How We Test
Every score on this site traces back to one of the three protocols below. They are versioned; articles state which version their data used.
AI 3D tool protocol (the one nobody else runs)
Benchmark set: 10 standard images (organic, mechanical, miniature, low-poly, high-detail) — identical inputs for every tool.
Metrics:
- Watertightness % — share of outputs passing a manifold check unmodified
- Minimum wall thickness — median across the set (mm)
- Print success % — share that slices and prints acceptably on both reference printers
- Generation time, effective cost per generation, commercial license terms
Reference printers (kept constant): Bambu Lab A1 (0.4 mm, stock profile) and Creality Ender 3 V3 SE (0.4 mm, stock profile).
Printer protocol
- Benchy at stock 0.2 mm — blind-scored quality (1–10)
- Speed test model at advertised max vs. practical settings
- Dimensional accuracy coupon — max deviation (calipers)
- Noise at 1 m during steady-state printing
- Reliability: successful print hours across the first 20 hours
Filament protocol
- Diameter tolerance — 10-point measurement per spool
- Dry vs. wet strength coupon (72 h at 60% RH)
- Temperature tower quality
- Price per kg at time of testing
What we have and haven't tested yet
Honesty first: this site launched recently. Some early reviews combine manufacturer specifications with community-reported data, and are labeled as such. Hands-on protocol data accumulates continuously, and articles are updated when it does. If a number on this site doesn't come from the protocols above, the article says where it came from.
About star ratings: ratings shown as "community est." are our editorial synthesis of community consensus — a starting point, not a measurement. When a product completes our protocol run, its rating switches to our own measured score.