This past summer the Illuminating Engineering Society published TM-30-2015, IES Method for Evaluating Light Source Color Rendition. Developed by consensus, this new calculational methodology reflects updated color science and addresses the escalating problems with color rendering index (CRI), the existing standard. (Who had seen an R9 specification prior to the rise of LED?) Some lighting designers and manufacturers have already embraced the dual metrics of the TM-30 methodology. But will it catch on?

The IES Color Metric Task Group formed in 2013 to evaluate the evolution of vision and color science since CRI was developed 50 years ago. The two members interviewed here (Michael Royer and Aurelian David) expressed confidence in the TM-30 method because lighting design, manufacturing, research and government were all represented in their small group. Their goal was not to produce new science, but to review the research and quickly come to a consensus on a new industry standard.

Greatly simplified, CRI provides a single number: an average of a light source’s color rendition performance compared to a continuous-spectrum reference illuminant (halogen incandescent and certain daylight phases). CRI mathematically analyzes the light source’s spectral power distribution (SPD) and evaluates fidelity: how closely will the light source render colors? Will colors look the same or different when lighted by the source, compared to their appearance under the reference illuminant?

HPS Histogram
The responses of TM-30’s 99 color samples locate deficiencies in the spectrum of this HPS source.

A single average score is easy to use on a specification, but it can hide an entire swath of outliers. Plus, the CRI fidelity metric is based solely on eight standard color samples, largely pastels. A light source may do a splendid job on those samples and receive a near-perfect CRI in the nineties.

But architectural design isn’t restricted to pastels.

This was not such a problem under legacy light sources. Designers and occupants could predict, pretty well, what a space would look like under fluorescent or HID; understanding that a higher CRI would provide a more-balanced SPD; less likely to have glaring deficits across the spectrum.

TM-30 metrics – computed from LM-79 data – from typical halogen, HPS and T8 lamps.

Enter LED

But LED is a whole different animal. “One of the interesting things with LED is you have more potential for spectral engineering, or creating a spectrum of your choice,” said Michael Royer, lighting engineer at Pacific Northwest National Laboratory. “When you get these carefully engineered, narrow spectral features it stresses the metric and it becomes more important what samples you use.” He points to both multi-channel emitters (e.g., RGBA) and narrow-band phosphors.

“It’s a new technology, and the spectral power distributions are a little different from old ones. There’s now a more diverse range of sources. People see a source and say, The number I’m seeing doesn’t really correspond to the experience I’m having.” A partial remedy, R9 is a ninth color sample, a vibrant red. In recent years we’ve seen individualized R9 scores added to LED specifications and lighting codes.

This careful engineering of LED spectra is being exploited to create tailored light sources for specific applications. “People assume that a source with a CRI of 100 is a perfect light source in terms of color,” said John Yriberri, VP Worldwide Business Development at Xicato. “In some cases it is OK because people want a natural looking environment. In some cases people would actually like some amount of color distortion, especially if it corresponds to more vivid, more saturated colors.”

Compare the responses of color samples: TM-30 vs CRI.

A butcher or tomato vendor might prefer an enhanced red spectrum. A dermatologist may focus on skin tones. A jeans store wants blues to pop. “Now this may be a good or a bad thing, depending on people’s personal preference,” according to Aurelien David, chief scientist at Soraa. “The difficulty is that CRI doesn’t care about that kind of nuance.” So, on average, colors will look different. But which colors? And are they brightened or dulled? “That’s the primary problem with CRI. It’s just one number, and one number doesn’t give you enough information,” David said.

“The other high-level problem with CRI is the color science – the math that’s used in computing CRI – is decades old. Some of the science is just obsolete, and there’s been a lot of progress over the last 50 years. Now we have much better tools to run these calculations. There’s is a need to bring the science up to date, so the result of the calculations are more accurate,” he said.

Updated color science and improved computing power come together to bring color rendering into the 21st century.

In response, TM-30 raises the number of standard color samples from eight to 99, and creates a new “how closely” calculation method: the Fidelity Index (Rf). But, Royer said that early on in the process, the task group determined that a single, average fidelity metric was insufficient.

Enter gamut

“With TM-30 you’ve got now two values to look at: Rf, which is the fidelity value; and then you have Rg, which is looking at gamut,” explained Yriberri. Gamut is a graphical measure, used in the printing and display industries, of a range of possible colors. Gamut area, or gamut area index, was developed in the lighting industry (though not widely adopted) to assess chromaticities: greater saturation equals greater gamut.

Using modern color science and new color samples, TM-30 has adopted Gamut Index, Rg: a single average level of saturation gauged relative to the reference illuminant. An Rg of less than 100 indicates that colors, on average, will be muted compared to the reference illuminant. Rg greater than 100 indicates richer colors.

But still, average values can hide important information. Which colors? How much brighter or duller? Is the tone shifting, from, say, red to orange-red?

Color Vector vs Color Distortion
The color vector graphic and color distortion graphics are standardized to depict shifts in both hue and saturation.

TM-30 standardizes a complementary graphical representation of hue and saturation changes – representing both Rf and Rg. A color vector graphic shows the shifts in chromaticity from the reference illuminant (dark circle) to the light source being evaluated (red circle). Where the light source extends beyond the boundary of the reference illuminant, those hues will be more vivid and saturated. Where the light source shows a shift to the interior of the reference illuminant colors will be faded and less saturated. A color distortion graphic shows the same results; simply depicting the comparative gamut backed by the circle of the reference illuminant.

Cue TM-30

“My impression is that it is catching on in some communities. Especially with lighting designers; people who are fairly educated about color rendition and who want to get more information than just CRI,” said David. “We think that for advanced users probably it’s going to be well received.”

David and some other members of the IES Color Metric Task Group are active with the color committees at the CIE (International Commission on Illumination), which “supports the study of the recently published IES Technical Memorandum TM-30.” The CIE is the global standardization body and expects to release its report on color fidelity some time in 2016.

“Ultimately, I think everyone would like an international reference standard,” said Royer. “Achieving agreement within CIE, it would make things easier for multinational companies, to have everyone working off the same page.” As an IES Technical Memorandum, TM-30 is not a required standard. Even if adopted by the CIE, the TM-30 methodology would not be required until cited by a standard or code: ANSI, ISO, IBC, Energy Star, Title 24, etc.

In addition to a push from code-making bodies, the TM-30 methodology will require pull from the specification community. “TM-30 allows the designer to be more specific and accurate in terms of the light source they’re going to specify and the lit effect that they’re going to get. The IES has published it as a metric, and you can specify it today. It uses the existing photometric data; you don’t need to wait for anything further,” said Yriberri.

“So if you put that minimum Rf and Rg value on the specification, a manufacturer is going to have to ensure that they have the data. That’s part of the whole process.”

Lois I. Hutchinson

About Lois I. Hutchinson

Lois I. Hutchinson is a freelance writer specializing in lighting and energy issues. She is also the content marketing mastermind behind Inverse Square LLC, a Los Angeles–based consultancy. Contact her via with your comments and any article ideas that concern the lighting community here in the Southwest.

Share This