Growing plants for virtual 3D environments

In a virtual environment, objects of various types may exist. Creating these objects and modeling their behaviour with a high degree of realism is of major interest. Static objects are usually simpler to model than dynamic ones, and perhaps, agents are the most common dynamic objects in a virtual environment. Other dynamic components may include avatars and a wide variety of other living organisms such as plants. In particular, the life cycle of plants takes place entirely within the environment in which they exist, and modeling their natural change represents a challenge hard to deal with. Among the methods proposed to date for plant modeling and simulation, L-systems have been widely used with remarkable results in unrestrictive environments. In this paper we propose a combination of L-systems with a genetic algorithm for plants growth emphasizing the incremental natural change of plants. We argue that L-systems on their own encourage inefficient rendering of the growth and change of plants over time. In our model, we also highlight the relationship among plant components (i.e. a plant limb has a parent, children, and occurs at a certain level in the overall structure). We also present preliminary results of a 3D prototype with new rules for the L-system and compare them with those generated by a standard L-System.