Video Url
A video overview of IES TM-30-15, which is a new system of several related measures and graphics that can be used together to effectively evaluate and communicate a light source’s color rendering properties.
U.S. Department of Energy

TM-30 is a new method for evaluating light source color rendition, developed by the Color Metrics Task Group, which was part of the IES Color Committee. And then it was later balloted by the Technical Review Council and the Board of Directors. The Color Metrics Task Force was a group of eight individuals, seven voting members. And we worked collaboratively over about a year and a half to really synthesize and bring in existing research on color rendering. This has been a popular area of interest for a couple of decades now, at least.

TM-30 can be used by specifiers to choose the right light source for an application, by lighting manufacturers to engineer new spectra for their light sources that they're trying to sell. It can be used for research purposes to characterize lighting in different ways, and vary the color rendering properties of light sources so you can learn what is, perhaps, the preferred light source for any given application, like an office, or school, or a retail application.

At its most basic, CRI is a measure of color fidelity, or how similar a light source renders colors compared to a reference. And it only tells us the similarity, or the magnitude of the difference. It doesn't tell us how the colors are varying. So with TM-30 we get a whole bunch of additional information. Not only do we learn about fidelity, but we learn about gamut, which is essentially the saturation or how those colors are becoming more or less saturated. We can have graphical outputs that tells us how individual colors or hue ranges are varying compared to the reference. Now on top of those differences in the outputs, there's also a lot of differences in the calculation engine, that's the under the hood part of the metric.

We have a new set of color samples, 99 instead of eight. We have new color spaces – the areas where we're performing the calculations, so the system we're using. And that really improves the accuracy of the calculations and makes them more appropriate and more representative of real world environments.

Rf is the TM-30 Fidelity Index. It functions in a very similar way to CRI. So it's just telling us the magnitude of the similarity between the test and the reference source. You could use Rf, for example, if you have a lighting installation where you have incandescent lamps and you're trying to match the exact color of that with LEDs. In TM-30, we're using the same reference sources as we are with CRI so at CCTs below about 4,500 K we're using Planckian or blackbody radiation. So you can think of that like an incandescent lamp.

Above 5500 K we're using a model of daylight, just like the sky outside. And in between the two, the one difference with CRI is that we're actually blending between 4500 and 5500 K, instead of a sharp cut off right at 5000 K. We have three variables when we're talking about color. We have hue, or red, blue, green, yellow. We have saturation, so is it a pale red or is it a really bright vibrant red. Then we also have lightness, so is it very dark to very light.

Rg is the TM-30 Gamut Index. And it's a relative measure of gamut. So again, we're using the same reference scheme as we would for the Fidelity Index where the reference varies with the CCT of the test source. So Rg will tell us, on average, are colors being saturated, or desaturated. So Rg, the neutral point, the neutral value is 100. If the value is greater than 100, we're increasing saturation, on average. And if the value is less than 100, we're decreasing saturation on average. It doesn't tell us whether we're increasing the saturation of reds or greens or yellows, or any particular hue, but it just tells us the average level of increasing or decreasing saturation.

The way we would use TM-30 together as a system to really pick out the most appropriate light source for an application would be to look at both the average values together, as well as some of the more detailed information. So for example, if I'm trying to increase the saturation slightly, but not get too far away from the reference, I'd look at maybe an Rf in the 80s, an Rg slightly above 100. And if I knew I wanted to really make the reds pop, then I would go to the Color Vector Graphic, and look at that and see where exactly the colors are being saturated.

This is an experimental room that we developed to sort of do some pilot testing to understand how exactly the values and outputs from TM-30 relate to human perception. So we conduct an experiment here. We have color-tunable luminaires. And we can create all kinds of different lighting conditions. In this case, we created 26 different conditions, all varying the color rendering without changing the color of the light itself, or the output. So, constant output and constant chromaticity.

So we asked participants to evaluate whether they liked or didn't like the lighting, and a number of other questions. We sort of created this room with five different groups of objects. So on the table we had natural objects. On this wall we had art work. And behind here, consumer goods, clothing behind me, as well as a mirror so the participants could look at their own skin tones. All the light sources were seven-channel LED systems, so seven different individual LEDs. In the luminaire we control each channel individually, by varying those, we're changing the spectral power distribution, and we're changing the color rendering properties of the fixture.

This is the highest fidelity setting we can create with this lighting system. It has an Rf value of about 94. Rg, when you're up at that high fidelity, has to be about 100. So this might be what you would target if you're trying to create the most natural appearance, the most familiar appearance. So I can change this. And we're going to go through the extremes now. So this will go to something with a very high level of saturation. So this has a fidelity value, Rf, around 65, an Rg value, gamut saturation, around 115.

So this is really, really saturated. You'll notice the reds really stand out here. For example, the orange pepper and the red pepper almost look much more similar now, harder to tell apart. This might not be something you'd want in a restaurant or a grocery store. Things would look a bit unnatural. But say you're in a party store, or an amusement park, and you really want to make that cartoonish appearance, this might be something you're interested in.

So we can go to another extreme now, to a source that's very desaturating. So again, this has the same Fidelity Index as the previous source, Rf about 65. But now the Rg value is down around 80. So this is really desaturating. Everything's a bit dull and depressing. Now you see the cabbage is almost like a dark purple color. It used to have sort of more fuchsia color to it. Tomato really dull. Now the orange and red pepper, again, they look very different in this case. So we're not getting the hue shift, more of a desaturating effect.

So in our experiment in this room, this is actually the least favored source of all. And the other important thing is it has the same fidelity value as the previous source. This is the issue with CRI is that you're only considering fidelity, these two would be rated the same. So we'll change it now back to a more typical source. So this is actually the most preferred source in this experiment that we conducted. So this has a fidelity value around 84, a gamut value around 101. Pretty neutral in terms of average saturation, but it's boosting the reds, which is something a lot of people look for. Seems to be an indicator of preference throughout this experiment and in many other experiments.

Fairly high fidelity. Now just to illustrate the importance of the Color Vector Graphic and understanding where saturation is occurring or isn't, I can go to another setting with the same values for Rf and Rg, but this one's instead of saturating the reds, it's desaturating the reds and increasing the saturation in the yellows, which our human vision doesn't seem really to be that sensitive to. So this is actually much, much lower in terms of its rank order in the experiment that we conducted. So this is something pretty typical you might see in an architectural interior.

So I can show one more here. And this is the second most preferred source. So this actually has a fidelity value around 78, but it's actually an Rg value of 116. So again, very saturated, but we don't see the same kind of hue shifts. The red and orange pepper look still distinct in this case, versus that one that was really too far saturated and too much hue shift. So this one's not going as far. Still that saturation, that boost that people tend to prefer, at least in this room.

So a bit lower fidelity than you might typically use in an architectural interior. But again, the importance of pairing the fidelity and gamut values together with the Color Vector Graphic to really get a better overall impression of what's going on in the space.