Computer-Generated Plants


For over thirty years, botanists and computer scientists have made considerable efforts to develop effective methods to synthetically generate natural objects. As early as in 1966 the first method for the simulation of a branching structure via a computer was introduced. This method used the so-called cellular au­tomatons, an arrangement of square cells on a grid that could adapt to different conditions directed by a computer program. Based on a simple rule mechanism, and provided the appropriate rules were applied, the model then grew from an initial cell into the adjoining cells and developed a branching structure. Con­trary to these discrete models, later models worked continuously. Since then, multiple different methods have been developed. Moreover, in parallel to the rapid development of computers, the complexity of digital design and the vi­sual quality of synthetically produced images has reached a point at which it is rather difficult to differentiate between computer simulations and actual pho­tographs.

Today, synthetically produced natural objects are frequently used in computer graphics and in related fields. Plants, for example, are the components of many computer images since it is almost impossible to create outdoor scenes with­out incorporating such natural objects. Thus, the corresponding synthetic plant models have become more and more integral parts of many modeling systems. At the same time, expectations for the quick and perfect rendering of virtual images have grown among a wide range of commercial users.

Outside of computer science, many other areas benefit from these models: land­scape designers are now able to visualize and predict the results of their plan­ning, architects enhance their simulations with computer-generated plants, and in botany models are used to determine physiological parameters. In model­ing and simulation as well as in the game industry more and more realistic plant and landscape models have become essential tools in designing realistic – looking environments.

In this book, we mainly focus on the modeling of plant vegetation and the pro­duction of images associated therewith. Natural landscapes consist of a multi­plicity of elements. Although these elements for the most part contain similar

Chapter 1

Computer-Generated Plants

mathematical characteristics, they can differ in their methods of production and in their representation. For example, in computer graphics objects such as rocks, clouds or trees are known as fractals. However, this is only a basic de­scription of a fundamental geometrical characteristic, which for each class of objects has to be converted separately into different production procedures. Although the methods for the simulation of clouds or mountains are related di­rectly to the underlying mathematical principles of fractals, vegetation is ren­dered completely differently. Here, various geometrical aspects of modelling play significant roles, and the fractal characteristics of the plants are, at best, to be found in the recursive structures of the models. What differentiates one plant from another is not so much the general mathematical parameters but rather the concrete detailed structure. Thus, the question of efficient modelling, including especially interactive systems and their evaluation, is the decisive factor.

The aspects treated in the following chapters are grouped into three major ar­eas: methods for the modeling of single plants, methods for the modeling of plant communities and landscapes, and rendering methods for such synthetic landscapes, whereby the first two areas are looked at from the botanical point of view as well as from the point of view of computer graphics.

Here, however, conflicts become apparent about what is regarded as impor­tant in each respective discipline. For example, for a botanist a geometrical plant model is interesting in two ways: it permits the visual validation of the underlying production process, and it is used to calculate mathematical char­acteristics, such as the interaction of light with the plant and the environment. The visual model as such is not of much concern here.

This is completely different in computer graphics: here the geometrical model is used because of its visual effect. The underlying processes are only important in as much as they must permit us to efficiently produce a complex geometry. The exclusive visual evaluation often leads to the production of botanically incorrect models so that in a certain situation a desired result can be achieved.

This divergence between the disciplines surfaces throughout the book. But since this book is primarily directed at readers in computer science, and partic­ularly in computer graphics, botanical aspects are often kept in the background. Hence, the botanical introduction in Chap. 2 is not meant to serve as a knowl­edge base for biologists, but rather to enable a general understanding of botany by the nonbiologist.

For the same reason, the technical descriptions of algorithms and geometry- producing methods are more detailed. Chapter 4 takes into account all essential stages for the modeling of plants using computers and introduces the models in the order in which they were developed. Aside from the purely bibliographic aspect, the variety of possible solutions becomes evident with the abundance of details provided. Later chapters describe rendering issues and demonstrate the results of various researchers and artists in the form of many synthetic images. Before getting into details, a short overview should outline some relevant ques­tions. In each of the following sections, specific problems involving the transla­tion of the general steps into practical applicable procedures are demonstrated.

This should help the reader to easier identify and coordinate the problems treated later on in the book.