In part I of this work [Opt. Express , 27 ,9276, (2019) [CrossRef] ], we carried out an experiment to investigate the effects of adapting luminance and correlated color temperature (CCT) on degree of chromatic adaptation. Under the highest white luminance Lw of 900 cd/m2, an incomplete chromatic adaptation was still found under the 2700 and 3500 K adapting conditions. This motivated us to further increase the adapting luminance to investigate whether a complete chromatic adaptation cannot happen under a low adapting CCT (e.g., 2700 K). In this experiment, we investigated the degrees of chromatic adaptation under 12 adapting conditions, comprising four CCT (i.e., 2700, 3200, 4000, and 6500 K) and three white luminance levels (i.e., Lw of 1200, 2100, and 3000 cd/m2), by asking human observers to adjust the color appearance of a stimulus to the whitest. Such luminance levels of the adapting conditions were never investigated in the past and are assumed to introduce a complete chromatic adaptation. The results clearly show that an incomplete chromatic adaptation still happened under the adapting condition having a CCT of 2700 or 3200 K, though the luminance was so high. The adapting luminance and CCT were found to jointly affect the degree of chromatic adaptation, with a higher degree of adaptation under a higher adapting CCT or luminance level. When the adapting CCT was low (i.e., 2700 or 3200 K), the increase of adapting luminance was found to be able to increase the degree of adaptation more effectively. These findings suggest the necessity to revise the chromatic adaptation transforms (CAT) and color appearance models (CAM) for better characterizing the color appearance of stimulus under different adapting conditions.
© 2021 Optica Publishing Group under the terms of the Optica Open Access Publishing Agreement
Chromatic adaptation is an important mechanism in the human visual system. It helps to automatically remove the color cast of the illumination on the illuminated objects, so that the perceived color appearance of the objects remains relatively constant under a wide range of illumination conditions . In order to accurately characterize and predict the color appearance of stimuli under different conditions, great efforts have been made to understand how various factors affect the chromatic adaptation. Chromatic adaptation transforms (CATs) were developed to characterize the relationship between the tristimulus values of a pair of corresponding colors (i.e., the colors having the same perceived color appearance under the two adapting conditions), in relative to the tristimulus values of the adapting conditions . These CATs were generally developed based on the von Kries coefficient law, which states that the three types of cones are individually scaled based on the adapting condition .
The chromatic adaptation, however, may be incomplete, meaning the color cast of the illumination cannot be completely removed. Thus, some CATs include a factor to characterize the degree of chromatic adaptation, with a value of 1 for a complete chromatic adaptation [1–3]. Though real-life experience suggests that the illumination color (i.e., the chromaticities) affect the color appearance of the objects through chromatic adaptation, the degree of adaptation factor is commonly only affected by the luminance of the adapting field. For example, the CAT02  degree of chromatic adaptation factor D is calculated using Eq. (1), with LA being the adapting luminance and F being a factor describing the surround condition.
Therefore, some recent studies specifically focused on how the chromaticities of the illumination affected the chromatic adaptation. For example, Smet and his colleagues carried out a series of experiments using a memory color matching technique, in which the human observers adjusted the color of a 3D object through a projector under various illumination chromaticities [4,5]. Zhai and Luo carried out an experiment using an achromatic matching technique by asking observers to select the reflective surface color sample having the white appearance and to adjust the color appearance of a stimulus produced by a smartphone display under the illumination having different chromaticities . Zhu and his colleagues carried out a memory color matching experiment by asking the human observers to adjust the color appearance of a series of familiar objects on a display at different display white points . These studies all clearly suggested the significant effect of the illumination chromaticities on the degree of chromatic adaptation. In particular, a lower degree of adaptation was found under the illumination having a lower CCT. The adapting luminance, however, was not varied in these studies, which cannot reveal whether the degree of chromatic adaptation is affected by the luminance and chromaticities of the adapting field separately or not.
In part I of the study , we systematically varied the luminance and CCT of the adapting field and asked the obsthe different adapting conditions through the degree of chromatic adaptation. A higher degree of chromatic adaptation was found to happen under a higher adapting CCT or luminance, with the effect of adapting luminance more obvious under a lower adapting CCT. More importantly, though the adapting luminance Lw of 900 cd/m2 was predicted to result in a degree of adaptation factor D to 0.996 according to Eq. (1), the adapting CCT of 2700 and 3500 K still did not introduce a complete adaptation under such an adapting luminance. In this experiment, we investigated the effects of adapting CCT and luminance on degree of adaptation by further increasing the luminance of the adapting field Lw up to 3000 cd/m2, which was never studied in the past.
The experiment methods and protocols were approved by the Institutional Review Board, and the experiment was carried out in Color and Illumination Laboratory at The Hong Kong Polytechnic University.
The experiment was carried out using a viewing booth, whose dimensions were 60 cm × 60 cm × 60 cm. The interior walls of the booth were painted using a Munsell N7 spectrally neutral paint. The booth was under a uniform illumination of a high power spectrally tunable LED device, which was placed above the booth. A 4.5 cm × 4.5 cm square opening was cut at the center of the back wall. A color stimulus was produced by uniformly illuminating the opening from the back of the viewing booth using a high power spectrally tunable LED projection light. A piece of fused silica was attached at the opening, at the back side of the wall, so that the stimulus had a diffuse distribution when viewed from the front. The front side of the booth was partially covered, with the bottom area open. A chin rest was mounted just outside the front side of the booth, centered on the opening, so that the observer cannot view the LED device directly and all the observers experienced a similar viewing geometry, with the stimulus having a field of view around 4.3°, during the experiment. Figure 1 shows the experiment setup, viewed from the observer’s eye position during the experiment.
Both devices contained three LED channels (i.e., red, green, and blue) with the intensities being individually controlled through a 16-bit DMX controller. In particular, the three channels in the projection light had good independence (i.e., the adjustment of individual channel does not affect the other channels) and linearity, allowing to develop a linear model to characterize the relationship between the input digital counts and the tristimulus values of the stimulus (i.e., the sum of the light produced by the projection light and the reflection from the fused silica). A control program was then developed based on the linear model to change the chromaticities of the stimulus along the two axes in the CIE 1976 u’v’ chromaticity diagram using the four arrow keys (i.e., ↓ for -v’, → for + u’, ↑ for + v’, and ← for -u’), with a step of 0.0015 units, at a fixed luminance level. These four directions corresponded to the change of the stimulus color appearance towards blue, red, yellow, and green, which was found to be easily understandable in our past experiments [8–10].
2.2 Adapting conditions and stimulus
Twelve adapting conditions were organized as a 3 × 4 factorial design, comprising four levels of CCT (i.e., 2700, 3200, 4000, and 6500 K) and three levels of adapting luminance (i.e., Lw ≈ 1200, 2100, and 3000 cd/m2). The two CCT levels (i.e., 3200 and 4000 K) were carefully selected to be slightly different from those in part I of our study  (i.e., 3500 and 5000 K), as the degree of chromatic adaptation under the 5000 K conditions were found very high and a significant effect of adapting luminance was found under the conditions at lower adapting CCT levels. In addition, the adapting luminance Lw levels were carefully selected, under which CAT02 predicts a complete chromatic adaptation (i.e., the degree of chromatic adaptation factor D = 1). The 1200 cd/m2 level was higher than 900 cd/m2, and the 3000 cd/m2 level was the highest level that can be achieved by the LED device in the viewing booth for the CCT ranges. The adapting conditions were calibrated by adjusting the digital counts of the three channels in the LED device. Table 1 summarizes the photometric and colorimetric characteristics of the adapting conditions based on the spectra measured using a JETI specbos 1211UV-2 spectroradiometer and a reflectance standard placed at the opening.
The stimulus was designed to have two relative luminance Y levels of 30 and 60, which helped to isolate the effect of chromaticities from lightness. More importantly, the levels of 30 and 60 were carefully selected. Firstly, these two levels made the stimulus to have a lower luminance level than the diffuse white, so that the stimulus was viewed in the surface mode and appeared reflective. Secondly, these two levels corresponded to the lightness values (i.e., J’ and L*) around 62 and 83 in CAM02-UCS and CIELAB. Though both lightness values were below 100, the lightness value of 83 seemed to result in a different viewing mode under the conditions at lower adapting CCT levels, according to part I and a recent experiment [8,11], due to the smaller color gamut of surface colors at such a lightness level. Thirdly, these two levels allowed a wide range of chromaticities to be adjusted by the observers, so that the adjustments would not be restricted. Under each adapting condition, the observer was asked to adjust the color appearance of the stimulus to the whitest using the four arrow keys on the keyboard.
2.3 Experimental procedure
Upon arrival, the experimenter explained the task and the procedure of the experiment, and the observer completed the general information survey and the Ishihara Color Vision Test. After that, the observer was seated in front of the viewing booth, with his or her chin fixed on the chin rest, and the general illumination in the experiment space was switched off. The LED device was then switched on to provide the first adapting condition to the viewing booth, and the observer was asked to look into the viewing booth for 90 seconds, which helped to achieve a stable degree of adaptation . After 90 seconds, the projection light was switched on to produce the stimulus whose color appearance was obviously different from white. The observer adjusted the color appearance of the stimulus to the whitest by changing the chromaticities of the stimulus using the four arrow keys on the keyboard without time limitation. When adjusting the chromaticities, the computer program simultaneously estimated the digital counts to produce such chromaticities based on the model developed and sent the control signals to the projection lights. Once the observer was satisfied with the adjusted color appearance, he or she pressed the enter key to confirm the adjustment, with the corresponding digital counts saved by the program. The same procedure was repeated for all the experiment settings, with the same 90-second adaptation period for each adapting condition. For evaluating the intra-observer variations, the observers repeated the adjustments under the 2700 K conditions with the stimulus relative luminance of 60. The entire experiment took around 45 minutes for each observer.
Eight observers (4 males and 4 females), with ages between 24 and 28, completed the experiment. All the observers had normal color vision as tested using the Ishihara Color Vision Test.
3. Results and discussions
Though the model was used to predict the digital counts that were needed to produce the desired chromaticities of the stimulus, the differences between the desired and the actual chromaticities could not be avoided. To better characterize the stimuli adjusted by the observers, the digital counts for producing the stimuli adjusted by the observers were sent to the device after the experiment. The spectroradiometer was used to measure the spectra of the stimuli under the corresponding adapting condition from the observer’s eye position, with the spectra used to derive the colorimetric quantities for the following analyses.
The average difference between the relative luminance of the stimuli derived from the measurements and the target levels was 2.87%, and the average difference between the chromaticities of the stimuli derived from measurements and those estimated from the model was 0.0120 units in the CIE 1976 u’v’ chromaticity diagram. The chromaticities of the stimuli adjusted by the observers were all far from the boundary of the gamut. These suggested the reliability of the control program. To allow a better comparison to the results from part I of the study , the CIE 1964 10° Color Matching Functions (CMFs) were used in the following analyses. Similar analyses were also performed using the CIE 1931 2° CMFs, which did not introduce significant differences.
3.1 Intra- and inter-observer variations
Both the intra- and inter-observer variations were characterized using the mean color difference from the mean (MCDM) values in the CIE 1976 u’10v’10 units. Specifically, the intra-observer variations were characterized based on the differences between the chromaticities of the repeated adjustments and the average chromaticities. The MCDM values ranged between 0.0006 and 0.0049, with an average of 0.0026. In particular, the MCDM values were 0.0032, 0.0022, and 0.0025 under the Lw of 1200, 2100, and 3000 cd/m2 respectively. The inter-observer variations were characterized based on the differences between the chromaticities of the adjustments made by each observer and the average chromaticities of the adjustments made by all the observers (i.e., an average observer). The MCDM values ranged between 0.0063 and 0.0132, with an average of 0.0081. Table 2 lists all the MCDM values for characterizing the inter-observer variations under each condition. These MCDM values were comparable to those in similar experiments, suggesting the reliability of the experiment results [8,13].
Figure 2 shows the 95% confidence error ellipses of all the adjustments made by the observers at the two stimulus relative luminance levels under the different adapting conditions. It can be observed that the chromaticities adjusted by the observers under the same conditions were generally distributed along the blackbody locus, suggesting the chromaticity region for producing the perception of white appearance. A lower adapting luminance generally resulted in a larger inter-observer variation, especially when the adapting CCT level was 2700 K. In addition, when the adapting luminance was fixed, a lower adapting CCT resulted in a larger inter-observer variation. Moreover, it can be observed that the relative luminance value of 60 caused a much larger inter-observer variation, when the adapting CCT was 2700, 3200, and 4000 K. This can also be observed in Table 2.
3.2 Average chromaticities in the CIE 1976 u’10v’10 chromaticity diagram and CAM02-UCS
The average chromaticities of the stimuli adjusted by the observers in the CIE 1976 u’10v’10 chromaticity diagram are shown in Fig. 3, with the chromaticity distance Δu’10v’10 to the chromaticities of the adapting conditions shown in Fig. 4. A smaller chromaticity distance corresponds to a higher degree of chromatic adaptation. Under a certain Lw level, a lower adapting CCT caused much larger chromaticity distances, with the chromaticities of the adjusted stimuli being much further from those of the adapting conditions. This suggested that the degree of chromatic adaptation was much lower under a lower adapting CCT. When the adapting CCT was either 2700 or 3200 K, the chromaticity distances became much smaller with the increase of the Lw level, with the chromaticities of the adjusted stimuli being shifted towards those of the adapting conditions. In contrast, the chromaticity distances under the 4000 and 6500 K adapting conditions were generally similar under the three Lw levels. This clearly suggested that the adapting luminance and CCT jointly affected the degree of chromatic adaptation, and the effect of adapting luminance was more obvious under the conditions having a lower adapting CCT.
Moreover, the chromaticity distances under all the 6500 K adapting conditions were smaller than 0.005. With a 1 unit of just-noticeable color difference (JND) approximately corresponding to a Δu’v’ value of 0.004, this implies that the white appearance of the stimuli adjusted by the observers were very similar to that of a diffuse white under the same adapting conditions. In other words, the three 6500 K adapting conditions seemed to introduce a complete degree of chromatic adaptation.
The average chromaticities were also calculated in CAM02-UCS, which is more uniform than the CIE 1976 u’10v’10 chromaticity diagram. Figure 5 shows the average chromaticities in the a’10-b’10 plane in CAM02-UCS. Since CAM02-UCS embeds a chromatic adaptation transform (i.e., CAT02), the chromaticities of the adapting conditions are the origin of the a’10-b’10 plane. Thus, the chromaticity distances between the stimuli adjusted by the observers and the adapting conditions become the chromaticity distances Δa’10b’10 between the stimuli and the origin, as shown in Fig. 6. According to CAT02, all the three Lw levels resulted in a complete chromatic adaptation, with the degree of chromatic adaptation factor D set to 1 in the calculations. The distributions of the chromaticities are generally similar to those in the CIE 1976 u’10v’10 chromaticity diagram in Fig. 3. Similar to Fig. 4, Fig. 6 shows the clear effect of adapting CCT on the degree of chromatic adaptation, with a lower degree of adaptation under a lower adapting CCT (i.e., 2700 and 3200 K). The effect of adapting luminance suggested by Fig. 6, however, is less obvious than suggested by Fig. 4, which is likely due to the fact that CAM02-UCS considers the effect of adapting luminance in the calculations.
In general, the effects of adapting luminance and CCT on the degree of chromatic adaptation shown in this experiment generally corroborated those in part I of the study .
3.3 Difference caused by the two stimulus luminance levels
In addition to the adapting CCT and Lw levels, the stimulus luminance levels also seemed to affect the chromaticities and the chromaticity distances, especially when the adapting CCT was 2700 and 3200 K, as shown in Figs. 3 to 6. This can also be observed in Figs. 3 to 6 in part I  of the study. Given the small size of the opening in relative to the adapting field, we believe that the opening was perceived as a stimulus. Coupled with the fact that the two luminance levels were close (i.e., lightness J’ value 62 versus 83), the stimulus luminance should not affect the degree of chromatic adaptation.
We speculate that the differences caused by the two stimulus luminance levels were due to the change of the viewing mode. As reported above, the lightness J’ value was around 62 and 83 for the relative luminance level of 30 and 60 respectively. A typical reflective surface color with a lightness J’ value around 62 can cover a very wide range of chromaticities, with the chromaticities producing the whitest appearance within such a range. This made the stimulus perceived as a typical surface color. In contrast, the chromaticities of a typical reflective surface color with a lightness J’ value around 83 only cover a very small range around the diffuse white, with the chromaticities for producing the whitest appearance outside such a range. To illustrate this, the chromaticity ranges of the 1269 Munsell samples with the lightness values above 80 under the four adapting conditions, together with the average chromaticities of the stimuli adjusted by the observers, in the a’10-b’10 plane are shown in Fig. 7. It can be observed that the average chromaticities of the stimuli adjusted by the observers under the 2700 and 3200 K conditions were outside the chromaticity ranges of the Munsell samples. In other words, though the stimulus adjusted by the observers had the luminance below the diffuse white, its chromaticities cannot be realized by typical reflective surface colors, which was likely to make the stimulus perceived as self-luminous. Such a speculation was also made by other researchers in the past. In particular, the study carried out by Uchikawa and his colleagues  suggested that brightness, instead of luminance, of a stimulus determines the boundary between the surface and self-luminous mode, but they only focused on the chromatic stimuli under a 6300 K adapting condition.
3.4 Degree of chromatic adaptation
With the speculation described in Section 3.3 in mind, it is more appropriate to investigate the degrees of chromatic adaptation under the different adapting conditions based on the chromaticities of the stimulus with the lightness level J’ of 62 (i.e., relative luminance level of 30), with Fig. 8 (a) showing the chromaticity distance Δu’10v’10 to the adapting chromaticities. For allowing a more comprehensive analysis, the results of the stimuli having the similar lightness level J’ around 60 in part I of the study  are combined in Fig. 8 (b). The effects of adapting luminance, CCT, and their interaction can be clearly revealed. Similar analyses were performed using the Δa’10b’10, as shown in Fig. 9.
It can be observed that the effect of the adapting luminance is less obvious for Δa’10b’10 in Fig. 9. Though this suggests that CAM02-UCS has a better performance in characterizing the effect of adapting luminance, improvements can still be made, especially on the conditions having a low adapting CCT. More importantly, the results clearly shows that the effect of adapting CCT is not accurately characterized. Even under the Lw of 3000 cd/m2 (i.e., illuminance around 10,000 lx), a complete chromatic adaptation still cannot happen under the low adapting CCT (e.g., 2700 and 3200 K).
In this experiment, human observers adjusted the color appearance of a stimulus to the whitest under 12 adapting conditions, comprising four adapting CCT (i.e., 2700, 3200, 4000, and 6500 K) and three adapting luminance levels (i.e., Lw of 1200, 2100, and 3000 cd/m2). Such high adapting luminance levels were never investigated in the past, and are assumed to introduce complete chromatic adaptations according to CAT02. In addition, the stimulus was designed to have different relative luminance levels (i.e., Y of 30 and 60) instead of absolute luminance levels in part I of the study , which aimed to avoid the possible effect caused by the stimulus luminance.
The adapting luminance, adapting CCT, and their interaction were found to significantly affect the chromaticities for producing the whitest appearance. Under the adapting conditions having a higher adapting CCT or luminance, the chromaticities were adjusted to be closer to the chromaticities of the adapting condition, suggesting a higher degree of chromatic adaptation. In addition, the increase of adapting luminance was more effective when the adapting CCT was 2700 or 3200 K. More importantly, an incomplete chromatic adaptation was still found under the 2700 and 3200 K conditions, even under such high adapting luminance levels. The results clearly suggested that improvements are needed in CAM02-UCS and CAT02 to consider the effect of adapting CCT on degree of chromatic adaptation.
Google Research Scholar Program; National Natural Science Foundation of China (61975170).
The authors declare no conflicts of interest.
Data underlying the results presented in this paper are not publicly available at this time but may be obtained from the authors upon reasonable request.
1. M. D. Fairchild, Color Appearance Models, 3rd edition. (John Wiley & Sons, 2013), pp. 450. [CrossRef]
2. CIE, “A review of chromatic adaptation transform,” in CIE 16x:2004, CIE, Vienna, Austria, (2004).
3. M. R. Luo and C. Li, “CIECAM02 and its recent developments,” in Advanced Color Image Processing and Analysis. (Springer, 2013).
4. K. A. G. Smet, Q. Zhai, M. R. Luo, and P. Hanselaer, “Study of chromatic adaptation using memory color matches, Part I: neutral illuminants,” Opt. Express 25(7), 7732–7748 (2017). [CrossRef]
5. K. A. G. Smet, Q. Zhai, M. R. Luo, and P. Hanselaer, “Study of chromatic adaptation using memory color matches, Part II: colored illuminants,” Opt. Express 25(7), 8350–8365 (2017). [CrossRef]
6. Q. Zhai and M. R. Luo, “Study of chromatic adaptation via neutral white matches on different viewing media,” Opt. Express 26(6), 7724–7739 (2018). [CrossRef]
7. Y. Zhu, M. Wei, and M. R. Luo, “Investigation on effects of adapting chromaticities and luminance on color appearance on computer displays using memory colors,” Color Res. Appl. 45(4), 612–621 (2020). [CrossRef]
8. M. Wei and S. Chen, “Effects of adapting luminance and CCT on appearance of white and degree of chromatic adaptation,” Opt. Express 27(6), 9276–9286 (2019). [CrossRef]
9. Y. Li, M. Wei, and X. Zhang, “Consideration of degree of chromatic adaptation for reproducing illuminated scenes,” Color Res. Appl. published online, DOI:10.1002/col.22760.
10. J. Wu, M. Wei, Y. Fu, and C. Cui, “Color mismatch and observer metamerism between liquid crystal displays and organic light emitting diode displays,” Opt. Express 29(8), 12292–12306 (2021). [CrossRef]
11. S. Chen and M. Wei, “Chromaticities for producing white stimuli depend on viewing mode rather than viewing medium: a pilot study,” LEUKOS 16(4), 255–265 (2020). [CrossRef]
12. M. D. Fairchild and L. Reniff, “Time course of chromatic adaptation for color-appearance adjustments,” J. Opt. Soc. Am. A 12(5), 824–833 (1995). [CrossRef]
13. S. Ma, P. Hanselaer, K. Teunissen, and K. A. G. Smet, “Impact of starting point chromaticity on memory color matching accuracy,” Opt. Express 27(24), 35308–35324 (2019). [CrossRef]
14. K. Uchikawa, K. Koida, T. Meguro, Y. Yamauchi, and I. Kuriki, “Brightness, not luminance, determines transition from the surface-color to the aperture-color mode for colored lights,” J. Opt. Soc. Am. 18(4), 737–746 (2001). [CrossRef]