🎬 Vid-Gen — AI Video Generation App

Full-stack AI video generation platform. Accepts text prompts, generates videos using Groq LLM + FramePack/Fal.ai, and serves them through a React TypeScript frontend with a FastAPI backend.

🎯 What It Does

1. User enters a text prompt describing a video
2. Groq LLM refines the prompt for optimal generation
3. FramePack / Fal.ai generates the video
4. Video is served back through the React UI with progress tracking

📁 Project Structure

vid-gen/
├── backend/
│   ├── app/
│   │   ├── main.py                     # FastAPI entry point
│   │   ├── config.py                   # Settings & env vars
│   │   ├── routes/video.py             # Video generation endpoints
│   │   ├── services/
│   │   │   ├── groq_client.py          # Groq LLM prompt refinement
│   │   │   └── framepack_client.py     # Video generation client
│   │   ├── models/schemas.py           # Pydantic schemas
│   │   └── utils/storage.py           # Output file management
│   ├── requirements.txt
│   └── test_fal.py
├── frontend/
│   ├── src/
│   │   ├── App.tsx                     # Root component
│   │   ├── components/
│   │   │   ├── PromptForm.tsx          # Prompt input UI
│   │   │   ├── VideoJobList.tsx        # Job status tracking
│   │   │   └── VideoPlayer.tsx         # Video playback
│   │   ├── pages/Home.tsx
│   │   └── api/client.ts              # API client
│   └── package.json
├── video-worker/
│   ├── main.py                         # Async video worker process
│   └── requirements.txt
└── outputs/                            # Generated video output dir

⚙️ Setup & Run

Backend

cd backend
pip install -r requirements.txt
# .env: GROQ_API_KEY, FAL_API_KEY
uvicorn app.main:app --reload

Frontend

cd frontend
npm install
npm run dev    # TypeScript + Vite

Worker

cd video-worker
pip install -r requirements.txt
python main.py

🛠️ Tech Stack

Component	Tech
Backend	FastAPI (Python)
LLM	Groq API
Video AI	FramePack / Fal.ai
Frontend	React + TypeScript + Vite
Worker	Python async worker

---

GitHub · LinkedIn