SOLVESPACE -- parametric 2d/3d CAD

(you are viewing a thread; or go back to list of threads)

tessellation problems (by VisibleCode)
I very quickly run into problems with SolveSpace's tessellations when there's more than a few intersections going on. Attached is a screenshot of an experiment with modeling a room with vaulted ceilings that didn't work out very well.

Here, in order to narrow down the problems, I've cut things down to just the arch profile, duplicated 3x with a translation group, and then duplicated once more with a rotation group at a 90 degree angle. Still pretty messy.

Is there a better approach to modeling this kind of thing which doesn't stress tesellation so much?
Wed Mar 22 2017, 04:19:53, download attachment screenshot.png
(no subject) (by Jonathan Westhues)
I think that's the NURBS operations failing, not the tessellation into triangles. You can switch to mesh operations ("force NURBS surfaces to triangle mesh") if you don't need STEP output, and/or try to avoid coincidences (e.g., a vertex or edge that lies on the surface of an existing volume, vs. definitely inside or outside it) when you create your model.
Wed Mar 22 2017, 05:01:11
(no subject) (by VisibleCode)
Forcing NURBS surfaces to triangles makes it slightly worse (though not significantly different), actually.
Thu Mar 23 2017, 01:40:31
(no subject) (by Jonathan Westhues)
Try adjusting model to avoid coincidences, or post the .slvs file?
Thu Mar 23 2017, 01:45:20
(no subject) (by VisibleCode)
Put together a marginally simpler example, attached. Not really sure how to avoid coincidences with this sort of construction.
Thu Mar 23 2017, 02:30:54, download attachment inverse-vault.slvs
(no subject) (by VisibleCode)
(Ideally, I'd shift the 90-degree-rotated group down by a tiny amount, but that doesn't seem to be possible with rotated groups?)
Thu Mar 23 2017, 02:34:06
(no subject) (by Jonathan Westhues)
I think you might have just found a bug in how "force to mesh" gets handled when we step and repeat a solid operation, and that despite your instructions the Boolean is getting performed as NURBS.

If you step or rotate the section instead of the solid model, then everything seems to work, as attached.
Thu Mar 23 2017, 02:51:30, download attachment vault.slvs
(no subject) (by VisibleCode)
That looks satisfactory; thanks!
Thu Mar 23 2017, 03:00:28
(no subject) (by whitequark)
BTW, this is a known bug, and it's going to be fixed soon.
Thu Mar 23 2017, 06:35:35
Post a reply to this comment:
Your Name:
Your Email:
(no HTML tags; use plain text, and hit Enter for a line break)
Attached file (if you want, 5 MB max):
© 2008-2018 SolveSpace contributors. Most recent update Nov 22 2018.