Code and Generative Art
Context Machines: A series of situated, outward-looking, self-organizing and generative artworks
by Ben Bogart and Philippe Pasquier
Abstract “Context Machines” (CMs) are a family of site-specific, conceptual and electronic media artworks that capture photographic images from their environment in the construction of generative compositions. These artworks are produced in the context of meta-creation, where artworks are systems constructed in order to exhibit creative behaviour. Their production began with a central question: Could a machine be constructed that found its own relationship to it's physical context, without the artist predetermining that relation?
Abstract “Resurfacing” was the first outward looking installation produced. The system captures images of the environment, at multiple moments in time, to produce interactive temporal landscapes. This project is a precursor to the CMs that follow, it was not intended to find a relation to its environment, and was not produced in the context of meta-creation.
Abstract “Memory Association Machine” integrates photographic images of its environment into an organized structure. This process is enabled by an artificial intelligence inspired by a model of human memory. The system free-associates through this structure as initiated by the most recent captured image. These free-associations are framed as the creative actions of the machine, and are meant to situate it in the physical world shared with the viewer. The process of free-association is enabled by a model of creativity as proposed by L. M. Gabora.
Abstract In “Dreaming Machine” these free-associations are framed as machine dreams. This interpretation of the work takes the naïve view that dreaming is a result of random activation in the brain, one conception of dreaming as proposed by Hobson, and therefore analogous to the concept of free-association.
Abstract The method of memory integration used in “Memory Association Machine” and “Dreaming Machine” is applied to thousands of prerecorded images in “Self-Organized Landscapes”. These landscapes are high resolution and intended for large-scale print reproduction.
Abstract CMs are produced at the intersection between art production, computer and cognitive science. Their application of cognitive models of memory, creativity, dreaming, and perception invite us to reconsider what is essentially human, how we relate to machines, and to look at ourselves anew.
Generative narrative in xTNZ and Senhora da Graça
by Rui Filipe Antunes and Frederic Fol Leymarie
We will look into the artistic practice of computational ecosystems, attempting to see behind its formal novelty and invention. Computational ecosystems operate in a distinct model of narrative. The concept of generative narrative
discussed here illustrates what Umberto Eco denes as 'open work'. First published in Italian in 1962, Eco's critical book Open Work, addresses questions which are found (also) transversely in Generative Art. Modern music, literature and art are said to operate in a state of potential, of unexplored possibilities which the work may admit. The open-ended nature of the works offer an unlimited range of possible readings, works are 'open' to continuous generation of internal relations, which the addressee must uncover and select in his act of perceiving the totality of incoming stimuli.
Generative narrative, is a concept we can nd in electronic literature. We extend this concept discussing it within the framework of the computational ecosystems. We will look into Lizbeth Klastrup's concepts of 'multi-user digital textuality' and 'interpretative framework' to assist in this project and understand how the material aspects of code participate in the narrative processes.
To illustrate these ideas, we discuss two case studies, xTNZ and Senhora da Graça, two computational ecosystems, where, we argue, this model of narrative conveys context and artistic meaning to the works.
Procedural Taxonomy: An Analytical Model for Artificial Aesthetics
by Miguel Carvalhais
This paper discusses an analytical model for the study of computational aesthetic artifacts. This work is motivated by the growing ubiquity of computational media, by the study of how remediation and procedurality transform the media, and by the understanding of the creative potential and uniqueness of computational tools. It also recognizes the need to define and establish a common terminology for all those that interact with these systems, either as consumers, producers, critics, educators, historians, etc.
The starting point to this work is Espen Aarseth's typology proposed in “Cybertext: Perspectives on Ergodic Literature” (1997), defining seven variables and their eighteen possible values. We studied its adequacy for the analysis of ergodic visual and audiovisual pieces and adapted it with new variables or possible values, tailoring it to this broader field. We tested the new model in samples representative of diverse approaches to procedural art, design and other contemporary clusters of creative activity and aesthetic communication and we developed a control analysis, in order to assert the usability and usefulness of the model, its capacity for objective classification and the rigor of our analysis.
We demonstrated the partial adequacy of Aarseth's model for the study of artifacts beyond text-based systems, and expanded it to better suit the objects in study and circumvent its shortcomings. The new model produces a good description of the pieces, clustering them logically, reflecting stylistic and procedural affinities that probably wouldn’t be found if the study was solely focused in their physical, sensorial or superficial structures and in the established aesthetic analyses that can be developed from them, and for which we already have well-established resources. The similitudes revealed by this model are structural and procedural, they attest to the importance of computational characteristics in the aesthetic enjoyment of the works and to the weight of procedurality, both as conceptual grounding and as aesthetic focus, as an aesthetic pleasure in itself.
Painting as Programming: Casey Reas and the Aesthetics of Generative Code
by Meredith Hoy
Software art has been frequently historicized in terms of the self-reflexive, philosophical, anti-aesthetic principles of conceptual art. Generative art is critiqued as being overly aestheticized, all about surface insofar as it conceals its algorithmic operations instead of bringing them to the forefront, as is the case in software art . In this paper, I consider the work of Casey Reas, an artist whose output hovers in an indeterminate range of picture-making between painting and computing, between surface and coded process. I propose that the seemingly insurmountable division between the conceptual and the aesthetic is not absolute , and argue that Reas’s practice, one that is fundamentally tied to the specific functions made possible by a digital computer, but whose designs exhibit a markedly painterly morphology, begins to offer an alternative to this artificial distinction. Not only does Reas up the ante of conceptual art by translating linguistic programs into a computationally executable form, but he also brings the emphasis back to the visual register. Reas’s programs, in all their visual richness, refuse the “dematerialized” anti-ocular ethos of 1960’s and 70’s conceptual art. Moreover, his desire to “minimize the technical aspects” of his work and to make working with software as fluid as drawing suggests a resistance to the determinate properties of computational media. Reas’s production hybridizes the analog and the digital; one the one hand, he is deeply invested in a computational, proceduralist logic, and on the other, the richness, ambiguity, and organic harmony of his output revel in painterly abstraction. Computers, taken both within and outside of an art context, are tools constructed to facilitate regular, repeatable, and efficient calculations. However, they are also tools that allow dynamic change to occur within their programmatic constraints. Reas addresses the question of what happens to information when it becomes painterly, and when that painterly quality becomes dynamic. Information becomes a living organism whose transformational capacities are revealed in the performative execution of code.
Aesthetic Agents: A Multi-agent System for Non-photorealistic Rendering with Multiple Images
by Justin Love, Philippe Pasquier, Brian Wyvill, George Tzanetakis, and Steve Gibson
The creation of expressive styles for digital art is one of the primary goals in non-photorealistic rendering (NPR). In this paper, we introduce a swarm-based multi-agent system that is capable of producing expressive imagery through the use of multiple bitmap images. At birth, agents in our system are assigned a digital image that represents their aesthetic ideal. As agents move throughout a digital canvas they try to realize their ideal by modifying the pixels in the digital canvas to be closer to the pixels in their aesthetic ideal. When agents with different aesthetic ideals occupy the same canvas, a new image is created through the convergence of their conflicting aesthetic goals. By varying the inputs and parameters of our multi-agent system we were able to emulate a variety of painterly styles including Impressionism, Futurism, Cubism and Montage. The ease of implementation and variety of results created through minor changes in variables and inputs makes a compelling argument for more research using multi-agent systems for NPR.