Level of Detail Introduction

So far, any volumes can be rendered, but terrain has some special needs because of its size. A typical use case is a very large terrain where the camera moves closely around. An example is a character wandering around, discovering the world.

To keep up with a decent level of performance, a Level of Detail (LOD) mechanism is needed. This means, that the volume near to the camera is more detailed than the parts far away being very small anyway. The amount of detail is reflected by the amount of used triangles.

All details of the developed mechanism are described in the following articles. This one gives again just an overview, how the single parts are connected.

The idea behind the LOD is somehow similar to Chunked Level of Detail Control [Ulr02]. The volume is split into multiple chunks. They vary in size and details and they are organized in a tree data structure. Close to the camera parts of the volume are represented by many small but detailed chunks, whereas far away, big and less detailed ones are used.

This gives the ordering of the following articles. First, it will be described how to connect chunks of different detail grade without any cracks or holes with Marching Square Skirts. Then the generation of the tree is presented, the so called chunk tree. And the last article shows, how to select the chunks to draw from this tree.

If you like this work and want to support it, feel free to donate something or click on any Flattr button!