Specifying Color Quality With TM-30

By now most of us have attended one or more seminars or webinars about IES TM-30 and understand that it is a method of measuring various color rendering properties of a light source and reporting those measurements. The thing that’s been missing is a recommended set of values that set minimums, maximums and/or tolerances for the various measurements. This has been true for two reasons. First, TM-30 is a method and as such was never intended to set recommended values. The second is that while the science behind TM-30 is solid, the science doesn’t offer any predictions of acceptability.

Good news! After almost three years of research and tests around the world we’re much closer to establishing a set of recommended values. At this year’s IES Annual Conference in Boston, Tony Esposito, Kevin Houser, Michael Royer and I will be presenting the seminar “Specifying Color Quality With TM-30” The description of the seminar is, “This presentation will discuss several research projects which have used the IES TM-30 color rendition framework, and whose results have been used to develop various specification criteria. We will discuss UFC 4-510-01, The Department of Defense Unified Facilities Criteria for Military Medical Facilities, which has already implemented IES TM-30-15 specification criteria.”

During the seminar we’ll review some TM-30 basics, look at several research projects that are helping to establish TM-30 thresholds, and review how to use the TM-30 calculator. Don’t miss it!

I’m Presenting at LEDucation 2018

This year LEDucation, the largest LED only trade show and educational forum in the U.S., will be on March 13 and 14. On Tuesday the 13th from 11 am to 12:30 pm I’ll be presenting a forum called How To Use TM-30, along with Dr. Michael Royer of Pacific Northwest National Laboratory and Wendy Luedtke of ETC. Here’s a summary of the event.

This presentation discusses a number of design trends that are currently shaping the industry and promising to redefine the role of lighting as we know it. New technologies are making dimensions such as color and dynamic behavior over time viable for main-stream lighting, and that constitutes exponential change. This presentation explores a few of the dimensions that appear to be gaining traction, and attempts to illuminate some of the reasoning behind their development. Implications to the user experience are discussed, along with the inadequacy of present lighting metrics.

See you there!

TM-30 Rg, The Gamut Index

In addition to an index that measures the fidelity of a light source to its reference source (Rf) IES TM-30 includes an index that indicates the change in saturation of colors called the Gamut Index and abbreviated Rg. Rg is calculated using the same Color Evaluation Samples (CES) and underlying calculation engine as Rf, which makes TM-30 a cohesive system.

Here’s how Rg works. An Rg value of 100 indicates that, on average, the light source in question does not change the chroma, or saturation, of the 99 CES when compared to the reference light source. An Rg value below 100 indicates that, on average, the light source renders colors as less saturated than the reference source, and an Rg value above 100 indicates that, on average, the light source renders colors as more saturated than the reference source.

Since Rg is an average it says nothing about the possible change in chroma for any individual hue angle bin or for any individual color evaluation sample. That’s ok, thought, because TM-30 also tells us the Rg values for each hue angle bin, and for each CES.

Here’s an example of the graphic for the hue angle bins using the same light source as the previous post on Rf.

TM-30 doesn’t recommend any particular Rg or set of Rg values. As with Rf, the interpretation of the information is left to the specifier. Acceptable or desirable values will vary by application. Rg doesn’t have a maximum or minimum value, but the possible range increases as Rf decreases, as shown below. The wedge to the left of the gray lines shows the range of possible Rg values, while the red dot represents the lamp we’ve been discussing.

The Rg values are also presented in a Color Vector Graphic (CVG), as shown below. The white circle is the normalized reference source. The black circle is the lamp in question. Where the black circle is inside the white, colors are desaturated. Where the black circle is outside of the white, colors have increased saturation. The colored arrows indicate the direction of saturation shift, and the direction of hue shift. Arrows that point straight in or out show only saturation shift. Arrows that show rotation left or right also indicate hue shift. I know! And, the next version of TM-30 will present a graph showing the hue shift!

Research is revealing that we shouldn’t treat all hue angle bins the same. Bins 1 and 16, which include the most red, are indicative of preference and it seems likely that they will take on increasing importance in that role. Some specifications are already acknowledging this. For example, the Department of Defense recently re-issued the Unified Facilities Criteria for Military Medical Facilities that establishes the following requirements for light sources:

Fidelity Index: Rf ≥ 80,

Relative Gamut Index: 97 to 110,

Fidelity Index, Hue-Bin 1: ≥ 78,

Chroma Shift, Hue-Bin 1: -9% to +9%.

Clearly, TM-30 permits us to be much more specific about the color rendering that is acceptable or desirable for a project. Why bother with CRI anymore?

Focal Point Introduces TM-30 Based “Preferred Light”

Today Focal Point Lights of Chicago, IL introduced a series of fixtures that feature what they call Preferred Light. Preferred Light is based on recent studies at PNNL and Penn State, plus their own study, and uses TM-30’s Rf, Rg, and Hue Bin 16 values to establish a balance of fidelity, saturation, and red rendering that is “visually appealing to humans.”

The overall idea is that people seem to prefer a light source that slightly over saturated most colors, especially red. “Using a custom LED mix, Focal Point defines Preferred Light using TM-30-15 metrics as having a fidelity (Rf) of 89, a gamut (Rg) of 107, and over-saturating Hue Bin 16, deep red content, by 9% at a [Correlated] Color Temperature of 3500K.” So, by using the statistical measures of TM-30 and applying them to the related topic of color preference Focal Point has identified an optimized set of LED products to meet their customers’ needs.

I’ll be the first to admit that it may be risky to base all of this on only three studies, but other studies have shown that the TM-30 results can be applied in this way, and are also showing us the relative importance of the various calculated values. I’m excited to see the industry using the tools, and am looking forward to seeing the Preferred Light for myself.

Samsung Introduces Chip-on-Board LED Packages Optimized for Commercial Lighting – Samsung Global Newsroom

Source: Samsung Introduces Chip-on-Board LED Packages Optimized for Commercial Lighting – Samsung Global Newsroom

An interesting bit of news from Samsung this week. They’ve developed an LED package especially designed to achieve an Rg value over 110, “a level that ensures lighting with outstanding color and whiteness.”

It’s important to note that increased saturation means decreased fidelity to the reference light source. This is a lighting solution that will be desirable in some applications, such as retail,and undesirable in others, such as medical facilities.

TM-30 Rf: So Big, So Strong, So Smart!

As we know, CRI Ra and TM-30 Rf are both measurements of color fidelity. That is, they compare a test light source to a known reference light source and measure how well the test source matches the reference source. One of the many shortcomings of CRI Ra is that it provides us with a single value. That single value is easy to use, but doesn’t tell us anything about what colors will have increased saturation, decreased saturation, hue change, or will be unaffected.

TM-30 is a tougher test than CRI, so how do Rf and Ra values relate? Lamps with Ra values below about 70 tend to have higher Rf values, while lamps with higher Ra values tend to have reduced Rf values. Of course, this doesn’t mean that the lamps we think of as better have suddenly become worse, it’s just that we’re scoring on a different scale. This means that we can’t draw direct comparisons. For example, Energy Star requires that lamps have a minimum CRI Ra of 80, but that doesn’t mean that they should also have a minimum Rf of 80. Different tests give different results and we have to be careful not to apply the meaning of one to the scores of the other.

IES TM-30’s Rf mathematically compares the appearance, under a test light source, of 99 color evaluation samples (CES) that are derived from real world objects, to the CES appearance under a reference light source of the same CCT. The distance of the color shift for each CES is measured in the CAM02-UCS color space and averaged. Throw in a lot of calculus (which we don’t need to get into) and voila, the Rf value. It’s important to remember that what we get is just a number. TM-30 doesn’t qualify any of the results as good or bad, desirable or undesirable. It presents information to the lighting specifier and allows the specifier to apply education, professional experience, and knowledge about the project to determine whether or not a given light source is appropriate.

As with Ra, the single value of Rf conveys limited information. It is more accurate, but still only tells us the average match or mismatch between the two light sources. What makes TM-30 so powerful and useful is that it tells us much more if we want to know. For example, using the Calculation Tool that can be downloaded with the purchase of TM-30 (which I wish the IES would make freely available), we can see that one common F32T8/830 has the following characteristics:

Rf 78
Rg 102
CCT 2943
Duv 0.0014
Ra 85

This lamp has moderately good fidelity (Rf), a slight increase in saturation (Rg), has a CCT of just under 3000K, and is slightly above the black body locus and therefore is slightly green (Duv). The Advanced Calculation Tool tells us that the R9 value is 2 and that the Rf for skin is 85. It also tells us the (x, y), (u, v), and (u’, v’) chromaticity coordinates (which, frankly don’t mean anything to me, but the information is there). This information is immediately useful and isn’t provided as part of the CRI calculation. In addition, most light source manufacturers don’t tell us the Duv, although understanding it is becoming increasingly important, especially now that NEMA has extended the chromaticity bins for LEDs in ANSI/NEMA C78.377 American National Standard for Electric Lamps – Specifications for the Chromaticity of Solid-State Lighting Products. That’s a post for another time.

As I’ve already discussed, IES TM-30’s color fidelity metric Rf provides us with as little, or as much, information as we want. If you just want top line information that the beige office you’re lighting will continue to look beige, you can have it. An Rf of 78 is probably just fine. If you want to see the fidelity of each of the 16 hue bins because you’re interested in the fidelity of a particular color range, it’s there. If you want to know the Rf value of all 99 color samples you can have that, too! What else? Well, would you like to see the chromaticity coordinates in (x, y) color space, the SPD vs the reference source, or a pictorial comparison of each of the 99 CES? No problem.

Pretty pictures but are they useful? Not as useful as the data given above, but lighting designers do like to see this information, even if it’s difficult to interpret. The CIE 1931 (x, y) color space isn’t perceptually uniform, so the distance we see between the reference source and the test source isn’t very informative. Seeing the SPD is interesting, but no one can read an SPD and know what the light looks like or how it renders colors. The CES Chromaticity Comparison is also interesting, but the red and black dots aren’t connected to one another. With some light sources it’s easy to tell how they relate so we can see chroma and hue shifts, but as Rf drops and color shift increases it gets harder and harder. What is useful are the next two graphics: the Rf value by Hue Angle Bin and by CES.

Now we can see how individual color ranges are affected by the lamp in question. This may be especially useful on certain projects were specific color ranges are present and need to be accurately rendered. The individual CES scores useful for the same reason. However, in my opinion if you want information at that level of detail you’re probably better off doing a mockup and looking at project specific color and material samples instead of the CES.

TM-30 arms the lighting specifier with as much or as little information as needed on a particular project. It also provides additional information that may be important (such as Duv). It then allows the specifier to apply experience and knowledge about the client and the project to determine whether or not a given light source is appropriate. Who could say no to that?

The Advantages of TM-30

In this series of posts about IES TM-30-15 I’ve discussed the problems with CRI and the resistance to adopting TM-30. In this post I’ll discuss the advantages of TM-30 over CRI, and what TM-30 is and isn’t.

Color Samples

Like CRI, TM-30 compares color samples rendered by a given test light source and a reference illuminant of the same correlated color temperature (CCT). The first advantage of TM-30 is the selection of color samples. CRI uses the eight samples show in Fig. 1, which are selected from the Munsell color system.

Figure 1 Colors used to calculate CRI Ra

All eight are of medium value and are not evenly distributed across the color space or across the visible spectrum. This allows lamp manufacturers to “optimize” lamp spectra to score a higher CRI Ra than visual evaluation of the light would indicate. TM-30 uses the set of 99 color samples shown in Fig. 2. These color samples range from pale tints to saturated colors, and are drawn from real world objects including textiles, plastics, skin tones, printed materials, natural objects, and paints.

Figure 2 Colors used in the TM-30 calculations

These colors have been selected from a database of about 105,000 objects. In reducing that number to one that is more manageable, the authors of TM-30 made sure that the color samples were even distributed across the most modern color space (CAM02-UCS) and that their reflectances were evenly distributed across the visible spectrum, as shown in Fig. 3.

Figure 3 Spectral reflectances of the TM-30 color samples

Spectral tuning (gaming the system to achieve a higher score) isn’t possible with these color samples, which means that the resulting scores are honest, and comparisons of light sources are apples-to-apples.

Color Space

The second advantage of TM-30 is the selection of color space. A color space is a model of a range of possible colors. In our case we are interested in a color space that encompasses the entire range of visible colors. CRI uses a color space called CIE 1964 (U*, V*, W*), which is no longer recommended for any other use. In other words, it’s very outdated. TM-30, on the the other hand, uses the most up-to-date color space CAM02-UCS. TM-30 isn’t locked in time, either. There is a new, more accurate color space under discussion at CIE. If it is approved, and increases the accuracy of TM-30, I expect it would be included in a future update.

Reference Light Source

Like CRI, TM-30 uses Plankian radiation (blackbody radiator) for lower CCTs and the CIE Daylight (D) Series for higher CCTs for the reference light source. The difference is that CRI Ra has a pronounced shift at 5000 K from one to the other, resulting in the possibility of a significant shift in Ra values between 4999 K and 5001 K. TM-30 overcomes this by using a proportional blend of Plankian radiation and the CIE Daylight (D) Series between 4000 and 4999 K, much the way a variable white LED fixture blends LEDs of two different colors to achieve its full range.

Calculation Results

Instead of a single fidelity value, as with CRI Ra, TM-30 give us a wealth of data about the color rendering of the light source in question. The first is the Fidelity Index Rf. Like Ra, it is a comparison of the color rendering of the test light source compared to the reference light source. However, with 99 color samples it is a tougher test that cannot be gamed. I’ll have more to say about Rf in a future post.

The second is the Gamut Index Rg. Rg indicates the average change in saturation of the 99 color samples as rendered by the test source compared to the reference source. I’ll have more to say about Rg in the future, too.

So, from the start TM-30 gives us more information, but it doesn’t stop there. It also divides the color space into 16 wedges, called hue angle bins, as shown in Fig. 4. The Rf and Rg values of each bin are also calculated and reported so that if a specifier is interested in the performance of a light source in a particular color range, that information is available. The information is also presented graphically by showing the average shift of each bin on the same graphic.

Figure 4 TM-30 hue angle bins

In addition, if you really want to dig down deep, the TM-30 calculation tool calculates the Rf and Rg values of the individual 99 color samples.

What TM-30 Is And Isn’t

TM-30 is a calculation procedure that takes an objective and statistical approach to analyzing two aspects of color rendering – fidelity to a reference source, and saturation shift relative to the same reference source. The calculation also produces information about hue shift, which is presented graphically. The calculation procedure is a consolidation of years of research by individuals and organizations around the world. Its authors come from the research, specification,and manufacturing areas of the lighting industry. Research since its introduction in 2015 has supported its validity as an accurate method of characterizing color rendering. The CIE has endorsed Rf for scientific use in CIE 224:2017 Color Fidelity Index for accurate scientific use. Unfortunately, they declined to endorse it for specification or other uses, as I’ve discussed here. However, quite a few manufacturers see the advantage of TM-30 and are including Rf and Rg information on their cut sheets.

TM-30 isn’t a color rendering guide. It doesn’t contain recommendations for acceptable values. It reports calculated values and leaves interpretation of those values to the specifier based on experience and the particulars of the project. (However, the IES is likely to publish guidance in the future.) It also doesn’t attempt to evaluate color perception or color preference. Those two aspects of color are application (and even situation) dependent, so again the specifier will use experience and understanding of the project to determine what values are appropriate and/or acceptable.

TM-30 provides significantly more information about the color rendering of a light source, and the information presented is far more accurate than CRI. The authors of TM-30, and the IES Color Committee in general, are open to improvements in the calculation and the presentation of its results. As additional scientific information becomes available, or improved or expanded means of calculation and presenting information are developed, it can be updated as needed or on a regular three to five-year cycle.

Who’s Afraid of TM-30?

As the Co-Chair of the IES Color Committee, I have seen too many statements that full-scale adoption of TM-30 is too difficult and will create confusion in the market. Often, these assertions come from major manufacturers who want to control market disruption, not be disrupted. In my professional lifetime there have been, and continue to be, significant changes in the lighting marketplace. When new products are introduced, designers are told about the wonderful benefits of using them. There has never been a time when large manufacturers or organizations with loud voices have said the market could not accept about a new product because doing so was too burdensome. For example,

The introduction and transition to electronic ballasts and transformers meant that we had to learn about reverse phase dimming and control protocols.
The T5 lamp meant we had to change our layout patterns to accommodate lamps that weren’t standard 2’, 4’, and 8’ lengths.
Metal Halide lamps, especially PARS, meant that in exchange for energy savings we had to learn about the color rendering of a new type of lamp, and give up dimming.
Daylight harvesting and daylight responsive designs meant we had to learn about daylight zones, photosensors, and daylight harvesting control systems.
White LEDs meant we had to learn about another light source and its specific pros and cons, including different color rendering properties due to its SPD.
Circadian lighting means we are all in the process of learning how and when to apply the most current scientific evidence to certain project types. Since the science is constantly advancing on this topic, we must be aware and continue to educate ourselves.
Regularly updated energy conservation codes mean that as we begin to memorize the lower LPDs and changes to control and daylighting requirements, we have to relearn that information because it changes every three years.
Most recently, we’re supposed to enthusiastically embrace IoT, adding huge complexity to our lighting control systems and opening them up to hacking.

But, I keep hearing that industry adoption of TM-30, allowing specifiers to have a much clearer idea of the color rendering properties of their light sources, is tooo haaaaard! This is especially maddening when so many professions, including lighting design if you have an LC or LEED credential, require continuing education that is supposed to be more than halfway paying attention to a webinar.

Manufacturers love introducing and promoting new products and technologies that will expand profits, and specifiers get the hard sell all the time. But some manufacturers don’t want to consider TM-30 for several reasons. First, there’s the fear that the R_f value, which is analogous to CRI R_a, will be lower than the R_a value. Even though it’s a different, and tougher, test they fear loss of sales if numbers change. I suspect the manufacturers who fear this the most are those who have most engineered their spectra to score well on R_a, but know that R_f can’t be gamed in the same way. Second, as one manufacturer flat out told me, they’d rather put their money into IoT (and other new and profitable products) instead of updating cut sheets and web pages.

Here’s the thing – as a designer and specifier I have no interest in being stuck in 1965 (the year CRI was unveiled) or even 1995 (the most recent update to CRI). We know that CRI is flawed, we know what the flaws are, and we know that the CIE has been unable to come to consensus on fixing the flaws. The IES has done a great job of developing a new, accurate, modern tool that gives us so much more information than CRI ever could. My design decisions, and my ability to learn about my profession so I can be better at it, are not driven by manufacturer profit masquerading as manufacturers worrying about specifier or consumer confusion. Research over the past two years has shown TM-30 to be more accurate, and we continue to learn more about how to effectively use it. Lighting specifiers should begin the transition to TM-30 by insisting that manufacturers provide them with R_f, R_g, and color vector graphics.

R.I.P. CRI

It’s been a little over two years since the IES released TM-30-15 IES Method for Evaluating Light Source Color Rendition. In that time TM-30 has seen growing support in the industry and a growing body of evidence for its accuracy and usefulness. We’ve nearly reached the moment when we can all agree that it’s time to retire CRI and fully adopt a modern, accurate system of measuring and describing the color rendering of light sources. What’s wrong with CRI? Quite a bit, so if you’re not up to date on the issue here’s an overview.

In 1948 The CIE first recommended a color rendering index based on a method developed in 1937. The 1937 method is a fidelity metric (that is, it compares a test light source to a reference light source) that divides the spectrum into eight bands and compares each band to a full spectrum radiator. In 1965 the CIE finally adopted CIE 13-1965 Recommended method of measuring and specifying color rendering properties of light sources, based on a test color sample method, what today we call CRI R_a or just CRI. From the start it was apparent that there were problems. In 1967 a committee was established to correct for adaptive color shift. Other problems were uncovered, and in 1974 a formal update was published. Errors were uncovered in the 1974 edition, resulting in a third version in 1994, which is the version we use today.

So far, so good. Errors are discovered in the method used and are eventually corrected, so what’s the fuss? The fuss is that the corrections were minor compared to the scope of the errors, and 23 years after the last correction we still don’t have an accurate, up to date system. In the early 1990s a proposal to update the formula and test color samples failed to gain consensus. Two subsequent attempts to improve the metric also closed without adoption. The current problems, as described in the 2011 IES Lighting Handbook, 10th Edition include:

Averaging the color shifts of the eight test colors says nothing about the rendering of any single sample. A large error in one color can be masked by accurate rendering of the other samples.
The test color samples are all of moderate saturation so the index doesn’t reveal color shifts in saturated colors. In addition, the test colors are not evenly distributed through the color space or the spectrum, so light source spectra can be engineered to score higher than visual observation would indicate.
The color space used, the 1964 UCS chromaticity diagram, is no longer recommended for any other use.
All chroma shifts are penalized, even though research shows that moderately increasing chroma is desirable in many applications.
The chromatic adaptation used has been shown to perform poorly and is no longer recommended for any other use.
A single number index gives no information about the direction or extent of color shift for any particular color or color range.

Why haven’t these problems been corrected in the past 23 years? I’m told that there are two issues. The small issue is that competing scientific interests on the committee advocate new metrics that they’ve developed as a replacement or supplement to CRI. The larger problem is that manufacturers on the committee don’t want to see any changes that would reduce the CRI of any of their lamps. From their perspective, it’s better to have a high score on an inaccurate test than a low score on an accurate one. It seems that internal politics has been preventing updates, corrections, and improvements.

Although many other color rendering metrics have been proposed over the years, none has been adopted by CIE, which has the most significant voice on this issue. The result is that the sole internationally accepted metric, which has also been written into product specifications and into codes, is CRI. That began to change in 2015 with the introduction of TM-30. I’ll have more to say about TM-30 in future posts, but for now let’s agree that CRI R_a is broken and CIE is in no hurry to fix it. A better system exists, and our industry should adopt it.

CIE Adopts TM-30 (sort of)

The CIE has issued a statement titled “CIE 2017 Colour Fidelity Index for accurate scientific use” in which they partially adopt the Rf fidelity metric of TM-30. Here are some of the details.

When the IES released TM-30-15 it seemed to be a wake up call for the CIE, who have understood the inaccuracies of CRI but haven’t been able to build the internal consensus to correct them. Two CIE committees soon began work, one on the issue of fidelity and one on other perception related issues. The committee that worked on fidelity (TC 1-90) has decided to adopt TM-30 Rf with a few minor changes to the calculation. They are calling this the CIE 2017 Color Fidelity Index, which will still be described as Rf, and are asking the IES to adopt the same changes to the calculation so that both Rfs are the same.

However, at this time there is no retirement date for CRI, even though they acknowledge that Rf is more accurate. The logic is a bit twisted. Since CRI is used for purposes other than the intended purpose, and since TM-30 doesn’t satisfy those purposes, we’re going to continue to mis-use CRI. Meanwhile, “replacement of the CRI will be a matter of future study and discussion.”

I’ll probably have more to say once I read the entire report. If you want to read the report yourself you can purchase it here.