Procedural terrain generation. A case study from the game industry

Abstract

This chapter shows how PCG can be used for landscape generation in games. A very brief introduction to value noise generation is provided. Any noise generator capable of generating cloud pictures can generate similar results with the new algorithm, for example the well-known Perlin noise or its derivation, the simplex noise. We then provide both basic algorithms and practical hints for generating different types of terrain. A new algorithm is presented which generates landscapes with islands of different size and levitation. This algorithm has been created for an industry game project to increase the variety of islands in an explorer game. We show in detail how noise-based images generate a 3D-Terrain, how this terrain can be manipulated so it looks realistic and how the landscape can be textured. The techniques used are not specific to any game engine-they can be implemented in any 3D engine capable of creating custom meshes at runtime.