I wanted a YouTube channel. I hated editing videos. So I built a machine that does it for me.
What It Does
AI Video Factory is a fully autonomous video production pipeline. It:
- 🧠 Researches trending topics
- ✍️ Writes a professional script
- 🎨 Generates 1080p slides with dynamic gradients and typography
- 🎵 Adds background music
- 🎬 Assembles everything into a polished video
All without a single human click. It runs on a cron job — every morning at 7 AM, a new video is ready.
The Stack
# The entire pipeline fits in one Python file
from PIL import Image, ImageDraw, ImageFont
import subprocess, json
# 1. Research topic (Bing API)
# 2. Generate 8-10 slides with Pillow
# 3. Assemble with FFmpeg
# 4. Ready for YouTube
What makes this different:
- 💰 $35 Raspberry Pi — no AWS, no cloud GPUs
- 📦 Python stdlib + Pillow + FFmpeg — zero exotic dependencies
- ⏰ 24/7 cron automation — set it and forget it
- 🌍 Offline-capable — no internet needed for rendering
Why I Built This on a Pi
Everyone told me: "You need a $200/mo cloud server to run AI video generation."
I proved them wrong. A $35 Pi sitting under my desk handles everything: topic research, script writing, slide generation, video assembly. The entire stack costs $0/month to run.
What I Learned
- Python stdlib is underrated. You do not need FastAPI, Docker, or Kubernetes to build real products.
- Cron is the best scheduler. Simple, reliable, zero-config.
- The best product is the one that runs itself. I was sleeping while my Pi made this video.
This Is Part of Something Bigger
AI Video Factory is one of 23 AI products I'm running on the same $35 Raspberry Pi. All open source. All 100% automated.
Check out the full stack at ulnit.github.io/agent-store.
If you found this interesting, I'd love a ⭐ on GitHub. It helps more than you know.
Top comments (0)