Growing Trees

The piece explains an implementation of a procedural tree-growth method based on Palubicki and colleagues’ 2009 self-organizing tree model. Growth proceeds in seasonal steps, built from metamers: stem segments with leaves and buds. A voxel grid estimates shading; each bud measures its available light and the direction of a less-shaded region. Light is summed from the tips toward the trunk, then converted into a growth resource that travels outward through the branches. Buds receiving enough resource form new shoots, while others remain dormant. A parameter, lambda, controls how resource is divided between a branch’s main continuation and its lateral growth. Small changes in this value can shift the result from a broad crown to a narrow tree with a dominant leader.

Shoot direction combines the existing stem direction, a tendency to seek light, vertical or horizontal tropisms, and noise. Bud arrangement around the stem also shapes branching patterns. The model can shed poorly lit branches, preventing crowded inner crowns, and estimates trunk thickness afterward with a pipe model; branch width and sag do not influence subsequent growth. The grown skeleton is turned into bark tubes, while clusters of leaves are represented by textured twig cards, with simpler geometry and baked views used at greater distances. The same system can also constrain growth with spatial marker points. Finally, a baked linden is instanced at varied sizes to create a forest, illustrating how a simulated tree can feed into a rendering workflow.