Texture Measurement, Modeling, and Computer Graphics
Following a definition by ASTM, texture can be defined as the visible surface structure depending on the size and organization of small constituent parts of a material, typically surface structure of a woven fabric. Many other materials that are used in the manufacturing industry also show texture, often of very different types. Examples are various types of wood, satin, paper, foods, and injection-molded plastics. This review focuses on effect coatings, since the characterization of their texture is advanced as compared to the case for most other materials. Also, effect coatings such as metallic and pearlescent coatings are widely used in the automotive, cosmetics, and packaging industry.
Some other parameters for describing the texture of effect coatings have been introduced as well, and some of them will be discussed further below. However, those parameters generally lack strict definitions.
Conventional colorimetric results such as the modern color difference equations assume that the so-called reference conditions hold. The Commission Internationale de l’Éclairage (CIE) recommended the investigation of nonreference conditions. According to one of the reference conditions, samples should be spatially uniform. Since effect coatings clearly do have a texture, their industrial importance explains the large body of research that has been conducted over the past decade on the texture of effect coatings.
Below, the progress is summarized that has been made on different subtopics in this area. Only recently, it became possible to objectively measure the texture of effect coatings. These developments are highlighted, since they also enabled the first quantitative studies on the influence of texture on perceived color differences between effect coatings. Because of these developments, it now has become possible to set tolerances on texture parameters. For color formulation and recipe prediction, models are needed that are able to predict texture values based on concentrations of colorants, similar to, e.g., the Kubelka-Munk theory for color prediction. Finally, it is discussed how the appearance of texture can be accounted for in computer graphics and 3D rendering of objects.
Measurement and Observation of Texture
As long as no clear definitions of texture parameters were introduced and an understanding of the relation between illumination conditions and the type of texture of effect coatings was lacking, no commercial instrument for measuring texture could be developed. After having derived clear definitions and constraints on illumination conditions, a large amount of visual data was collected on different types of texture: diffuse coarseness for diffuse illumination conditions and glint impression for intense directional lighting. This body of data was obtained for hundreds of different effect coatings observed under different angles and illumination conditions. The visual data was correlated with texture parameters that were obtained from image analysis applied on CCD images of the same samples. In this way, algorithms were obtained to train and test prototypes of an instrument [1, 2].
The resulting instrument is the BYK-mac™, which is commercially available from BYK-Gardner since 2007. Similar studies are being conducted by other manufacturers of optical instruments as well , whereas also the development of the underlying image analysis algorithms has been taken up by the academic world .
Based on feedback from the automotive OEM industry, the set of texture parameters considered to be best applicable for the OEM industry has gradually evolved over the years. A graininess parameter Gr was introduced that has a high correlation with the diffuse coarseness parameter mentioned before. Apart from that, glint impression (sparkle) is considered to have two separate dimensions, called sparkle intensity Si and sparkle area Sa . An overall parameter sparkle grade (SG) is no longer thought to be important.
Recently, also lighting booths became available that focus on observing the different types of texture under various geometries .
Influence of Texture on Color Perception
Using artificially generated images of textures, several studies showed that the presence of texture has an influence on perceived color differences . Images exhibiting textured lightness variations have been shown to increase the visual tolerance for lightness deviations. This explains why, for effect coatings, a color difference equation like dECMC is best used with l = 1.5; c = 1.0; for solid (untextured) colors, best results are typically obtained for l = 1.5; c = 1.0.
However, quantitative studies on effect coatings only became possible after 2007 as outlined in the previous section. Based on a large study on 405 effect sample pairs, an expression was derived that quantitatively describes how differences in texture and color combine into a total appearance difference. The validity of this approach was confirmed in later research .
Setting Tolerances on Texture
Being able for the first time to measure texture parameters of effect coatings, manufacturing industries decided to include consistency in texture in their criteria for quality control. Among the first industries to investigate how to set industrial pass-fail tolerances on texture are the automotive industry and the plastics industry, where texture matching used to be a long-standing problem for quality control .
Most of the global automotive manufacturers are currently setting texture tolerances, as evidenced, for example, by multiparty meetings of the Detroit Color Council in the USA and BYK-Gardner user meetings in Europe. However, no specific tolerances on texture parameters have been published yet.
Based on perception thresholds of texture parameters, a proposal for realistic tolerances was recently published for different sets of texture parameters: diffuse coarseness, glint impression, graininess, and sparkle grade . Only for the parameters sparkle area and sparkle intensity, the lack of published data makes it not yet possible to derive tolerance values.
Predictive Models for Texture
For color matching, optical models are widely used to optimize color formulations. In the 1940s, Kubelka and Munk developed a model that is still widely used in the paint and textile industry. In order to optimize the concentrations of colorants such that also the texture of an effect coating is matched, a similar model for texture is needed. Studies show that texture parameters like sparkle intensity and graininess can vary greatly with paint composition . However, an accurate optical model for predicting texture based on a physical model like Kubelka-Munk has not been developed yet. The first example of an accurate model for texture uses a very different derivation . It is based on a statistical approach for finding those mathematical terms that contribute significantly to the texture properties that are to be predicted.
Rendering of Texture in Computer Graphics
Digital rendering of three-dimensional objects has become very common in the cinematic and computer game industries. However, for accurate rendering of effect coatings also, the texture aspect needs to be taken into account. Most current approaches either completely ignore texture aspects or utilize a very crude and nonquantitative approach by adding sparkles to the image afterward .
Recently, two different approaches were published that do make accurate texture rendering possible. Using a few measured texture values as input as well as reflection measurements from an ordinary multiangle spectrophotometer, it was shown that accurate renderings can be produced . No full BRDF and BTF measurements are needed. The core algorithm was derived from procedures used for determining texture parameter values based on images captured from effect coatings. By reversing these procedures, an algorithm was obtained that produces images resembling effect coatings, based on measured texture parameter values. The algorithm does not aim at simulating the reflection from each and every flake in a scene, but it accounts for their combined effect in an effective way.
In contrast to conventional methods for rendering, the derivation of the algorithms and the verification of the accuracy of the visualizations were both determined using visual tests in which rendered images were directly compared to physical objects. Thus, a truly appearance-based method for rendering color and texture was obtained, with greatly reduced demands on computation power and disk storage.
A different approach was published very recently by Ferrero et al. . It uses an analytical model to visualize flake reflectance, in which, for example, the aperture angle of the light source and the diameter of the eye pupil of the observer are used as input parameters. In this way, a model is obtained that is able to generate images varying from purely directional until purely diffuse lighting and any illumination condition in between.
Despite all the work that has been done on texture research over the past decade, this area is still in development. There is a clear need for more accurate image analysis algorithms to extract parameters from CCD images that correlate well with visually perceived texture. For manufacturers of optical instruments, this is an absolute necessity for developing instruments that can form a next generation after the commercial launch of the BYK-mac™ in 2007.
Perception studies on the effect of texture on perceived color, or on perceived color differences, is also expected to be an active area of research for the coming decade . Currently, color difference equations are used that are derived from very different types of samples, such as solid colors or textiles. At present, texture can only be accounted for through parametric factors.
Setting tolerances on texture parameters and predicting texture properties based on paint composition are both areas of great industrial importance. Yet, these areas have hardly been explored yet.
For accurate rendering of textured paints, there are a few first publications, but also here the potential for improvement is enormous, and there are many obvious industrial applications in, e.g., the movie and the game maker industry.
- 1.Kitaguchi, S., Westland, S., Owens, H., Luo, M.R., Pointer, M.R.: Surface texture – a review. NPL Report DQL-OR 006 (2004); Kirchner, E.J.J., van den Kieboom, G.J., Njo, S.L., Supèr, R., Gottenbos, R.: The appearance of metallic and pearlescent materials. Col. Res. Appl. 32, 256–266 (2007)Google Scholar
- 2.Kitaguchi, S., Westland, S., Kirchner, E.J.J., van den Kieboom G.J., Luo, M.R.: Computational model for perceptual coarseness prediction. In: Proceedings Conference on Graphics, Imaging and Vision, pp. 267–282 (2006); Huang, Z., Xu, H., Luo, M.R.: Camera-based model to predict the total difference between effect coatings under directional illumination, Leeds (UK), Chin. Opt. Lett. 9, 093301–093305 (2011)Google Scholar
- 3.Kigle-Böckler, G.,Weixel, S.: New measurement system for characterizing the total color impression of effect coatings. In: Proceeding Conference of AIC, Newcastle (UK), pp. 209–212 (2013); Ellens, M.S., Lamy, F.: From color to appearance in the real world. In: Proceeding of Human Vision and Electronic Imaging XVII, SPIE, Burlingame (USA), vol. 8291 (2012)Google Scholar
- 4.Klein, G.A.: Industrial Color Physics. Springer, New York (2010); Kigle-Böckler, G.: New and innovative testing technologies for effect finishes. In: Proceeding Conference of AIC, Sydney (2009)Google Scholar
- 5.Cramer, W.R.: Die neue Version der Gonio-Vision Box. Phänomen Farbe, 38–39 (2012); Martínez-Verdú, F., Perales, E., Viqueira, V., Chorro, E., Burgos, F.J., Pujol, J.: Comparison of colorimetric features of some current lighting booths for obtaining a right visual and instrumental correlation for gonio-apparent coatings and plastics. In: Proceeding of CIE x037:2012 Lighting Quality and Energy Efficiency, Hangzhou (2012)Google Scholar
- 6.Huertas, R., Melgosa, M., Hita, E.: Influence of random-dot textures on perception of suprathreshold color differences. J. Opt. Soc. Am. A 23, 2067–2076 (2006); Xin, J.H., Shen, H.L., Lam, C.C.: Investigation of texture effect on visual colour difference evaluation. Col. Res. Appl. 30, 341–347 (2005); Montag, E.D., Berns, R.S.: Lightness dependencies and the effect of texture on suprathreshold lightness tolerances. Col. Res. Appl. 25, 241–249 (2000)Google Scholar
- 7.Dekker, N., Kirchner, E.J.J., Supèr, R., van den Kieboom, G.J., Gottenbos, R.: Total appearance differences for metallic and pearlescent materials: contributions from color and texture. Col. Res. Appl. 36, 4–14 (2011); Huang, Z., Xu, H., Luo, M.R., Cui, G., Feng, H.: Assessing total differences for effective samples having variations in color, coarseness and glint. Chin. Opt. Lett. 8, 717–720 (2010)Google Scholar
- 8.Longley W.V.: Automotive color certification. Col. Res. Appl. 20, 50–54 (1995); Ariño, I., Johansson, S., Kleist, U., Liljenström-Leander, E., Rigdahl, M.: The effect of texture on the pass/fail colour tolerances of injection-molded plastics. Col. Res. Appl. 32, 47–54 (2007)Google Scholar
- 10.Rentschler, T.: Measuring sparkling blues without blues. Eur. Coat. J. 11, 78–83 (2011)Google Scholar
- 11.E. Kirchner, E.J.J., Ravi, J.: Predicting and measuring the perceived texture of car paints. In: Proceeding 3rd International Conference on Appearance. “Predicting Perceptions”, Edinburgh (UK) (2012)Google Scholar
- 12.Meyer, G., Shimizu, C., Eggly, A., Fischer, D., King, J., Rodrigues, A.: Computer aided design of automotive finishes. In: Proceeding of AIC Colour, pp. 685–688 (2005); Ershov, S., Kolchin, K. and Myszkowski, K.: Rendering pearlescent appearance based on paint-composition modeling. Eurographics 2001, Granada (Spain), Comp. Graph. For. 20, C221-C238 (2001)Google Scholar
- 13.Van der Lans, I.B.N., Kirchner, E., Half, A.: Accurate appearance-based visualization of car paints. In: Proceedings of the European Conference on Colour in Graphics, Imaging and Vision, Amsterdam, pp. 17–23 (2012)Google Scholar
- 15.Kandi, S.G., Tehran, M.A.: Investigating the effect of texture on the performance of color difference formulae. Col. Res. Appl. 35, 94–100 (2010)Google Scholar