You are here

The frequently asked questions shown below cover TM-30 details.

37. What are the reference illuminants?

Knowledge Level: *

TM-30 is a reference-based measure that compares the color rendition of a test source to a reference illuminant—a mathematically modelled light source—at the same correlated color temperature (CCT). For sources with a CCT of 4500 K or less, the reference is a Planckian radiator—at the right CCT this is essentially an incandescent lamp. At 5500 K or above, the reference illuminant is the CIE D Series, a mathematical model of daylight. Between 4500 K and 5500 K, the reference illuminant is a proportional blend of Planckian radiation and the D Series illuminant, each at the specified CCT. For example, at 4750 K, the reference illuminant is 75% Planckian radiation (at 4750 K) and 25% CIE D4750.

38. How were the reference illuminants chosen?

Knowledge Level: **

Broadly, the IES Color Metrics Task Group chose to use the same reference illuminants as were used for CIE CRI. Importantly, these illuminants are mathematically derived, and are representative of non-electric, “natural” lighting. Some people have suggested that the references should be all Planckian radiation or all CIE D Series, but in fact this distinction is fairly minor. The sun is a Planckian radiator, and daylight only differs from Planckian radiation because the Earth’s atmosphere slightly filters some wavelengths. At any given CCT the two spectral power distributions are very similar, and thus using one instead of the other would have very little effect on scores.

Some people have also suggested that the reference sources should not be on the Planckian locus, because off-Planckian sources (i.e., those with a negative value for Duv) may be viewed as more neutral or be more preferred, according to some research. Unfortunately, there are no defined standard illuminants with chromaticity in this range, which makes establishing reference sources more difficult. While the maximum achievable Rf value is marginally lower as Duvincreases, the range of possible Rg values increases. In short, TM-30 is not an impediment to the development of off-Planckian sources.

Some questions in the development of color rendition metrics are choices with no right or wrong answer. With TM-30, the reference illuminants are one of these choices, especially because TM-30 no longer only emphasizes fidelity, where a higher number requires greater similarity to the reference. In other words, the references are anchoring points, and there is a clear way in which sources can be superior to the reference, even if they don’t render colors similarly to the reference.

39. Can I use my own reference source?

Knowledge Level: ***

TM-30 specifies the reference (see question 37) that must be used for calculation at each CCT. Defined references enable appropriate comparisons between products, and are thus essential for commerce.

Importantly, the reference source is not necessarily the ideal source for an application, or the most important point of comparison. For example, if one wants to render the colors in a painting to match what the artist saw while painting under a different lighting condition, the artist’s lighting condition can be used as a reference and a unique fidelity value could be calculated. Going one step further, one could calculate the special fidelity value for the actual colors in the painting, rather than the generic sample set used in TM-30. It is possible to use the TM-30 framework to perform these types of calculations, but this is not explicitly defined in the TM; it must be a custom calculation performed by the user.

40. What are the 99 color evaluation samples?

Knowledge Level: *

The 99 color evaluation samples are a set of spectral reflectance functions for real objects. These objects, or subsets of them, are used in determining all the measures included in TM-30. The objects include paints, textiles, inks, skin tones, natural objects (e.g., food, flowers, foliage), and plastics. They were specifically chosen (see question 41) to represent a useful set of objects independent of any given architectural space.

41. How were the 99 color samples chosen?

Knowledge Level: **

Color samples are an important part of any color rendering metric. They serve as a standard by which sources can be compared, but the applicability of the comparison depends on the samples being relevant to a given environment.

The 99 color evaluation samples (CES) were downselected from a set of more than 100,000 object spectral reflectance measurements. The large set of samples was considered to represent the extent of possible colors, but because it was comprised of several specific databases, it was not an even sampling of all colors. Thus, a process was established for generating a subsample that evenly covered the volume of possible colors, and included a balance set of spectral features. Applying these formulas, a set of approximately 5,000 reflectance functions was established.

While the set of 5,000 samples had desirable properties for being an average characterization of the colors in the world, the TM-30 developers felt it was too large for rapid calculations. The group felt that a tolerance of plus or minus one point for calculating fidelity for a smaller set versus the set of 5,000 samples was acceptable. An optimization routine was used to establish the final set of samples that met this criterion, and the set of 99 samples was established. Note that due to their relative ubiquity, two skin samples were separately selected; the two samples provided the greatest correlation to a larger set of skin reflectance measurements.

The samples are numbered according to their hue angle under the 5000 K reference illuminant.

Despite the effort that went into selecting the samples, TM-30 users must recognize that the objects in any given environment differ from the 99 CES to some degree. In some cases, indices for specific hues may provide more useful information than average values, due to an uneven balance of hues in the space. In most cases, the objects in a space will not exhibit spectral neutrality, but because each space is different and there is no way to characterize an average architectural space, the developers felt a spectrally neutral sample set was the most appropriate.

42. Does TM-30 include saturated samples?

Knowledge Level: **

Yes. TM-30 includes a wide variety of samples, both saturated and unsaturated, in all hues. The TM-30 samples were chosen (see question 41) from a large library of more than 100,000 measured object spectral reflectances, which were considered to represent the extents of possible colors. Some of the most saturated samples were excluded from the selection process to ensure the validity of color difference formulae, and to make sure the samples were representative of typical architectural environments. Still, the figures below show that the TM-30 samples cover the range of samples included in CIE CRI or CQS, as well as common consumer goods and natural objects. More importantly, the samples have a large variety of spectral features, which are important when determining color shift for the sample. A less saturated sample can have the same spectral feature as a more saturated sample; that is, rendition for saturated samples can typically be predicted with somewhat less saturated samples.

43. How many points of difference for Rf and Rg are noticeable?

Knowledge Level: *

There is no value that represents a universal threshold for noticeability of color rendition differences. TM-30 characterizes a light source, not an application. That is, it does not take into account luminance, scene makeup, proximity, or any other factor that may contribute to our perception of color difference in a space. In one situation, a certain difference in fidelity may be noticeable, but in another, it may not. In fact, because both TM-30 Rf and TM-30 Rg are average measures, it’s possible to have two sources with identical values for both measures that render colors significantly differently [1].

1. Royer MP, Wilkerson AM, Wei M, Houser KW, Davis R. Human Judgements of Color Rendition Vary with Average Fidelity, Average Gamut, and Gamut Shape. Submitted for Publication.

44. Are negative values possible for TM-30?

Knowledge Level: **

Although not common, values in the CIE CRI system can be negative. This is not intuitive, and zero is often mistakenly claimed as the lower limit for the scale. For the fidelity values in TM-30 (including Rf, the sub-indices for skin or hue-specific fidelity), zero is set as the lower limit using a logarithmic transformation, so the full range of possible values is 0 to 100. The goal was to make the scale more intuitive, even though it has little to no effect on sources that are practical for interior architectural applications; only fidelity values less than about 30 see a meaningful change due to the transformation.

Also note that TM-30 sub-index values do not suffer from the same scaling issues that occur with R8 or R9 in the CIE CRI system. This is because CAM02-UCS, used for TM-30, is a more uniform color space than CIE U*V*W, which is used for CIE CRI.

45. Does TM-30 address white rendering?

Knowledge Level: **

TM-30 includes gray and near-white samples, but it does not address the effect of optical brightening agents, which contribute to the appearance of many white objects. This may be addressed in separate metrics currently under development by a different IES committee.

46. How is TM-30 related to light source chromaticity?

Knowledge Level: **

Light source chromaticity and color rendering are independent characteristics, although both can influence the appearance of objects in a lighted environment. Light source chromaticity characterizes the color of the light, which can in turn affect the appearance of objects, although the human visual system has the ability to adapt to a consistent white point, much like the adjustment that can be made on a camera. This enables colors to appear consistent, within limits, even if the light has a different chromaticity.

Light source chromaticity—or CCT, which is derived from chromaticity—and color rendering can be specified together to create an appropriate luminous environment. Because the TM-30 measures are based on a reference illuminant at the same CCT, the two attributes can be considered independently. That is, the range of values for the TM-30 measures does not change at different CCTs.