Hello! I have worked with these files: 1.3dm (402.7 KB) 1.gh (20.2 KB)
I am pretty happy with it so far, but there are often weird surface irregularities. I have also tried to add many different mesh smooth components, but none worked well.
This looks like it is caused by the input curves rather than the downstream processing of the mesh.
To make smooth clean pipes, it helps to have an input without very short curves, tiny loops, or very tight angles.
I would recommend to fix this by simplifying the curve network before piping it.
I’m not familiar with the MeshEdges component that you use, but judging from your screenshot, you only seem to be utilizing the naked edges of your mesh. Try the AE output instead.
I also don’t get why you cast the MP output, which is a subdivision surface, to a brep (or polysurface) and then to a mesh. You can simply extract the subd control polygon, which is a mesh. If you want a smoothed mesh instead that more closely resembles the multi-pipe result, you can subdivide it with Catmull-Clark-subdivision from for instance the Weaverbird plug-in.
Looks right to me! Now, if you want it smoother you need to dial up the subdivision levels - input L - of the Catmull-Clark-component. Two is usually a good compromise between smoothness and performance. Three is the maximum and super smooth.
If you see any dark blotches after smoothing, you can try to unify the mesh normals before the preview component.
I think these are the node points where different curves come together. Is there a way to detect these points and tween them smoothly into the other mesh?
Can you upload your file, but with the Crv component data internalised? I don’t have the plug-ins you have so I can’t inspect the definition otherwise.
Thanks, MultiPipe simply doesn’t seem to be up to the task in your case. For high valence vertices in your line graph - points with many adjacent edges -, it produces kinda undesirable geometry that leads to the darker blotches.
To remedy this, I’ve remeshed it with QuadRemesh, which gives a nice result in my opinion and the blotches are gone.
This isn’t actually specific to MultiPipe - in all SubD surfaces (either as actual SubD objects or subdivided meshes) high valence vertices do not give great tangent/curvature continuity around the vertex, particularly when the surrounding faces are also elongated. This is a known limitation of Catmull-Clark subdivision.
I know there has been some work going on looking at improving this continuity when converting SubD to NURBS.
QuadRemeshing is a good way of reducing the issue here though.