Treffer: FluxZayn: An Extension for Stable Diffusion WebUI Forge

This practice-led, auto-ethnographic case study documents the conceptualisation, development, and release of FluxZayn, a Python-based extension for the Stable Diffusion WebUI Forge platform. Motivated by a creative workflow challenge – specifically generating 2D transparent (.png) game development assets more efficiently – the study investigates “vibe coding,” an intuitive artist- centric development process mediated by generative AI coding assistants. Over the course of the project, the author sequentially collaborated with ChatGPT, Claude, and Gemini Pro (AI Studio) to iteratively prototype the tool. The resulting open-source extension integrates FLUX.1 model support and LayerDiffuse-based background removal, enabling one-click transparent image generation in WebUI Forge. In addressing this project, the study examines whether generative AI coding tools can build a functional prototype from a designer’s perspective, identifies barriers encountered during AI-assisted prototyping, explores the practical realisation of a Python GUI under “vibe coding,” and compares the utility of ChatGPT, Claude, and Gemini as coding partners. The findings confirm that AI assistants (particularly Gemini Pro) can indeed empower a non-programmer to create complex software, though challenges in communication, context, and domain knowledge remain. This work highlights the pedagogical and democratizing potential of AI-assisted development in creative practice, while also noting the dependency of success on the underlying AI tool’s capabilities.