|dc.description.abstract||When we look at the physiologies of creatures in life simulation games, we can see that they are often extremely simplified and not simulated beyond very simple preprogrammed appearances. In this thesis, we propose that this might be caused by three issues: the difficulty in creating models that generate creature physiologies that are neither chaotic nor ordered, the inherent loss of control by a game’s designer when using procedural techniques, and issues with performance when using such complicated models in a real-time gaming environment. We try to solve these problems by using a model from the field of artificial development, and adapting it to give back control to the designer and to improve its performance.
The artificial development model gives us the ability to grow a creature from a simple DNA string through an iterative process. However, these models are optimised to be used in an offline genetic algorithm, and are often expressly designed to not require any human input. We adapt an artificial development model to allow designers to easily influence specific features, such as the range and threshold of the morphogenic fields, and their specific association with regulatory units. We also introduce a way for designers to have more control over the development of a creature during its lifetime, by allowing them to introduce bias in the phenotypical properties of creatures. These features combined also allow for a better performance during the growth process.
We show through a small experiment that the ability to influence specific parts of a creature is both important as well as very hard to do using traditional artificial development models, and that our model alleviates some of these issues.||