FrameFind

AI B-Roll Asset Search

Open Source

Open-source AI tool that turns a video script into ranked B-roll photos and videos from stock libraries and optional social platforms.

Key contributions

Built a full-stack TypeScript monorepo with a Next.js frontend and NestJS API for script-to-visuals search.
Used Gemini/Groq to split narration into scenes, generate visual keywords, and rank B-roll matches by relevance.
Integrated a pluggable provider layer across 9 sources (stock libraries and optional social platforms) with rate limiting and quota-safe toggles.
Added filters for media type, orientation, and duration so creators can find usable photos and videos faster in one workflow.

Quick start

Clone the repo and install dependencies (Node.js 20+).
Copy the env examples, then add at least a Gemini or Groq key plus one stock API key (e.g. Pexels).
Run the app, open the UI, paste a script, and hit Analyze script to get scenes + ranked B-roll.

git clone https://github.com/Mahdi015/FrameFind.git
cd FrameFind && npm install
cp apps/backend/.env.example apps/backend/.env
cp apps/frontend/.env.example apps/frontend/.env.local
npm run dev

Frontend: http://localhost:3000 · Backend: http://localhost:3001

View on GitHub