Default Offset Setting

[ParvGuptaSpCT]

Hello Devs,

I'm trying to replicate the mesh offset as show in the image below, can you tell me which offset method has been used and what are the default settings used in MeshInspector UI

Thanks!

[Grantim]

Hello!

Expand/Shrink is doubleOffsetMesh: the mesh is offset outward by the value, then back inward by the same value (Shrink/Expand swaps the signs). The result is then cleaned up with postprocessMeshFromVoxels. With the settings from your screenshot:

from meshlib import mrmeshpy as mm
from meshlib import mrcudapy as mc

def expand_shrink(mesh: mm.Mesh, offset: float) -> mm.Mesh:
    box = mesh.computeBoundingBox()
    # "Standard" resolution: ~100M voxels in the box grown by the offset
    size = box.size() + mm.Vector3f.diagonal(2 * offset)
    voxelSize = float(f"{(size.x * size.y * size.z / 1e8) ** (1 / 3):.3g}")

    params = mm.GeneralOffsetParameters()
    params.voxelSize = voxelSize
    params.signDetectionMode = mm.SignDetectionMode.HoleWindingRule

    if not mesh.topology.isClosed():
        params.fwn = mc.FastWindingNumber(mesh)  # "Use NVIDIA GPU"
        mesh = mm.generalOffsetMesh(mesh, 0, params)  # close open mesh first

    res = mm.doubleOffsetMesh(mesh, offset, -offset, params)

    # "Decimate" + "Reduce Total Angle"
    post = mm.MeshFromVoxelsPostProcessingParams()
    post.voxelSize = voxelSize
    post.decimate = True
    post.reduceAngleNumIters = 3
    post.tinyEdgeLength = box.diagonal() * 1e-4
    mm.postprocessMeshFromVoxels(res, post)
    return res

mesh = mm.loadMesh("input.stl")
mm.saveMesh(expand_shrink(mesh, 0.25), "output.stl")

CUDA only speeds up closing an open mesh; a closed mesh is processed on CPU anyway. Without an NVIDIA GPU, just remove the mrcudapy lines.

This gives the same result as MeshInspector (needs meshlib 3.1.2 or newer).

[ParvGuptaSpCT]

Thanks @Grantim