Validation & Coverage

Every check the audit runs, named in plain language — so you can tell the difference between "we looked and it passed" and "this check didn't apply to your file".

Numbers reflect the current engine build — we refresh this page each release.

44
checks on a typical FDM audit (resin: 36)
3
groups: Topology · Readiness · Inspection
5 slicers
Orca · Prusa · Bambu · CHITUBOX · Lychee — daily sync

The check lists below cover both FDM and resin runs. Some checks only apply to one technology — resin-only checks are marked (resin), FDM-only ones (FDM). Whatever doesn't apply to the file you uploaded is reported as skipped, not silently dropped.

Topology — 11 checks

Is the mesh itself sound? Independent of which printer you'll use. Runs on every audit, including the free browser Quick Check.

  • Is the mesh fully closed?
  • Edges shared by more than two faces
  • Overlapping surface pairs
  • Empty / zero-area faces
  • Duplicate faces
  • Open edge loops (cracks)
  • Disconnected shell count
  • Bounding box sanity
  • Triangle count
  • Vertex count
  • Edge-length distribution

Readiness — 16 checks (FDM) · 14 (resin)

Will this specific printer + material handle the mesh? Requires a recipe (printer + material profile). Some checks fire only for one technology.

  • Fits the build volume
  • Wall thickness vs profile minimum
  • Thin wires / small features
  • Minimum feature size vs nozzle (FDM) / pixel (resin)
  • Overhang angles + support need
  • Support lands on the model surface
  • Bridge spans
  • Pre-supported mesh detection
  • Scan / organic mesh detection
  • Stress check (optional, opt-in)
  • Layer adhesion (FDM)
  • Stability & tip-over risk
  • Print orientation
  • Suction cups / drain holes (resin)
  • Enclosed cavities / trapped volume
  • Volume + weight estimate

Inspection — 17 checks (FDM) · 11 (resin)

Looks at the slicer output — the G-code your printer will actually run. Most accurate when you upload your own slicer output. A mesh-only audit still returns a first-pass estimate, useful for screening before you slice.

  • Layer count
  • Total print height
  • Layer height
  • Estimated print time
  • Resin weight estimate (resin)
  • Slicer detected
  • Nozzle temperature (FDM)
  • Bed temperature (FDM)
  • Max feedrate vs printer limit (FDM)
  • Volumetric flow vs hotend capability (FDM)
  • Minimum layer time (cooling headroom) (FDM)
  • First-layer speed (FDM)
  • First-layer island count (FDM)
  • Retraction pattern (FDM)
  • Travel ratio (FDM)
  • Nozzle collision with printed part (FDM)
  • Print-orientation suitability
  • Floating islands (resin)per-layer connected-component, support detection
  • Vacuum traps (resin)3D voxel flood drains hollow + drain shells — eliminates 2D false positives
  • Peel-force spikes (resin)>3× cross-section jump per layer
  • Printer-profile mismatch (resin)file-recorded build dims vs selected profile
  • Lift-cycle divergence (resin)file lift / retract speed vs material recommendation
  • Coverage of the inspection check list

How we measure

  • Test corpus~1,000 real-world meshes — hobbyist prints, CAD exports, AI-generated outputs, scanned figurines
  • Regression suite965 test cases / 108K assertions on every release
  • Slicer parityFilament-weight median +4% vs OrcaSlicer on 71 stratified samples; 75% within ±30%
  • Slicer-aware G-code parsingAMS / prime-line / wipe calibration phases (Bambu / Orca / Prusa) are gated out of the first-layer-island, first-layer-speed, and retraction-count rules. Measured: a single Bambu X1C print produces 1027 false-positive islands + 17k+ false retractions under a naive parser; admeshio counts only what fires after the first model-feature tag.
  • Published datasetAn independent validation dataset with per-check precision / recall is on the roadmap

Each report tells you the share of checks that actually ran on your file — so you always know how much of the list applied to the upload.

Questions or a check you'd like us to add? Contact us →