Researchers develop semi-automated AI pipeline to convert 2D medieval manuscript illustrations into 3D models for virtual reality and 3D printing.

Framework tested seven image-to-3D methods (TripoSR, SF3D, SPAR3D, TRELLIS, Wonder3D, SAM 3D, Hi3DGen) on 69 manuscript figures from two collections

Hi3DGen identified as best starting point, balancing topological quality with detailed surface geometry through normal bridging approach

Pipeline combines SAM segmentation, Hi3DGen mesh generation, expert refinement in ZBrush, and AI-assisted texturing for enhanced accuracy

Results reveal trade-off between volumetric expansion and geometric fidelity when reconstructing medieval miniatures

Processed case studies from Gothic illuminations in the Decretum Gratiani manuscript held at Vatican Library