 Have you ever thought what it would be like to live in a black and white world, one that has no color? Color-coded traffic signals tell us when to go and when to stop. The color of certain foods gives us a clue as to their freshness. Color even warns us of danger. Without color, our lives certainly would be different. Color is as important indoors as it is outside. For the most part, we are accustomed to the way things look under daylight or incandescent lighting conditions. To save energy though, the use of different types of lighting systems has increased in recent years and this change has created some problems. Energy-efficient lights can cause colors to shift from what we are used to seeing to something very different. To demonstrate, split-screen pictures such as this one show how a particular scene looks under incandescent or tungsten light on the left and under examples of more energy-efficient lights on the right. During this presentation, we will talk about the three variables that affect color perception and describe some of the work that has been done at the National Bureau of Standards on how energy-efficient lighting systems can affect the way colors look. We will also provide some helpful suggestions on things to consider when designing energy-efficient lighting systems for commercial, institutional and industrial buildings. When designing a building lighting system, decisions you make about lighting and color will affect the system's initial cost and ongoing operating expenses. They will also affect owner and user satisfaction with the building. One obvious way to reduce lighting costs is to use daylight when possible to replace artificial lighting. Another way is to use lights that are more energy-efficient, such as the examples of high-intensity discharge type that you see here. You also should be aware of such economic benefits as longer lamp life and of negative factors, such as the detrimental effects these lights can have on people due to color distortion. Your decisions about trade-offs between good color rendering and efficient lighting should be guided by the requirements of the visual tasks involved. Color distortion can make pairs of colors that need to be told apart look too similar to each other, make single colors that need to be identified in an absolute sense, look the wrong cue, or finally make the whole pattern of colors look peculiar in a way that some people don't like. Three major variables are associated with the perception of color. The first is a person's tendency to adapt to colors. That is, the appearance of colors you are looking at now is influenced by colors that you just viewed. This is referred to as chromatic adaptation. The second major variable affecting perception is the colorant, which gives an object its color. The third variable is the spectral energy distribution of the light source. To demonstrate chromatic adaptation, stare at the small black spot in the middle of the screen where the three colors overlap. Try not to let your eyes wander at all. If things start to fade or look misty, that's a good sign. Keep staring a little longer. Now stare at the black spot in the middle of this white field. The after image, which should appear in complementary colors, illustrates how adaptation of the eye and brain are critical in the perception of color. The colorant in an object, the second variable, is usually a pigment or dye which selectively absorbs more light in some wavelength bands than in others, and reflects what is left in each wavelength band back to the eye. The curve indicating the fraction of light at each visible wavelength reflected back to the eye is called the reflectance spectrum of the object. It is important to be aware that many different reflectance spectra can lead to the perception of exactly the same color in the eye. Each different type of light source has its own characteristic spectral energy distribution, or spectrum, the third variable. A spectrum may be continuous, as in an incandescent lamp, have energy emitted only at discrete wavelengths, as in some energy-efficient lamps, or it may have a combination of both, as in other energy-efficient lamps. The second and third variables interact because the light that is reflected by an object to a greater or lesser extent at each wavelength is part of the light that falls on the object from the light source. The National Bureau of Standards has been studying these variables to better understand various lighting applications and the limitations of energy-efficient lights. This research is conducted in the Illumination Color Experimentation Laboratory, part of the Bureau's Center for Building Technology. The U.S. Occupational Safety and Health Administration provided funds for constructing the laboratory and for related research. This laboratory provides a realistic environment for studying changes in color rendering or the way colors look under energy-efficient lighting systems. Changeable walls allow the researchers to vary wall colors and reflectance values. The chamber was designed so that different lights can be switched on and off for comparison purposes. The output of different lights can be combined in various proportions to study how a mixture of lights affects color rendering. The laboratory contains several of the many types of high-intensity discharge lamps available on the market today, low-pressure sodium, high-pressure sodium, clear mercury, and metal halide. It also has incandescent and fluorescent lighting. Each type of lamp has a unique color spectrum. Now, using the chamber, let's examine the perceptual color changes that can occur when these light sources are used. Here are several U.S. Department of Transportation color tolerance charts, which correspond to the current ANSI standard safety colors of the American National Standards Institute. Each chart shows a safety color standard in the middle and around the standard a number of colors representing maximum allowable color differences in manufacturing. Suppose we take into daylight a spectroradiometer, an instrument which measures the spectrum or amount of light coming off an object, wavelength by wavelength. Assume that we measure the central or standard color on each of the five basic safety color tolerance charts in daylight, the red, orange, yellow, green, and blue standards. From these spectra, we can calculate the daylight color of each standard and then plot the five measurements on a special color or chromaticity diagram. The locations of the five standard safety colors can be connected to form an area or gamut of coloration. The chromaticity diagram is a standard way to quantitatively represent the color of any object, aside from the lightness or brightness aspect. The diagram is based on the fact that any color can be matched by a mixture of three fixed colors called primaries. They are red, green, and blue. We are talking here about mixing lights, not mixing colored materials such as paints. Given any color, we can calculate how much of each primary color would be needed in a mixture to match that particular color. Then we can plot the fraction of red in the mixture as the horizontal or x-axis in the chromaticity diagram. The fraction of green may be plotted as the vertical or y-axis. Whatever remains is the fraction of the blue primary in the mixture. For example, to match whites, more or less equal amounts of all three primaries are needed, so the whites are colors located in the middle of the diagram. Another example is the standard blue color, represented as the point of intersection of lines extending from the x and y values on the chromaticity diagram. The gamut of coloration is constructed by connecting lines between points representing the five basic safety colors. Now, let's measure the five standard safety colors under an artificial light source, in this case an incandescent light, and again plot the results in the x-y diagram. The gamut has rotated slightly toward the long wavelengths, but is still roughly the same shape and size. Since the distance between the points is roughly the same as before, there is no trouble distinguishing one color from another. Now, what happens when the standard colors are measured under clear mercury light? The gamut has become shifted and distorted. The red standard is no longer in the red region, but is instead in the yellow region, whereas the blue standard does not shift much. From this, we can predict that blue colors will show up well under mercury, but reds will not, and in fact this corresponds with what people see. Now, let's repeat the measurement, this time under high pressure sodium light. Again, the gamut of coloration has shrunk and shifted, so we would predict that colors such as reds, oranges, and yellows will look more similar to each other under high pressure sodium than under daylight. That is a valid prediction. However, these predictions can only be trusted so far. The chromaticity diagram is based only on the light reaching our eye from the sample, and it makes no allowance at all for our first basic variable, the observer's state of adaptation. Under high pressure sodium light, all five of the standard safety colors have been physically shifted into the region of the chromaticity diagram that we normally identify as orange. Even the blue standard has been pushed physically into the orange region because of the heavy concentration of light energy in the orange band of wavelengths. What gives a blue sample the potential for appearing blue is that it reflects much of any light from the blue part of the spectrum that falls on it, and reflects much less of the light in other parts of the spectrum. Since ordinary white illumination contains energy in all parts of the spectrum, including the blue, when we shine such light on a blue sample, the sample reflects to the eye a substantial amount of blue energy and less energy in most other parts of the spectrum. However, high pressure sodium does not have much energy in the blue end of the spectrum. So, although the blue sample is good at reflecting blue light, there is little blue light for it to reflect. On the other hand, although the blue sample is poor at reflecting orange light, much of the high pressure sodium light is in the orange range, so a good deal of the light reflected from the sample is in the orange part of the spectrum. Despite all this, because of chromatic adaptation, nominally blue samples such as this plate actually look blue under high pressure sodium light and not orange. Thus, only the crudest sorts of predictions can be made from chromaticity alone, without somehow allowing for adaptation. Unfortunately, finding a good formula for accurately predicting the effect of adaptation is currently one of the chief unsolved problems of color theory. Now, let's examine the effect of these and other lamps on the colors of familiar objects. Here is the spectral energy distribution of incandescent light. Here's what our laboratory looks like under incandescent light on the left, compared with various energy-efficient lights on the right. Is there any systematic approach for assessing or predicting how well colors are rendered under particular light sources? Remember, color perception is affected by three major variables, chromatic adaptation, the specific colorant, and the spectral energy distribution of the light. A number of rating systems are described in the technical literature, but none of them is totally satisfactory. We will describe the most widely used system, the color rendering index, which is based on predicting the average change in appearance of eight standard color chips, when illumination is shifted from a standard reference source to the one being evaluated, in this case, high pressure sodium. The definition of this index involves gauging appearance shifts by corresponding shifts in chromaticities. For example, this diagram shows the color rendering of a test lamp through the chromaticities of the eight standard colors as illuminated by the lamp, seen as circles, and by the reference source, seen as squares. There is such a close agreement that this lamp has a color rendering index of 95 out of a possible 100. The UV diagram seen here is similar to the XY chromaticity diagram shown earlier when shifts in the gamut of coloration were demonstrated. The name UV refers to the variables plotted on the axes and has no connection with ultraviolet light. The UV and XY diagrams incorporate the same facts about mixtures of colored lights except for the spacing between colors, which is different. In this diagram, the distance separating any two points is approximately proportional to the amount of difference that people see between the corresponding colors. This is true for the UV diagram, but not for the XY. So UV is used when making predictions about differences between colors. Now let's examine another test lamp and see how it renders the standard colors, the circles, compared with the reference source, the squares. To the eye, the color of the light from this test lamp is the same as the color of the light from the good color rendering lamp we just considered, but the spectra of the two lamps are very different. As you can see, the test chromaticities here are a long way from the reference colors. The color rendering index of this lamp is only 18 out of 100. When the color shifts are large, the index number is low, as in this case. Often lamps are advertised on the basis of their color rendering index, but this provides only part of the information needed. Lamps such as high pressure sodium have relatively low color rendering indices, and yet some people perceive colors under them to be truer than one would predict on the basis of the color rendering index. The reason is the great amount of adaptation that takes place in the human visual system. Thus in areas where critical color recognition work is not required, you can consider using an energy-efficient lamp, even though it has a low color rendering number. In addition to putting out more light per unit of electricity, high efficiency lamps last much longer than familiar incandescence, a distinct advantage where replacing lamps is difficult. On the other hand, where color recognition work is a key factor, high intensity discharge lamps of the types currently available should not be used. For example, in a new federal building in New England, which includes a medical facility, the lighting that was installed shifted the colors so much that doctors were unable to see the true color of a patient's skin or tongue. Had experts from the lighting and medical communities been consulted prior to installing high pressure sodium lamps, the lighting system in the medical area would not have had to be replaced after the building was completed. Some other disadvantages of energy-efficient lights are that initial cost is high, and there is sometimes an initial consumer resistance. Many energy-efficient lighting systems, when first installed, may be objectionable to the user because of the unusual color rendering of familiar objects. This perception can be overcome somewhat through the use of carefully chosen interior colors. We have seen that it's not currently possible to predict with much accuracy how colors will look under energy-efficient lights using chromaticity diagrams or formulas. It remains better today to view interior colors, objects, and fabric samples under the same type of light source as the one proposed for installation before making final selections. This approach would have helped in Atlanta, Georgia, where employees complained that they couldn't eat in a company cafeteria because the food had a strange, unappetizing color. Had officials viewed beforehand how food looked under the type of lighting installed, they might have avoided the expense of changing the lighting, as had to be done. In designing a building, the colors of safety signs and hazard markings are also important to keep in mind. Sometimes the perception of a red or orange warning color is required, such as in a hazardous workplace. Tests at NBS have shown that red or orange fluorescent paints can be recognized under any of the currently used light sources on the market, even low-pressure sodium. That is often not true of other types of paints, however. Under some lights, ordinary paints in bright red colors appear dull brown, for instance, thereby losing their safety importance. Another concept to consider is the use of transition zone lighting to help overcome objections of building occupants to drastic color differences when passing from a room exposed to daylight to an interior room lighted with energy-efficient lamps. For example, if most of a building contains high-pressure sodium lights, you can light a transition area with fluorescent lights, as well as high-pressure sodium, changing the mix progressively from one end of the area to the other. The gradual change between the two light sources will help occupants adapt. Another technique that can help ease problems with color distortion is to mix some proportion of fluorescent lamps with energy-efficient lamps uniformly throughout the building. In order to determine a suitable proportion, a pretest before installing the lighting is almost a necessity. During this presentation, we have examined three major variables affecting color perception, chromatic adaptation, colorants, and the spectral energy distribution of lights. We have described the National Bureau of Standards research aimed at helping solve problems associated with the effects of energy-efficient lighting on color. And we have presented a few suggestions on how to minimize these problems, such as previewing samples under the lighting system to be installed. Over the next several years, as part of NBS's lighting technology program, the Bureau plans to pursue further theoretical and experimental research on color rendering, color appearance, and their practical implications for lighting design.