Text-to-Video Multi-Agent System
A LangGraph pipeline that turns a story prompt into a narrated short video through shot planning, media generation, quality checks, and editing.
Studio interface · Story and film settings
Project overview
An end-to-end video workflow that plans a story, creates scene media and narration, validates each stage, and assembles a final edit. The project is under active development.
What I built
Designed a multi-agent graph with human review checkpoints, per-shot retries, visual QA, and resumable runs. It can generate animated shots and use FFmpeg motion when animation is unavailable.
How I delivered
Connected FastAPI and a browser UI to LangGraph orchestration, media providers, SQLite checkpoints, and FFmpeg-based assembly to produce reviewable video artifacts.