Slicer-independent generation of customizable non-planar toolpaths for fused filament fabrication

This paper presents a slicer-independent Python workflow for localized non-planar finishing in fused filament fabrication. Candidate surface patches are selected from STL geometry using an inclination criterion, processed independently, populated with a planar zig-zag reference path, vertically projected onto the selected mesh, and exported as non-planar G-code. A separate integration stage registers the non-planar block in XY, inserts it at a user-selected position in conventionally sliced planar G-code, and adds transition moves where required. The workflow is an interoperable proof-of-concept rather than a general-purpose non-planar slicer or optimized computational-geometry kernel. Experimental validation with PLA produced executable finishing passes with no observed collisions and visibly reduced staircase artifacts on shallow, externally accessible surfaces. For four representative geometries, the fraction of scanned upper-surface points within ± 0.05 mm of the nominal CAD surface increased by 0.6–31.0 percentage points after non-planar finishing. The experiments used a modified three-axis printer equipped with a 0.4 mm nozzle extended by 5 mm, while the stock part-cooling fan and duct were removed to increase clearance. Compatibility therefore primarily concerns standard slicer, firmware, and G-code process chains, while printhead accessibility and cooling-system design remain hardware constraints. Runtime measurements identified pointwise ray–mesh projection as the main scalability bottleneck at high mesh and trajectory resolutions.