3D Print My STL

What it actually tests.

No black box. Here is every check, why it matters, and what to do when it fails.

Watertight and manifold

The tool welds vertices onto a 0.0001 mm grid — exporters rarely write the same corner identically twice — then counts how many triangles use each edge.

Exactly twoCorrect. The surface is closed there.
Only oneAn open edge. The solid has a hole in it.
Three or moreNon-manifold: surfaces meeting in a way no real object can. Usually overlapping bodies that were never merged.

What to do: most CAD packages have a repair or stitch command. If it will not close, the geometry underneath is usually the problem, not the export.

Face direction

Triangle winding order decides which side is "out". We compare every shared edge between neighbouring faces: in a correctly wound mesh the two faces traverse that edge in opposite directions. We also compare against the normal vector the exporter recorded, which sometimes disagrees with the geometry it wrote.

A model can also be entirely inside out — consistent with itself but inverted. Many slicers cope silently. Some do not, and the failure is baffling when it happens.

Separate shells

We group connected triangles to count how many distinct objects the file holds. Two is fine if you meant two. It is a problem if you meant one, because parts that merely touch on screen are not joined.

Degenerate triangles

Triangles with effectively zero area carry no surface and can confuse slicers. A handful is normal, especially around the poles of a sphere. Thousands means the export went wrong.

Scale and units

This is the most common real-world mistake and the easiest to catch. STL has no units, so everything assumes millimetres.

Build volume

Checked against a 256 × 256 × 256 mm cube, which is what our machines run. We sort your part's dimensions against the bed's, so a part that only fits when turned is reported as fitting — rotating it is our job, not yours.

Thin overall geometry

A 0.4 mm nozzle needs roughly three perimeters to make a wall worth having, so anything under about 1.2 mm is flagged.

An honest limitation: this measures the part's smallest overall dimension, not the thinnest wall inside it. Proper wall-thickness analysis needs a full solid sampling pass, which is more than a browser tab should be doing. A thin internal rib can still pass. If a part is right at the edge, ask us and we will look at it in the slicer.

Weight and time

Volume comes from summing the signed volume of the tetrahedron between each triangle and the origin — exact for a closed mesh, and accurate to within about 0.05% on a smooth model. We then apply the shell-and-infill ratio for your chosen layer height, multiply by PETG's density of 1.27 g/cm³, and estimate machine time from the layer height and typical speeds.

It is the same arithmetic our quoting tool uses, so the figure here matches the price you would be given.

What it does not check