Visual and Instrumental Assessments of Color Differences in Automotive Coatings

J_ID: COL Customer A_ID: COL21964 Cadmus Art: COL21964 Ed. Ref. No.: COL-14-101.R3 Date: 8-April-15 Stage: Page: 1

1 1 1 AQ4 Omar Gomez , * Esther Perales, Elısabet Chorro, Francisco J. Burgos,2 Valentın Viqueira,1 2 1 AQ1 Meritxell Vilaseca, Francisco M. Martınez-Verdu, Jaume Pujol2 AQ2 1Color & Vision Group, University of Alicante, Carretera De San Vicente Del Raspeig S/N 03690, Alicante, Spain

2Center for Sensors, Instruments and Systems Development (Cd6), Technical University of Catalonia (UPC), Rambla De Sant Nebridi, 10, Terrassa 08222, Spain

Received 22 October 2014; revised 31 March 2015; accepted 31 March 2015

Abstract: The interest in gonioapparent pigments (metal- correlation being worse. In particular, it was checked lic, pearlescent, interference, or diffractive) has increased that observers accepted a larger number of color pairs, in the last few years, especially for applications in the that is, the visual color difference was smaller than the automotive industry. To assure a proper characterization tolerance demanded by the industry (derived from of colors with gonioapparent pigments, commercial devi- AUDI2000). VC 2015 Wiley Periodicals, Inc. Col Res Appl, 00, 000– ces have appeared to characterize the color in different 000, 2015; Published Online 00 Month 2015 in Wiley Online Library geometries, which are called multiangle spectrophotome- (wileyonlinelibrary.com). DOI 10.1002/col.21964 ters. As the gonioapparent pigments and multiangle instruments are relatively new, no studies exist regarding Key words: vision; color; color measurement; perception the instrumental-based procedure followed in the indus- psychology; psychophysics; industrial inspection try, and if the results provided are in agreement with the observer perception. Consequently, the main objective of this study was to INTRODUCTION examine the correlation of the instrumental color differences with visual assessments. The instrumental color dif- The interest in gonioapparent pigments applied in many ference was calculated with the color difference formula industries (cosmetics, inks, etc.) continues increasing day AUDI2000 (speciﬁc for this sector) between the pairs of by day, especially in the automotive sector. This sector similar samples of three types of coated panels (solid, has undergone many changes and updates in the last 1,2 metallic, and pearlescent). The values measured by a tel- years from employing only paints without visual effect espectroradiometer in a directional lighting booth and to use many of them containing gonioapparent pigments the colorimetric values obtained by means of a multian- providing new and attractive visual effects by combining gle spectrophotometer BYK-mac were considered for this variable color and texture according to irradiation and purpose. Additionally, a visual experiment was conducted viewing directions. to quantify the color difference by using the gray-scale Goniochromatism is a change in any or all attributes of method. color of a specimen on change in angular illuminating– The results revealed that an acceptable instrumental cor- viewing conditions but without change in light source or 3 relation existed despite the visual and the instrumental observer. This happens when the material that is under evaluation includes gonioapparent pigments. Therefore, it is possible to classify the materials into three types *Correspondence to: Omar Gomez(e-mail:[email protected]) according to the pigment recipe and its colorimetric Contract grant sponsors: EMRP, EMRP participating countries within EUR- 4–6 AMET, the European Union. behavior : solid, metallic, and pearlescent or interference coatings (Fig.1, right). The solid color includes F1 VC 2015 Wiley Periodicals, Inc. only scattering pigments, and hence, the perceived color

Volume 00, Number 00, Month 2015 1

ID: thangaraj.n Time: 20:21 I Path: N:/3b2/COL#/Vol00000/150022/APPFile/JW-COL#150022 J_ID: COL Customer A_ID: COL21964 Cadmus Art: COL21964 Ed. Ref. No.: COL-14-101.R3 Date: 8-April-15 Stage: Page: 2

C O L O R Fig. 1. Left: Color appearance of the 60 assessed color pairs. Right: Colorimetric behavior according to the pigment recipe.

does not show signiﬁcant changes owing to the variation lighting booths.15–17 In other words, there are no studies of the direction of lighting and/or detection. The metallic to check if the procedure followed in the industry, mainly coating includes metallic pigments that provide a color based on the instrumental measures of a multiangle spec- shift mainly in lightness. The pearlescent coating includes trophotometer, would match the decision that an observer in its recipe pearlescent pigments (even solid pigments) would give.18 Therefore, the conventional approach using that provide a color shift mainly owing to a hue and a a diffuse lighting booth and an integrating sphere spec- chroma change. troradiometer is not valid for gonioapparent colors, and Currently, in the automotive sector, color quality of auto- hence it is necessary to use a commercial directional motive coatings with gonioapparent pigments is performed lighting booth such as the gonio-vision box from Merck by characterizing the color under different measurement or the byko-spectra effect cabinet from BYK-Gardner. geometries.5,7–10 For this, commercial devices, called mul- The purpose of this study is to study the correlation of tiangle spectrophotometers, are able to characterize the the visual and instrumental assessment of the color differ- color in different geometries, such as BYK-mac, or X-Rite ences between pairs of similar samples. The instrumental MA98, which satisfy ASTM and DIN standards.11–14 measurements were made with two different measurement Although multiangle spectrophotometers are commonly devices: BYK-mac multiangle spectrophotometer and PR- used in the automotive industry, there are few studies 650 telespectroradiometer. In addition, a psychophysical focused on the correlation between instrumental measure- experiment that included a directional lighting booth was ments and visual assessments by using rightly directional designed to obtain the visual assessments to quantify the

2 COLOR research and application

color difference by the comparison with a standard gray ci ʦ {15 ,...,110 }. To obtain a weighting for the mea- scale for color differences. surement angle of 110, we use half the value of the flop term from the angle, 75. For the negative effect angle 215, we specified to use the weighting for the 15 angle MATERIALS AND METHODS multiplied by a factor 1.2..20 A total set of 60 pairs of different samples from solid, In this study, the instrumental evaluation is carried out metallic, and pearlescent pigments were selected (Fig. 1, using two different strategies. On the one hand, the color left). For this study, knowing previously the pigment for- difference was calculated by using the measurement data mulation of this panels was not important though it could performed by the commercial multiangle spectrophotome- be interested in other future studies, because we only ter BYK-mac. This is a device that measures the samples focused in the optical and colorimetric behavior, not by contact in six different geometries. This methodology structural (pigments type, size, shape, etc.). Figure 1(a) is called “direct measurement,” and it is the usual proce- (left) shows the a*b* and L*C* chromatic diagrams with dure in the automotive sector. the representation of the 20 solid color pairs. The differ- On the other hand, a system consisting of the telespec- ent colors indicate the different measurement geometries, troradiometer PR-650, and the directional lighting booth, 21 and the different symbol, circle, and square, the two color the byko-effect spectra cabinet (from BYK-Gardner) samples of one color pairs. Figure 1(b) (left) shows the was used to obtain the real spectral color stimulus and representation at both chromatic diagrams for the metallic the corresponding colorimetric values of all samples with- color pairs. Finally, Fig. 1(c) (left) shows the pearlescent out contact, never used in the automotive sector, in spite color pairs. In this figure, it can be seen that the color of having directional lighting booths in this sector. The gamut associated with the selected color samples and the main advantage of this indirect system is that it collects behavior of each kind of pigment, solid, metallic, and exactly what the observer would perceive, the PR-650 has pearlescent. a telecentric lens, with a measuring area of 1 and the The color differences between pairs of samples were distance to the spot area was about 55 cm. For this rea- calculated by the color difference Formula (1) son, this methodology was called “(true) visual simu- AUDI2000,19,20 based on the CIELAB color space. This lation.” In addition, a reference white placed at the same color difference formula was developed by and is used by position as the samples was measured with the spectrora- AUDI, without being endorsed by any standardization diometer to allow transformations to CIELAB color space 2 body till now, with the main purpose to manage the color by means of the absolute XYZ values (in cd/m ), ena- tolerances varying depending on the specific application, bling us to apply absolute colorimetry into relative color- 22 paint batch acceptance, color matching for add-on parts imetry as it is usual in color appearance models. The in the car body, refinish, and so forth. byko-effect cabinet was selected owing to the fact that it AUDI2000 color difference formula was especially allowed measuring the same geometries of illumination/ designed for materials with gonioapparent pigments; then, observation that the BYK-mac. It is important to mention it considers the changes that these materials show depend- that the luminaire of the byko-effect spectra cabinet is a ing on the angle of illumination and observation (flop) as not a good D65 simulator if not it is closer to a D50 sim- 23,24 follows: ulator. We referred to both methodologies as direct measurement and visual simulation; however, it is impor- sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2 2 2 DLc DCc DHc tant to point out that both arise from different instru- DEc ¼ 1 1 (1) ments. But, obviously, the method described above can sDL;cgDL sDC;cgDC sDH;cgDH be applied to other directional lighting booths, with or not jL 2L j 2=3 the same nominal geometries common with other multi- ci ci11 sDL ¼ 10:002C 10:33 (2) angle spectrophotometers. ci c 2c 45 i11 i The measurement geometries used were fixed by the multiangle spectrophotometer and the lighting booth: ¼ sDLci;Flop1sDLci;Solid10:33 (3) 45as-15, 45as15, 45as25, 45as45, 45as75, and 45as110, jC 2C j ci ci11 written by following ASTM standards (Fig.2). These F2 sDC ¼ 1:478 10:014C 10:27 (4) ci 45 measurement geometries can be rewritten by following ci112ci the CIE nomenclature25 as: 45x:260,45x:230, ¼ sDCci;Flop1sDCci;Solid10:27 (5) 45 x:220 ,45x:0 ,45x:30 , and 45 x:65 , respectively. The illumination angle was constant for all the measure- jCc 2Cc j s ¼ 0:800 i i11 10:004C 10:30 (6) ment geometries. DHci 45 ci112ci For the visual evaluation, a psychophysical experiment was conducted by using a gray chart of the Society of ¼ sDHci 1sDHci 10:30 (7) ;Flop ;Solid Dyers and Colourists26 for qualifying the color difference Where “s” are the weighting functions and “g” are the perceived in the byko-spectra effect cabinet (Fig.3, left). F3 parametric factors for each term, lightness, chroma, and The gray chart used consisted of nine neutral gray chips hue, and c is referred to the measurement geometry, with pairs, which are ordered in increasing color difference.

Volume 00, Number 00, Month 2015 3

measurement geometries were randomly presented to the observers. Before each session, the observer spent 3 min to adapt to the illuminance level and cabinet environment. The observer’s task was to quantify the color difference in the test pair by comparing with the color differences shown in the several gray pairs (Fig. 3, left). The STRESS index (standardized residual sum of C squares) was used to evaluate the performance of the O color difference formula with respect to the visual assess- L 28,29 O ments obtained. As a consequence, the lower the R STRESS values, the better the correlation. In addition, the Fig. 2. Schematic representation of the illumination/mea- STRESS index was used not only to estimate the color surement geometries used with the BYK-mac device for difference performance but also to know how reliable the color characterization. results were by determining the intra- and interobserver variability.30 The intraobserver variability refers to the differences The gray scale was measured both with the BYK-mac for between the results obtained when the same observer the experiment called “direct measurement,” and with the reported more than once on the same issue. It was meas- PR650 for “visual simulation” experiment. To obtain true ured by the STRESS index, calculated from three repeti- values for the calibrated visual assessment, that is, the tions for each observer at the same experiment. The visual color difference DV which is the subjective answer interobserver variability refers to the differences between provided by the observers from the visual assessments, a the results produced by different observers. It was also transformation had to be done to the values that related to calculated by means of the STRESS index, considering the color differences calculated from the instrumental the total average of the 10 observers and the mean value measurements of the gray scale. The CIELAB color dif- of three replicates of each observer. ference (DE*ab) was selected for this purpose as it is tra- ditionally used for this transformation.27 The gray pairs RESULTS were measured at each angle, following the same methodology applied in the previous study,19 and a fourth poly- The psychophysical assessments and the instrumental data nomial fitting was found to be the best to adjust the allowed determining the instrumental and visual correla- values in the scale color differences obtained through tion regarding the experiment executed. instrumental measurements for each of these observation angles (Fig. 3, right). Thanks to this adjustment, the vis- Observer Variability ual color difference (DV) for a given angle can be inter- polated and directly related to DE*ab. For this The average intraobserver variability obtained in this experiment, 10 observers were included (five men and experiment was 23.65 STRESS units and the average five women) with normal color vision. In each 30-min interobserver variability was 28.25 STRESS units. session, the observer performed 120 evaluations, having a Although there were two observers who showed high total of 1080 evaluations done in nine sessions (three rep- intravariability, the calculated values were good compared etitions in each session). The color pairs and the different with the previous studies.19,31

C O L O R Fig. 3. Left: Which gray pair is closer to the color difference seen in the color pair? Right: CIELAB color differences for each one of the nine color pairs in the SDC gray scale. Fourth degree polynomial ﬁt, R2 (215) 5 0.9849, R2 (15) 5 0.9938, R2 (25) 5 0.9975, R2 (45) 5 0.9998, R2 (75) 5 1, and R2 (110) 5 1.

4 COLOR research and application

C O L O R Fig. 4. Instrumental correlation obtained from the color differences of the instruments for the geometries 45as-15, 45as15, 45as25, 45as45, 45as75, and 45as110. Different colors are used to identify the sample type (blue, solid; red, metallic; and green, pearlescent pigment).

Instrumental Correlation erated with both measuring instruments for each of the measurement geometries. In these conditions, a pair of Before analyzing the visual and instrumental correla- panels would exhibit similar color differences by using tion, an instrumental correlation between the two mea- the BYK-mac multiangle spectrophotometer (DE ) surement methodologies (direct measurement and visual BYK-mac and by using the PR-650 telespectroradiometer and the simulation) was done. The AUDI2000 color difference cabinet (DE ) although they were analyzed with the (DE ) was calculated by considering the measure- PR650 AUDI2000 BYK-mac and the telespectroradiometer PR-650 plus the ments from the BYK-mac multiangle spectrophotometer, lightning booth, respectively. In addition, the correlation fixating the D65 illuminant, and the measurement from would be linear in spite the fact that the method that the PR-650 telespectroradiometer from the true spectral involves the telespectroradimeter and the lightning booth stimuli, that is, the combination of the true lamp into the contains higher variability in the position/direction of the booth and the spectral reflectance of the panel for this instrument, orientation of panels, and the spectral quality measurement geometry. Next, a correlation of these of the daylight D65 simulation installed into the lighting instrumental data (BYK-mac and PR-650) was performed booth. by using a graphical analysis in which the values of Plots in Fig.4 show the correlations in terms of the F4 DE obtained in the different conditions were AUDI2000 color differences (DE and DE ) for the 60 compared. PR-650 BYC-mac color pairs classified by measurement geometry. There Ideally, a linear correlation of slope close to 1 in was a good correlation between both instruments and DEAUDI2000 values was expected. These values were gen-

Volume 00, Number 00, Month 2015 5

ID: thangaraj.n Time: 20:22 I Path: N:/3b2/COL#/Vol00000/150022/APPFile/JW-COL#150022 J_ID: COL Customer A_ID: COL21964 Cadmus Art: COL21964 Ed. Ref. No.: COL-14-101.R3 Date: 8-April-15 Stage: Page: 6

specular peak, the intensity of the reﬂected light decreased the goniochromatic effects, there was less spreading in characterizing the samples and the results of color difference matched for both procedures.

Visual and Instrumental Correlation The next step was the evaluation of the visual and C instrumental correlation. As it was mentioned before, the O L correlation was evaluated by the STRESS index. There- O fore, this index was calculated for both instrumental color R differences (BYK-mac and PR-650) by considering the Fig. 5. STRESS values calculated for both instrumental visual differences obtained from the psychophysical color differences. experiment. In Fig.5, a histogram is shown for the F5 STRESS values calculated for both color differences and methods although it was not perfect. The differences measurement geometry. As it can be seen, the results found might be owing to several factors. First, the fact were very similar for both color differences. The STRESS that the light source (fluorescent lamp) was not the same. average for all geometries was of 39.78 and 40.87 units First, although using the BYK-mac the reference illumi- for BYK-mac and PR-650 data, respectively. This result nant is the D65, the light source of the byko lighting was already expected as the correlation between both booth, both belonging to the same company, is closer to instruments was acceptable as shown previously. the D50 illuminant.30,31 Second, the illumination/observa- A set of color pairs was identified having high devia- tion conditions may be slightly altered between various tions between the visual assessment and the instrumental measurements, and this is especially relevant in gonio- measurement, even for both methodologies. These are chromatic samples where color strongly depends on the types of samples that would cause problems in the indus- measurement configuration. try. Therefore, the multiangle spectrophotometer would It is worth to point out that the correlation between not accept the color pair as a good color matching, but an both methodologies was poorer for the geometry 45as-15. observer would. Although the 45as-15 and 45as15 geometries were Furthermore, color pairs that generated conflicts, that is equally close to the specular direction, the geometry of the response of the instrument was different from the 45as-15 was experimentally the most conflicting as in this observer’s decision, were identified and selected. There- direction the intensity of the reflected light was higher fore, a new analysis was done to show for each color than in other more distant specular geometries. In general, pair, the correlation of both processes, instrumental and a great spreading was found for the geometries 45as-15 visual evaluation, in each measurement geometry. This and 45as110. In addition, some color pairs did not follow analysis consisted of doing a representation (Fig.6) by F6 a linear behavior as expected. As these measurement geo- denoting with zones the criteria established by the AUDI metries are the extreme ones in both instrumental config- color tolerances. In this way, it was easy to visualize if urations, this could be the reason why a worse correlation the samples presented acceptable color differences or not. is found, specifically owing to some optomechanical Based on the AUDI tolerance criteria, all samples with design problems for applying telespectroradiometry on a DE > 1.7 units would be placed within the red zone and tilted scene. However, for 45as25, 45as45, and 45as75 would be considered as different; samples with color dif- geometries which progressively move away from the ferences between 1.4 and 1.7 would be at the yellow

C O L O R Fig. 6. Color differences for all the measurement geometries for (a) solid sample, (b) pearlescent sample, and (c) metallic sample following the criterion Passes/Not Passed. The green region means almost indistinguishable color differences, the red zone means rejected color pairs, and the yellow zone means critical color differences; samples need to be visually assessed by expert colorists.

6 COLOR research and application

zone, where samples should be studied with other visual AUDI2000 color difference with the visual assessments tests to establish if they pass or not; finally, the green done by observers was evaluated by the STRESS index. area would contain samples with DE < 1.4, which could The results revealed that it would be possible to obtain a be considered almost indistinguishable. better performance by improving the color difference for- Once the most troubled samples were identified and mula as the observer variability was lower than the per- analyzed, three pairs of characteristic samples were formance of the color difference. A detailed analysis was selected as shown in Fig. 6. In this case, measurement done for each color pair. A number of color pairs were data from the BYK-mac were used as it is the most com- identified to have instrumentally color differences far mon instrument used in the automotive industry. away from visual color differences. It was determined Figure 6(a) shows the color differences for all the mea- that the general tendency was that observers gave lower surement geometries for a solid sample. As it can be color difference values than those provided by the devi- noticed, there was a good correlation between visual and ces. This implies that observers accepted a larger number instrumental assessments, which means that for both meth- of valid color pairs compared with those accepted by the odologies the color difference of the color pair was indis- devices, despite being “experts” in color although not tinguishable (green region). This situation should be the color automotive engineers. Therefore, it could be con- real one, where the information provided by the instrumen- cluded that there was not a perfect correlation between tal analysis and visual evaluation would be equivalent. instrumental and visual evaluation and that it would be Figures 6(b) and 6(c) show an opposite situation. These interesting to analyze which parameters can influence, examples correspond to pearlescent and metallic samples, and how, in this correlation failure related with commer- respectively. In the pearlescent sample, the color pair was cial color devices typically used in the automotive sector. accepted by the BYK-mac (for all the measurement geome- One approach would be to improve the AUDI2000 color tries) but rejected by the average observer. On the contrary, difference formula or to propose a new one to obtain a for the metallic sample the average observer accepted the better performance. Taking into account that the observer color pair as a good color matching or approval, whereas variability was lower than the performance of the color the color difference calculated from the BYK-mac meas- difference, it would be possible to improve the urements was big enough to consider the color pair as not AUDI2000 color difference formula to find a better corre- good or fail. Therefore, it would be interesting for car mak- lation with visual evaluations. Other strategies, even com- ers and coating providers to know what type of automotive bining with the first approach, would be to use a better colors or recipes (by hues, flops, etc.) are prone to be asso- D65 simulator, a more powerful lamp for a higher level ciated with the cases shown in these figures. of illumination, better lighting uniformity, or even use a It is important to mention that with the PR-650 data, the specified gray card with smaller steps to compare direc- results were similar to those obtained using the BYK-mac as tional level colors. This experiment should also be done in the majority of samples, pairs were accepted by observ- with professional colorists of the automotive industry and ers, but rejected by the instrument, as for the BYK-mac. with normal observers (with no experience in colorimetry), to see whether the level of experience affects DV. Therefore, this justifies that it is necessary to apply both methodologies, visual and instrumental evaluation, to CONCLUSIONS avoid this kind of discrepancies. Otherwise, working The correlation between visual and instrumental evalua- hardly in the improvement of directional lighting booths, tion was performed with a set of automotive samples. using better D65 simulators and high illuminance levels, Two instrumental evaluations were conducted, a direct either based on current fluorescent lamps or based on new measurement with the BYK-mac multiangle spectropho- solid-state lighting sources as LEDs, OLEDs, and so tometer, and a visual simulation with the PR-650 tele- forth. spectroradiometer and the byko-effect lighting cabinet, In conclusion, the new main contributions of this arti- with the same measurement geometries setup than in the cle is the use of telespectroradiometry for testing the vis- BYK-mac. The visual evaluation was conducted by a ual and instrumental correlation in automotive coatings at conventional psychophysical experiment where observers the realistic way, showing that this instrumental technique quantified the color difference in color pairs by using the for measuring the true spectral stimuli is necessary to gray-scale method. The observer variability showed that understand and manage doubtful color pairs for the final the observer responses were consistent enough. color matching/approval. It can be applied to new experi- First, a comparison between both instrumental evalua- ments and studies crossing structural (formulation, pig- tions was done. The results exhibited an acceptable corre- ment size, etc.) and colorimetric data. spondence between both methodologies. Nevertheless, a slight measurement mismatch that could be owing to the illuminant–light source differences or different geometries ACKNOWLEDGMENTS configurations was found. The authors thank the Ministry of Economy and Competi- Second, the study of the visual and instrumental corre- tiveness for the coordinated project “New developments in lation was carried out. The performance of the visual optics, vision and color technology” (DPI2011-

Volume 00, Number 00, Month 2015 7

30090-C02). Francisco Javier Burgos also thanks the 17. Burgos FJ, Perales E, Gomez O, Chorro E, Viqueira V, Martınez- Autonomous Government of Catalonia for his predoctoral Verdu FM, Pujol J. Instrumental and visual correlation between a fellowship grant and Omar Gomez to the Ministry of Econ- multiangle spectrophotometer and a directional lighting booth, in Proceedings AIC Colour 2013, Vol. 4. Newcastle, UK; 2013. p 1373– omy and Competitiveness for his predoctoral fellowship 1376. (FPI BES-2012-053080). 18. Longley WV. Automotive color certification. Color Res Appl 1995;20: 50–54. 1. Brock T, Groteklaes M, Mischke P. European Coatings Handbook. 19. Melgosa M, Martınez-Garcıa J, Gomez-Robledo L, Perales E, Hannover: Vincentz Network; 2010. Martınez-Verdu FM, Dauser T. Measuring color differences in automo- 2. Streitberger HJ, Dossel€ KF. Automotive Paints and Coatings, 2nd edi- tive samples with lightness flop: A test of the audi2000 color-difference tion. Wienheim: Wiley-VCH; 2008. formula. Opt Express 2014;22:3458–3467. 3. ASTM E284-13b. Standard Terminology of Appearance. Consho- 20. Dauser T. Audi Color Tolerance Formulas (Private communication). AQ3 hocken: American Society for Testing Materials. 2012. 4. Maile FJ, Pfaff G, Reynders P. Effect pigments—Past, present and 21. Experience the byko-spectra effect. Available at: https://www.youtube. future. Prog Org Coat 2005;54:150–163. com/watch?v=7j6oBkDKnjE. 5. Klein GA. Industrial Color Physics. New York: Springer; 2010. 22. Fairchild MD. Color Appearance Models, Published Online: 18 June 6. Pfaff G. Special Effect Pigments, 2nd edition. Hannover: Vincentz Net- 2013. DOI: 10.1002/9781118653128.ch20. work; 2008. 23. Perales E, Chorro E, Werner C, Viqueira V, Gomez O, Martınez-Verdu 7. Chorro E, Martınez-Verdu F, campos J, Pons A. Colorimetric and spec- FM. Influence of spectral power distribution of light sources on the tral evaluation of the optical anisotropy of metallic and pearlescent color appearance of goniochromatic colours. CIE Centenary Confer- samples. J Mod Opt 2009;56:1457–1465. ence, Paris; 2013. 8. Schanda J. Colorimetry: Understanding the CIE System. New York: 24. Martınez-Verdu F, Perales E, Viqueira V, Chorro E, Burgos FJ, Pujol John Wiley & Sons; 2007. J. Comparison of colorimetric features of some current lighting booths 9. Baba G, Suzuki K. Gonio-spectrophotometric analysis of white and for obtaining a right visual and instrumental correlation for goniochromatic reference materials. Anal Chim Acta 1999;380:173–182. apparent coatings and plastics. Proceedings of CIE 2012 Lighting Qual- 10. Kirchner EJJ, Ravi J. Setting tolerances on color and texture for auto- ity and Energy Efficiency. 2012. p 692–705. motive coatings. Color Res Appl 2012;39:88–98. 25. 25. Commission Internationale de l’Eclairage (CIE), Colorimetry, 3rd 11. DIN-6175-2. Tolerances for Automotive Paint. Part 2: Goniochromatic edition. CIE Publication 15:2004, Vienna: CIE Central Bureau, 2004. Paints. Berlin, Germany: Deutsches Institut fur€ Normung; 2001. 26. http://www.sdcenterprises.co.uk/products/sdc-assessment-aids/grey-scale/ 12. ASTM E2194-12. Standard Practice for Multiangle Color Measurement See also ISO 105–A02:1993. Test for colour fastness—Part A02: Gray of Metal Flake Pigmented Materials. Conshohocken: American Society scale for assessing change in colour. Geneva, Switzerland: International for Testing Materials. Organization for Standardization. 13. ASTM E2539-12. Standard Practice for Multiangle Color Measurement 27. Guan SS, Luo MR. Investigation of parametric effects using small col- of Interference Pigments. Conshohocken: American Society for Testing our differences. Color Res Appl 1999;24:331–343. Materials. 28. Garcıa PA, Huertas R, Melgosa M, Cui G. Measurement of the rela- 14. Perales E, Chorro E, Viqueira V, Martınez-Verdu FM. Reproducibility tionship between perceived and computed color differences. J Opt Soc comparison among multiangle spectrophotometers. Color Res Appl Am A 2007;24:1823–1829. 2013;38:160–167. 29. Kirchner E, Dekker N. Performance measures of color-difference equa- 15. Huang M, Chen G, Liu H, Cui G, Luo MR, Liao N, Melgosa M, tions: Correlation coefficient versus standardized residual sum of Zhang Y, Zheng C.A comparison of different psychophysical methods squares. J Opt Soc Am A 2011;28:1841–1848. for color-difference evaluation, in Proceedings AIC Colour 2013, New- 30. Melgosa M, Garcıa PA, Gomez-Robledo L, Shamey R, Hinks D, Cui castle, UK; 2013. G, Luo MR. Notes on the application of the standardized residual sum 16. Gomez O, Burgos FJ, Perales E, Chorro E, Viqueira V, Martınez- of squares index for the assessment of intra- and inter-observer variabil- Verdu FM, Pujol J. Preliminary comparative performance of the ity in color difference experiments. J Opt Soc Am A 2011;28:949–953. AUDI2000 and CIEDE2000 color difference formulas by visual assess- 31. Huang M, Liu H, Cui G, Luo MR, Melgosa M. Evaluation of threshold ments in directional lighting booth, in Proceedings AIC Colour 2013, color differences using printed samples. J Opt Soc Am A 2012;29: Vol. 4. Newcastle, UK; 2013. p 1545–1548. 883–891.

8 COLOR research and application

AQ1: Please confirm that all author names are OK and are set with first name first, surname last.

AQ2: Please provide division/department for the given afﬁliations.

AQ3: Please provide the year of publication for Refs. 3, 12, 13.

AQ4: Please conﬁrm that given names (red) and surnames/family names (green) have been identiﬁed correctly.

ID: thangaraj.n Time: 20:22 I Path: N:/3b2/COL#/Vol00000/150022/APPFile/JW-COL#150022