The Shell command currently uses the same basic algorithm as OffsetSrf with sharp corners. With complex objects, this usually fails to make a solid, and often leaves quite a mess… There are no options for this command other than thickness.
OffsetSrf with the round corner option is currently much more reliable, producing much more usable results requiring many fewer corrections. Therefore I think it would be good to offer the round-corner option in Shell as well.
Below a moderately complex object made with planar surfaces, the results of Shell and OffsetSrf/Sharp are pretty poor, while the result with OffsetSrf/Round is nearly perfect, leaving only two small naked edges which are easily fixed.
The OffsetSrfs are made after having deleted the bottom face, with flip to inside and Solid=Yes. In the images below, the backfaces are red, indicating open, unjoined surfaces.
I don’t recommend doing this for anything but testing, but for now you can access it with testSelectShellingType. I really don’t know what to expect if you do. In your example, it does a better job. I put the test command in there when I first started messing with shelling, and don’t recall if any code has been written since that assumes surfaces will be extended rather than blended.
Thanks. Still not working like OffsetSrf though, if I run testSelectShellingType and set it to “blend”, then try to shell the object labeled “original” in the file above with a thickness of 2, I get a message “Thickness is too large for this polysurface.” and nothing happens. It starts working at 1.7 - where it actually makes a perfect closed shell - and under, where it works mostly, but still leaves open spots or naked edges.
If you say this code has not been developed, that’s too bad, as currently sharp extensions are not working with anything very complex. Of course, that would be ideal, but for these types of objects - 3D printing parts - a shell that nearly works with rounded edges is far more useful to me than a mess with straight edges…
