“Stable Diffusion Denoising Path Orientation: Implementing L-System in Generative AI Model” presented by Şalcı
Symposium:
Session Title:
- Virtual (full papers)
Presentation Title:
- Stable Diffusion Denoising Path Orientation: Implementing L-System in Generative AI Model
Presentation Subtheme:
- Speculative practices
Presenter(s):
Venue(s):
Abstract:
This paper invests the integration of the L-system algorithm with Stable Diffusion, a method called ‘Denoising Path Orientation,’ to create complex, lifelike structures in digital art, named Cybernetic Flora. Influenced by Michel Serres’ theoretical lens, this paper analyzes the complexities of noise, disruption, and transformation in the generative phases of digital art. Utilizing Serres’ concept of the parasite, the study interprets the denoising phase as a transformative disruption, crucial in system development. The L-system, originally a tool for simulating biological structures, iteratively generates intricate patterns. When fused with Stable Diffusion’s denoising capabilities, these patterns undergo refinement, resembling natural evolutionary processes. This fusion not only improves the visual quality but extends interdisciplinary applications, particularly in AI convolution processes. This paper demonstrates how disruptions or ‘parasitic’ elements in the creative process can lead to richer, more complex artistic expressions, challenging conventional notions of order and disruption in digital art creation.
