A study of generative systems for modeling natural phenomena


Joakim Karl Johnson


anarule.gif (1534 bytes)


Nature is a fantastic phenomenon. When examined at a close range, huge creations such as clouds, mountains and ocean are extremely detailed. No two clouds, plants or mountains are identical, although some may look similar.

It is a challenge to attempt to model these phenomena with enough realism for use in pictures or films. To create geometric models "by hand" using normal geometrical forms is a very tiresome and difficult task. So we will like to use tools that would generate the natural phenomena automatically.

Fractals are a useful tool when highly detailed large objects are needed. They are an excellent modeling tool, when parts of the object resemble the complete model (self-similar property). For example clouds and mountains exhibit this property of self similarity. We present new methods for generating waves, clouds (Foggy circles, Smear circles, L-system based).

In addition, the Mandelbrot fractal, Iterated function system and Hopalong fractals are covered, and their usefulness in creating realistic models is also discussed. When attempting to model plants and trees, a more powerful tool is needed as only some parts of plants and trees are self-similar. One method is the Lindenmayer system (L-systems), which is discussed in detail in this thesis. We present new methods of generating clouds and mountains using L-systems.


Table of Contents

1 Introduction

2 Survey of fractals

3 Water

4 Mountains

5 Clouds in the Sky

6 Plants

7 Lindenmayer Fractals

8 Conclusion


anarule.gif (1534 bytes)

To contact me: