partforge

Mesh rebuild scan

hard tier · create

The request

Attached is a rough phone scan (STL) of a small block with one hole through it. Rebuild the clean part this is a scan of, as one sub-part named `block`, standing where the scan stands (Z up, same origin, resting on the same plane), with controls `length` (±10 mm) and `boreDiameter` (3–12 mm). The scan is noisy and may not be watertight — measure it any way you can and prefer round numbers where the scan is ambiguous. Keep the mesh imported in the part while you work.
A render of the mesh the model was given
Mesh · made with Blender 4.5.1 LTS (voxel remesh at 0.45 mm, clouds displacement at 0.10 mm, collapse decimation to 0.55) · CC0-1.0

What the AI judge checks

  • The overall form matches the reference: the same kind of object, the same silhouette from the matching angle
  • Proportions match: the relative sizes of the major dimensions agree with the reference within what the eye can judge
  • Every major feature visible in the reference (holes, slots, bosses, tabs, lips) is present in the result
  • Those features sit where the reference has them
  • Nothing is present that the reference does not show — no decorative or unexplained geometry
  • A clean block with flat faces and one round hole — the scan's noise is gone
  • The hole is visible from both the top and bottom views

Measured checks

  • one sub-part named block (gate)
  • bounding box within 0.5 mm of the mesh (gate)
  • volume within 5 % of the mesh (gate)
  • symmetric difference under the measured floor
  • the bore is 10 mm
  • the bore goes all the way through
  • the imported scan is still declared as an import
  • asked the oracle at most once

Ranked by hand

One reviewer put one attempt from each model in order, best first, ties left tied.

  1. #1
    Opus 5
    Opus 5's ranked attempt
  2. #2 tie
    Gemini 3.8 Flash
    Gemini 3.8 Flash's ranked attempt
  3. #2 tie
    GPT-5.6 Sol
    GPT-5.6 Sol's ranked attempt
  4. #2 tie
    GPT-6 Astra
    GPT-6 Astra's ranked attempt
  5. #2 tie
    Fable 5.1
    Fable 5.1's ranked attempt

Ordering by hand decides Human rank and Human rate on the leaderboard; it never affects Overall or Strict.

Every model's attempts

Run baseline-2026-092026-09-10commit 41aec7daspend $339.70 + $48.02 AI judging, 7 unpricedOther runs: 2026-09-04T20-02-10-stage2-hard-cases, 2026-09-04T19-48-31-screen-14-fixed