Local revision inspection
Compare Two STL Files
Inspect before and after exports in one coordinate frame, then measure sampled distances to the other surface. Models are never silently scaled to match.
Choose revisions
Reviewing a size-reducer candidate? Compare it inline in the reducer using this same method while retaining the original and candidate. This pair stays under your control. For centre alignment here, deliberately load the original and downloaded candidate.
Up to 5 MiB and 100,000 triangles per STL. Unit assumptions apply independently; results are in millimetres.
Original coordinates retain placement differences. Explicit centre alignment translates bounding-box centres only; it does not rotate, rescale or register features. Record its offset if placement is intentionally irrelevant.
Ready for revision inspection.
Alignment: original. Applied after translation: 0, 0, 0 mm.
Try an example
Can different triangles describe the same surface, and can an outside face really move?
Settings: Use millimetres and original coordinates. Loading is local and replaces only after your explicit action; no server analysis is submitted.
Before: baseline-original.stl. Both revision slots will be replaced.
Downloads this public control, then opens it locally. Your own file can still be chosen above.
Read the explanation · Download and compare controls · Methods and processing
Revision metrics
Dimensions, triangle count and surface area alone cannot prove shape equality. Use the batch checker for screened closed-model volume; comparison does not assume a valid solid.
Read the comparison result
| Output | What it answers | Next action |
|---|---|---|
| Outer bounds | Overall span changes | Check expected scale and placement |
| Bidirectional sampled distances | Differences at the sampled surface locations | Inspect that region and unsampled critical features |
| Samples under tolerance | Those sampled points met the entered tolerance | Do not infer whole-model equality |
| Error, cancelled or unavailable attempt | No completed comparison for this attempt | Resolve the state or use another suitable application |
Deterministic area-distributed points are measured against triangles in both directions. Finite sampling can miss small features, and the sampled maximum is not a guaranteed upper bound over the whole surface. Automatic registration, hole recognition, Boolean difference and CAD acceptance are not performed.
Same count and bounds, changed opening
The owned 40 × 30 × 10 mm pair has 32 triangles and 1,684 binary bytes on each side, but its square opening changes from 8 to 6 mm. This is a deliberate mathematical edit, not a decimator output or physical fit test.
Use the known-revision examples above, then inspect the opening directly. The revision guide retains both measured directions, settings, downloads and a reusable decision record.
Check supported mesh findings separately, then review the exact saved derivative in the receiving slicer.