. The fix: use a phosphor-converted LED with a violet pump, not a blue pump. Costs 15% more but saves the artwork.

Scenario 3: Textile Conservation Lab. Here, you're matching threads to original dyes. The CIE 177:2007 standard says Ra ≥ 95 and R13 (skin tones, but also wool and silk) ≥ 95. You'll want a 5000K source at 1000 lux for detailed work, but with a UV filter below 400 nm. I've used the Xicato XTM series for this—they hit Ra 98 with R13 97. The catch: they're expensive, but one re-weave mistake costs more than the fixture.

Scenario 4: Photography Gallery with Digital Prints. Prints on baryta paper have a narrow tonal range. You need Ra ≥ 98 and R9 ≥ 95 to avoid metamerism—where the print looks different under gallery lights vs. the artist's studio. Use a 5000K source with a CRI of 98 and a TM-30 Rf of 95. I've seen a $50,000 print returned because the gallery lights made the blacks look purple. Don't be that gallery.

Design Guidelines

First rule: never trust a datasheet that only lists "CRI > 90." Ask for the full CRI report with R1–R15 values. I've seen fixtures claiming Ra 92 that had R9 = 12. That's a red flag—literally.

Second: use the IES TM-30-18 metric alongside CRI. For art, target Rf ≥ 88 and Rg between 100 and 105. If the manufacturer can't provide TM-30 data, walk away. It's 2025, not 2005.

Third: consider the color temperature. For oil paintings, 3000K is standard because it matches the warm light of the original studios. For contemporary works, 4000K is common. But here's a trick: use a tunable-white fixture (2700K–5000K) with constant CRI ≥ 95 across the range. That way you can adjust for different exhibitions without relighting the whole room.

Fourth: don't forget the UV and IR. Per CIE 157:2004, total UV content should be < 75 µW/lm for sensitive materials. Most quality LED fixtures are below 10 µW/lm, but cheap ones can hit 200. I've seen a Degas pastel fade in 18 months because the "museum-grade" LED had no UV filter. The spec sheet said "UV-free" but the test report showed 150 µW/lm. Always ask for the test report.

Fifth: calculate the color consistency. You want a 3-step MacAdam ellipse (SDCM ≤ 3) for all fixtures in the same gallery. If you mix 2-step and 5-step bins, the same painting will look different on adjacent walls. I've seen this happen in a $50M gallery renovation. The fix: order all fixtures from the same manufacturing batch and verify with a spectrophotometer on site.

Key Takeaways