I began this project during the summer between 10th and 11th grade, when generative image tools were still developing quickly and maintaining character consistency across longer sequences was difficult.
I independently built a web-hosted system that could turn text, chapters, or ideas into connected AI-generated comic sequences. Users could control art style, characters, character designs, pacing, and scene details, then generate and edit full comic sequences.
As I tested the system, consistency became one of the biggest problems, so I built a prompt-rewriting layer and character controls to help preserve important visual details across panels and longer chapters.
Website/interface
Image-generation pipelines
Script-to-panel generation
Character and background controls
Editing and regeneration tools
Continuity controls between panels
Full-chapter generation
The workflow begins with story input and continues through AI-generated panel planning, moving from narrative text to a structured comic sequence.
New Project Setup: Users can create a project from raw text or an uploaded file and choose the number of comic panels to generate.
AI panel generation: The script editor converts the story into structured panel descriptions, helping turn written scenes into visual sequences.
Each panel can be reviewed and edited before generation, including its visual description, character references, background context, system prompt, and additional instructions.
The generator produces multiple visual options for each panel, allowing the user to select, regenerate, or refine an image before adding it to the sequence.
Generated panels remain connected to the underlying script, allowing a full sequence to be reviewed panel by panel alongside its scene descriptions and dialogue.
Example from a longer generated sequence. (panel 30 image)