Artwork visualization using a solid-state lighting engine with controlled photochemical safety

Arūnas Tuzikas; Artūras Žukauskas; Rimantas Vaicekauksas; Andrius Petrulis; Pranciškus Vitta; Michael Shur

doi:10.1364/OE.22.016802

1. Introduction

The illumination of artworks is a complex problem, which necessitates both meeting conservation requirements and pursuing adequate visual impression [1,2]. Each photon absorbed by a pigment, binder, varnish, or substrate of an artwork might invoke an irreversible photochemical reaction contributing to the alteration of the color (typically, the probability of the photochemical reaction increases with the photon energy). The radiant heating effect, which is less wavelength dependent, might also contribute to the artwork damage. Therefore, artwork conservation requires as short as possible duration of exposure to as dim as possible light having as low as possible short-wavelength spectral content and no non-visible spectral components. On the other hand, the color discrimination of the human eye decreases with reducing luminance [3] and white light with a low short-wavelength spectral content, i.e. generally having low correlated color temperature (CCT) and a shrunk chromatic gamut area [4], renders colors differently from daylight, which the most of artworks have been created at. Moreover, the visual reconstruction of an aged artwork might require the use of short-wavelength enriched light [5], which is photochemically harmful. The approaches to the conservation and visualization of artworks must resolve this conflict by finding a reasonable compromise.

Early work on conserving illumination of artworks [6–9] led to the establishment of the recommended illuminance of 50 lx and 200 lx, for materials with high and medium responsivity (e.g. textiles, water-colors, silk, low-grade paper) and for materials with low photo-responsivity (e.g. oil and tempera paintings, frescos, bone, and wood), respectively. Such recommendations followed from that the color discrimination ability drops dramatically below 50 lx and increases slowly above 200 lx. In addition, appropriate optical filters were recommended for the light sources that generate violet-UV radiant flux (with the wavelengths below 400 nm) per unit luminous flux in excess of 75 μW/lm, which is a characteristic of a common incandescent lamp [10]. Later on, the requirement for the violet-UV radiant flux was tightened to the limit of 10 μW/lm [11]. Also, a common practice is to minimize infrared radiation exposure by using, for example, dichroic mirrors for tungsten radiators [11,12].

A straightforward approach to increasing photochemical safety is to use light with the highest luminous efficacy of radiation (LER), which is the ratio of luminous flux to radiant flux [12]. However the spectral responsivity in respect of light-induced color shifts for various artwork materials must be taken into account. The spectral responsivity can be approximated by a “damage” function, which almost exponentially decreases with increasing wavelength [13]. Krochmann et al. [14] investigated the damage functions and effective radiant exposures causing just noticeable color shifts for several groups of artwork materials (low-grade paper, rag paper, oil paints on canvas, textiles, and water colors on rag paper). With some limitations due to more peculiar spectral behavior of actual responsivities of individual pigments [15], the exponential damage functions can be used for estimating the photochemical damage index of light, which is the ratio of the damage flux (i.e. the radiant power weighted by a damage function) to the luminous flux [11,16,17]. Minimizing the damage index is more meaningful in terms of artwork conservation than maximizing LER, which does not distinguish between light with different wavelengths posing different photochemical hazard.

Research on finding the optimal light sources for the visualization of artworks was focused on the selection of the most appropriate CCT and color rendition and color discrimination properties of light. The optimal CCT was searched by finding subjective preferences to visual impression using tungsten radiators [18,19], the monitor simulations based on hyperspectral images [20,21], and the Xe lamp based digital light processor [22]. Typically, the most preferred CCTs (3600 to 5700 K) were below that of daylight (6500 K) at illuminances within the recommended “safe” range of 50-200 lx. This partially validated the Kruithof hypothesis [23,24], which claims that lower CCTs are required to make illumination appear “pleasing” at reduced illuminances. However, the above experiments on finding the optimal CCT did not fully account for the conservation issues, since the variation of CCT results in the variation of the photochemical damage potential of light [11,17].

Selecting the most appropriate color rendition and color discrimination properties of light for artwork illumination is a complex issue. A common requirement is to use light sources with high color rendering ability, which is usually understood as a degree of fidelity of the color rendition with respect to a blackbody or daylight-phase reference illuminant. Recommended sources must have the general color-rendering index R_a of the Commission Internationale de l’Éclairge (CIE) [25] of at least 85 [10,11]. However, color fidelity can be partially sacrificed for the sake of photochemical safety [26]. For instance, optical filters can be used to reduce the photochemical damage potential while maintaining appropriate color fidelity of light sources [27]. On the other hand, increased color saturation (gamut area) at an expense of color fidelity might be desirable in order to compensate the reduced color discrimination of the human eye at lower (and thus photochemically safer) illuminances [28,29]. Meanwhile, reduced color saturation might be useful to simulate the appearance of artworks originally located in low-light conditions, such as in churches or caves [29]. Narrow-band (450 nm-540 nm-610 nm) “prime-color” fluorescent lamps having both sufficiently high color fidelity and increased gamut area have been proposed years ago [4,28]. Cuttle has demonstrated an application of a narrow band red-green-blue (RGB) illuminant for artwork illumination with reduced color fidelity and increased color discrimination [12]. A computational comparison of a large number of daylight metamers against color discrimination ability for hyperspectral images of artworks has shown that more colors can be resolved with illuminants having structured spectra [30]. However, despite the above findings, no versatile approach to the control of photochemical damage while varying the color rendition and color discrimination properties of light has been developed so far. In addition, the progress in this direction is hindered by the lack of a consensus on color rendition metrics [27,31].

The solid-state lighting technology based on direct-emission and phosphor-converted light-emitting diodes (LEDs) [32] considerably increased the flexibility of the selection of the optimal light spectrum for artwork illumination. In comparison with common incandescent, fluorescent, and discharge light sources, LEDs have important advantages, such as high efficacy, very low spectral power in the unwanted regions of the spectrum, the ease of assembling into polychromatic clusters with independent control of the output for each primary-color component, and instantaneous switching and low-voltage power supplies that facilitate intelligent control using computers and smart devices. These advantages and the maturity of solid-state lighting technology allow for the development of LED based painting-specific lighting, which can meet both conservation and visualization needs of an individual artwork [27,29].

Of particular interest for advanced illumination of artworks are polychromatic clusters of colored LEDs, which are known to provide the tunability of CCT [33–36]. Besides, their peak wavelengths and relative radiant fluxes can be optimized for attaining the desired color rendition properties of composite light. The computational optimization of the composite spectral power distributions (SPDs) can be performed by searching within a multi-parameter space [37,38] and using objective functions with trade-offs between certain figures of merit [37,39,40] and/or constraints on some properties [41,42]. Using such approaches, the SPDs of high-fidelity [29,35,36,43], color-saturating [29,38,39], and color-desaturating (dulling) [29,42] LED clusters have been optimized. Typically, color saturating LED illuminants are trichromatic red-green-blue (RGB) clusters and color-dulling illuminants are either dichromatic yellow-blue (YB) or trichromatic red-yellow-blue (RYB) or amber-green-blue (AGB) clusters. High-fidelity LED illuminants (with the general color rendering index in excess of 90) typically feature a tetrachromatic red-yellow-green-blue (RYGB) or red-amber-green-blue (RAGB) design [43,44]. Important examples of the applications of polychromatic LED clusters in artwork illumination are matching the non-white LED blends to halogen and fluorescent lighting [45], enhancing the faded colors of museum artefacts [46], and finding cultural preferences to color quality of light (CCT and color saturating ability) for artwork illumination [47]. Vazquez-Molini et al. attempted to find a trade-off between the color contrast and damage index for a cave painting [40]. However usually, adjusting the SPD of polychromatic LED clusters for meeting the required color quality of artwork visualization has been performed without accounting for the variation of the photochemical damage index of light.

In this work, we present a practical approach of using tunable LED clusters for artwork illumination with the control of photochemical safety. Our approach is based on an intelligently controlled color rendition engine [48] appended with a built-in dimming functionality for instantaneously maintaining a constant damage irradiance with the variation of CCT and color rendition properties. We also demonstrate the operation of the engine in artwork-specific lighting in the three modes: (1) selecting CCT, (2) tuning chromatic saturation, and (3) shifting chromaticity outside the white locus.

2. Lighting engine

2.1 Principle of operation

Our concept of a solid-state lighting engine for artwork illumination is based on a tetrachromatic RAGB LED cluster. The radiant fluxes of each of the four primary-color components are set to produce the composite light with desired painting-specific spectral properties, such as chromaticity and color saturating ability. Meanwhile, the generated radiant flux is instantaneously adjusted in such a way that the net photochemical damage irradiance provided by the actual composite light equals to that of the reference light for a particular type of the photosensitive material.

The relative partial radiant fluxes of the four components, $c_{i}^{RAGB}$ (i = R, A, G, B), are derived from the chromaticity coordinates (CIE 1931 x,y) and color saturating ability of the desired light in two steps. First, the relative partial radiant fluxes $c_{i}^{RGB}$ (i = R, G, B) and $c_{i}^{AGB}$ (i = A, G, B) for two metameric trichromatic sets of primary LEDs, RGB and AGB, respectively, are calculated (by default $c_{A}^{RGB} = c_{R}^{AGB} = 0$ ). These calculations are based on the solution of two respective systems of color mixing equations, which follow from the principle of additive color mixing [41,49],

{\begin{matrix} \sum_{i = R, G, B} c_{i}^{RGB} X_{i} = x \sum_{i = R, G, B} c_{i}^{RGB} (X_{i} + Y_{i} + Z_{i}), \\ \sum_{i = R, G, B} c_{i}^{RGB} Y_{i} = y \sum_{i = R, G, B} c_{i}^{RGB} (X_{i} + Y_{i} + Z_{i}), \\ \sum_{i = R, G, B}^{n} c_{i}^{RGB} = 1, \end{matrix}

and

{\begin{matrix} \sum_{i = A, G, B} c_{i}^{AGB} X_{i} = x \sum_{i = A, G, B} c_{i}^{AGB} (X_{i} + Y_{i} + Z_{i}), \\ \sum_{i = A, G, B} c_{i}^{AGB} Y_{i} = y \sum_{i = A, G, B} c_{i}^{AGB} (X_{i} + Y_{i} + Z_{i}), \\ \sum_{i = A, G, B}^{n} c_{i}^{AGB} = 1, \end{matrix}

where X_i, Y_i, and Z_i are the tristimulus values of each primary-color component of unit radiant power.

Second, the four relative partial radiant fluxes of the tetrachromatic RAGB LED cluster are calculated using the six computed values of the relative partial radiant fluxes ( $c_{i}^{RGB}$ , i = R, G, B and $c_{i}^{AGB}$ , i = A, G, B), two default values ( $c_{A}^{RGB} = c_{R}^{AGB} = 0$ ) and the saturation (gamut) factor, Γ,

c_{i}^{RAGB} = Γ c_{i}^{RGB} + (1 - Γ) c_{i}^{AGB}, i = R, A, G, B .

The saturation factor 0 ≤ Γ ≤□1 is the weight of the trichromatic RGB blend in the resulting tertrachromatic RAGB blend. (The corresponding weight of the AGB blend is 1 - Γ.) The variation of the saturation factor allows for traversing all possible RAGB metameric blends, starting with the AGB blend at Γ = 0 and ending with the RGB blend at Γ = 1 [48]. At the RGB endpoint, the generated light has the highest ability to saturate the colors of illuminated objects, whereas at the AGB endpoint, the highest ability of color dulling is attained. The RAGB blends with the highest color fidelity, as well as with subjective preferences to the color quality of visualized objects and with the highest color discrimination are in between the endpoints. The peak values of the different color fidelity indices, such as R_a [25], statistical Color Fidelity Index (CFI) [50], and Color Fidelity Scale Q_f of the Color Quality Scale (CQS) [51] are obtained at very similar values of the saturation factor [47,48]. However, different color fidelity indices with values considerably smaller than the peak values have different meaning and cannot rate color rendition when used separately. For instance, two light sources with the same low value of a color fidelity index (R_a, CFI, or Q_f) might have opposite abilities to saturate and desaturate colors. (This is the main reason of the ambiguity of color rendition metrics based on a single color fidelity index.) Therefore, our approach relies on interactive finding subjective preferences to color quality of light through continuously tuning the saturation factor [47,48] rather than on setting magnitudes of certain color rendition indices. This approach is very useful for resolving the ambiguities of the color rendition metrics.

The four values of the relative partial radiant fluxes of the RAGB cluster are used for the calculation of the actual partial radiant fluxes of each of the four primary-color components. For the actual partial radiant fluxes measured in watts (W),

Φ_{i} = F c_{i}^{RAGB} \times 1 W, i = R, A, G, B,

where F is the actual dimming coefficient, which ensures that the photochemical damage irradiance provided by the actual light is kept equal to that of the reference light. The calculation of the actual dimming coefficient F is based on the reference relative partial radiant fluxes (

c_{i, ref}^{RAGB}

, i = R, A, G, B), the type of the photosensitive material, and the value of the reference dimming coefficient, F_ref. The latter defines the partial radiant fluxes of the reference light,

Φ_{i, ref} = F_{ref} c_{i, ref}^{RAGB} \times 1 W, i = R, A, G, B .

The value of the actual dimming coefficient is obtained by calculating the relative damage fluxes (per unit radiant flux) of the actual and reference lights, which are

D = \int_{300 nm}^{780 nm} D_{b} (λ) S (λ) d λ

and

D_{ref} = \int_{300 nm}^{780 nm} D_{b} (λ) S_{ref} (λ) d λ,

respectively. Here, D_b(λ) is the damage function that can be approximated as [14,17]

D_{b} (λ) = \exp [- b (λ - 300 nm)]

where the value of the coefficient b depends on the type of the photosensitive material being illuminated [11]. It equals 0.038, 0.0125, 0.0115, and 0.010 for low-grade paper, rag paper, oil paints on canvas and water colors on rag paper, and textiles, respectively.

The SPDs of the actual and reference lights are

S (λ) = \sum_{i = R, A, G, B} c_{i}^{R A G B} S_{i} (λ)

and

S_{ref} (λ) = \sum_{i = R, A, G, B} c_{i, ref}^{R A G B} S_{i} (λ)

respectively, where S_i(λ) are the SPDs of the primary-color LED components normalized to unit power.

Eventually, the actual dimming coefficient is calculated from the condition FD = F_refD_ref, which equates the damage irradiance of the actual and reference lights. With Eqs. (6)–(10) taken into account, the dimming coefficient equals

F = F_{ref} \sum_{i = R, A, G, B} c_{i, ref}^{RAGB} d_{b i} (b) / \sum_{i = R, A, G, B} c_{i}^{RAGB} d_{b i} (b)

where d_bi is the damage flux of the i-th component for a particular photosensitive material per unit radiant flux

d_{b i} = \int_{300 nm}^{780 nm} D_{b} (λ) S_{i} (λ) d λ, i = R, A, G, B .

The actual dimming coefficient found from Eq. (11) is used for setting the four values of the actual partial radiant fluxes defined by Eq. (4).

2.2 Practical implementation

Our practical light engine comprises two basic modules: an external smart device (e.g. tablet computer or mobile phone) with uploaded software and a tetrachromatic LED luminaire containing a microcontroller-based four-channel current source with firmware. The two modules are connected by a wireless link based on the Bluetooth standard. Figure 1 shows the block diagram of the engine.

Fig. 1 Block diagram of the light engine.

Download Full Size | PDF

Figure 2 displays the screenshot of the graphic user interface (GUI). The GUI allows the user to input painting-specific data (chromaticity and/or CCT, saturation factor, type of photosensitive material) and to fix the spectral composition and dimming coefficient of light selected to provide the reference damage irradiance. The software also contains a calculation script and the database of LED parameters (normalized SPDs, tristimulus values, and damage indices for different photosensitive materials). Upon inputting the chromaticity coordinates of the desired light (CIE 1931 x,y or those derived from CCT) and the saturation factor, the software computes the four relative partial radiant fluxes of the tetrachromatic RAGB cluster using Eqs. (1)–(3).

Fig. 2 Snapshot of the GUI of the smart device that is used for the control of light engine.

Download Full Size | PDF

Upon setting the spectral properties of the reference light (e.g. CCT = 3000 K, saturation factor Γ = 0.47) and the type of photosensitive material (e.g. oil paints on canvas), the relative damage factor of light (RDF) [16] is calculated as the ratio of the damage factors of the source and the CIE standard illuminant A,

RDF = \frac{D / LER}{D_{A} / {LER}_{A}},

where D and D_A are the relative damage fluxes for the RAGB blend and CIE standard illuminant A, respectively, defined by Eq. (6) and LER and LER_A are the respective luminous efficacies of radiation. The value of the RDF (0.80 in this case) is displayed on the GUI.

Then the reference dimming level (flux) is adjusted based on the indicated RDF (e.g. 50 lx measured by an auxiliary light meter at the surface of the illuminated artwork under the same engine at RDF = 0.80) and the reference relative partial radiant fluxes and the reference dimming coefficient are recorded using the GUI fixation checkbox. With further tuning the chromaticity and color saturating factor, the software instantaneously recalculates the value of the dimming coefficient and the resulting values of the partial radiant fluxes of the actual light, using Eq. (11) and Eq. (4), respectively. The latter values are sent to the luminaire via the wireless connection.

Based on the input data, the luminaire microcontroller generates the pulse-width modulation (PWM) signals for the four-channel current source driving the four groups of the colored LEDs. The output of each group is digitally regulated against the drift of the LEDs junction temperature using feed forward compensation [33] based on the calibration data of the LEDs output as a function of the heat sink temperature (measured by a thermistor) and driving current. The calibration data is stored in the memory of the firmware of the microcontroller.

The primary-color LEDs for the tetrachromatic RAGB cluster were selected from a range of standard commercial devices. The main criteria for the selection of particular types of LEDs were attaining as high as possible diversity in color rendition properties [44] and high luminous efficacy of the cluster at CCTs of 3000-4000 K. These CCTs fall within the most preferred range [20–22] and might result in a lower photochemical damage index. The selected LEDs were as follows: direct-emission red LED (Cree model XP-C XPCRED, 639 nm peak wavelength), green LED (Cree model XP-E XPEGRN, 523 nm), and blue LED (Cree XP-E2 XPEBRY, 452 nm) and phosphor converted amber LED (Philips Lumileds model LXM2-PL01, 597 nm). (These primary LEDs differ from those used in our previous versions of the engine [46,47].) The calibration data on the LED SPDs were obtained using the Labsphere model Illumia Pro system.

Table 1 displays the color rendition indices and photochemical damage parameters for the selected set of primary-color LEDs for three values of CCT (3000 K, 4500 K, and 6500 K). Each value of CCT is represented by three limiting blends, AGB (Γ = 0), RGB (Γ = 1), and high-fidelity tetrachromatic blend with the indicated value of Γ. These blends have the highest color-dulling, color-saturating, and color-fidelity abilities, respectively. Also shown are the characteristics typical of tungsten radiators, CIE standard illuminant A (2856 K blackbody) and that with the SPD truncated below 400 nm. The latter SPD mimics an incandescent lamp with a UV filter widely used in museum lighting.

Table 1. Color Rendition Indices and Photochemical Damage Parameters of the RAGB LED Cluster and Tungsten Illuminants

View Table | View all tables in this article

The color rendition indices in Table 1 are presented for three color rendition approaches: the CIE general color rendering index R_a and Gamut Area Index (GAI) used in the two-metrics approach [52], Color Saturation Index (CSI), Color Dulling Index (CDI), and Color Fidelity Index (CFI) used in the statistical approach [50], and Color Fidelity Scale (Q_f) and Color Gamut Scale (Q_g) used in the CQS approach [51]. The RGB blends have the highest ability to saturate the colors of illuminated objects, which is represented by the highest values of GAI, CSI and Q_g. In contrast, the AGB blends have the highest desaturating ability, which is represented by the lowest values of GAI, CSI and Q_g and the highest value of CDI. The high-fidelity RAGB blends, which correspond to the peak values of CFI, also have high values of R_a and Q_f.

The photochemical damage indices in Table 1 are presented by LER, the UV radiant flux per unit luminous flux, and RDF. Table 1 shows the RDF values for two photosensitive materials: low-grade paper (b = 0.038) and oil paint on canvas (b = 0.0115).

As seen, the LED cluster has LERs of up to twice as high as those of the tungsten illuminants; its UV radiant flux per unit luminous flux is more than twice lower than that of the unfiltered tungsten illuminant. However within a more consistent approach based on damage functions, the RDF of light generated by the LED cluster at 3000 K is comparable with that of the filtered tungsten illuminant. The RDF increases with CCT showing the necessity of reducing illuminance in order to maintain damage irradiance at a constant value. Also, the illuminance is to be adjusted while varying the color rendering properties, since the RDF varies while tuning from the color-dulling AGB endpoint to the color-saturating AGB endpoint (either increases or decreases depending on the material type).

3. Examples of the tunable visualization of artworks with controlled photochemical safety

Below we present examples of application of our light engine in three different operation modes for adjusting the visual appearance of artworks and museum exhibits under constant damage irradiance. The purpose of the demonstration is to prove the effect of varying visual impression due to the control of photochemical safety rather than to psychophysically assess or find subjective preferences to color quality of artwork illumination, which are artwork-specific and even depend on the cultural background of the observer [47].

The first example refers to the CCT control, which is a well-known approach to advanced artwork illumination [18–21,47]. In this example, the highest color fidelity is maintained while tuning CCT. In the second example, we demonstrate the control of color rendition by continuously tuning the saturation factor under metameric conditions. Such tuning is required for the compensation of reduced color discrimination under dim light (alternatively, for mimicking dim light conditions), enhancing faded colors or meeting other artwork-specific needs in chromatic representation [12,29,40,46,47]. The third example shows shifting the chromaticity out of the white light locus along an isotemperature line (assuming that the CCT and color saturation factor have been already established). Such shifting might be helpful for the visual reconstruction of aged artworks and/or for visualization using non-white light [5,45].

Each example is presented for an individual artwork or museum exhibit that is selected on the basis of its relevance to a particular mode of the engine operation. For revealing the effect of the photochemical damage control on artwork visualization, two series of photoimages are displayed in each case: the series at the constant illuminance of 50 lx and the other series at the constant damage irradiance that equals that under the reference light with the illuminance of 50 lx. The visual differences between the series of images are to be traced accounting for the limited color accuracy of the photoimages.

3.1 Color temperature selection

The selection of the CCT with controlled photochemical safety was demonstrated using a contemporary water color painting by E. Kuokštis. The artwork displays a medieval downtown scene in Vilnius, Lithuania and has been painted in 2004 under daylight. According to the artist, the colors of the artwork underwent minor changes since the date of creation. Therefore at first attempt, the optimization of the illumination conditions was restricted to the CCT selection at the conditions of the highest color fidelity, i.e. without the visual correction of the hue and chroma of the colors. While varying CCT, the chromaticity point was kept as close as possible (within a distance of Δxy ≤ 0.006) to the blackbody locus (for CCT below 5000 K) or to the CIE daylight locus (for CCT above 5000 K).

Figure 3(a)-(e) show the SPDs of the light engine for different CCTs in the range of 2500-6500 K with the saturation factor set for the highest color fidelity (for specificity, here we use R_a as a scale for color fidelity, although the statistical CFI and CQS Q_f attain the peak values at almost the same value of Γ). The SPDs are seen to continuously cover the visible spectrum and differ in the proportion between the components: with increasing CCT, the relative parts of the blue and green components increase, whereas those of the red and amber components decrease.

Fig. 3 SPDs of the light engine for CCTs of 2500 K (a), 3500 K (b), 4500 K (c), 5500 K (d), and 6500 K (e) with the saturation factor (indicated) tuned to the highest values of the general color rendering index R_a.

Download Full Size | PDF

Table 2 shows the RDFs and color rendition indices for the used high-fidelity tetrachromatic blends with different CCTs in the range of 2500-6500 K. The RDFs were estimated for different materials having spectral damage functions specified in Ref. 11. As seen, the values of the RDF increase with CCT by a factor of 2 for low-grade paper, which has the steepest damage function (the highest value of coefficient b) and by a factor of 1.67 for textiles, which have the flattest damage function (the lowest value of coefficient b). The color rendition indices are displayed for the three metrics discussed above (two-metrics, statistical, and CQS). The color fidelity indices (R_a, CFI, and Q_f) are seen to have very high values. The GAI, which shows the gamut area in respect to that of the CIE standard illuminant E, increases with CCT and attains the value of about 100 at daylight chromaticity. The CQS Q_g shows the gamut area similar to that for reference illuminants with the same CCT. The statistical indices CSI and CDI, which show the percentage of test color samples with distorted chroma, are low.

Table 2. Relative Damage Factor for Different Materials and Color Rendition Indices of Different Metrics as Functions of CCT for the Light Engine Operating in the High-fidelity Regime

View Table | View all tables in this article

Figure 4(a)-(e) show the images of the painting for different CCTs under the constant illuminance of 50 lx. The CCT and RDF of the applied light estimated for water colors (b = 0.0115) are shown above each image. Increasing CCT is seen to enhance the color palette of the painting. At the CCT typical of daylight (6500 K; Fig. 4(e)), the conditions are the closest to those under which the artwork was created. With increasing CCT from 2500 K to 6500 K, the RDF increases by a factor of 1.8 indicating a substantial variation of damage irradiance.

Fig. 4 (a)-(e) Images of a water-color painting “Downtown Vilnius” (E. Kuokštis, 2004) for different CCTs at the constant illuminance of 50 lx. (f)-(j) Images of the same painting for the respective CCTs with the damage irradiance kept constant at the value equal to that at the CCT of 3500 K. The illuminance is indicated in the bottom-right corner of the images.

Download Full Size | PDF

In order to avoid increasing damage irradiance with increasing CCT, the product of illuminance and RDF must be kept constant. Figure 4(f)-(j) show the images for the respective CCTs with the illuminance varied in such a way that the damage irradiance is maintained at the constant value equal to that for the 50 lx illuminance at the CCT of 3500 K. Now increasing CCT leads to a substantial dimming effect, which converts to the illuminance of only 36 lx at the CCT of 6500 K. In contrast, reducing CCT to 2500 K increases the illuminance to 65 lx. Such a noticeable variation in illuminance might influence the selection of the preferential lighting conditions in favor of CCTs that are lower than that of daylight.

3.2 Tuning the color saturating ability

In order to demonstrate color rendition tuning with controlled photochemical safety, we selected a folk-art exhibit, a Lithuanian national ribbon. The ribbon has been woven in 1977 and since that time was being exposed to daylight (for 37 years). Due to aging, the colors have partially lost vividness; in particular, the dominant red color of the woollen threads dyed with a natural pigment has darkened. In order to visually compensate the darkening, the optimization of illumination can be performed by selecting an appropriate saturation factor. While varying the saturation factor, the CCT was kept constant at the value of 3000 K, which is typical of halogen lamps, widely used in museum lighting.

Figure 5(a)-(d) show the metameric SPDs of the light engine for different values of the saturation factor at the CCT of 3000 K. With increasing the saturation factor, the amber component is gradually replaced by the red one and the green component increases as well. Figure 5(e)-(h) depict the corresponding distributions of the color-shift vectors in respect of the metameric reference light (blackbody) for the 218 Munsell samples of value /6 in the a*−b* chromaticity plane of the CIELAB color space. The open circles in Fig. 5(e)-(h) denote chromaticites of the color test samples that have the color-shift vectors residing within the three-step MacAdam ellipses, i.e. rendered with high fidelity. The arrows schematically represent the vectors of the samples that stretch out of the three-step MacAdam ellipses. Figures 5(e), 5(f), 5(g), and 5(h) show the distribution of the color shift vectors for the color dulling AGB end-point (Γ = 0), high-fidelity RAGB blend (Γ = 0.47), “color-preference” RAGB blend (Γ = 0.64) [48], and color saturating RGB end-point (Γ = 1) that correspond to SPDs shown in Figs. 5(a), 5(b), 5(c), and 5(d), respectively. At Γ = 0, a substantial portion of the vectors is seen to be directed inwards, i.e. many colors have reduced chroma. At Γ = 0.47, the maximal number of colors are rendered with high fidelity. With a further increase of Γ, the number of vectors directed outward, i.e. showing increased chroma of the samples, starts increasing at the expense of the samples rendered with high fidelity. At Γ = 0.64, the preferential color rendition with a still high number of the high-fidelity colors can be established. At Γ = 1, the number of samples having increased chroma attains the maximum value (also for this blend, many color test samples have perceptually noticeable distortions of hue [50]).

Fig. 5 SPDs of the light engine for the values of the saturation factor of 0 (a), 0.47 (b), 0.64 (c), and 1 (d) at the constant CCT of 3000 K. (e)-(f) Corresponding distributions of the color-shift vectors for 218 Munsell samples of value /6 in the a*−b* chromaticity plane of the CIELAB color space. Open circles, chromaticites of the color test samples rendered with high fidelity; arrows, schematic representation of the color-shift vectors of the samples that are rendered with perceptually noticeable color distortion.

Download Full Size | PDF

Table 3 shows the RDFs and color rendition indices for the used tetrachromatic blends with different saturation factors at the CCT of 3000 K. The RDF decreases with the saturation factor for low-grade paper (by a factor of 0.82). For other materials, it increases by factors ranging from 1.16 to 1.25, depending on the material. The color fidelity indices (R_a, CFI, and Q_f) are reduced with moving from the high-fidelity value of Γ in both directions. With increasing the saturation factor, the GAI, CQS Q_g, and statistical CSI increase, whereas the statistical CDI decreases.

Table 3. Relative Damage Factor for Different Materials and Color Rendition Indices of Different Metrics as Functions of the Saturation Factor for the Light Engine at the CCT of 3000 K

View Table | View all tables in this article

Figures 6(a)-6(d) display the images of the ribbon for different values of the saturation factor under the constant illuminance of 50 lx. The saturation factor and RDF estimated for textiles (b = 0.010) are indicated above each image. As seen, with increasing Γ above the high fidelity value of 0.47, the colors of the ribbon, especially the red one, become more vivid, which indicates the compensation of the darkening effect. At Γ = 0, the colors appear even duller than under the high-fidelity conditions. Within the entire range of the variation of the saturation factor from 0 to 1, the RDF is seen to increase by a factor of about 1.25. At constant illuminance, this corresponds to the increase of the damage irradiance by the same factor.

Fig. 6 (a)-(d) Images of a Lithuanian national ribbon for different saturation factors at the CCT of 3000 K and at the constant illuminance of 50 lx. (e)-(h) Images of the same ribbon for the respective saturation factors with the damage irradiance kept constant at the value equal to that at Γ = 0.47. The illuminance is indicated in the bottom-right corner of the images.

Download Full Size | PDF

The effect of the variation of damage irradiance with increasing Γ can be avoided via the corresponding control of illuminance. Figure 6(e)-(h) show the images for the respective values of the saturation factor with the illuminance varied in such a way that the damage irradiance is maintained at the constant value equal to that for the 50 lx illuminance at Γ = 0.47. This results in the illuminance increase to 53 lx at Γ = 0 corresponding to the highest color-dulling effect, and in the drop of illuminance to 42 lx at Γ = 1 corresponding to the highest color-saturating effect. As seen, such a variation of illuminance noticeably affects the variation of visual impression provided by the ribbon. This means that with varying the color saturating ability of illumination, the subjective preferences to color rendition established at constant illuminance and at constant damage irradiance might vary as well.

3.3 Shifting chromaticity along an isotemperature line

The photochemically safe shifting of the chromaticity out of the white light locus is demonstrated on a painting “Funeral symphony” by Lithuanian artist ant mystic M. K. Čiurlionis (1875–1911). The artwork has been painted in 1909 with pastel on paper. After exhibiting for more than one hundred years, the colors of this painting partly faded and the substrate underwent some brownishing. The latter effect resulted in the overall shift of the chromaticity gamut, which can be partially compensated by shifting the chromaticity of the light source out of the white light locus. Prior to such a shifting, the most appropriate CCT and saturation factor are to be selected. In order to maintain a constant value of CCT, we performed shifting along the isotemperature line within the color space.

Figure 7 shows a segment of the CIE 1931 chromaticity diagram with the blackbody locus (bold line) and an isotemperature line (thin line). The CCT was selected to equal that of the standard illuminant B (4870 K).

Fig. 7 Segment of the CIE 1931 chromaticity diagram with the Planckian (blackbody) locus (bold line) and an isotemperature line for the CCT of 4870 K (thin line) shown. The isotemperature line stretches out of the Planckian locus by Δxy = 0.05 in both directions.

Download Full Size | PDF

Figure 8 shows the SPDs of the light engine for three chromaticities of the selected isotempertaure line: (a) below the Planckian locus (Δxy = −0.05), (b) right on the locus (Δxy = 0), and (c) above the locus (Δxy = + 0.05). These SPDs provide different overall shifts of the chromaticity gamut. In addition, the constant saturation factor value of 0.5 was selected in order to partially compensate the fading of colors by increasing saturation (gamut area), rather than by providing the high fidelity blend (Γ = 0.28). While moving upward along the isotemperature line (i.e. from the purple region to the white point and further to the yellow-green region), the red and blue components are seen to decrease, whereas the amber and green components increase.

Fig. 8 SPDs of the light engine tuned along an isotemperature line (CCT = 4870 K) at the saturation factor Γ = 0.5. The deviation from the Planckian locus Δxy is –0.05 (a), 0 (b), and + 0.05 (c), respectively.

Download Full Size | PDF

Table 4 shows the RDF and color rendition indices for the used tetrachromatic blends with different chromaticity shifts along the isotemperature line of CCT = 4870 K at Γ = 0.5. With the variation of the chromaticity shift from –0.05 to + 0.05, the RDF decreases by a factor of 0.67 to 0.76, depending on the material. The color rendition indices (shown for the white chromaticity at Δxy = 0) indicate the increased color saturating effect at the expense of the color fidelity.

Table 4. Relative Damage Factor for Different Materials and Color Rendition Indices as Functions of the Chromaticity Shift along the Isotemperature Line for CCT = 4870 K at the Saturation Factor Γ = 0.5

View Table | View all tables in this article

Figures 9(a)-9(c) display the images of the painting for different chromaticity shifts under the constant illuminance of 50 lx. The chromaticity shift and RDF estimated for b = 0.010 (equal to that of oil paint on canvas or water colors on rag paper) are indicated on top of each image. For the chromaticity shift of Δxy = –0.05, the brownishing effect on the colors of the painting is seen to become even more pronounced in respect to that under white light (Δxy = 0) due to the emerging purplish hues. For the chromaticity shift of Δxy = + 0.05, the brownishing effect is noticeably compensated. With the variation of the chromaticity shift from –0.05 to + 0.05, the RDF is seen to decrease by a factor of about 0.75, i.e. with the damage irradiance decreasing accordingly.

Fig. 9 (a)-(c) Images of a pastel painting “Funeral symphony” (M. K. Čiurlionis, 1909) for different shifts of the chromaticity of incident light along an isotemperature line of 4870 K CCT at the constant illuminance of 50 lx. (d)-(f) Images of the same painting for the respective shifts of the chromaticity with the damage irradiance kept constant at the value equal to that at Δxy = 0. The illuminance is indicated in the bottom-right corner of the images.

Download Full Size | PDF

The decrease of RDF with the favored shift of the chromaticity of light in the upward direction can be exploited for increasing illuminance without increasing damage irradiance. Figures 9(d)-9(f) show the images for the respective values of the chromaticity shift with the illuminance varied in such a way that the damage flux is maintained at the constant value equal to that for the 50 lx illuminance at Δxy = 0. This results in the illuminance decrease to 43 lx at Δxy = –0.05, which corresponds to the enhancement of the undesirable brownishing effect, and in the increase of illuminance to 57 lx at Δxy = + 0.05, which corresponds to the partial compensation of the brownishing effect. Such a variation of illuminance with the shift of the chromaticity of incident light might affect the selection of subjective preferences to color rendition. Note that once the Δxy = + 0.05 option is selected for the painting under consideration, the illuminance can be reduced in order to reduce the damage irradiance in respect of that for white light.

4. Conclusions

We have developed and implemented the concept of a solid-state lighting engine for artwork-specific illumination with the control of photochemical safety. By using an instantaneous dimming functionality, the driving software allows for maintaining the damage irradiance specific to a particular type of the photosensitive artwork material at a constant value, while varying the chromaticity and color rendition properties of the generated light. The engine is based on a tetrachromatic (RAGB) cluster of colored LEDs wirelessly controlled via an external smart device.

We have demonstrated the effect of maintaining constant damage irradiance on the visual impression from artworks by operating the lighting engine in three modes: selecting the color temperature, tuning the color saturating ability, and shifting the chromaticity outside white light locus along an isotemperature line. These operating modes can be used for attaining the most appropriate visual appearance of artwork, visual reconstruction of the colors altered by aging, and meeting specific visual needs of observers (e.g. color deficient or elderly people). The variation of chromaticity and color rendition properties was being accompanied by an instantaneous adjustment of the illuminance with respect to the RDF of the incident light estimated using the spectral damage functions of particular photosensitive materials. In all operating modes, such a control of photochemical safety resulted in a perceptually noticeable alteration of visual impression due to the variation of illuminance in comparison with the constant-illuminance conditions.

For the particular set of four primary LEDs used in the RAGB illuminant of our engine, the most noticeable visual effect due to maintaining constant damage irradiance was dimming by a factor of up to 1.8 when increasing CCT in the range of 2500-6500 K. Tuning the color saturating ability in between the color-dulling (AGB) and color-saturating (RGB) endpoints of the RAGB illuminant required the illuminance variation up to 25%, depending on the CCT and the type of the photosensitive material. Also, the sign of the illuminance variation was found to depend on the material. For example, for the low-grade paper, increased color saturating ability results in an increase of illuminance at constant damage irradiance, whereas for other materials, the effect is reverse. Shifting the chromaticity along the isotemperature line at the CCT of 4870 K resulted in the illuminance increase by up to 15% per upward chromaticity increment of Δxy = 0.01.

To summarize, the replacement of incandescent lamps by LEDs in museum lighting allows for the introduction of smart functionalities that are much more sophisticated that just improving the efficacy and longevity of light sources. The ease of dynamically composing SPDs and smart means of control offered by the solid-state lighting technology allow for instantaneously adjusting illuminance in accordance with the spectral damage function of a photosensitive material rather than just following the recommended illuminance values of 50 lx and 200 lx. In particular, a versatile trade-off between the conservation and visualization of artworks can be implemented within the introduced technological approach using tunable LED clusters with the software-based control of the chromaticity and color rendition of the composite light and maintaining constant photochemical damage irradiance.

Acknowledgments

The work at VU was funded by the Research Council of Lithuania under Grant MIP-098/2012. The work at RPI was supported primarily by the Engineering Research Centers Program (ERC) of the National Science Foundation under NSF Cooperative Agreement No. EEC-0812056 and in part by New York State under NYSTAR contract C090145.

References and links

1. G. Thomson, The Museum Environment (Butterworth-Heinemann, 1986).

2. C. Cuttle, Light for Art's Sake: Lighting for Artworks and Museum Displays (Butterworth-Heinemann, 2007).

3. W. R. J. Brown, “The influence of luminance level on visual sensitivity to color differences,” J. Opt. Soc. Am. 41(10), 684–688 (1951). [CrossRef] [PubMed]

4. W. A. Thornton, “Fluorescent lamps with high color-discrimination capability,” J. Illum. Eng. Soc. 3(1), 61–64 (1973). [CrossRef]

5. R. H. Lafontaine, “Seeing through a yellow varnish: a compensating illumination system,” Stud. Conserv. 31(3), 97–102 (1986). [CrossRef]

6. G. Thomson, “A new look at colour rendering, level of illumination, and protection from ultraviolet radiation in museum lighting,” Stud. Conserv. 6(2-3), 49–70 (1961). [CrossRef]

7. G. Thomson, “Annual exposure to light within museums,” Stud. Conserv. 12(1), 26–36 (1967). [CrossRef]

8. B. H. Crawford, “Just perceptible colour differences in relation to level of illumination,” Stud. Conserv. 18(4), 159–166 (1973). [CrossRef]

9. D. L. Loe, E. Rowlands, and N. F. Watson, “Preferred lighting conditions for the display of oil and watercolour paintings,” Lighting Res. Tech. 14(4), 173–192 (1982). [CrossRef]

10. Illuminating Engineering Society of North America, Museum and Art Gallery Lighting: A Recommended Practice (IESNA, 1996).

11. Commission Internationale de l’Éclairage, “Control of damage to museum objects by optical radiation,” Pub. CIE 157, 2004.

12. C. Cuttle, “A proposal to reduce the exposure to light of museum objects without reducing illuminance or the level of visual satisfaction of museum visitors,” J. Amer. Inst. Conserv. 39, 229–244 (2000).

13. L. S. Harrison, Report on the Deteriorating Effects of Modern Light Sources (Metropolitan Museum of Art, 1953).

14. G. S. Hilbert, S. Aydinli, and J. Krochmann, “Zur Beleuchtung musealer Exponate,” Restauro 97, 313–321 (1991).

15. D. Saunders and J. Kirby, “Wavelength-dependent fading of artist’s pigments,” in Preventive Conservation Practice, Theory and Research, A. Roy and P. Smith, eds. (International Institute for Conservation of Historic and Artistic Works, London, 1994), pp. 190–194.

16. C. Cuttle, “Lighting works of art for exhibition and conservation,” Lighting Res. Tech. 20(2), 43–53 (1988). [CrossRef]

17. C. Cuttle, “Damage to museum objects due to light exposure,” Lighting Res. Tech. 28(1), 1–9 (1996). [CrossRef]

18. M. Scuello, I. Abramov, J. Gordon, and S. Weintraub, “Museum lighting: Why are some illuminants preferred?” J. Opt. Soc. Am. A 21(2), 306–311 (2004). [CrossRef] [PubMed]

19. M. Scuello, I. Abramov, J. Gordon, and S. Weintraub, “Museum lighting: Optimizing the illuminant,” Color Res. Appl. 29(2), 121–127 (2004). [CrossRef]

20. P. D. Pinto, J. M. M. Linhares, and S. M. C. Nascimento, “Correlated color temperature preferred by observers for illumination of artistic paintings,” J. Opt. Soc. Am. A 25(3), 623–630 (2008). [CrossRef] [PubMed]

21. P. D. Pinto, J. M. M. Linhares, J. A. Carvalhal, and S. M. C. Nascimento, “Psychophysical estimation of the best illumination for appreciation of Renaissance paintings,” Vis. Neurosci. 23(3-4), 669–674 (2006). [CrossRef] [PubMed]

22. S. M. C. Nascimento and O. Masuda, “Best lighting for visual appreciation of artistic paintings—experiments with real paintings and real illumination,” J. Opt. Soc. Am. A 31(4), A214–A219 (2014). [CrossRef]

23. A. A. Kruithof, “Tubular luminescence lamps for general illumination,” Philips Tech. Rev. 6, 65–73 (1941).

24. F. Viénot, M.-L. Durand, and E. Mahler, “Kruithof's rule revisited using LED illumination,” J. Mod. Opt. 56(13), 1433–1446 (2009). [CrossRef]

25. Commission Internationale de l’Éclairage, “Method of measuring and specifying colour rendering properties of light sources,” Pub. CIE 13.3, 1995.

26. C. Richardson and D. Saunders, “Acceptable light damage: A preliminary investigation,” Stud. Conserv. 52(3), 177–187 (2007). [CrossRef]

27. M. F. Delgado, C. W. Dirk, J. Druzik, and N. WestFall, “Lighting the world’s treasures: approaches to safer museum lighting,” Color Res. Appl. 36(4), 238–254 (2011). [CrossRef]

28. W. A. Thornton, “The high visual efficiency of prime color lamps,” Lighting Des. Appl. 5, 35–41 (1975).

29. R. S. Berns, “Designing white-light LED lighting for the display of art: A feasibility study,” Color Res. Appl. 36(5), 324–334 (2011). [CrossRef]

30. P. D. Pinto, P. E. R. Felgueiras, J. M. M. Linhares, and S. M. C. Nascimento, “Chromatic effects of metamers of D65 on art paintings,” Ophthalmic Physiol. Opt. 30(5), 632–637 (2010). [CrossRef] [PubMed]

31. J. Druzik and B. Eshøj, “Museum lighting: its past and future development,” in Museum Microclimates, T. Padfield and K. Borchersen, eds. (National Museum of Denmark, Copenhagen, 2007), pp. 51–56.

32. A. Žukauskas, M. S. Shur, and R. Gaska, Introduction to Solid-State Lighting (Wiley, 2002).

33. S. Muthu, F. J. P. Schuurmans, and M. D. Pashley, “Red, green, and blue LEDs for white light illumination,” IEEE J. Sel. Top. Quantum Electron. 8(2), 333–338 (2002). [CrossRef]

34. I. Speier and M. Salsbury, “Color temperature tunable white light LED system,” Proc. SPIE 6337, 63371F (2006). [CrossRef]

35. G. He and L. Zheng, “Color temperature tunable white-light light-emitting diode clusters with high color rendering index,” Appl. Opt. 49(24), 4670–4676 (2010). [CrossRef] [PubMed]

36. P. Zhong, G. He, and M. Zhang, “Spectral optimization of the color temperature tunable white light-emitting diode (LED) cluster consisting of direct-emission blue and red LEDs and a diphosphor conversion LED,” Opt. Express 20(S5Suppl 5), A684–A693 (2012). [CrossRef] [PubMed]

37. A. Žukauskas, R. Vaicekauskas, F. Ivanauskas, R. Gaska, and M. S. Shur, “Optimization of white polychromatic semiconductor lamps,” Appl. Phys. Lett. 80(2), 234–236 (2002). [CrossRef]

38. O. Masuda and S. M. C. Nascimento, “Lighting spectrum to maximize colorfulness,” Opt. Lett. 37(3), 407–409 (2012). [CrossRef] [PubMed]

39. A. Žukauskas, R. Vaicekauskas, and M. Shur, “Solid-state lamps with optimized color saturation ability,” Opt. Express 18(3), 2287–2295 (2010). [CrossRef] [PubMed]

40. D. Vazquez-Molini, A. Alvarez Fernandez-Balbuena, A. Garcia Botella, J. A. Herraez, M. Del Egido, and R. Ontañon, “Advanced LED lighting system applied to cultural heritage goods,” in Colour and Light in Architecture, P. Zennaro, ed. (Knemesi, Verona, 2010), pp. 341–348.

41. R. Vaicekauskas and A. Žukauskas, “LEDs in lighting with tailored color quality,” Int. J. High Speed Electron. Syst. 20(02), 287–301 (2011). [CrossRef]

42. A. Žukauskas, R. Vaicekauskas, and M. Shur, “Color-dulling solid-state sources of light,” Opt. Express 20(9), 9755–9762 (2012). [CrossRef] [PubMed]

43. A. Žukauskas, R. Vaicekauskas, F. Ivanauskas, H. Vaitkevičius, and M. S. Shur, “Rendering a color palette by light-emitting diodes,” Appl. Phys. Lett. 93(2), 021109 (2008). [CrossRef]

44. A. Žukauskas, R. Vaicekauskas, and M. S. Shur, “Colour-rendition properties of solid-state lamps,” J. Phys. D Appl. Phys. 43(35), 354006 (2010). [CrossRef]

45. S. Boissard and M. Fontoynont, “Optimization of LED-based light blendings for object presentation,” Color Res. Appl. 34(4), 310–320 (2009). [CrossRef]

46. F. Viénot, G. Coron, and B. Lavédrine, “LEDs as a tool to enhance faded colours of museums artefacts,” J. Cult. Herit. 12(4), 431–440 (2011). [CrossRef]

47. A. Liu, A. Tuzikas, A. Žukauskas, R. Vaicekauskas, P. Vitta, and M. Shur, “Cultural preferences to color quality of illumination of different artwork objects revealed by a color rendition engine,” IEEE Photonics J. 5(4), 6801010 (2013). [CrossRef]

48. A. Žukauskas, R. Vaicekauskas, P. Vitta, A. Tuzikas, A. Petrulis, and M. Shur, “Color rendition engine,” Opt. Express 20(5), 5356–5367 (2012). [CrossRef] [PubMed]

49. G. Wyszecki and W. S. Stiles, Color Science: Concepts and Methods, Quantitative Data and Formulae (Wiley, 2000).

50. A. Žukauskas, R. Vaicekauskas, F. Ivanauskas, H. Vaitkevičius, P. Vitta, and M. S. Shur, “Statistical approach to color quality of solid-state lamps,” IEEE J. Sel. Top. Quantum Electron. 15(6), 1753–1762 (2009). [CrossRef]

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

52. M. S. Rea and J. P. Freyssinier-Nova, “Color rendering: A tale of two metrics,” Color Res. Appl. 33(3), 192–202 (2008). [CrossRef]

Correlated color temperature (K)		2500	3500	4500	5500	6500
RDF for different materials	Low-grade paper	0.23	0.31	0.36	0.41	0.46
	Rag paper	0.69	0.92	1.04	1.18	1.29
	Oil paint on canvas, water colors on rag paper	0.71	0.92	1.04	1.16	1.28
	Textiles	0.73	0.91	1.01	1.12	1.22
Two-metric indices	General color rendering index R_a	96	95	95	93	93
Two-metric indices	Gamut area index GAI	48	75	90	98	104
Statistical indices	Color saturation index CSI (%)	0	1	1	1	1
	Color fidelity index CFI (%)	95	93	89	86	84
	Color dulling index CDI (%)	1	2	2	2	2
CQS indices	Color fidelity scale Q_f	90	93	92	90	89
CQS indices	Color gamut scale Q_g	103	106	105	105	104

Saturation factor Γ		0	0.47	0.64	1
RDF for different materials	Low-grade paper	0.28	0.27	0.27	0.23
	Rag paper	0.75	0.79	0.80	0.91
	Oil paint on canvas, water colors on rag paper	0.77	0.80	0.81	0.89
	Textiles	0.76	0.80	0.82	0.95
Two-metric indices	General color rendering index R_a	78	96	89	16
Two-metric indices	Gamut area index GAI	55	62	66	79
Statistical indices	Color saturation index CSI (%)	0	1	27	78
	Color fidelity index CFI (%)	39	93	60	7
	Color dulling index CDI (%)	46	1	0	0
CQS indices	Color fidelity scale Q_f	80	91	87	35
CQS indices	Color gamut scale Q_g	90	102	108	132

Chromaticity shift Δxy		−0.05	0	+ 0.05
Chromaticity coordinate x		0.341	0.348	0.35
Chromaticity coordinate y		0.302	0.352	0.401
RDF for different materials	Low-grade paper	0.42	0.37	0.28
	Rag paper	1.34	1.13	0.98
	Oil paint on canvas, water colors on rag paper	1.31	1.12	0.98
	Textiles	1.27	1.09	0.96
Two-metric indices	General color rendering index R_a	−	84	−
Two-metric indices	Gamut area index GAI	−	103	−
Statistical indices	Color saturation index CSI (%)	−	34	−
	Color fidelity index CFI (%)	−	44	−
	Color dulling index CDI (%)	−	1	−
CQS indices	Color fidelity scale Q_f	−	84	−
CQS indices	Color gamut scale Q_g	−	113	−

Correlated color temperature (K)		2500	3500	4500	5500	6500
RDF for different materials	Low-grade paper	0.23	0.31	0.36	0.41	0.46
	Rag paper	0.69	0.92	1.04	1.18	1.29
	Oil paint on canvas, water colors on rag paper	0.71	0.92	1.04	1.16	1.28
	Textiles	0.73	0.91	1.01	1.12	1.22
Two-metric indices	General color rendering index R_a	96	95	95	93	93
Two-metric indices	Gamut area index GAI	48	75	90	98	104
Statistical indices	Color saturation index CSI (%)	0	1	1	1	1
	Color fidelity index CFI (%)	95	93	89	86	84
	Color dulling index CDI (%)	1	2	2	2	2
CQS indices	Color fidelity scale Q_f	90	93	92	90	89
CQS indices	Color gamut scale Q_g	103	106	105	105	104

Saturation factor Γ		0	0.47	0.64	1
RDF for different materials	Low-grade paper	0.28	0.27	0.27	0.23
	Rag paper	0.75	0.79	0.80	0.91
	Oil paint on canvas, water colors on rag paper	0.77	0.80	0.81	0.89
	Textiles	0.76	0.80	0.82	0.95
Two-metric indices	General color rendering index R_a	78	96	89	16
Two-metric indices	Gamut area index GAI	55	62	66	79
Statistical indices	Color saturation index CSI (%)	0	1	27	78
	Color fidelity index CFI (%)	39	93	60	7
	Color dulling index CDI (%)	46	1	0	0
CQS indices	Color fidelity scale Q_f	80	91	87	35
CQS indices	Color gamut scale Q_g	90	102	108	132

Artwork visualization using a solid-state lighting engine with controlled photochemical safety

Abstract

1. Introduction

2. Lighting engine

2.1 Principle of operation

2.2 Practical implementation

3. Examples of the tunable visualization of artworks with controlled photochemical safety

3.1 Color temperature selection

3.2 Tuning the color saturating ability

3.3 Shifting chromaticity along an isotemperature line

4. Conclusions

Acknowledgments

References and links

Cited By

Figures (9)

Tables (4)

Equations (13)

Optics Express

CCT (K)	Light source	Color rendition metrics						LER (lm/W)		Irradiance below 400 nm per unit luminous flux (μW/lm)	Relative damage factor RDF
		Two-metric		Statistical			CQS				Relative damage factor RDF
		Color rendition indices									low-grade paper	oil paint on canvas
		R_a	GAI	CSI (%)	CFI (%)	CDI (%)	Q_f	Q_g			low-grade paper	oil paint on canvas
3000	AGB (Γ = 0)	79	55	0	39	46	80	90	343	38	0.28	0.77
	RAGB (Γ = 0.47)	96	62	1	93	1	91	102	296	36	0.27	0.80
	RGB (Γ = 1)	16	79	78	7	0	35	132	231	27	0.23	0.89
4500	AGB (Γ = 0)	89	84	0	64	20	87	97	301	38	0.37	1.03
	RAGB (Γ = 0.28)	95	90	1	89	2	92	105	290	36	0.36	1.04
	RGB (Γ = 1)	30	125	70	9	1	41	138	236	28	0.33	1.12
6500	AGB (Γ = 0)	92	100	1	74	8	87	100	281	38	0.46	1.27
	RAGB (Γ = 0.15)	93	104	1	84	2	89	104	276	38	0.46	1.28
	RGB (Γ = 1	38	147	67	9	2	44	138	231	30	0.43	1.35
2856	CIE A	100	56	0	100	0	98	98	156	82	1.00	1.00
2848	CIE A filtered	100	56	0	100	0	98	98	157	0	0.17	0.85