Make 370 Hours of Video Searchable for Just $10 in GPU Time

vanstriendaniel · x · 2026-07-31

A developer demonstrated how to turn massive video archives into searchable databases using AI. By processing 1,864 public-domain films from the Prelinger Archives, cutting them into 60-second chunks, and running an open-source 2B video model for multimodal feature extraction.

The entire index cost only about $10 in GPU compute. Users can now search with natural language queries (e.g., "typing on a computer keyboard") and jump directly to the exact moment the action occurs on screen.

Related event: AI Makes Massive Public Domain Film Archive Searchable(2 posts)→

Original post →

More from coding & agent

coding & agent channel →