Computational Aesthetics: On the Complexity of Computer-Generated Paintings

Abstract

Ma ny argue that with the rapid advances of modern technology, particularly digital display and computer graphics, digital art is more expressive than traditional visual art. This increased expressive power, compounded with highly advanced artifi cial intelligence, has created tremendous opportunities for the realization of computational aesthetics and even simulated creativity [1]. If the potential of computational aesthetics is achieved, we would see a profound impact on various application domains where computer technology has traditionally played only an assistive role, such as graphic and industrial design. This is not to advocate the possibility of replacing human creativity with computational creativity; rather, the advancement of computational aesthetics would further extend human creativity by providing inspiration to artists and graphic/industrial designers. I. emergIng new dIsCIplInes In Arts And ComputIngWhen combining visual art with digital technology, two lines of questioning tend to emerge. On the one hand are questions such as: "How can the computer automatically generate various forms of visually aesthetic expressions?" On the other hand, some ask: "How can the theory and techniques of traditional visual art help beautify modern technology outputs and products and enhance their usability?" Addressing such questions, two interdisciplinary areas have emerged in recent years: computational aesthetics and aesthetic computing [2]. Both computational aesthetics and aesthetic computing aim at bridging computer science, philosophy, cognitive science and the fi ne arts through analytic and synthetic investigation. Figure 1 illustrates a conceptual map of the relationships between various computer science areas (in rectangular nodes) and arts (in oval nodes), moving from theories to applications. The dotted arrows in Fig. 1 help to conceptualize that computational aesthetics aims at answering the fi rst question above, that is, "How can the computer automatically generate various forms of visually aesthetic expressions?" In other words, computational aesthetics investigates how modern technology helps the arts. This technology in fact serves to create tools that can enhance the expressive power of visual art and heighten human understanding of aesthetic evaluation, perception and meaning. In the reverse direction, illustrated by the dashed arrows in Fig. 1, aesthetic computing addresses the second question-"How can the theory and techniques in the traditional visual arts help beautify modern technology outputs and products and enhance their usability?" This issue includes the aesthetic design of computer algorithms, simulation, visualization, human-machine interfaces and high-tech products, so that users are highly engaged and thus usability is enhanced. An interesting example of aesthetic computing is the application of Kandinsky's aesthetics to Java programming [3]. In 2006, Malina [4] highlighted the aesthetic computing activities published in Leonardo over the previous 40 years.We have previously ex...

A program for Victory Boogie Woogie

“…Zhang et al [29] argue that digital art is more expressive than traditional visual art and creates opportunities for computational aesthetics, which is quite distinct from my claims which are that programming is a tool for better understanding and for developing vocabulary. Their blueprint for generative art is to have an arrangement of primitive shapes, which forms an initial aesthetics, and then disrupt it with randomized perturbations.…”

Generation of Kandinsky Art

“…Such an approach would assist and extend designers' creativity through machine's intelligence and automation, leading to the generation of more sophisticated aesthetic forms at Level 4 [4]. The analysis of Kandinsky aesthetics would certainly benefit the future design of information visualization [14].…”

“…Zhang et al [4] made an initial attempt in automatic generation of Kandinsky abstraction paintings, which included very few styled patterns and generated uniform backgrounds and no sophisticated combination of geometric objects. We have since conducted an in-depth and comprehensive experiment on generating several styles of Kandinsky's paintings.…”

“…One of the most successful computer generated art forms has been the fractal art [3]. According to a recently proposed 4-level classification in terms of computational power being utilized [4], fractal art is considered Level 2. The classification measures computer generated abstract paintings by their computational complexity, i.e.…”

Generating abstract paintings in Kandinsky style

“…One of the most successful computer generated art forms has been the fractal art [Barnsley 1993]. According to a recently proposed 4-level classification in terms of computational power being utilized [Zhang et al 2012], fractal art is considered Level 2. The classification measures computer generated abstract paintings by their computational complexity, i.e.…”