Platform: Unity3D
Language: C#
Year: 2019
Algorithm: Generative Design, Genetic Algorithm
Language: C#
Year: 2019
Algorithm: Generative Design, Genetic Algorithm
Optimize the conventional Game of life program and increase more possibilities. There are two steps. Firstly, search the different start patterns and different rules of Game of life, and then analyze its density, age, and the trend of geometry change. The next step integrates a gens system like a genetic algorithm. Each trial has its DNA with used Game of life rules and other conditions like which level to change different rules according to various situations. With this generative design system, we can control the basic rules and let computers generate the design result.
We start to use genetic algorism to search the best result. We set a DNA which includes indexes of images, maximum and minimum of density, and rules. Besides, we set fitness which is the aim that we want to reach from the combination. So we define some methods to evaluate the model.
First method is aging which would generate tall straight towers in the model. Density is the method to generate cubes as many as it can. Next method of Two Connection restricts the number of neighbours from each cube with two. Therefore, this combination would have more branches.
This method is in order to limit the number of neighbours to a certain domain.
Structural stability is to make sure that there is definitely a cube under the cube that computer is going to create. It makes model connect more from downside to upside. Growth rate is a method to control the relationship between density and layers. You can see in the picture, it helps to defer increase rate of density.
Layering method is a totally automatically Algorism which let program to choose which law is the most suitable based on density for the fitness. We divided the tower with 5 parts and each part would find the best laws to meet the requirement of density.
Further more, we try to combine different methods and rules to search the combination, it is clearly that rules would change with different conditions.
