Type a prompt. Watch morphogenesis run. Living geometry generated from biological growth rules.
The engine translates natural language into growth parameters — branching probability, upward bias, twist, noise — and simulates the structure in three dimensions.
The architecture mirrors how biological systems actually work — local rules producing global form through iteration.
You describe a structure in plain language. Keywords map to growth mode and parameter modifiers.
The prompt engine extracts mode (tree / coral / spiral), then adjusts branching, bias, noise, and twist based on descriptors.
Growing tips advance step by step, steered by directional bias, noise, and branching rules — producing emergent 3D geometry.
Three.js renders the structure in the browser. Segments are colored by depth — gold trunk, green canopy.
The generator calls a live Python morphogenesis backend via FastAPI, falling back to the client-side engine if the API is offline. GRN-driven growth is next.
Each mode encodes a different spatial logic. Modifiers shift parameters within that logic.
Strong upward bias. Moderate branching angle. Trunk thickens at base and tapers toward canopy. The architecture of phototropism.
Outward radial spread with higher noise. Dense, chaotic branching. The structure of reef ecosystems — maximizing surface area.
Helical upward growth with tangential twist bias. Architectural, minimal, precise. The logic of nautilus shells and plant phyllotaxis.