Yeah, I’ve played with that idea. The key is to treat each plant as a node in a graph or a tree, with a clean adjacency list that stores parent‑child relationships. Keep the node data in a small struct—position, growth stage, maybe a “wildness” flag. Then run a recursive growth function that picks a random angle or a Perlin‑noise offset at each step, so the structure stays mathematically tidy but the output looks alive. If you separate the deterministic rules (the data structure) from the stochastic seed (the noise), you get both order and that wild fractal feel.