V6 vs V5 Issue

The resulting surface from Sweep2 (also loft) give undesirable and different result from v5 vs v6. The v5 result is good. The v6 result has issue. The shading of the surface even looks odd in v6. srf_issue.zip (40.8 KB)

Hello - use the Refit rails setting in the Sweep dialog box to change this.


That works, thanks.

Hi @pascal, without the refit rails what is behind
the wierd surface surface created?
Does it tak into account curvature and pushes the result outward? in this case for example?

Actuall, what does the refit do?

The surface is not weird, just different.

(Note: these comments are based on experimentation with Rhino and general knowledge of NURBS modeling and surface modeling. I am not associated with McNeel.)

Sweep2 has to decide how locations along the two rails correspond with the resulting surface isoscurves running between the corresponding locations. At the sections it is clear that the location of each rail at a particular section correspond. But what is the correspondence in between the sections? Sweep2 with the Simple sweep option in V5, and Sweep2 without options in V6 use the input rail curves’ parameterizations to determine the relationship between the rails. If the input rail curves have similar parameterizations then the resulting sweep will usually be what is “expected”. However if the parameterizations differ significantly the resulting sweep may appear to be unusual.

Rebuilding the rails before creating the surface, which is default in V5 and optional in V6, changes the parameterization of the rails and makes it similar for both rails, and in general about proportional to distance along each rail. The result is a surface which usually appears as “expected” but which generally has more control points.

“Slashes” is a method in Sweep2 for the user to directly influence the relationship between the two curves without or without rebuilding the rails. The curves are rebuilt such that surface is generated with an isocurve extending between the points on the rails which were at the end of the each slash.

Example with several different versions of Sweep2 results in V5 and V6: Sweep256.3dm (78.5 KB)

Note that when rails are rebuilt the edges of the resulting surface usually will not be exactly on the original input rails. The difference should be smaller than the absolute tolerance.

@davidcockey Is there a good rule to follpw when making slashes? I just put them if surfaces crosses each other or when isocurves gets too close to eatch other. But they do change the resulting surface a lot, so just wondering where not to put slashes.

Other quesation I had is if the rails are edges of surfaces instead of curves. Wouldn’t altering the incoming edges isocurves create surfaces that are not tangent even if you select tangency option?

The purpose of adding slashes is to make the surface better. Better means having a structure more fit for the purpose it is serving. In general the most robust and useful NURBS surfaces that you can get from sweep2 have isocurve structure where each isocurve is as close to planar as possible and the U and V isocurves cross each other as close to 90 degrees as possible. Adding a slash can be helpful in straightening out areas which you see in the preview are less than ideal.

I can’t picture how slashes help in those situations. Can you post an example?

1 Like

Hello -

At the risk of ‘TMI’ - ‘traditionally’, Sweep2 has always refit the rails - that is, it does not pay attention to the exact point and knot structure of the rails, it makes new, compatible rail curves that will work and be within some tolerance of the input curves. In short, it never cared whether the two rail curves selected by the user were clean and compatible. That’s good in that it did not make demands on the user to pay attention to anything but the 3d shape of the curves, but one undesirable consequence of this is that Sweep2 generally makes pretty dense surfaces with double knots at the location where the sweeper adds (internally) cross section curves to generate the surface.

In V5 Lowell added the ability to make sweeps that do pay attention to the rail curve structure if conditions are right: The rails do match in structure and any cross section inputs are located exactly on matching pairs of edit points. That is the ‘Simple sweep’ option in V5 Sweep2.

V6 simply makes the ‘siimple sweep’ if conditions allow it and makes the user choose otherwise with 'refit rails."
I’m still on coffee #1 but I believe the above is more or less accurate.


1 Like

Thank you.
I was thinking V5 simple was sort of rebuilding the curves and sort of refitting. So I rarely used that…

For V6 I was wondering why the refitted option was not default.
I guess it’s best to go about doing a sweep2 without refit rail if it works, but if not try refit.

Best way to make input curves would be to have equal number of control points and as @jim mentioned making control point layout so that resulting surface segments would be planar as possible and UV isocurves close to 90deg as possible?

It helps to know what the tool is looking for so I can try to be more careful in creating curves. Thanks

The old behavior has been in Rhino since the start. I don’t understand why it has changed. Seems more confusing for the user now.