Research · Compiling intent to fabrication

Grading the slicing process

Toolpath planning and printer settings that vary continuously through a part, compiled into ready-to-slice projects for the slicers people already use.

The key idea. Functional grading usually means varying an object’s geometry or its materials. Here the design grades the slicing and printing process itself: infill, surface texture, nozzle temperature, flow, and color mixing change from point to point inside the part, and a compiler turns those fields into settings that existing slicers already understand.

Overview showing functionally graded design fields passing through the proposed compiler into a ready-to-slice .3mf project, with three examples, graded infill density, graded foaming rate through nozzle temperature on a bunny, and a cyan-to-magenta color gradient, each beside its printed artifact
Three kinds of process grading from one compiler. Toolpath planning settings (graded infill density), printer state (nozzle temperature, which sets how much a foaming filament expands), and color halftoning, each compiled into a ready-to-slice .3mf project and printed.

The problem. Slicers already control many of the things that shape how a printed part behaves, such as infill density and pattern, surface texture, nozzle temperature, flow, and material mixing. But they apply them to whole regions that users have to cut out and configure by hand, so grading a setting smoothly through a part means hundreds of manual steps or hand-editing G-code.

The idea. I treat the slicer as a compiler target. A design describes each setting as a spatial field. The compiler divides those fields into regions, extracts matching sub-meshes, and writes a native project file for PrusaSlicer or OrcaSlicer. The slicer then does its own toolpath planning, supports, and preview as usual, so the method works with existing printers and slicer profiles instead of requiring a custom slicer.

Three rows showing a designed field, its segmented regions, a .3mf project file, and the sliced G-code in an existing slicer, for infill density, Shore hardness through temperature and flow, and an RGB color image
How it works. The compiler partitions each designed field into regions and writes them into a .3mf project as settings-mesh regions (top, infill density), virtual-extruder assignments that carry nozzle temperature and flow (middle, Shore hardness), or mixture-ratio assignments (bottom, color). Everything to the right of the divider is done by an existing slicer, unchanged.

Key result: grading the process, not just the material

A single bunny combines three process fields. Surface texture is assigned by anatomical region and blended smoothly, while nozzle temperature and flow control how much a foaming TPU expands, which sets the local hardness between 65A and 85A. In calibration, measured hardness followed the designed gradient with a mean absolute error of 0.5 Shore A.

Bunny renders colored by semantic region and by designed and computed fields for surface texture, hardness, temperature, and flow rate
Fields for one bunny. (a) Semantic regions, (b) blended into smooth volumetric fields. (c, d) Fuzzy-skin thickness and point distance. (e) Designed Shore hardness, translated into (f) nozzle temperature and (g) flow rate.
A blue printed bunny with close-up insets showing foamed and unfoamed surface texture on the ear, foot, and tail
The printed result. Surface texture and foaming-TPU hardness vary together. Lighter regions are printed hotter, so they foam more and are softer; darker regions are denser and harder.

This work is with Devon Beck (Draper) and Robert MacCurdy, and the compiler is part of OpenVCAD.

Publications

  1. Functionally Grading the Slicing Process by Compiling Design Intent into Slicer Projects
    Functionally Grading the Slicing Process by Compiling Design Intent into Slicer Projects
    Charles Wade, Devon Beck, and Robert MacCurdy
    SCFProceedings of the 11th ACM Symposium on Computational Fabrication (SCF ’26), 2026