Video feedback pointing a camera at its own output produces self-similar, fractal-like patterns from simple recursion

When a camera captures the screen it is feeding into, the recursion creates fractal-like structures — each frame contains a smaller copy of itself, repeated until resolution limits. In Hydra this is replicated in code: blend a shape with its own output, then repeat the chain, so each iteration is a scaled copy of the previous. Small parameter changes (angle, brightness) determine whether the feedback converges to an ordered pattern or diverges into chaos. Crutchfield’s 1980s video feedback paper showed the same simple analog loop could compute complex physical models (fluid dynamics, reaction-diffusion).

Examples

Hydra: blend a shape with src(o0) (its own output), then .repeat() — produces fractal self-similarity at near-zero cost. Pointing a webcam at itself creates real-time analog feedback that is ‘very chaotic’ and sensitive to small changes.

Assessment

Explain why blending a Hydra output with itself and then repeating produces a fractal-like result. What determines whether the feedback is stable or chaotic?