In this paper, we present a case study exploring the potential use of Generative Artificial Intelligence (GAI) to address the real-world need of making the design of embroiderable art patterns more accessible. Through an auto-ethnographic case study by a disabled-led team, we examine the application of GAI as an assistive technology in generating embroidery patterns, addressing the complexity involved in designing culturally-relevant patterns as well as those that meet specific needs regarding detail and color. We detail the iterative process of prompt engineering custom GPTs tailored for producing specific visual outputs, emphasizing the nuances of achieving desirable results that align with real-world embroidery requirements. Our findings underscore the mixed outcomes of employing GAI for producing embroiderable images, from facilitating creativity and inclusion to navigating the unpredictability of AI-generated designs. Future work aims to refine GAI tools we explored for generating embroiderable images to make them more performant and accessible, with the goal of fostering more inclusion in the domains of creativity and making.

翻译：本文通过一项由残障人士主导团队开展的自我民族志案例研究，探讨生成式人工智能在满足现实刺绣艺术图案设计需求方面的潜在应用。我们研究了GAI作为辅助技术在生成刺绣图案中的具体实践，重点关注其在处理文化相关性图案设计以及满足细节与色彩特定需求时所面临的复杂性。我们详细阐述了为生成特定视觉输出而定制GPT模型的提示工程迭代过程，特别强调实现符合真实刺绣要求的理想结果所涉及的细微差别。研究结果揭示了运用GAI生成可刺绣图像的混合成效：既展现了促进创意与包容性的潜力，又需应对AI生成设计固有的不可预测性。未来工作将致力于优化我们所探索的GAI工具，提升其生成可刺绣图像的效能与可及性，以促进创意与制造领域更具包容性的发展。