Pareidolia: Faces in the Static

This simulation generates a field of soft, randomly placed light and dark blobs — visual noise in the spirit of clouds, static, or wood grain — and runs a simple, genuine face-likelihood detector over it: the algorithm hunts for pairs of dark blobs positioned like eyes, checks whether a contrasting region sits below them where a nose or mouth would be, and scores the resulting arrangement. Tune the noise scale to change how big and how numerous the candidate blobs are, and the contrast/threshold slider to control how confident the detector must be before it draws a circle. Turn the overlay off first and look for faces yourself, then turn it on to see exactly which accidental arrangements a coarse, real face-detection heuristic — the same "two eyes above a mouth" template your own fusiform face area uses — picked out as face-like.