Correcting ray distortion in tomographic additive manufacturing

Light-based additive manufacturing techniques enable a rapid transition from object design to production. In these approaches, a 3D object is typically built by successive polymerization of 2D layers in a photocurable resin. A recently demonstrated technique, however, uses tomographic dose patterning to establish a 3D light dose distribution within a cylindrical glass vial of photoresin. Lensing distortion from the cylindrical vial is currently mitigated by either an index matching bath around the print volume or a cylindrical lens. In this work, we show that these hardware approaches to distortion correction are unnecessary. Instead, we demonstrate how the lensing effect can be computationally corrected by resampling the parallel-beam radon transform into an aberrated geometry. We also demonstrate a more general application of our computational approach by correcting for non-telecentricity inherent in most optical projection systems. We expect that our results will underpin a more simple and flexible class of tomographic 3D printers where deviations from the assumed parallel-beam projection geometry are rectified computationally.