T e c h n i c a l a r T i c l e Affective Quantification of Color Gestalt Modeling of Affective Factors for Combinative Color Design G u o S h e n G h u Integrated combination of multiple colors as it is, color gestalt invariably is largely infeasible, especially in regard to affective factors troubles researchers as they seek to understand and analyze the of multicolor relations. In general, there are three perplex- complex constructions of affective factors. The author proposes a set of ing obstacles to current research on multicolor gestalt: in- quantification methods for measuring color-affective factors of harmony, ABSTRACT preference and structural composition. He also describes quantified terdisciplinary discrepancy (i.e. the differences between two models and interactive toolkits he built for both experimental purposes given disciplines in terms of tools, approaches, etc.), the lack and user study. The author proposes a holistic method for understanding of a method for quantifying color combination and single- color combination that stands in contrast to conventional methods. color focus (the tendency of researchers to take the colors Through this method, perception of color gestalt becomes computable as isolated objects when studying color affective factors in and analyzable, inviting further research on interactive color computing multicolor combinations). and parametric design. • Interdisciplinary discrepancy. This discrepancy occurs when computers are brought into color Color design generally involves interlacing colors into com- research and application. Such a discrepancy is due plex patterns, and both creating and analyzing color designs to differences in the ways artists and computer engi- can be challenging. The termcolor gestalt, which refers to the neers define and analyze color. Generally speaking, integrity of multicolor design, generally hints at subtle in- artists are accustomed to grasping and judging colors terrelations of both color components and structural modes as a whole (color gestalt), while computer scientists that determine how well the colors combine together [1]. tend to deduce parameterized relations from color Although there is clearly a conceptual relationship be- appearances. Artists may describe a color gestalt us- tween color gestalt and Chevreul’s color principles of simul- ing metaphorical expressions, such as “sparkling,” taneous contrast and harmony, the term gestalt as used here “vivid,” “elegant,” “gloomy” and so on. Computer stems from gestalt psychology of cognition of the early years scientists may intuitively weigh the individual factors of the twentieth century [2]. Gestalt means that humans see that determine the overall color features. This dis- objects with all of their elements taken together as a global crepancy leads to obstacles for interchange between construct, e.g. they see the overall interrelations between the the instinctive synesthesia of artists and the pro- figure in the foreground and the background, not only the gramming languages used by computer engineers— individual component parts. computers cannot interpret the affective feeling of It is difficult to measure and analyze color gestalts because artists merely through synesthesia and metaphors. of the complicated interactions of their affective factors. For This discrepancy is a common problem for both this reason, even though color researchers have studied af- artists and computer scientists in metaphorical- fective factors for nearly two centuries, unquantified general- mathematical mapping. ization still dominates their descriptions of color principles. • Color standardization. In color systems like Mun- Without quantifiable methods, one cannot reliably restore sell’s, Ostwald’s, CIE, etc., colors are formatted as or reconstruct color gestalt by managing color combination specific values of elements. The digitized element rules. To this point, color research on computer platforms values allow color exchange between different systems. To some extent, color standardization implements color quantification and exchange in Guosheng Hu (artist, educator), China Academy of Art, Hangzhou, Zhejiang, China. Email: [email protected]. pigment mixing, color lighting and computer color See www.mitpressjournals.org/toc/leon/52/2 for supplemental files associated simulation. This constitutes progress over previous with this issue. conceptual systems, such as Goethe’s color wheel and ©2019 ISAST doi:10.1162/LEON_a_01197 LEONARDO, Vol. 52, No. 2, pp. 157–163, 2019 157 Downloaded from http://www.mitpressjournals.org/doi/pdf/10.1162/leon_a_01197 by guest on 25 September 2021 Runge’s color sphere; however, these systems do little ModelinG of ColoR Harmony PRinCiPleS to quantify color-affective factors and expression ef- Color harmony is pleasing affective or aesthetic content fects. Standardized color systems made progress in that is produced by colors when combined together [6]. Re- color digitalization but did nothing to further under- searchers after Goethe proposed many ways to generalize standing and quantifying of color gestalt as integral the principles of color harmony. Ostwald insisted that color expression. harmony lies in ordered relations within elements, i.e. that • Single-color focus. In view of perceptual wholeness, the constitutive elements of harmonious colors are identical color gestalt lies in the interactions of each compo- to chromatic spans, such as gradation, isotones, isotints and nent in sight. In contrast, traditional color researchers so on [7,8]. Itten proposed color harmony principles based have usually taken colors as isolated objects, seldom on a series of graphic experiments, and he explained these perceiving color combinations as integral gestalts principles using the metaphor of musical harmony [9]. Moon and leaving interrelations of color components un- and Spencer divided the principles of harmony into catego- touched. Affective description and evaluation of color ries of similarity, ambiguity, contrast and glare, in accordance gestalt has remained a scarcely cultivated territory up with classical color harmony theory [10]. They also applied to this point. Meanwhile, the concept of color in an Birkhoff’s aesthetic measure M = O / C (where M stands for art context is more likely to refer to multicolor com- value of aesthetic measure; O for order as identity, similarity, bination than to single-color stimuli. Previous artistic contrast and area balance; and C stands for complexity as investigations under gestalt theory have been unani- summation of color amount and pairs of colors with element mously based on multicolor compositions [3–5]. In difference) [11]. Pieters constructed a similar measure based short, isolated colors make no sense in an art context on his own experimental results: Color Harmony = Satura- unless they are organized in certain interrelations that tion Harmony × (Hue Harmony + Lightness Harmony) [12]. form an integral object. Hård introduced a theoretical model for color harmony in which he divided harmony factors into three categories: color For these reasons, color researchers should seek methods interval, color chord and color tuning [13]. of affective factor quantification that can reconstruct and -ma However, all the aforementioned methods emerge from nipulate color gestalts. Each affective factor of color gestalt conceptual generalizations rather than computable mea- should be quantified and modeled for computing purposes. sures. Moreover, many existing methods are confined to However, it is not enough to merely quantify individual af- harmony principles of hue, leaving other element dimen- fective factors; composing interrelations between each factor sions—say, saturation or lightness—unaddressed. Well- is also essential. Though it is complicated and challenging to rounded harmony systems should be built on these all these do so, the integrity of different affective factors and the inter- factors together. Taking hue, saturation and lightness into relations of each should be considered, too. Another matter consideration, I built an integral model of harmony princi- is the diversity of individual observers. Although common ples by combining familial elements and rhythmic contrasts. sense exists in color perception, personal differences in re- Lenclos proposed the concept of familial colors to refer to sponse to color combinations can still be remarkable due those colors with identical elements such as hue, saturation to the observers’ personal instincts or cultural backgrounds. or lightness, because color combinations with corresponding To avoid the interference of these subtle factors, researchers elements tend to be harmonious [14]. In fact, Oswald had should make sure that each item under survey is confined already proposed a similar idea, but his work emphasized the in specific scope in order to isolate the affective factor from principles of contrasting elements, the opposite of familial irrelevant subjects. The abstract organization of affective fac- elements [15]. According to familial color theory, one can tors can then be restored and represented in a quantified set one or two of the three elements—i.e. hue, saturation or data model. lightness—to be identical for all the colors to make a color The main concept of my method consists of classifying combination harmonious. However, if all three elements are affective factors and constructing quantified models for each set in accordance with the familial rule, colors will become individual factor.