Design and Implementation of Asymptotically Optimal Mesh Slicing Algorithms Using Parallel Processing

Mesh slicing is the process of taking a three dimensional model and reducing it to 2.5 dimensional layers that together create a layered representation of the model. The process is used in layered additive manufacturing, three dimensional voxelization, and other similar problems in computational geometry. The slicing process is computationally expensive, and the time required to slice an object can inhibit the viability of layered manufacturing in some industries. We designed and developed a fast implementation of the slicing process, called Sunder, that uses new asymptotically optimal algorithms and takes advantage of parallel processing platforms. To our knowledge, no other slicing implementation leverages massive parallel execution hardware, such as graphics processing units (GPUs), leaving significant potential for improvement. Furthermore, no published set of slicing algorithms completes all three major steps in the slicing process (preprocessing, slicing, and contour assembly) in linear time complexity, which our design achieves. Therefore, our implementation improves the current state of the art in mesh slicing