Type a prompt. Get a full comic book. Instantly.

ComicWand started with a simple frustration: creating a comic book required either years of drawing skills or a big budget to hire an artist. I wanted to change that. The goal was to build a open source tool that lets anyone - writers, kids, content creators turn a plain text prompt into a fully illustrated comic book in seconds. The result is a web app that takes a user's input, breaks it into scenes, generates panel-by-panel artwork using AI image models, and assembles the pages into a downloadable comic layout - complete with speech bubbles and panel borders.

The core challenge was turning a single open-ended prompt into a structured narrative. I built a prompt decomposition layer that uses an LLM to break the user's input into individual scenes — each with a setting, characters, and action — before any image generation happens.

Each scene is fed into an image generation API with a custom comic-art style prompt. The output images are then placed into a grid-based panel, which dynamically sizes panels based on the narrative weight of each scene.

Dialogue is extracted from each scene description and rendered as overlaid speech bubbles using SVG, positioned automatically to avoid covering key parts of the image. Comic-style fonts give it an authentic feel.

Users can download their comic as a PDF or share a public link. The layout is rendered as a full-resolution page sequence, optimized for both screen reading and print.

Next.js 14 App Router

FastAPI

PostgreSQL

Redis

Celery

Replicate

Anthropic Claude API

Cloudflare R2

Clerk