openstory-so/openstory

Open-source AI-powered video sequence platform built with TanStack Start

What it solves

OpenStory automates the process of turning written scripts into styled video productions. It removes the manual effort of prompt engineering and compositing by coordinating AI generation of images, motion video, and audio while maintaining visual consistency across different scenes.

How it works

The platform uses a pipeline of AI models to process a script:

  1. Script Analysis: It breaks down a script into scenes, identifying camera angles, mood, and continuity requirements.
  2. Generation: It uses Fal.ai for image and video generation and OpenRouter for LLM-based script analysis.
  3. Consistency Management: It ensures that characters, locations, and color palettes remain consistent throughout the production.
  4. Edge Infrastructure: The entire application is deployed on Cloudflare Workers, utilizing D1 for the database, R2 for storage, and Cloudflare Workflows for durable execution of generation tasks.

Who it’s for

Content creators and production teams who want to rapidly prototype or produce AI-generated videos from scripts without needing to manually manage complex AI prompts for every frame.

Highlights

  • Automated Scene Breakdown: Automatically extracts camera angles and mood treatments from text scripts.
  • Style Continuity: Maintains consistent visual identity for characters and locations across multiple scenes.
  • Image-to-Video: Converts still AI-generated frames into motion video clips.
  • Edge-Native Stack: Built with Bun and Cloudflare Workers for global deployment and low latency.

Related

  • Project
  • Project
  • Project
  • Project
  • Project