Is Ra=100 really the ideal color rendering?

Color is one of human feelings, and it is always related to the observer's personal subjective experience. The feelings that each person has after seeing a color are difficult for others to know. Therefore, the study of color is always filled with mysterious imagination. At the same time, color makes the world colorful; visual arts, image display and transmission, textile printing and dyeing, color printing, etc., all rely on the study of color. Thus, the research on color and the objective quantitative description of color have become the subjects of study for many scientists.

In 1664, Newton used a prism to disperse white sunlight into different shades of the spectrum, laying the physical foundation for the color of light. In 1860, Maxwell created light from white light to various colors using red, yellow, and green light of different intensities, establishing the foundation of trichromatic colorimetry. Based on this, in 1931, the International Commission on Illumination established the CIE colorimetry system and continuously improved it. Today, the CIE color system is widely used to quantitatively express the color of light.

Color cannot be separated from illumination; only under light can objects display color, and the color of light has a significant impact on people's psychology. Professor Yang Gongxia from Tongji University has made a very brilliant description in the fifth chapter of his book "Vision and Visual Environment."

Under different light sources, the same object can display different colors. For example, green leaves appear bright green under green light but nearly black under red light. This shows that the light source plays an important role in the manifestation of the object's color. The ability of a light source to fully display the color of the illuminated object is called the color rendering ability of the light source.

In 1965, the International Commission on Illumination recommended using the general color rendering index Ra in the CIE color system to describe the color rendering ability of light sources. The general color rendering index Ra has been successfully applied and widely accepted in the lighting industry, but there are also some issues. This article will introduce the evaluation methods for the color rendering ability of light sources and recent progress.

1. General Color Rendering Index Ra

The evaluation method for the color rendering ability of light sources aims to be both simple and practical. However, simplicity and practicality are often two contradictory requirements. In the CIE color system, the general color rendering index Ra is such a compromise: it is relatively simple, requiring only a numerical value within 100 to express the color rendering performance of the light source, with Ra=100 considered ideal.

However, sometimes people's feelings are not so. For example, leaves under incandescent light do not appear very bright. Where is the problem? Let's discuss what the general color rendering index is.

For simplicity, we will only discuss the main composition method of the general color rendering index Ra here, without discussing its specific calculation method.

In fact, we often test the color rendering ability of light sources in our daily lives. Many people have the experience that careful women often check the color of clothes in outdoor sunlight when shopping in malls. What they are doing is actually testing the color rendering ability of the mall's light source: observing the same piece of clothing under the mall's lighting and under sunlight to see what color differences exist. Therefore, describing the color rendering ability of a light source requires two additional elements: sunlight (reference light source) and clothing (colored object).

In the CIE color system, to determine the color rendering ability of the light source to be tested, a reference light source must first be selected, assuming that the color of the illuminated object can be perfectly displayed under the reference light source. The CIE color system stipulates that when the correlated color temperature of the light source to be tested is below 5000K, a black body with a color temperature close to it is used as the reference light source; when the correlated color temperature of the light source to be tested is above 5000K, a D light source with a similar color temperature is used as the reference light source. Here, the D light source is a series of color coordinates that can be represented numerically and are related to color temperature.

After selecting the reference light source, a colored object also needs to be selected. Due to the diversity of colors, a set of standard colors must be chosen to adequately represent commonly used colors. The CIE color system has selected 8 colors, which have various hues and moderate brightness and chroma values.

In the u-v color system, the difference in color coordinates of each standard color plate under the test light source and the reference light source, referred to as color displacement ΔEi, can be used to determine the special color rendering index Ri for that color plate. (Ri=100—4.6ΔEi)

The arithmetic average of the special color rendering index Ri measured for the 8 standard color plates gives the general color rendering index Ra. It can be seen that the maximum value of the general color rendering index Ra for a light source is 100, indicating good color rendering ability.

2. Limitations of the General Color Rendering Index Ra

Although the general color rendering index Ra is simple and practical, it shows serious deficiencies in many aspects.

First, color is a subjective feeling of people, not an inherent property of objects. It is related to lighting conditions, observers, irradiance, illuminance, surrounding objects, and observation angles, and there is no such thing as a so-called 'true color'.

However, since Ra is defined in the CIE system to reach a high value of 100 under near-black body radiation, bulb manufacturers consciously design bulbs to make their color rendering ability when illuminating objects as close as possible to that of black bodies or sunlight. This means that when the spectral distribution of the light source deviates from that of black bodies or sunlight, the color rendering index will decrease. For example, white light LEDs composed of red, green, and blue monochromatic LEDs may have a low general color rendering index Ra, but their color rendering ability is not necessarily poor.

However, researchers Judd, Thornton, and Jerome have confirmed that people do not necessarily prefer the colors illuminated by the reference light sources defined by the CIE. For example, using incandescent light with a very low color temperature to illuminate green leaves is not necessarily a good choice. The stipulation that the color rendering index is best at Ra=100 under black body or sunlight illumination is questionable.

The reference light sources defined by the CIE are black bodies or sunlight with correlated color temperatures close to the light source to be tested. They are sources of continuous spectral radiation with various spectral components of color. When the color temperature is at 6500K, the spectral power distribution of long and short waves is relatively balanced, making it reasonable as a reference light source. However, when the color temperature is below 4000K, the spectral power distribution is severely asymmetric, with the short-wave spectral power of blue being much lower than the long-wave spectral power of red, resulting in a color bias towards red, which raises questions about its suitability as a reference light source.

In the CIE color system, the 8 standard color plates are all at moderate brightness and color saturation, spaced equidistantly in the u-v system. They can be considered to adequately represent various commonly used colors for indoor lighting. However, in outdoor lighting, there are often some colors with higher color saturation, and these 8 standard color plates may not adequately represent commonly used colors.

Many scholars believe that the number of standard color samples is too few, which is another drawback of the general color rendering index. Although CIE has 6 color samples with higher saturation numbered 9-14, they are not included in the general color rendering index Ra. In lighting practice, the colors that people are familiar with, such as skin, leaves, and food, are extremely important, but they are all excluded from the general color rendering index.

Seim once proposed using 20 standard color samples, but this was rejected because it would make calculations too complex. Currently, with the widespread use of computers, it seems that this proposal needs to be reconsidered.

Due to these two major issues in the evaluation of the color rendering of light sources, many other evaluation methods have gained widespread interest, and this article will provide a brief introduction based on the author's knowledge.

3. Flettre Index Rf

Research shows that people tend to remember the colors of familiar objects, particularly when they are vivid and highly saturated. This memory color often aligns with preferred colors and tends to shift towards higher saturation. For example, the memory color of human skin tends to shift towards red, while the color of leaves shifts towards green. This is clearly different from the Ra method in CIE.

Rf is actually a modification of Ra, which includes two aspects:
First, Rf is defined as 90 under the illumination of the reference light source, and only under the illumination of an imaginary 'perfect light source' can Rf be 100.
Second, 10 standard color samples are selected, which include the original samples numbered 1-8, plus samples 13 and 14, corresponding to skin color and leaf color.

At this point, the 'perfect light source' refers to a light source that can shift the colors of the 10 standard color samples towards the preferred direction.

It can be seen that the corresponding color coordinates of the 'perfect light source' for each standard color sample are different and can be determined experimentally. This also indicates that such a 'perfect light source' can only be imaginary.

The calibration method of Rf is similar to that of Ra, but there are two differences:
1. For each standard color sample, the color coordinates of the reference light source need to be adjusted according to the experimentally determined color coordinates of the 'perfect light source'. Then, when illuminated by the test light source, the color difference of each sample is obtained by comparing it with its corresponding 'perfect light source'.
2. When calculating Rf, the average color difference of the 10 samples is taken, but each sample has a different weight. Sample 13 is skin color with a weight of 35%, sample 2 is 15%, sample 14 is 15%, and the rest are 5% each. This particularly emphasizes the importance of skin color. Therefore, the Rf of the test light source can be higher than the reference light source Rf=90, but less than 100.

4. Color Preference Index (CPI)

The Color Preference Index CPI (colour preference index) utilizes the concept of preferred colors proposed in the previous section, defined under D65 light source illumination, where the color preference index CPI = 100.

Thus, the CPI of the test light source can be obtained as follows: under the illumination of the test light source, calculate the difference in color coordinates of 8 standard color samples from the preferred color coordinates, and find the average of their vector sum: CPI=156-7.18.

The above calculations are performed in the CIE UV color system.

Although both CPI and Rf utilize the concept of preferred colors, there are significant differences between the two:
1. In calculating Rf, 10 standard color samples are used, including samples 1-8 and 13, 14, while CPI only uses samples 1-8.
2. In technical Rf, the color difference (ΔE) is taken as 1/5 of the experimental value, while CPI uses the original experimental value.
3. When calculating Rf, the weights of each color sample are different, while CPI takes the same weight.
4. According to the definition, the maximum value of Rf is 100, while the maximum value of CPI is 156.

It should be noted that the researchers who proposed the two indices Rf and CPI both determined the preferred colors through experiments, which used daylight color illumination. There is now evidence that preferred colors are related to the correlated color temperature of the light source. Therefore, when using Rf and CPI to quantify color rendering, it is only applicable to light sources with high color temperatures.

5. Color Discrimination Index (CDI)

Using Ra, Rf, or CPI to describe the color rendering of a light source, the reference light source must have the same color temperature as the test light source. The Color Discrimination Index CDI (colour discrimination index) overcomes this limitation.

This index is proposed based on the assumption that the greater the ability to distinguish colors under a certain light source, the better the color rendering of that light source. Under the illumination of a specific light source, the area surrounded by 8 standard color samples in the CIE UV color diagram is: GA = 0.5Σ(UiVj-UjVi) for i,j=1,2,…8; i≠j.

Under C light source illumination, this area GA=0.005, defining CDI=100 at this time, so under the illumination of the test light source, its color discrimination index is: CDI=(GA/0.005)×100.

Conclusion  

From the above discussion, it can be seen that there are many methods for evaluating the color rendering of light sources, and they are constantly developing and improving. The methods introduced in this article are just a part of them, each with its own advantages and disadvantages. Even the widely used general color rendering index Ra has many shortcomings.

Its main drawback is the selection of the reference light source: the reference light source is a spectrally continuous light source, which is not very suitable as a standard for measuring spectrally discontinuous light sources. The color temperature of the reference light source must be close to the correlated color temperature of the test light source, but in fact, for certain lighting tasks, the color temperature itself has a significant impact on color rendering, and this method is limited to conditions where the color temperature of the light source has been determined.

Its second drawback is the selection of standard color samples: for indoor lighting, it can be considered that 8 standard color samples can adequately represent various commonly used colors. However, in outdoor lighting, for some colors with higher saturation, they cannot adequately represent commonly used colors.