Select this rectilinear object with radiussed corners
Turn on Solid Object Points
Drag points orthoganally
Rhino thinks and thinks.
Chaos Ensues.
Any suggestions?
The project and a video of the problem is below.
PointsDragDistortsObject.3dm (930.4 KB)
Hello- run DivideAlongCreases > SplitAtTangents=Yes on the object first to split single faces at curvature discontinuites.
-Pascal
1 Like
Thanks Pascal.
- Any thoughts on how one gets surface discontinuities from sweeping the section shape around a rectangle to make this door frame? (Could the units have been too coarse for the fineness of the curves that were swept?)
- The āWhatā command says the object is a valid polysurface. Is that usual for an object with surface discontinuities?
The object is valid but some things work less well or not at all if a surface has curvature discontinuities - that is, it has spans that are only tangent continuous with neighboring spans - surfaces prefer to be curvature continuous. I donāt know what the inputs were but very likely SimplifyCrv on these could have helped. In general I try to keep surfaces split up - things just work better. There has been a good deal of back and forth over the years about how to handle this - sometimes it does make sense not to split things up. Just keep that DivideAlongCreases command in your hip pocket.
-Pascal
I"ve been using Rhino for four years now, but I donāt understand the concept of splitting in this context. I guess Iād better learn more. Not sure how to go about it. I take it āDivideAlongCreasesā somehow āsplitsā an object at these tangent-continuous-but-not-curvature-continuous joints. (Are they creases? I thought creases were coincident edges between spans with neither sort of continuity.) And this is a sort of splitting which doesnāt seem to compromise the closedness of a joint. I took a look at the help files for these commands, but got into a circular loop which seemed to require more understanding than I have.
Hello- there will be āfully multipleā knots at these locations - creases in wating as it were - if you were to move a control point that is there, well, other than in a constrained way perhaps, a crease (G0) would develop.
-Pascal
Thanks. This leads me to attempt to understand what ādividingā does to the geometry to solve this problem. It doesnāt seem to separate the surfaces, since the objects remain closed.
If the command were āReplaceAndRepairAllFully MultipleKnotsā (like with micro edges) Iād sort of get it, but this command seems to be doing more of a transformation.
Hello- it makes a single curvature-discontinuous surface into multiple faces - these are joined but explodeable into separate surfaces. These behave better in many ways and (in my experience) are generally preferable if there is not a need for a single surface.
-Pascal
Thanks Pascal. This has been helpful. I think I need to clarify what a crease is. Iāve been under the impression that itās a boundary of discontinuous curvature between surfaces. Based on your input here, it sounds like it can be a sharp kink in a single surface. This brings up an area of weak understanding for me - the difference between a knot, a crease, and a kink (if kinkās a Rhino term.) Iāve searched Help for definitions, but only operations instructions arise. If you can point me in the right direction, Iād appreciate it.
I think creases and kinks are basically the same - I guess a kink might be for curves and creases perhaps more referring to surfaces - I donāt know if that is strictly true. Where there are degree number of knots in the same place on a curve or surface (= āfully multiple knotā), the geometry can be and often is, smooth across that location but itās fragile - edits to the conrol point that is there (it will be on the surface or curve) will or can, cause a tangent discontinuity ( = an actual kink or crease)
Smooth, in this case G2
But there is a fully multiple knot in the middle:
Moving that point makes a kink form
That is why point-edting a sphere, for example, gets messy very quickly unless you Rebuild it first (removes the fully multiple knots that are at each quadrant, at the expense of true-sphereness), or constrain the point movements to a very small subset.
-Pascal