FlowBoard

Landing Page 1
Landing Page 2
Canvas
Video player
Long storyboard
Tech Stack

🧐 Inspiration

Traditional animation is slow. You have to redraw every frame and refine motion by hand. We built FlowBoard to automate that process. You sketch once, refine your image with AI, write a quick prompt, and AI generates the moving sequence for you.

🤯 What it does

FlowBoard is an AI-powered storyboard tool that turns sketches and text prompts into context-aware connected video clips that extend forever.

Users can:

Draw directly on 16:9 frames on an interactive canvas
Add written prompts to generate video clips from each frame
Connect frames with arrows to build a sequence
Enhance frame drawings with AI
Merge multiple clips into one final output

It turns rough visual ideas into coherent videos of any length in minutes.

👷‍♂️ How we built it

Frontend:

React + TypeScript
Tldraw 3.15 for the interactive canvas and custom frame shapes
Tailwind CSS 4.1 and Radix UI for styling
Vite for development and builds
IndexedDB storage

Backend:

Python (BlackSheep async web framework)
Gemini AI
- Veo 3.1 for video generation
- Nano Banana Pro Flash for frame enhancement
Google Cloud Storage for file output
Redis for job queue and caching
Uvicorn ASGI server

🤦‍♂️ Challenges we ran into

Enhancing images and creating accurate videos based on user's description
Making sure context is saved and preserved across scenes
Managing deployment architecture

😊 Accomplishments that we're proud of

Seamless end-to-end AI video generation
A polished UI that feels like a real storyboard editor
A fully functional drawing canvas with custom shapes
Video thumbnails displayed directly on connecting arrows
Smooth merging of multiple generated clips into a final render

🧠 What we learned

How to manage long-running AI operations using Redis and polling
How to orchestrate a full pipeline between drawing → image → video
The importance of UX features like loading states, blur overlays, and thumbnail previews