To what Extent can Signed Distance Fields Benefit Destruction on Mesh Data

To create destruction in real-time applications, video game engines depend heavily on the Voronoi approach. This method creates additional data, the level of detail depends on the number of subdivisions, and is mainly executed as a pre-process. This thesis attempts to explore a new way of applying destruction in a non-destructive way, without altering the mesh, using signed distance fields on mesh data. During this exploratory research, we started with a semi-structured interview to establish our destruction criteria for evaluation. Following this, an ideal solution was outlined, and a prototype was constructed. We then evaluated our progress with a continuous comparison by measuring and analysing the performance for every iteration. We then concluded with a visual comparison against the Voronoi fracturing method by using Q-Sort, the picture-sorting technique. The gathered data during the iteration progress concluded that our proposed method is efficient in real-time video game applications, even on low-end hardware. From the Q-sorting results, we concluded that our proposed signed distance field method's visual outcome is more visually appealing than the Voronoi method. While our method shows potential as a destruction method, we need to delve deeper and investigate how our approach performs in different in-game situations, assets, and hardware setups.