ClipFarm.TV

ClipFarm: Make Money, Feed Your Family. Auto-clip Twitch VODs into IPFS-backed segments, then curate, edit, and export the best highlights to any social platform—automagically.

Comment

Inspiration

I’m an active listener. I heard Neon, a streamer, say in a YouTube Short that he pays his clip farmers $10,000 a month. I thought: there’s got to be a way to help average people make money clip-farming—and, with this automated pipeline, give them a source of income to feed their families. ClipFarm.TV was born from that spark.

What it does

ClipFarm.TV automates the discovery and extraction of high-engagement 30-second MP4 clips from any Twitch VOD:

  1. Ingest

    • Accepts a Twitch VOD URL
    • Downloads the full video via a Cloud Run–hosted pipeline

  2. Analyze

    • Splits the VOD into consecutive 30-second segments
    • Annotates each segment with chat speed (messages per second)
    • Calculates the overall average chat speed across the VOD

  3. Extract

    • Filters for segments where chat speed > average
    • Produces individual 30-second MP4 files for each above-average segment

How we built it

  • Language & Framework: Python 3.11
  • Containerization: Docker (multistage build)
  • Compute: Google Cloud Run for scalable, serverless processing
  • API & Orchestration: Supabase → Google Cloud Functions
  • Video Processing → Google Container Registry → Cloud Run

Challenges we ran into

  • Day 24 “Debugging Hell”: Crashed on HD playback—logs were cryptic.
  • Moov Atom Mystery: After 5 days (day 29), discovered ffmpeg’s +faststart flag was needed to relocate the “moov atom” to the front. HD clips then played flawlessly.
  • Concurrency Tuning: Balancing Cloud Run concurrency and task-queue throughput to avoid throttling during multiple simultaneous VOD processing jobs.

Accomplishments that we’re proud of

  • Serverless, End-to-End Pipeline: From Twitch URL input → annotated 30-second MP4 clips output.
  • Resilience Under Load: Survived bursts of VOD ingestion without errors or timeouts.
  • Personal Win: Pushed through doubt when I felt like I’d let my family down. Even without external praise, I proved to myself that persistence pays off.
  • Made a Demo walk through with eleven labs ( didnt get an eleven labs code in my pack)

- Made an ad with Google Veo 3 (see link on page)

What we learned

  • MP4 Internals: Deep dive into moov atoms and faststart flags.
  • Cloud Run Best Practices: Concurrency settings, memory allocation, cold-start mitigation.
  • API Design with Supabase: Lightweight, secure function triggers.
  • Debugging Strategies: Incremental logging, local repro, community resources.
  • Growth Mindset: Every blocker is a stepping stone to mastery.

What’s next for ClipFarm.TV

  1. Web Dashboard: Cleanup interface, small things matter.
  2. Platform Expansion: Add Facebook Gaming, YouTube Live support, Tik Tok, Google Drive, One Drive.
  3. AI Tagging: labels (e.g., “funny,” “epic”) for easier browsing

“ponerse las pilas mijo”

Built With

  • 40-mini-chatgpt
  • algorand
  • bolt.new
  • elevenlabs
  • googlcloud
  • netlify
  • picaos
  • python
  • revenuecat
  • stripe
  • supabase
  • vite
Share this project:

Updates