Tile-Based Procedural Terrain Generation

Dominik Scholz
Tile-Based Procedural Terrain Generation
[thesis]

Information

Abstract

In recent years, procedural content generation became a valuable tool to create virtual experiences. In a multitude of computer graphics applications, it aids designers to generate vast, detailed worlds. Previous methods for procedural content generation, for example, terrain generation, only cover certain distinct domains where highly specialized algorithms are used. These specialized algorithms can only be configured by programmers or artists with a technical background. With a new approach to procedural generation, called Model Synthesis, it is possible to generate a variety of content, like textures or models with a general purpose algorithm that can be configured by users without domainspecific knowledge of grammars or logical expressions. In this thesis, a proof-of-concept implementation is created to show how this new algorithmic approach can be utilized to generate infinite terrains. A widely used Model Synthesis algorithm named Wave Function Collapse is incorporated into a system providing the complete pipeline needed for the use in a production environment. This pipeline starts with processing of the input: three-dimensional blocks of geometry called tiles. In the terrain generation, they are placed in a lattice such that their geometry matches with the neighboring ones. To create this matching information, the developed system analyzes the geometry of the tiles to generate neighborhood constraints. The terrain generator then uses this information to solve small patches of terrain with instances of the Wave Function Collapse algorithm. Merging these patches to an arbitrary large terrain that can be extended on runtime is also achieved with the developed system.

Additional Files and Images

Additional images and videos


Additional files

Weblinks

No further information available.

BibTeX

@bachelorsthesis{scholz_2017_bac,
  title =      "Tile-Based Procedural Terrain Generation",
  author =     "Dominik Scholz",
  year =       "2019",
  abstract =   "In recent years, procedural content generation became a
               valuable tool to create virtual experiences. In a multitude
               of computer graphics applications, it aids designers to
               generate vast, detailed worlds. Previous methods for
               procedural content generation, for example, terrain
               generation, only cover certain distinct domains where highly
               specialized algorithms are used. These specialized
               algorithms can only be configured by programmers or artists
               with a technical background. With a new approach to
               procedural generation, called Model Synthesis, it is
               possible to generate a variety of content, like textures or
               models with a general purpose algorithm that can be
               configured by users without domainspecific knowledge of
               grammars or logical expressions. In this thesis, a
               proof-of-concept implementation is created to show how this
               new algorithmic approach can be utilized to generate
               infinite terrains. A widely used Model Synthesis algorithm
               named Wave Function Collapse is incorporated into a system
               providing the complete pipeline needed for the use in a
               production environment. This pipeline starts with processing
               of the input: three-dimensional blocks of geometry called
               tiles. In the terrain generation, they are placed in a
               lattice such that their geometry matches with the
               neighboring ones. To create this matching information, the
               developed system analyzes the geometry of the tiles to
               generate neighborhood constraints. The terrain generator
               then uses this information to solve small patches of terrain
               with instances of the Wave Function Collapse algorithm.
               Merging these patches to an arbitrary large terrain that can
               be extended on runtime is also achieved with the developed
               system.",
  month =      jan,
  note =       "1",
  address =    "Favoritenstrasse 9-11/E193-02, A-1040 Vienna, Austria",
  school =     "Research Unit of Computer Graphics, Institute of Visual
               Computing and Human-Centered Technology, Faculty of
               Informatics, TU Wien ",
  keywords =   "procedural modelling, model synthesis, terrain generation",
  URL =        "https://www.cg.tuwien.ac.at/research/publications/2019/scholz_2017_bac/",
}