Real-time image generation for interactive art : developing an artwork platform with ComfyUI and TouchDesigner
Kuoppala, Peetu (2025)
Julkaisun pysyvä osoite on
https://urn.fi/URN:NBN:fi:amk-2025053018415
https://urn.fi/URN:NBN:fi:amk-2025053018415
Tiivistelmä
In this thesis is presented the development of an adaptable and reusable artwork platform model. This platform, called P2KTool, is targeted towards artistic exploration of generative AI tools with a specific focus on real-time interactivity. The platform model was designed as a project template for off-the-shelf, consumer-grade PC hardware and the software tools ComfyUI and TouchDesigner. These tools and the relevant image generation technology are evaluated in the context of integrating them into a real-time application.
“ANTIMAGIA,” a proof-of-concept interactive installation utilising this platform, is presented with a technical description of the physical setup of the piece, as well as an overview of its central theme, the perception of artifice, and artificiality. Conceptual motivations and stylistic decisions regarding the presentation of generated imagery, as well as the role of interactivity and participation are discussed.
Finally, audience reactions to the exhibited installation are assessed together with insights gained in the process of working with generative AI models. Markedly positive and humorous impressions, as well as general curiosity about the technical implementation of the installation suggest that audiences are receptive to the interactive artform presented. In view of these findings, a proposal for the direction for continued development of P2KTool, future installations and other potential applications is given.
“ANTIMAGIA,” a proof-of-concept interactive installation utilising this platform, is presented with a technical description of the physical setup of the piece, as well as an overview of its central theme, the perception of artifice, and artificiality. Conceptual motivations and stylistic decisions regarding the presentation of generated imagery, as well as the role of interactivity and participation are discussed.
Finally, audience reactions to the exhibited installation are assessed together with insights gained in the process of working with generative AI models. Markedly positive and humorous impressions, as well as general curiosity about the technical implementation of the installation suggest that audiences are receptive to the interactive artform presented. In view of these findings, a proposal for the direction for continued development of P2KTool, future installations and other potential applications is given.