Background By using Voronoi diagrams to clip polygons in 2D, destruction of meshes can be done fast enough for interactive applications running in real-time. This is achieved by lifting results back into 3D after calculating the polygon shapes of fractured pieces. This technique, however, has a limited amount of mesh shapes which can be fractured due to calculations mainly being done in 2D.

Objectives This thesis aims to implement a proposed method for mesh fracturing in real-time which allows for unique results and dynamic impact locations, and measure its computational speed relative to the desired frame-times of interactive applications such as games.

Methods The computational speed of the proposed method is measured by its implementation as a function in a test application. Metrics are gathered between a range of 5 to 1000 meshes produced from a fracture event.

Results Gathered times show that the proposed method is suitable for real-time in 60 frames per second when producing less than 500 meshes. Smaller results always perform better with larger results being more suitable for lower frame rates.

Conclusions The proposed method is suitable for real-time interactive applications but has a limited amount of destructible shapes, and may not produce results as realistic as simulation-based solutions.