news.ycombinator.com• Feb 12, 2026• 1 min read
Camera based true random number generator BetaI’m an electromechanical engineer who usually tinkers with RPis and Arduinos more than software, but a few weeks ago I went down the rabbit hole of randomness. I read about Cloudflare’s lava lamp wall and thought: if they’re ultimately using cameras, why not use the image sensor noise directly?Modern CMOS sensors produce a lot of grainy, unstable pixels—even when pointed at a static scene. You can film the same white wall at 30 fps and never get an identical frame. That felt like a decent physical entropy source, so I built a TRNG around it and called it Aegis Optikon(.com).The pipeline right now looks like this: Capture frames from a camera (client devices) Extract “noisy” pixel data Whiten and compress it Send entropy packets to the server Mix multiple streams together Extract with BLAKE3 Attach timestamps and hashes per packet for basic verifiability I’m not a seasoned backend/security engineer—I leaned heavily on AI tools (Copilot/DeepSeek) to get the code into shape, then iterated
