Abstract

Previous analyses have shown that optimizing illuminants’ spectral power distributions for object reflectance can yield energy savings in excess of 40% by reducing the light lost to absorption. Here, commercially available LEDs and real objects, instead of theoretical spectra and test sample colors, are investigated. Simulations show that energy savings of up to 15% are possible when illuminating common objects with mixtures of narrowband LEDs, compared to illumination by reference phosphor-coated white LEDs, without inducing changes in color appearance. Experiments show that higher energy savings are achievable without degrading object appearance. Object optical properties impact the success of this approach.

© 2017 Optical Society of America

Introduction

Electric lighting systems are widely used in the built environment, and the quality of light they provide has a significant impact on occupant satisfaction. In the past few decades, more efficient lighting technologies, such as solid-state lighting (SSL) devices, emerged. These technologies behave quite differently from their predecessors in a number of respects and present a range of opportunities for changing the way that lighting is provided within architectural spaces.

The spectral power distribution (SPD) of a light source, the power emitted as a function of wavelength, and the reflectance characteristics of materials determine objects’ color appearance [1]. Objects reflect, absorb and/or transmit the light incident on their surfaces. Reflected light reaches observers. However, absorbed light transforms into heat and remains within the object, having no impact on human vision. For example, a yellow rubber duck predominantly reflects light from middle and longer wavelengths, while much light of shorter wavelengths are absorbed by the duck and do not reach the observer. Energy absorbed in the objects is useless for illumination and can be considered a source of loss. Therefore, an SPD-adjustable lighting system could leverage the spectral properties of SSL devices by minimizing the unnecessary spectral power emitted by the light source to reduce absorption by illuminated objects.

Though a previous publication has speculated on one way in which lighting systems intended to reduce the light lost to absorption might be designed [2], the engineering of such a system cannot be undertaken until the spectral design considerations are better understood, which is the aim of this research. It is reasonable to envision, however, that sensors could detect the surface colors of all objects within a space and luminaires could distribute light in spatially precise ways, much like video projectors (but without filters). By doing this, individual objects within the space could be illuminated by light spectrally tailored to minimize absorbed light, while still facilitating the appropriate object color appearance. A related approach, that is more spatially simple, involves optimizing the light spectrum for more than one object color [3], which sacrifices some of the possible reduction in absorbed light for likely increased ease of implementation.

Recent research has shown that computationally generated SPDs can minimize the absorption of light by optimizing light source spectrum for object reflectance, without inducing perceptible color differences [2,4]. In these previous studies, color differences were calculated for the 15 reflective samples used in the color quality scale (CQS) [5], when illuminated by a reference illuminant and test light source, in CIE 1976 L*a*b* color space using the CIE 2° standard colorimetric observer [6]. Results showed that spectrally optimized single-peak SPDs could reduce the energy used to light the objects between 38% and 44%, without reducing luminance or inducing visible color shifts [2]. Although an imperceptible difference of ΔE*ab<1.0 [7] was the focus of the research, the analysis also showed that increased energy savings are possible, between 43% and 62%, with higher tolerances for color shifts (ΔE*ab<10). Using a similar research approach, theoretical two-peak test SPDs have been shown to result in energy savings of up to 71%, without inducing perceptible color shifts [4]. In both studies, incandescent and equal-energy radiators were used as reference illuminants, and theoretical test SPDs were simulated across a broad range of spectral possibilities. Therefore, the test SPDs that resulted in the greatest energy savings had spectral shapes that cannot be easily manufactured with current commercialized technologies. These earlier studies established the theoretical feasibility of reducing the energy consumed by lighting, by minimizing absorption, without consideration of current technological limitations. Building on the previous research, this project focuses on the possibility of nearer-term implementation of this approach, by quantifying the potential energy savings when commercially available light sources are optimized for the reflective characteristics of commonly found objects, to minimize the loss of light to absorption. Absorption-reducing, spectrally optimized SPDs were used in experiments to investigate observers’ judgments of the naturalness and attractiveness of object color appearance.

Methods

Computational analyses

The color appearance of ten objects was determined when illuminated by reference white light sources and various combinations of nine different real narrowband light emitting diodes (LEDs) in CIE 1976 L*a*b* color space [6]. Ten objects that have generally recognizable color appearances were selected across five different hues (red, orange, yellow, green and blue). The spectral reflectance of a Coca-Cola can and a tomato for the red hue, a mandarin and a carrot for the orange hue, a 3M Post-it Note and a lemon for the yellow hue, a Granny Smith apple and a lime for the green hue, and a container of Nivea Creme moisturizer and a blueberry for the blue hue were measured with a Konica Minolta CM-2600d spectrophotometer. Figure 1 shows the spectral reflectance factors of these objects.

 

Fig. 1 Spectral reflectance as a function of wavelength for the 10 reflective objects.

Download Full Size | PPT Slide | PDF

In previous studies [2,4], object color appearance was compared when illuminated by computationally generated test SPDs and reference illuminants (incandescent and a theoretical equal-energy radiator). In this research, commercially available light sources, in both the test and reference conditions, are investigated to analyze the real-world implementation of the concept. Two white light sources that are readily available in the lighting market were chosen as reference illuminants; their SPDs are shown in Fig. 2. The color appearance of objects was compared when illuminated by either a phosphor-coated LED (pcLED) (CCT = 4101 K, CRI Ra = 81, R9 = 15, CQS Qa = 80) or a phosphor-coated LED with an additional red peak (pcLED + red) (CCT = 3061 K, CRI Ra = 90, R9 = 64, CQS Qa = 89) and when illuminated by test SPDs that were created by mixing the SPDs of commercially available narrow peak LEDs.

 

Fig. 2 Power as a function of wavelength for the two reference illuminants.

Download Full Size | PPT Slide | PDF

For the test light sources, the SPDs of nine different LEDs were measured with a Photo Research Spectrascan PR-730 Spectroradiometer. These SPDs were arranged according to peak wavelength and assigned a channel number. Channels 1, 2, 3, 4, 5, 7 and 8 belong to Source Four LED Profile x7 Color System theatrical lights, and channels 6 and 9 are from Innovations in Optics Lumibright LED engines. The spectral properties of the narrowband LEDs that were combined to generate test SPDs are shown in Fig. 3.

 

Fig. 3 The spectral power distributions of the nine narrowband LEDs that were mixed to simulate test SPDs.

Download Full Size | PPT Slide | PDF

Test SPDs were iteratively simulated by mixing these nine components, after their peak intensities were normalized, in different proportions. For example, a test SPD could be generated by combining 100% of channel 1, 0% of channel 2, 20% of channel 3, 10% of channel 4, 60% of channel 5, 70% of channel 6, 0% of channel 7, 100% of channel 8, and 50% of channel 9. In the iterative process, each step added 10% relative power to a channel until it reaches 100%, and then the next channel would be increased iteratively. As a result, numerous variations were tested for each object to identify the optimal SPD—the SPD that results in the smallest shift in object color and the largest decrease in absorbed light, relative to the reference source. Once the channels were combined, the overall magnitude of the test SPD was rescaled so that the amount of light reflected from the object was the same for both the reference and test light source. This means that the object would have the same luminance in both conditions, making the calculation of the amount of energy saved with the test SPD meaningful. Finally, the test SPD was used to calculate energy consumption, object color coordinates, object chroma difference and total color difference between illumination by the test and reference SPDs. The calculations assumed that the observer’s white adaptation point (Xn Yn Zn) was the chromaticity of the reference light source.

In the first analysis, only test SPDs that yielded a color appearance of the illuminated object that was imperceptibly different from its color under the reference light, with a ΔE*ab<1.0, were considered. A higher, but likely not greatly disturbing color difference of ΔE*ab<5, was also analyzed. Since several studies suggest that users prefer increased saturation of the colors of illuminated objects [8–10], chroma differences were restricted to dC*ab>0 (a signed measure of chroma difference in which positive values indicate increased chroma when illuminated by the test SPD). Hue and lightness changes were limited to be imperceptible, such that ΔE*ab - dC*ab<1.0. Test SPDs were selected in which energy consumption and the differences in hue and lightness were minimal, and chroma differences were positive.

Psychophysical experiments

Following the computational analyses, two experiments were conducted in the University of Sydney Lighting Lab to investigate the perceived naturalness and attractiveness of objects under selected test SPDs. Five of the ten objects that were computationally investigated were chosen as stimuli (i.e., tomato for red, mandarin for orange, lemon for yellow, Granny Smith apple for green, blueberry for blue).

Twenty-one naive participants, aged between 18 and 40 years, took part in each of the two visual experiments, which used a two-alternative forced choice method. All participants had normal color vision as tested by Ishihara plates. Participants were asked to judge the naturalness of object color appearance in the first experiment, and attractiveness in the second experiment. These two experiments were conducted several weeks apart, to mitigate possible association and confusion between perceived attractiveness and naturalness of object appearance. Participants were asked to make judgments related to the color appearance of the presented stimuli in two adjacent black booths within four seconds. Booths were placed 1.5 m from the observers. The interiors of the booths were covered with Protostar© flocked light trap material with 0.4% average reflectivity at 0° incidence angle.

One of the booths was illuminated by the reference light source (either pcLED or pcLED + red), and the other by one of the test SPDs selected from the computational analysis. Six different test SPDs, yielding gradually increasing color differences, ΔE*ab<1.0, ΔE*ab = 1-3, ΔE*ab = 3-5, ΔE*ab = 5-7, ΔE*ab = 7-9 and ΔE*ab = 9-15, were used to illuminate each object. The characteristics of the test SPDs when the reference light source was the pcLED and the pcLED + red are shown in Table 1 and Table 2, respectively. As stated previously, hue and lightness changes were imperceptible and limited to ΔE*ab - dC*ab<1.0 for these SPDs. For each object and each reference light source, each participant reported their judgments for six test SPDs, which were presented ten times each in a random sequence, resulting in 60 experimental trials. Since there were five objects and two reference light sources, each participant completed a total of 600 trials for each experiment.

Tables Icon

Table 1. The narrowband LED channels that were mixed to generate the test SPDs, and the resulting color differences ΔE*ab, energy consumption relative to the reference source, and chroma difference dC*ab values, when the reference light source was the pcLED.

Tables Icon

Table 2. The narrowband LED channels that were mixed to generate the test SPDs, and the resulting color differences ΔE*ab, energy consumption relative to the reference source, and chroma difference dC*ab values, when the reference light source was the pcLED + red.

Both of the reference light source SPDs, pcLED and pcLED + red, and all of the test SPDs were generated by mixing the seven channels of the Source Four LED Profile x7 Color System theatrical lights. These luminaires were mounted above the booths, with their light projecting through a hole in the top panel of the booth, as shown in Fig. 4. Using seven channels from one light source, instead of nine channels from two separate light sources (as were used in the computational analyses), allowed the lighting in the adjacent booths to be easily alternated between the reference and test SPDs, reducing bias. The intensity of each light source was adjusted for each object, so that the intensity of reflected light (luminance) was the same for both stimuli.

 

Fig. 4 The test and reference SPDs were generated by the Source Four LED Profile x7 Color System theatrical lights, which were positioned on top of the two adjacent black booths.

Download Full Size | PPT Slide | PDF

Results and discussion

Computational results

The data from the computational study demonstrated that it is possible to optimize the spectrum, using a combination of existing narrowband LEDs, to reduce the energy lost to absorption, without inducing shifts in the color appearance of illuminated objects. Among all combinations of test SPDs that yielded the specified object color appearance, those that resulted in the lowest energy consumption were identified for each object. Figure 5 shows an example: the test SPD that was optimized for the green Granny Smith apple’s spectral reflectance factor when compared to a reference pcLED source. When the green Granny Smith apple is illuminated by a combination of 40% of channel 2, 20% of channel 3, 20% of channel 4, 100% of channel 6 and 80% of channel 9, energy consumption is reduced by 8% without inducing any visible changes in color or luminance.

 

Fig. 5 When the green Granny Smith apple (dashed gray line; right y-axis) is illuminated by the test SPD (continuous black line; left y-axis), which consists of five different LEDs, 8% of energy is saved, compared to illumination by the reference pcLED (continuous gray line; left y-axis), without inducing any noticeable color shifts (ΔE*ab<1.0).

Download Full Size | PPT Slide | PDF

Similarly, energy consumption decreases by 17%, compared to the pcLED reference, when the green Granny Smith apple is lit by a combination of 40% of channel 2, 60% of channel 7 and 70% of channel 9, as shown in Fig. 6. However, in this case, the appearance of the green apple shifted slightly (ΔE*ab = 3.6).

 

Fig. 6 When the green Granny Smith apple (dashed gray line; right y-axis) is illuminated by the test SPD (continuous black line; left y-axis), which consists of three different LEDs, 17% less energy is required than when it is illuminated by the reference pcLED (continuous gray line; left y-axis) if a slight color shift (ΔE*ab = 3.6) is allowed.

Download Full Size | PPT Slide | PDF

The results indicate that spectrally optimized SPDs can result in energy savings between 11% and 15% when the reference illuminant is a pcLED with a ΔE*ab<1.0, as shown in Table 3. There is a greater variability between the results for different objects, 2% - 15%, when test SPDs were compared to the pcLED + red reference. When a slight difference in color appearance was tolerated (ΔE*ab<5), higher energy savings were achieved, as expected: 14% - 19% when compared to the pcLED and 9% - 16% when compared to the pcLED + red.

Tables Icon

Table 3. Percentage of energy savings for illumination of individual objects, compared to reference light sources, when color difference is imperceptible (ΔE*ab<1.0) and detectable, but not large (ΔE*ab<5).

Although the difference between the magnitudes of energy savings for illumination of individual items is not noteworthy for the pcLED reference, there is a clear pattern when the pcLED + red reference is considered. Understandably, objects with red-orange hues absorb relatively little light when illuminated by the pcLED + red reference, due to its sizable spectral peak in the long wavelengths. Therefore, the absorption decreases less when illuminated by the optimized test SPD. This pattern is more clear when objects are grouped into their associated hues, as presented in Table 4. When test SPDs were compared to the pcLED reference, optimized energy savings for different object colors were quite similar. When the reference was the pcLED + red, optimization of SPDs for red-orange object reflectance yielded the smallest energy savings, due to the high reflectance of longer wavelength light by these objects.

Tables Icon

Table 4. Average percentage of optimized energy savings for illumination of each hue group of objects, compared to the reference light sources, when color difference is imperceptible (ΔE*ab<1.0) and detectable (ΔE*ab<5).

Experimental results

Both experiments used a two-alternative forced choice task and the percentage of the trials in which observers selected the test SPD as rendering the object more naturally or attractively (k) was calculated. For these experiments, a k of 50% corresponds to chance and suggests that participants judge neither the test nor reference light source to render the object color more naturally/attractively than the other. A k of 25% or 75% corresponds to threshold and suggests that participants judge the test (75%) or reference (25%) light source to render the object color more naturally/attractively. This percentage is shown, as a function of dC*ab of the test SPD for each object when compared to the pcLED reference in Fig. 7. Analogous results, when judgments were made against the pcLED + red reference source, are shown in Fig. 8.

 

Fig. 7 Percentage of trials in which participants judged object color to appear more natural (long dashed line; left y-axes) and attractive (dotted line; left y-axes) as a function of the difference in chroma (dC*ab), when the object was illuminated by the test SPDs, than when illuminated by the pcLED reference light source. Error bars show the standard error of the mean (SEM). Generally, increases in object chroma were associated with increased energy savings (continuous black line; right y-axes).

Download Full Size | PPT Slide | PDF

 

Fig. 8 Percentage of trials in which participants judged object color to appear more natural (long dashed line; left y-axes) and attractive (dotted line; left y-axes) as a function of the difference in chroma (dC*ab), when the object was illuminated by the test SPDs, than when illuminated by the pcLED + red reference light source. Error bars show the standard error of the mean (SEM). Generally, increases in object chroma were associated with increased energy savings (continuous black line; right y-axes).

Download Full Size | PPT Slide | PDF

The results for the green Granny Smith apple suggest that, for both reference light sources, the range of test SPD dC*ab values tested in these experiments did not impact perceived naturalness or attractiveness. Aside from a reduction in judgments of naturalness for one of the apple’s test SPDs (dC*ab = 5.3) compared to the pcLED + red reference, the k value was quite close to 50% for all test SPDs.

When the chroma of blueberries was very large, participants judged them to appear unnatural and unattractive, particularly compared to the pcLED reference, as shown in Fig. 7. This was likely because the pcLED has significant power in the shorter wavelengths, rendering blue colors vividly. When test SPDs further increased the blueberries’ chroma, the appearance was unnatural and unattractive. A similar, but less pronounced, pattern of results is seen for judgments of naturalness when compared to the pcLED + red. For this reference light source, no marked differences in attractiveness were reported across the range of dC*ab values tested.

The lemon appeared equally attractive across the chroma values under the optimized spectra and the reference pcLED + red. With the possible exception of the highest chroma condition, observers did not report differences between naturalness and attractiveness when comparing with this reference light source. When compared to the pcLED, participants judged the lemon to appear more attractive than natural for test SPDs of all chroma values.

The color appearance of mandarin appeared equally natural and attractive under all optimized test spectra when compared with the pcLED + red reference, as shown in Fig. 8. When compared to the pcLED reference, test SPDs yielding lower chroma differences (dC*ab<4) resulted in judgments of reduced naturalness and attractiveness. Objects with higher chroma (dC*ab>4) were judged to be slightly more attractive, but slightly less natural, than when illuminated by the pcLED reference.

When compared to the pcLED, the naturalness and attractiveness of the tomato was indistinguishable when illuminated by the rest and reference light sources at lower chroma differences (dC*ab<6). At higher chroma differences, participants judged the pcLED to render the tomato slightly more naturally and attractively than the test SPDs. Observers consistently judged all test light sources to render the tomato less naturally and less attractively than the pcLED + red reference. Based on comments from the participants and observations of the experimenters, the translucent properties of the tomato appeared to interact with the test SPDs in ways that were not predicted by colorimetric calculations. Participants verbally reported perceiving hue shifts in the appearance of the tomato, despite the fact that test SPDs were chosen to render imperceptible hue and lightness changes (ΔE*ab - dC*ab<1.0). Problems with the perceptual uniformity (i.e., distances within color spaces corresponding to magnitudes of perceptual color differences) of the CIE 1976 L*a*b* color space for transparent objects might explain these observations [11]. This, combined with the strongly directional nature of the light sources used in this experiment, appeared to result in a visual artifact in which the top half of the tomato appeared significantly different from its bottom half, making the task very difficult for observers.

Naturalness vs. attractiveness

Participants were more likely to select the object illuminated by the test SPD as appearing more attractive than natural, especially when dC*ab values were higher. This result parallels other studies [8–10], reinforcing suggestions that observers prefer saturated colors. Many researchers believe that naturalness and attractiveness are two distinct characteristics of the color rendering ability of a light source [12]. Judgments of these two aspects also depend on the reference light source used for comparison. There was a notable difference in k values for judgments of naturalness and attractiveness when the reference light source was the pcLED. When the reference light source was the pcLED + red, k values were very similar, especially for the lower chroma test SPDs (dC*ab<5), except for the tomato.

Reference light source

Computational analysis showed that the two reference light sources, pcLED and pcLED + red, had a slightly different impact on the ability to reduce energy consumption with optimized spectra, especially for the red-orange hue objects. Resulting energy savings for these objects were lower, due to the additional red peak in the pcLED + red reference light source’s spectrum.

Experimental data showed that observers’ judgments of the naturalness and the attractiveness of object color appearances were very similar and consistent among dC*ab values when the reference light source was the pcLED + red. When the reference was the pcLED, there was a more pronounced difference in the observers’ evaluations of the naturalness and attractiveness of the stimuli, except for the tomato. Observers judged the color appearance of the tomato to be equally natural and attractive when compared to this reference light source.

Discussion

In this study, computational analyses and psychophysical experiments were conducted to investigate the potential use of real narrowband light sources to minimize the light energy absorbed by common objects to create more efficient lighting systems. Following previous research, which showed that energy could be reduced by up to 71% with theoretical SPDs [4], the computational simulations performed here demonstrate that currently commercially available LEDs can also be optimized for object reflectance, to save between 2% and 15% energy, depending on the object reflectance characteristics and reference light source being considered. Simulations showed that higher savings, between 9% and 19%, were achievable if slightly greater color differences, ΔE*ab<5, were deemed acceptable. The amount of energy that could be saved with this approach is expected to increase as lighting technology evolves and spectral design flexibility is increased.

Subsequently, test SPDs optimized in the computational analysis were physically realized and used in a laboratory setting to investigate the perceived color appearance of real objects with these energy-saving light sources. The experimental data further supports the proposition that, by optimizing the light spectrum for object reflectance, the energy consumed by lighting can be reduced without negatively impacting the visual appearance of illuminated objects. Although the object color appearance under most of the test SPDs resulted in naturalness and attractiveness comparable to illumination by the reference light sources, especially when the chroma increase was low, the translucency of the tomato negatively impacted the results. Data from the experiments showed that higher energy savings can be achieved, even when the color difference was as high as ΔE*ab = 15, as long as the hue change was minimized (ΔE*ab - dC*ab<1.0) and an increase in chroma was maintained (dC*ab>0), with a natural and attractive object color appearance.

References and links

1. W. R. McCluney, Introduction to Radiometry and Photometry (Artech House Publ., 1994).

2. D. Durmus and W. Davis, “Optimising Light Source Spectrum For Object Reflectance,” Opt. Express 23(11), A456–A464 (2015). [CrossRef]   [PubMed]  

3. J. Zhang, R. Hu, B. Xie, X. Yu, X. Luo, Z. Yu, L. Zhang, H. Wang, and X. Jin, “Energy-Saving Light Source Spectrum Optimization by Considering Object’s Reflectance,” IEEE Photonics J. 9(2), 1–11 (2017).

4. D. Durmus and W. Davis, “Absorption-Minimizing Spectral Power Distributions,” Light, Energy and the Environment (Optical Society of America, OSA Technical Digest (online), 2015), paper JTu5A.2. [CrossRef]  

5. W. Davis and Y. Ohno, “Color quality scale,” Opt. Eng. 49(3), 033602 (2010). [CrossRef]  

6. CIE, Colorimetry (CIE 15, 2004).

7. R. W. G. Hunt and R. M. Pointer, Measuring Colour (John Wiley & Sons, 2011).

8. G. Smets, “A tool for measuring relative effects of hue, brightness and saturation on color pleasantness,” Percept. Mot. Skills 55 (3f), 1159–1164 (1982). [CrossRef]   [PubMed]  

9. N. Camgöz, C. Yener, and D. Guvenc, “Effects of hue, saturation, and brightness on preference,” Color Res. Appl. 27(3), 199–207 (2002). [CrossRef]  

10. Y. Ohno, M. Fein, and C. Miller, “Vision experiment on chroma saturation for colour quality preference,” Light Eng. 23(4), 6–14 (2015).

11. K. McLaren, “CIELAB hue-angle anomalies at low tristimulus ratios,” Color Res. 5(3), 139–143 (1980). [CrossRef]  

12. S. Jost-Boissard, P. Avouac, and M. Fontoynont, “Assessing the colour quality of LED sources: Naturalness, attractiveness, colourfulness and colour difference,” Light. Res. Technol. 47(7), 769–794 (2015). [CrossRef]  

References

  • View by:
  • |
  • |
  • |

  1. W. R. McCluney, Introduction to Radiometry and Photometry (Artech House Publ., 1994).
  2. D. Durmus and W. Davis, “Optimising Light Source Spectrum For Object Reflectance,” Opt. Express 23(11), A456–A464 (2015).
    [Crossref] [PubMed]
  3. J. Zhang, R. Hu, B. Xie, X. Yu, X. Luo, Z. Yu, L. Zhang, H. Wang, and X. Jin, “Energy-Saving Light Source Spectrum Optimization by Considering Object’s Reflectance,” IEEE Photonics J. 9(2), 1–11 (2017).
  4. D. Durmus and W. Davis, “Absorption-Minimizing Spectral Power Distributions,” Light, Energy and the Environment (Optical Society of America, OSA Technical Digest (online), 2015), paper JTu5A.2.
    [Crossref]
  5. W. Davis and Y. Ohno, “Color quality scale,” Opt. Eng. 49(3), 033602 (2010).
    [Crossref]
  6. CIE, Colorimetry (CIE 15, 2004).
  7. R. W. G. Hunt and R. M. Pointer, Measuring Colour (John Wiley & Sons, 2011).
  8. G. Smets, “A tool for measuring relative effects of hue, brightness and saturation on color pleasantness,” Percept. Mot. Skills 55 (3f), 1159–1164 (1982).
    [Crossref] [PubMed]
  9. N. Camgöz, C. Yener, and D. Guvenc, “Effects of hue, saturation, and brightness on preference,” Color Res. Appl. 27(3), 199–207 (2002).
    [Crossref]
  10. Y. Ohno, M. Fein, and C. Miller, “Vision experiment on chroma saturation for colour quality preference,” Light Eng. 23(4), 6–14 (2015).
  11. K. McLaren, “CIELAB hue-angle anomalies at low tristimulus ratios,” Color Res. 5(3), 139–143 (1980).
    [Crossref]
  12. S. Jost-Boissard, P. Avouac, and M. Fontoynont, “Assessing the colour quality of LED sources: Naturalness, attractiveness, colourfulness and colour difference,” Light. Res. Technol. 47(7), 769–794 (2015).
    [Crossref]

2017 (1)

J. Zhang, R. Hu, B. Xie, X. Yu, X. Luo, Z. Yu, L. Zhang, H. Wang, and X. Jin, “Energy-Saving Light Source Spectrum Optimization by Considering Object’s Reflectance,” IEEE Photonics J. 9(2), 1–11 (2017).

2015 (3)

S. Jost-Boissard, P. Avouac, and M. Fontoynont, “Assessing the colour quality of LED sources: Naturalness, attractiveness, colourfulness and colour difference,” Light. Res. Technol. 47(7), 769–794 (2015).
[Crossref]

Y. Ohno, M. Fein, and C. Miller, “Vision experiment on chroma saturation for colour quality preference,” Light Eng. 23(4), 6–14 (2015).

D. Durmus and W. Davis, “Optimising Light Source Spectrum For Object Reflectance,” Opt. Express 23(11), A456–A464 (2015).
[Crossref] [PubMed]

2010 (1)

W. Davis and Y. Ohno, “Color quality scale,” Opt. Eng. 49(3), 033602 (2010).
[Crossref]

2002 (1)

N. Camgöz, C. Yener, and D. Guvenc, “Effects of hue, saturation, and brightness on preference,” Color Res. Appl. 27(3), 199–207 (2002).
[Crossref]

1982 (1)

G. Smets, “A tool for measuring relative effects of hue, brightness and saturation on color pleasantness,” Percept. Mot. Skills 55 (3f), 1159–1164 (1982).
[Crossref] [PubMed]

1980 (1)

K. McLaren, “CIELAB hue-angle anomalies at low tristimulus ratios,” Color Res. 5(3), 139–143 (1980).
[Crossref]

Avouac, P.

S. Jost-Boissard, P. Avouac, and M. Fontoynont, “Assessing the colour quality of LED sources: Naturalness, attractiveness, colourfulness and colour difference,” Light. Res. Technol. 47(7), 769–794 (2015).
[Crossref]

Camgöz, N.

N. Camgöz, C. Yener, and D. Guvenc, “Effects of hue, saturation, and brightness on preference,” Color Res. Appl. 27(3), 199–207 (2002).
[Crossref]

Davis, W.

Durmus, D.

Fein, M.

Y. Ohno, M. Fein, and C. Miller, “Vision experiment on chroma saturation for colour quality preference,” Light Eng. 23(4), 6–14 (2015).

Fontoynont, M.

S. Jost-Boissard, P. Avouac, and M. Fontoynont, “Assessing the colour quality of LED sources: Naturalness, attractiveness, colourfulness and colour difference,” Light. Res. Technol. 47(7), 769–794 (2015).
[Crossref]

Guvenc, D.

N. Camgöz, C. Yener, and D. Guvenc, “Effects of hue, saturation, and brightness on preference,” Color Res. Appl. 27(3), 199–207 (2002).
[Crossref]

Hu, R.

J. Zhang, R. Hu, B. Xie, X. Yu, X. Luo, Z. Yu, L. Zhang, H. Wang, and X. Jin, “Energy-Saving Light Source Spectrum Optimization by Considering Object’s Reflectance,” IEEE Photonics J. 9(2), 1–11 (2017).

Jin, X.

J. Zhang, R. Hu, B. Xie, X. Yu, X. Luo, Z. Yu, L. Zhang, H. Wang, and X. Jin, “Energy-Saving Light Source Spectrum Optimization by Considering Object’s Reflectance,” IEEE Photonics J. 9(2), 1–11 (2017).

Jost-Boissard, S.

S. Jost-Boissard, P. Avouac, and M. Fontoynont, “Assessing the colour quality of LED sources: Naturalness, attractiveness, colourfulness and colour difference,” Light. Res. Technol. 47(7), 769–794 (2015).
[Crossref]

Luo, X.

J. Zhang, R. Hu, B. Xie, X. Yu, X. Luo, Z. Yu, L. Zhang, H. Wang, and X. Jin, “Energy-Saving Light Source Spectrum Optimization by Considering Object’s Reflectance,” IEEE Photonics J. 9(2), 1–11 (2017).

McLaren, K.

K. McLaren, “CIELAB hue-angle anomalies at low tristimulus ratios,” Color Res. 5(3), 139–143 (1980).
[Crossref]

Miller, C.

Y. Ohno, M. Fein, and C. Miller, “Vision experiment on chroma saturation for colour quality preference,” Light Eng. 23(4), 6–14 (2015).

Ohno, Y.

Y. Ohno, M. Fein, and C. Miller, “Vision experiment on chroma saturation for colour quality preference,” Light Eng. 23(4), 6–14 (2015).

W. Davis and Y. Ohno, “Color quality scale,” Opt. Eng. 49(3), 033602 (2010).
[Crossref]

Smets, G.

G. Smets, “A tool for measuring relative effects of hue, brightness and saturation on color pleasantness,” Percept. Mot. Skills 55 (3f), 1159–1164 (1982).
[Crossref] [PubMed]

Wang, H.

J. Zhang, R. Hu, B. Xie, X. Yu, X. Luo, Z. Yu, L. Zhang, H. Wang, and X. Jin, “Energy-Saving Light Source Spectrum Optimization by Considering Object’s Reflectance,” IEEE Photonics J. 9(2), 1–11 (2017).

Xie, B.

J. Zhang, R. Hu, B. Xie, X. Yu, X. Luo, Z. Yu, L. Zhang, H. Wang, and X. Jin, “Energy-Saving Light Source Spectrum Optimization by Considering Object’s Reflectance,” IEEE Photonics J. 9(2), 1–11 (2017).

Yener, C.

N. Camgöz, C. Yener, and D. Guvenc, “Effects of hue, saturation, and brightness on preference,” Color Res. Appl. 27(3), 199–207 (2002).
[Crossref]

Yu, X.

J. Zhang, R. Hu, B. Xie, X. Yu, X. Luo, Z. Yu, L. Zhang, H. Wang, and X. Jin, “Energy-Saving Light Source Spectrum Optimization by Considering Object’s Reflectance,” IEEE Photonics J. 9(2), 1–11 (2017).

Yu, Z.

J. Zhang, R. Hu, B. Xie, X. Yu, X. Luo, Z. Yu, L. Zhang, H. Wang, and X. Jin, “Energy-Saving Light Source Spectrum Optimization by Considering Object’s Reflectance,” IEEE Photonics J. 9(2), 1–11 (2017).

Zhang, J.

J. Zhang, R. Hu, B. Xie, X. Yu, X. Luo, Z. Yu, L. Zhang, H. Wang, and X. Jin, “Energy-Saving Light Source Spectrum Optimization by Considering Object’s Reflectance,” IEEE Photonics J. 9(2), 1–11 (2017).

Zhang, L.

J. Zhang, R. Hu, B. Xie, X. Yu, X. Luo, Z. Yu, L. Zhang, H. Wang, and X. Jin, “Energy-Saving Light Source Spectrum Optimization by Considering Object’s Reflectance,” IEEE Photonics J. 9(2), 1–11 (2017).

Color Res. (1)

K. McLaren, “CIELAB hue-angle anomalies at low tristimulus ratios,” Color Res. 5(3), 139–143 (1980).
[Crossref]

Color Res. Appl. (1)

N. Camgöz, C. Yener, and D. Guvenc, “Effects of hue, saturation, and brightness on preference,” Color Res. Appl. 27(3), 199–207 (2002).
[Crossref]

IEEE Photonics J. (1)

J. Zhang, R. Hu, B. Xie, X. Yu, X. Luo, Z. Yu, L. Zhang, H. Wang, and X. Jin, “Energy-Saving Light Source Spectrum Optimization by Considering Object’s Reflectance,” IEEE Photonics J. 9(2), 1–11 (2017).

Light Eng. (1)

Y. Ohno, M. Fein, and C. Miller, “Vision experiment on chroma saturation for colour quality preference,” Light Eng. 23(4), 6–14 (2015).

Light. Res. Technol. (1)

S. Jost-Boissard, P. Avouac, and M. Fontoynont, “Assessing the colour quality of LED sources: Naturalness, attractiveness, colourfulness and colour difference,” Light. Res. Technol. 47(7), 769–794 (2015).
[Crossref]

Opt. Eng. (1)

W. Davis and Y. Ohno, “Color quality scale,” Opt. Eng. 49(3), 033602 (2010).
[Crossref]

Opt. Express (1)

Percept. Mot. Skills (1)

G. Smets, “A tool for measuring relative effects of hue, brightness and saturation on color pleasantness,” Percept. Mot. Skills 55 (3f), 1159–1164 (1982).
[Crossref] [PubMed]

Other (4)

W. R. McCluney, Introduction to Radiometry and Photometry (Artech House Publ., 1994).

D. Durmus and W. Davis, “Absorption-Minimizing Spectral Power Distributions,” Light, Energy and the Environment (Optical Society of America, OSA Technical Digest (online), 2015), paper JTu5A.2.
[Crossref]

CIE, Colorimetry (CIE 15, 2004).

R. W. G. Hunt and R. M. Pointer, Measuring Colour (John Wiley & Sons, 2011).

Cited By

OSA participates in Crossref's Cited-By Linking service. Citing articles from OSA journals and other participating publishers are listed here.

Alert me when this article is cited.


Figures (8)

Fig. 1
Fig. 1 Spectral reflectance as a function of wavelength for the 10 reflective objects.
Fig. 2
Fig. 2 Power as a function of wavelength for the two reference illuminants.
Fig. 3
Fig. 3 The spectral power distributions of the nine narrowband LEDs that were mixed to simulate test SPDs.
Fig. 4
Fig. 4 The test and reference SPDs were generated by the Source Four LED Profile x7 Color System theatrical lights, which were positioned on top of the two adjacent black booths.
Fig. 5
Fig. 5 When the green Granny Smith apple (dashed gray line; right y-axis) is illuminated by the test SPD (continuous black line; left y-axis), which consists of five different LEDs, 8% of energy is saved, compared to illumination by the reference pcLED (continuous gray line; left y-axis), without inducing any noticeable color shifts (ΔE*ab<1.0).
Fig. 6
Fig. 6 When the green Granny Smith apple (dashed gray line; right y-axis) is illuminated by the test SPD (continuous black line; left y-axis), which consists of three different LEDs, 17% less energy is required than when it is illuminated by the reference pcLED (continuous gray line; left y-axis) if a slight color shift (ΔE*ab = 3.6) is allowed.
Fig. 7
Fig. 7 Percentage of trials in which participants judged object color to appear more natural (long dashed line; left y-axes) and attractive (dotted line; left y-axes) as a function of the difference in chroma (dC*ab), when the object was illuminated by the test SPDs, than when illuminated by the pcLED reference light source. Error bars show the standard error of the mean (SEM). Generally, increases in object chroma were associated with increased energy savings (continuous black line; right y-axes).
Fig. 8
Fig. 8 Percentage of trials in which participants judged object color to appear more natural (long dashed line; left y-axes) and attractive (dotted line; left y-axes) as a function of the difference in chroma (dC*ab), when the object was illuminated by the test SPDs, than when illuminated by the pcLED + red reference light source. Error bars show the standard error of the mean (SEM). Generally, increases in object chroma were associated with increased energy savings (continuous black line; right y-axes).

Tables (4)

Tables Icon

Table 1 The narrowband LED channels that were mixed to generate the test SPDs, and the resulting color differences ΔE*ab, energy consumption relative to the reference source, and chroma difference dC*ab values, when the reference light source was the pcLED.

Tables Icon

Table 2 The narrowband LED channels that were mixed to generate the test SPDs, and the resulting color differences ΔE*ab, energy consumption relative to the reference source, and chroma difference dC*ab values, when the reference light source was the pcLED + red.

Tables Icon

Table 3 Percentage of energy savings for illumination of individual objects, compared to reference light sources, when color difference is imperceptible (ΔE*ab<1.0) and detectable, but not large (ΔE*ab<5).

Tables Icon

Table 4 Average percentage of optimized energy savings for illumination of each hue group of objects, compared to the reference light sources, when color difference is imperceptible (ΔE*ab<1.0) and detectable (ΔE*ab<5).

Metrics