Sequential image generation for illustrating stories
LLMs for creative writing
Personalising diffusion models
Research Questions
RQ1: How can an AI image generation application assist writers with overcoming writer’s block?
RQ1a: How can a generative AI application support creative writers with limited AI knowledge in fine-tuning diffusion models for consistent character generation?
User Requirements
Formative Interviews
Interview Design
Thematic Analysis
Initial user requirements and sitemap
Testing generative AI for writing tasks
Writing task #1: Generating images of scenes from a novel
Writing task #2: Generating images of characters for a fantasy novel
Writing task #3: Generating images of objects for a fantasy novel
Writing task #3: Generating images of settings for a fantasy novel
Conclusions from using generative AI
Updated user requirements
Methodology
Model Selection
Addressing Model Limitations
Solution
Overview of Process
Prompt expansion
Consistent character image generation using Dall E 3 and GPT-4
Creating the storyboard
Generating plot ideas
System Architecture
Limitations
Design
Wireframe Design
User Interaction Flow
Website Pages
Prototype Evaluation
Procedure
Exercises: Storyboarding and Image Generation
Semi-Formal Interview Questions
Participants
Results
Quantitative analysis/summary of questionnaire responses
Thematic Analysis of Results
Theme #1
Theme #2
Theme #3
Discussion
Areas for Future Work
Limitations of Work
Conclusion
Reflection on objectives achieved or failed to achieve
Dissertation outline
Abstract
Acknowledgements
Table of Contents