Uploading an STL for a measured DFM check

Overhang angle is measured exactly from the mesh's face normals — every face steeper than the process's support-free limit is found, not estimated.

Wall thickness is sampled — rays cast through the mesh at points across its surface, with the 5th-percentile thinnest reading reported. This catches the thin spot that matters (a wall that's mostly 2 mm but drops to 0.6 mm at one rib) rather than an average that would hide it, but it's a sample, not an exhaustive CAD measurement — the report says so explicitly.

When geometry is measured, DFM trusts it over anything stated in the brief's text — a candidate marked as coming from a measured mesh skips the text-based feature scan entirely. That's deliberate: if you uploaded a real part, guessing from prose would be a downgrade.

• 8 MB per file in the workspace (4 MB on the free instant check) — for a bigger mesh, decimate it first; DFM only needs enough resolution to sample wall thickness and face angles, not display fidelity.
• Binary and ASCII STL are both fine.
• A non-watertight mesh still analyses — the report flags it, since volume and wall-thickness sampling are less reliable on an open mesh.

You don't need a finished model. A rough block-out with the real wall thicknesses and rib geometry gets you a real, specific DFM read — worth doing before you invest more time in a polished model that might just need those same walls thickened.