Evolving Images with Random‑Mutation Hill Climbing I wanted to see how far I could get toward approximating a target image using nothing but overlapping translucent shapes and the Canvas 2D API. No machine learning, no long startup times, just something that kicks in instantly, stays responsive, and feels alive as you watch it evolve. The approach I settled on is called Random-Mutation Hill Climbing (RMHC), or the (1+1) Evolutionary Strategy. The idea is simple: Keep a single best-so-far so...
