Adaptive Voxelization for Rapid Projection Generation in Computed Axial Lithography

Computed axial lithography (CAL) is a tomographic additive manufacturing technology that offers exceptionally fast printing in a wide range of materials. CAL involves pre-computing a sequence of light patterns to be projected into a photopolymer. For a uniform spatial discretization of the target geometry, computational time scales inversely with the cube of the discretization pitch, which makes it challenging to exploit the full space-bandwidth product of available spatial light modulators. This work introduces an adaptive voxelization approach to reduce computational expense. Using one of several proposed mesh-based complexity analyses, a CAD model is recursively subdivided into stacked sub-meshes of varying voxel resolution. These complexity methods can be tailored to emphasize complexity in particular regions. Each sub-mesh is then independently voxelized before projections are generated and optimized in parallel. On a four-core CPU, this method results in a 2 − 6 × speedup with applications in high-precision CAL and other voxel-based additive manufacturing computations.