Synthetic Images: Databased Aesthetics
Rosemary Lee2023, Live Interfaces
Sign up for access to the world's latest research
checkGet notified about relevant papers
checkSave papers to use in your research
checkJoin the discussion with peers
checkTrack your impact
Abstract
This essay examines how data-based practices contribute to new perspectives on the empirical value of images. Recent methods employing machine learning enable visualisations to be produced based on the large-scale analysis of data but that are detached from direct sensorial observation, subverting the forms of visual objectivity traditionally associated with technical and scientific methods of image-making. This research aims to develop insights into the forms of visual knowledge that these methods may give rise to, as well as facilitating critical discourse on the grounding of visual practices in relation to technical and scientific methods.
Figures (3)
Related papers
Machine Learning and Notions of the Image
2020
This thesis addresses how current notions of image production remain tied to historical ideas which often prove inadequate for the description of visual artefacts of machine learning (ML). ML refers to the notion of simulating the process of information acquisition computationally, and when applied to the generation of images, it enables visual content to be influenced based on the statistical analysis of data. The increasing use of ML in image production highlights several aspects which have been present in older forms of media, but which now take on new forms and relevance, especially within artistic contexts. This research seeks to clarify the mediating role played by visual technologies and to demonstrate how images produced using ML offer new ways of approaching theories of the image. Images exist at the interstices between human perceptual experience and its technological mediation, which is especially relevant as the development and implementation of technologies offers new possibilities to produce visualisations from data. In so doing, technological mediation tangibly augments relationships between how images are produced, experienced and interpreted. The present incorporation of ML into various forms of visual media offers insight into this issue by enabling images to be produced as the result of the statistical analysis of datasets. Computational relations which are extracted and inferred between features within images help to construct learned representations which are in turn used to generate new images. This results in a form of computationally-determined representation which is informed by the interpretive processes performed by machines. Artists have taken great interest in the potential of ML, in an aesthetic, but also a processual capacity, often considering its relation to human vision. Their productions offer insight into novel aspects of ML in the creation of images through experimental practice which is informed by theory and by art history. Using and reflecting on ML, often in novel or reactionary ways, artistic and humanistic perspectives provide vital counter-narratives to those of computer science (CS), and which facilitate cross-disciplinary understanding.
Art as Data Set, Data Set as Art: Training machine vision systems to see as artists
Electronic Workshops in Computing
The process of an artist 'speaking' to an audience by illustrating meaning underlying a body of work on a subject is commonplace within contemporary art. In the age of mechanical reproduction artists can quickly and simply assemble and collate coherent sets of visual materials as a means of non-verbal investigation. In the world of technical images (Flusser 2011) the meaning in a photograph or technically produced image is buried beneath the surface. Artists often assemble bodies of work, as collections of related materials or multiples, sometimes in the style of the visual essay. This allows the artist to illustrate, through type and repetition and focus and observation, that which is important in the pattern of things to an audience. This process assists the audience to see 'what relates to what' between items in the body of work in a non-verbal visual, semantic or cognitive way. This collation and relation of pattern, or similarity, of defining an unspoken meaning with exemplar work is also the foundation of how computer vision systems classify and 'understand' the world. This classification comes from human 'trainers' assembling materials as data sets and ascribing meaning to the data sets in an ontological way. In this paper I describe the process of creating a data set for machine learning from a photographic collection. The photographic materials collected as part of a larger artwork. In the text I explain how this dataset is an extension of the artwork and also an artwork in it's own right. As datasets used for training machine vision systems are commonly centrally generated for commercial/industrial processes I argue that the idea of an artist generated artwork /dataset is the beginning of a process of taking control of the means of surveillance, teaching machine vision systems to see not just how the surveillance capital commercial hegemony sees but how artists see. This creation of artist's data sets and data sets as art is in itself an intervention as art work. We are proposing an evolution of the arguments originating in the industrial revolution of seizing control of the means of production. Now in the age of the surveillance society and surveillance capitalism artists are proposing 'Seizing control of the means of surveillance'.
Synthetic Images and Creative AI: A discussion on the nature and production of images in the era of deep learning
2021
The emergence of new AI algorithms in recent years, especially those concerning deep learning, brings new challenges to the sphere of art, changing how artists creatively use computer systems. Although AI is not new in the universe of art, the new scenario makes it possible for algorithms to produce new types of automated images. Given this picture, this paper proposes to shed some theoretical and practical lights on the processes employed in the generation of visual art using AI. We start exploring the very nature of computer images, having as a theoretical framework the ideas of Dietmar Kamper (1936-2001), Hans Belting (1935-), Christoph Wulf (1944-), and Vilém Flusser (1920-1991). Next, building on this conceptual exploration, we describe the process of using deep learning techniques to generate self-portraits, which are synthetic images pointing to an external index. Fabrizio Augusto Poltronieri [email protected] De Montfort University, Leicester, United Kingdom
The New Value of the Archive: AI Image Generation and the Visual Economy of 'Style'
IMAGE. The Interdisciplinary Journal of Image Sciences, 2023
Text-to-image generators such as DALL•E 2, Midjourney, or Stable Diffusion promise to produce any image on command, thus transforming mere ekphrasis into an operational means of production. Yet, despite their seeming magical control over the results of image generation, prompts should not be understood as instructions to be carried out, but rather as generative search commands that direct AI models to specific regions within the stochastic spaces of possible images. In order to analyze this relationship between the prompt and the image, a productive comparison can be made with stock photography. Both stock photography databases and text-image generators rely on text descriptions of visual content, but while stock photography searches can only find what has already been produced and described, prompts are used to find what exists only as a latent possibility. This fundamentally changes the way value is ascribed to individual images. AI image generation fosters the emergence of a new networked model of visual economy, one that does not rely on closed, indexed image archives as monetizable assets, but rather conceives of the entire web as a freely available resource that can be mined at scale. Whereas in the older model each image has a precisely determinable value, what DALL•E, Midjourney, and Stable Diffusion monetize is not the individual image itself, but the patterns that emerge from the aggregation and analysis of large ensembles of images. And maybe the most central category for accessing these models, the essay argues, has become a transformed, de-hierarchized, and inclusive notion of 'style': for these models, everything, individual artistic modes of expression, the visual stereotypes of commercial genres, as well as the specific look of older technical media like film or photography, becomes a recognizable and marketable 'style', a repeatable visual pattern extracted from the digitally mobilized images of the past.
Unmaking AI Imagemaking: A Methodological Toolkit for Critical Investigation
arXiv (Cornell University), 2023
AI image models are rapidly evolving, disrupting aesthetic production in many industries. However, understanding of their underlying archives, their logic of image reproduction, and their persistent biases remains limited. What kind of methods and approaches could open up these black boxes? In this paper, we provide three methodological approaches for investigating AI image models and apply them to Stable Diffusion as a case study. Unmaking the ecosystem analyzes the values, structures, and incentives surrounding the model's production. Unmaking the data analyzes the images and text the model draws upon, with their attendant particularities and biases. Unmaking the output analyzes the model's generative results, revealing its logics through prompting, reflection, and iteration. Each mode of inquiry highlights particular ways in which the image model captures, "understands," and recreates the world. This accessible framework supports the work of critically investigating generative AI image models and paves the way for more socially and politically attuned analyses of their impacts in the world.
Artistic Style Meets Artificial Intelligence
Journal of Perceptual Imaging, 2021
Recent developments in neural network image processing motivate the question, how these technologies might better serve visual artists. Research goals to date have largely focused on either pastiche interpretations of what is framed as artistic “style” or seek to divulge heretofore unimaginable dimensions of algorithmic “latent space,” but have failed to address the process an artist might actually pursue, when engaged in the reflective act of developing an image from imagination and lived experience. The tools, in other words, are constituted in research demonstrations rather than as tools of creative expression. In this article, the authors explore the phenomenology of the creative environment afforded by artificially intelligent image transformation and generation, drawn from autoethnographic reviews of the authors’ individual approaches to artificial intelligence (AI) art. They offer a post-phenomenology of “neural media” such that visual artists may begin to work with AI techno...
Seven Arguments about AI Images and Generative Media (Chapter 5 of Artificial Aesthetics)
Artificial Aesthetics, 2023
I describe a number of characteristics of AI visual generative media in its current forms that I believe are particularly significant or novel. Some of my arguments also apply to generative media in general, but most focus on visual media. The analysis reflects my own experience of using a few popular AI image tools, such as Midjourney and Stable Diffusion, almost every day from the middle of 2022 until now.
06 INTERNATIONAL CONFERENCE OF PHOTOGRAPHY AND THEORY 2022. Computer Vision and Art Theory: The (Forgotten) Premises of Iconology in Today’s Visual Field
If the work of the art historian has been organized from photographs since the origins of the modern Kunstwissenschaft, the use of digital photographs, and computer vision software, is radically reshaping the DNA of the discipline in unexpected directions. Computer vision software represents an incredible opportunity: it is a tool through which images can be described, organized, studied and shared. In this process there are however a variety of dynamics at play, which have to do with theoretical assumptions, historical categories, technological constraints and ideological stances: a set of premises which calls for a closer methodological survey. We propose an account which uses art theory and visual culture studies to scrutinize the different steps of computer vision analysis. Our intuition is that art photography databases provide a “protected environment” in which to observe how old problems, inherent to the discipline, interact with new problems created by the way we consume and design software. Which images are we talking about? Which research questions are we asking? Which linguistic and political logics are at play? What will emerge is an account of computer vision software which appears to be far from ‘neutral’ or ‘objective’ in its extremely layered functioning, built in the midst of diverse stakeholders’ interests and procedural false steps. Granted that these technologies are however contributing to build the visual culture of our time, we detect a series of overlooked assumptions along the way through the lenses of art theory, hoping to contribute to the design of a clearer view.
Computational Beauty: Aesthetic Judgment at the Intersection of Art and Science
Springer Verlag, 2014
In part one of the Critique of Judgment, Immanuel Kant wrote that "the judgment of taste . . . is not a cognitive judgment, and so not logical, but is aesthetic [1]." While the condition of aesthetic discernment has long been the subject of philosophical discourse, the role of the arbiters of that judgment has more often been assumed than questioned. The art historian, critic, connoisseur, and curator have long held the esteemed position of the aesthetic judge, their training, instinct, and eye part of the inimitable subjective processes that Kant described as occurring upon artistic evaluation. Although the concept of intangible knowledge in regard to aesthetic theory has been much explored, little discussion has arisen in response to the development of new types of artificial intelligence as a challenge to the seemingly ineffable abilities of the human observer. This paper examines the developments in the field of computer vision analysis of paintings from canonical movements within the history of Western art and the reaction of art historians to the application of this technology in the field. Through an investigation of the ethical consequences of this innovative technology, the unquestioned authority of the art expert is challenged and the subjective nature of aesthetic judgment is brought to philosophical scrutiny once again.
Machine Learning and the Mediating Tendencies of the Image
figshare, 2021
The participation of images in the technological mediation of human perceptual experience is especially relevant as the development and implementation of technologies offers new possibilities to produce visualisations from data. In so doing, technological mediation tangibly augments relations between how images are produced, experienced, and interpreted. The present incorporation of machine learning into various forms of visual media offers insight into this issue by enabling images to be produced as the result of the statistical analysis of datasets. Computational relations which are extracted and inferred between features within images help to construct learned representations which are in turn used to generate images. This results in a form of computationally-determined representation which is informed by the interpretive processes performed by machines. Existing notions of technically-produced images often prove inadequate for the description of the visual artefacts of machine learning (ML), leaning heavily on historical narratives regarding the technical production of images and even perpetuating inaccuracies. These tend to misconstrue images either as accurate reflections of reality or as the product of artificial perception and genius by virtue of their engagement with technological processes. Through a media archaeological approach, this paper addresses how current notions of image production remain tied to historical ideas which anthropomorphise and which overestimate the role played by machines, while minimising the role played by humans therein. It seeks to clarify the mediating role played by visual technologies and to demonstrate how images produced using machine learning offer new ways of approaching theories of the image.
Related topics
Related papers
Data-Driven Art -Another Renaissance? The Boundlessness of "Imagined" Creation
2024
Present writing aims to provide the reader with an overview of the ongoing paradigm shift in art in our time brought about by the age of data and artificial intelligence and to present the possible forms of application of artificial intelligence in art practice covering various branches of art as well as the fields of design and fashion.
Dissolving Reality: Exploring the Erosion of Photographic Authenticity in the Age of AI
Transilvania, 2024
This study explores the intricate dynamics of visual communication in the context of AI generated visuals. The question of picture interpretation, starting with the subjective interpretation of images, contextual constraints, and cultural influences of individuals that may affect the perception of an image, is at the heart of this work. The study examines a range of viewpoints on the perception of pictures from writers including Roland Barthes, John Berger, Susan Sontag, and Jean Baudrillard in order to investigate these concerns. The paper also explores the significance of AI-generated graphics, as well as Jean Baudrillard's simulacra notion. It draws attention to how simulacra, or representations without a true relationship to reality, are embodied by AI-generated visuals. In the context of AI-generated imagery, Baudrillard's phases of the image are analyzed, indicating the separation from reality and the formation of pure simulacra. The importance of critically engaging with images becomes evident, as viewers' personal experiences, cultural backgrounds, and the intricate complexities of visual media intertwine to shape meaning. The insights derived from this study contribute to a deeper understanding of the multi-faceted nature of visual communication in contemporary society, emphasizing the need for a nuanced and critical approach to the interpretation of visual content.
Periphery Vision: data as image
In this paper I argue that in an age of data visualisation, how we understand the form of the digital image should be re-considered as being neither purely visual nor purely perceptual. Instead, we may interpret digital images as being figural, as having a greater reliance on associations and connections. This figural reading of the image interrupts and disrupts the established framework of discursive reading and seeing. The narrative of the digital image becomes the story of information and data. I suggest that digital images themselves may be better understood in the 21st century, as interfaces located in the gaps between information, data and experience. I examine the embedded correlation between algorithm and image, between data as image and images of data. I propose, in an age of ‘Big Data,’ semiotics is no longer the most appropriate tool for approaching visual forms and the indexical relation of the digital image is to ones and zeros rather than to material objects.
Data, data everywhere, not a lot in sync: reconciling visual meaning with data
ENQUIRY: The ARCC Journal |Special Edition: Urban Data Assemblage, 2019
Up to 100 billion devices will be seeking to visually map out our existence over the internet by 2020 (UK Government Chief Scientific Adviser 2014). Just as the urban is a forcefield “of spatial transformations… that takes many different morphological forms” (Brenner 2014), this paper explores another underlying forcefield: our visual relationship with data. The most important piece of data, the individual, exists in the city as both prey and predator; having evolved from a “passive aesthetic view of the city” (Appleyard 1979, 144); transformed through shared territory (Evans and Jones 2008); and forged into impressively intricate sets of power relations through collective intentionality (Searle 2011). Through the presentation of self (Goffman, 1969, cited in Appleyard 1979, 146) we inhabit another home: the digital; in which we are simultaneously co-existent and removed by synchronisation of data. Traditionally, the software authoring the physical production of ‘space/hardware’ has been value driven (Raban, 1974, 128, cited in Appleyard 1979, 146). In a parallel universe, algorithms drive the data. For Ellis (2012) it is in the software, that meaning resides. What then is the allure of data to the individual? And what is the allure of the individual to data? It lies arguably in the perception of power and control through meaning (Appleyard, Searle et al.). We seek in the new reality to “discover where the real power lies” (Appleyard 1979, 146). Curiously, the power of data appears to increase the irrelevancy of ownership, between “ours” and “theirs” (Appleyard 1979, 152). This paper analyses past, present, and future states of data production. The data we get from data; data produced from objects; and objects produced from data. In closing, a speculative working hypothesis is presented of visual data production, which hopefully encourages further research reconciling data with meaning in the context of visual sustainability.
The Work of Art in the Age of AI Image Generation: Aesthetics and Human-Technology Relations as Process and Performance
Journal of Human-Technology Relations, 2023
Artificial intelligence (AI); DALL-E; art; aesthetics; philosophy of technology; process philosophy; performance AI image generators such as DALL-E 2 are deep learning models that enable users to generate digital images based on natural language text prompts. The impressive and often surprising results leave many people puzzled: is this art, and if so, who created the art: the human or the AI? These are not just theoretical questions; they have practical ethical and legal implications, for example when raising authorship and copyright issues. This essay offers two conceptual points of entrance that may help to understand what is going on here. First it briefly discusses the question whether this is art and who or what is the artist based on aesthetics, philosophy of art, and thinking about creativity and computing. Then it asks the question regarding humantechnology relations. It shows that existing notions such as instrument, extension, and (quasi) other are insufficient to conceptualize the use of this technology, and proposes instead to understand what happens as processes and performances, in which artistic subjects, objects, and roles emerge. It is concluded that based on most standard criteria in aesthetics, AI image generation can in principle create art, and that the process can be seen as poietic performances involving humans and non-humans potentially leading to the emergence of new artistic (quasi)subjects and roles in the process.
Feeling Data: New Perspectives for the Aesthetic Experience
Aisthesis 18(2), 2025
The production of unprecedented amounts of data across all sectors of society stands out as the defining feature of the present age. Thanks to an all-reaching net of pervasive technologies, it is now possible to draw out data from every entity or event on the planet. Artistic practice provides a suitable stage for the attempt to isolate specific expressive and signifying features out of the indistinct mass of data flowing through the digital realm. This article focuses on a relatively under-explored strand of research, where technology interacts with abstract data in order to extract their "aesthetic sense". Such an expression addresses the peculiar dynamics enabling art to move beyond the purely informative function of data, towards a different goal-designing experiences that turn the audience into perceptive participants, engaged in the otherwise imperceptible events and relations that are recorded and communicated by data. This kind of aesthetic experience presents interesting implications for philosophical enquiry. Through expressive means that are constantly reshaped by the interaction with digital technologies, contemporary art provides fertile ground for a philosophy of events and relations.
INTERSECTING HORIZONS: EXPLORING THE DUAL PATHWAYS OF ART AND SCIENCE IN IMAGE CREATION AND CONSUMPTION THROUGH DATA VISUALIZATION
European Chemical Bulletin, 2023
Data analysis has long relied on visual representation as a means of understanding and conveying explanations. With the continuous advancement of computing technology, storage and processing capacities have significantly improved. Simultaneously, the democratization of technology has allowed for an increase in artistic contributions to data visualization, resulting in novel creative forms and alternative approaches compared to those found in the realm of science. This article examines the reciprocal influences between art and science within the field of data visualization. By analyzing literature from primary bibliographic databases and exploring the thriving community of practitioners that encompasses scientists, designers, artists, and other professionals, we provide a comprehensive overview of data visualization. Drawing upon historical examples from prehistory to the present, we explore various instances where art and science have jointly contributed to advancing the communication of phenomena and the formulation of data-driven inquiries. Additionally, we reflect on the challenges faced by data visualization and the opportunities that arise from them.
Digital Image: The Role of Apparatus, Artificial Intelligence, and Machine Learning in Visual Communication in the Digital Age
Rotura - Revista de Comunicação, Cultura e Artes, 2024
The paper aims to provide a comprehensive analysis of the current state of digital imagery and the implications of recent technologies, including artificial intelligence (AI), in shaping the future of communication in the digital age. The impact of digital transformation on photography has been significant, with a shift from representing reality to serving as a source of raw data for autonomous learning machines. The generative capabilities of AI systems signify a profound shift in creative practices. These systems, capable of producing highly realistic images that often are indistinguishable from human-made ones, have significant implications for both individual creativity and the broader visual culture landscape. This transition has led to questions regarding the role of AI and machine learning in visual communication. As the landscape of creativity evolves with AI, it is imperative to navigate these complexities thoughtfully, ensuring that the integration of AI into creative domains enriches rather than diminishes the human creative experience. In this paper, we also incorporated a set of images generated by Midjourney of an ongoing project to illustrate the remarkable ability of this AI tool to produce visuals that can be deceptively similar to real photographs. Therefore, a comprehensive analysis of the role of apparatus, AI, and machine learning in visual communication, in the digital age is needed to understand the complexities of relying solely on machines and the risks of unconditional trust in technology. Finally, we propose to understand how these technologies are changing the way we communicate and consume images in the twenty-first century, and while the ability to employ computer vision, continuous image streams, and machine learning has beneficial implications, it also presents ethical challenges.
Image-as-data and visual methods
Elgar encyclopedia of technology and politics, 2023
Today, much of the communication in contemporary online environments such as social media is accompanied by or even based on images. Image-as-data refers to any kind of still or moving image such as news images, advertisements, artefacts, television programmes, symbols, social media imagery, and films/videos. This chapter presents approaches for analyzing the image itself, qualitative visual content analysis and photo elicitation, and for examining audiences perceptions of images, semiology and eye tracking with facial recognition.
Loading Preview
Sorry, preview is currently unavailable. You can download the paper by clicking the button above.