The last procedural method discussed in this chapter generates climbing plants, which are actually in an entirely different category. Ned Greene  deals with the question of how the interaction of such plants with the environment and the incidence of light can efficiently be rendered, and at the same time how these plants can be rendered to grow realistically along walls.
Greene divides the scene into voxels (square volume elements, see [48, 107]) and denotes those voxels in which the climbing plant can grow. A parameteri – zable probabilistic algorithm allows the plants to grow, starting at a manually specified seed point.
Using a search strategy, internodes are placed into the possible voxels, whereby the algorithm reacts to the directly entering sunlight and the diffused lighting. The determination of the direct light incidence takes place by computing the incident sun light for each voxel. The computed value indicates for how much time the voxel was illuminated by the sun. To compute the diffuse light, for each voxel it is defined how much can be seen of the sky in each case. Both values are clamped on the interval [0,1].
Although Greene speaks of a rule system for the growth computation, he does not supply a formalized description, but offers only a listing of the parameters to control the probabilistic algorithm. Included are the lengths of the internodes, the numbers of internodes between branches, the branching angle, the strength of the phototropism, the growth strength in relation to the light (weighted sum of direct and diffused sunlight), and the number of leaves per branching.
The geometry of the plants is produced in polygonal form; all important parameters are, however, assigned for each voxel, which reduces the computational effort. In Fig. 4.16 examples are given. Rule-based climbing plants are also found in .