| Name: | Description: | Size: | Format: | |
|---|---|---|---|---|
| 9.83 MB | Adobe PDF |
Authors
Advisor(s)
Abstract(s)
Procedural generation is a method of algorithmically generating data instead of manually doing so. There is an increasing opportunity for the use of procedural generation techniques, mainly in the ever growing video-game and movie industries, due to the necessity of creating virtual content. Even though the manual creation of such content offers an higher level of control, it is also usually a long process with the necessity of one or more technical experts, making the possibility of automation for these processes something desirable. When it comes to the video-game industry, replayability is a term used to assess a video-game’s potential for continued play value after its first completion. Using procedural generation it is possible for a game to offer very different experiences each time it is played, potentially increasing the game time, which is something commonly wanted in the industry. An example where this is implemented successfully is in the game Minecraft [12]. The possibility to generate different worlds each time offers unique experiences to each player, not only turning it into a more personal experience but also increasing the level of replayability as well. In the case of procedurally generating terrain in a virtual world we must take into account not only its shape and height points but also the type of terrain being created. Whether it be a beach or a mountain, the decision of what type of terrain to generate depending on the context can be as important as its shape. The intent of this dissertation is to develop a procedural generator of virtual worlds, so the results of the application of a decision tree evolved through genetic programming can be visualized. The decision to be done by the decision tree will be regarding the type of terrain to be generated. To perform the genetic evolution a small set of decision trees will be generated to be evolved simultaneously, with each generation the terrains resultant being shown to the user, allowing them to perform a choice according to their own criteria of which trees they wish to crossover for future generations.
Description
Trabalho de projeto de mestrado, Engenharia Informática (Interação e Conhecimento) Universidade de Lisboa, Faculdade de Ciências, 2020
Keywords
Geração Procedimental Mundos Virtuais Programação Genética Árvores de Decisão Inteligência Artificial Trabalhos de projeto de mestrado - 2020