Rhino ShrinkWrap: choose settings and check what changed

Use ShrinkWrap when a new mesh around the selected geometry is an acceptable way to reconstruct the shape. Preserve the input, identify the dimensions and openings that must survive, and compare those features before accepting the result. A closed mesh alone does not establish that the reconstruction kept your design intent.

This is a documentation-based guide to Rhino 8 for Windows. No model was reconstructed or measured for this article, and no print result is claimed. Settings must be selected against the actual geometry and receiving requirements.

Choose reconstruction or local repair

An existing mesh with a limited, understood defect may be better served by a targeted change. The MeshRepair panel exposes errors and repair tools so that you can inspect the actual problem rather than replace the whole shape.

ShrinkWrap creates a new wrap from inputs including surfaces, SubD, meshes and point data. McNeel presents it for combined geometry and damaged scans, among other uses. That makes it worth evaluating when reconstruction is acceptable, while feature fidelity remains something you must inspect.

Scroll the table horizontally to read every column.

For NURBS source problems, the geometry QA guide provides broader context. This procedure focuses on accepting a reconstructed mesh rather than diagnosing every source defect.

Preserve a reference and define the comparison

Save the editable input and work on another revision. Keep the original geometry on a reference layer and leave Delete input objects disabled for this comparison workflow. Rhino documents that this option removes inputs when output creation succeeds; it is not a backup mechanism. See the Rhino 8 ShrinkWrap command reference.

Before previewing, write down:

• Input types, Rhino build and model units.
• Required components, openings and spaces between parts.
• Critical dimensions, small details and their reference geometry.
• Agreed acceptance limits and who can resolve an uncertain feature.

Confirm the model units before interpreting distance-based controls. Avoid changing units during a comparison without recording whether geometry was scaled and checking a known dimension again.

Connect each setting to a design risk

Start from the actual preview and investigate one relevant change at a time. This matrix links documented behavior to a question about the design; it supplies no universal preset.

Scroll the table horizontally to read every column.

The command reference defines these controls and interactions. Fill holes has no effect for point/point-cloud input or while Inflate Vertices and Points is enabled. A smaller target edge length is not a promise that an intended opening or dimension will survive every other setting.

McNeel's worked tutorial shows how smoothing can shave convex corners and fill concave ones. Its incomplete-scan example also prepares large missing areas before wrapping. Treat its values and patches as particular demonstrations. They do not define a preset or the unknown intended surface of your scan.

If a feature is lost, return to the risk column: check the input reference, applicable filling/inflation mode, smoothing and resolution. Compare the changed result with the same feature before trying another adjustment. If no acceptable reconstruction preserves the required form, revisit the source or repair approach rather than approve closure alone.

Compare topology, dimensions and detail separately

First inspect the new mesh with Check, the repair panel and ShowEdges as relevant. Record structural errors, open boundaries and non-manifold findings against the required output. These checks answer different questions from dimensional acceptance.

An intended through-hole is a passage in the shape; it is not necessarily an unjoined boundary. Do not remove every opening merely to eliminate a visual hole. ShowEdges identifies open and non-manifold edges, while your reference defines which passages must remain.

Next compare repeatable landmark dimensions and important features with the preserved input. Distance reports the separation of two selected points. Record how those points were chosen and the units. A matching pair of landmarks does not inspect the entire surface, so examine small recesses, gaps, component separation and missing regions independently.

For an optional sampled comparison, PointDeviation can measure the reconstructed mesh's vertices against a preserved supported target: a curve, surface, polysurface, extrusion or mesh. Do not select a point cloud or SubD directly as that target. If no appropriate reference exists, record the limitation rather than invent a substitute reference.

Retain the measured object, target, units, point count, settings and statistics. Ignore distance can omit points, while Proximity angle qualifies the displayed hairs. Record those settings and the Bad point setting when interpreting the comparison; the visible hairs are not an unfiltered surface test. This is a directional sample comparison, not a continuous global or bidirectional error certificate. A favorable sampled maximum does not establish that a missing recess was recreated, and isolated matching points do not prove feature preservation.

Use an acceptance record for the reconstructed revision

Scroll the table horizontally to read every column.

Consider a hypothetical clip whose drainage channel must stay open and whose mating faces have a defined separation. The record should identify both requirements before wrapping. After each candidate reconstruction, the reader records the actual channel inspection and face measurements. Leave those observations blank until performed. A closed result would leave both functional questions unanswered without those checks.

After accepting the reconstructed revision, verify the saved export and its interpretation in the receiving application as a separate handoff task. Keep the native reference, accepted mesh and export identifiers together. The Rhino handoff guide explains the broader ownership and packaging questions.