Nakamura,N., Inoue,Y., Teshima,Y., Nishio,K., Kobori,K.
Abstract:
Boundary representation model and spatial partitioning model have been used in CG field. Each model has different advantages and disadvantages. Thus, if a user can transform between each model bidirectionally, a user can utilize advantages of each model. It is necessary to decide whether each voxel is on the inside or outside of the shape when a user transforms the boundary representation model into the voxel model. The processing load is large.
In order to solve this problem, we propose a fast method to transform the boundary representation model into the voxel model by using GPU; our method can decide the condition of inside of the shape with rendering engine. GPU is the specialized hardware for drawing shapes, and it has been used for fast processing or improvement of the accuracy of mathematical calculation recently. Also, users can alter a part of the function of GPU in accordance with their needs. First, our method renders an input shape with orthographic projection along z-axis. Next, it gets a particular color plane from the result. Thus, it generates a part of voxel data. By repeating this process as many times as the resolution of voxel space along z-axis, our method can generate voxel data rapidly. In addition, we propose an optional method which improves the accuracy of the transformed shape with fragment shader of GPU.