Stable Diffusion for Narrative Illustration: Transforming Text into Visual Art

In today's digital age, where technology continuously reshapes creative expression, this paper presents a system that transforms written narratives into fully illustrated comic books by combining Natural Language Processing and Stable Diffusion techniques. The proposed solution employs Google's Gemini model to analyze and summarize user-submitted stories, breaking them into key segments for illustration. Each segment is used to generate contextually accurate images via the Stable Diffusion XL model, ensuring visual consistency and relevance. Unlike traditional generative approaches such as Generative Adversarial Networks, which often require extensive training and struggle with multi panel coherence, Stable Diffusion delivers high-quality illustrations with lower computational overhead. The system is implemented as a web application, allowing users to upload stories, receive summarized captions, and view illustrated outputs either as a downloadable comic book or a narrated visual sequence. Key contributions of this work include a text-to-comic generation pipeline integrating large language models and text-toimage diffusion, a novel approach for context-preserving prompt generation at the sentence level and a resource-efficient implementation requiring no dedicated GPU. The results demonstrate the system's effectiveness in producing stylistically consistent illustrations with minimal human intervention. This framework holds significant promise across domains such as education, entertainment, accessibility, and digital marketing, enabling inclusive and scalable visual storytelling.