STL print preparation guide
How to Check STL Wall Thickness for 3D Printing
Substantially updated 2026-10-16. First published 2026-09-05.
Confirm the STL's physical size before interpreting a wall result. Then choose a review threshold, inspect measured and unmeasured samples separately, and verify critical features in the source model and sliced layers.
1. Set the scale before the threshold
STL stores coordinates without a declared physical unit. A cube spanning one coordinate unit could be read as 1 mm or 25.4 mm under different assumptions; neither proves the author's intended size. Check a known dimension in STL Dimensions Checker. In the STL Wall Thickness Checker, choose the STL coordinate unit. Its Review threshold and reported samples use that selected unit. A physical 0.04-inch target equals 1.016 mm; enter 0.04 with inches selected or 1.016 with mm selected. Changing the unit clears the earlier result and requires a new run.
2. Run the known thin-bar control
Download the closed thin-bar STL or load it on the checker. The three designed bars are 20 mm long, 10 mm high and 0.3, 0.45 and 0.9 mm wide. Select mm, enter a 1.2 mm review threshold and run the local analysis. The current centroid-ray implementation, checked 16 October 2026, returned:
| Attempted | Measured | Below 1.2 mm | Unmeasured | Smallest measured sample |
|---|---|---|---|---|
| 36 | 36 | 12 | 0 | About 0.29999 mm |
The source widths and measured ray distances answer different questions. Faces along a long dimension can return a larger distance; the roughly 0.3 mm sample includes the ray offset and is not a metrology accuracy claim. The result samples these 36 triangles, not every point in each bar. Reproduction settings and the original slicer exports are retained with the thin-feature example.
3. Read the result without over-interpreting it
| Observation | What to check |
|---|---|
| One or more measured samples below threshold | Find the marked region; compare intended CAD material with the sliced paths. The count is rays, not separate defects. |
| Some sampled locations unmeasured | Inspect open or ambiguous geometry. An absent opposing hit is neither zero thickness nor a pass; it can coexist with thin measured samples. |
| No measured sample below threshold | Review critical details and unmeasured locations. Finite input-order sampling can miss a narrow rib. |
| Unit changed | Rerun after confirming a known dimension. The old result is invalid for the new physical interpretation. |
The open-face cube is an optional no-hit control. At the same mm assumption and 1.2 mm threshold, the current checker attempted 10 samples, measured eight and left two unmeasured; none of the measured samples was below threshold. Removing one face does not make every ray miss: inspect which sampled directions still find another surface. Use a visual cross-section when an internal passage is the concern.
4. Change the feature that needs changing
In the source CAD, increase a specific wall or rib when its role requires more material. Global scale changes every dimension: a 1.2 mm wall scaled to 75% becomes 0.9 mm, while a 10 mm hole becomes 7.5 mm. That is dimensional arithmetic, not an observed print. Prefer a local parametric edit when holes, mating faces or clearances must remain fixed. Use the wall-thickness decision guide to judge the feature's process and use.
5. Confirm the saved copy and sliced layers
Export a new STL copy, recheck its physical dimensions and topology, then inspect the same critical region with the sampled checker. Slice that copy with the intended profile and inspect native perimeter paths layer by layer. In a retained PrusaSlicer 2.9.6 software experiment on the thin bars, Classic and Arachne produced different second-layer paths under recorded settings. The full comparison explains its settings and limitations. Neither software preview is a physical print test.