Try this operation with the new patch and compare the speed.
ok
patch-test tubi.3dm (624.5 KB)
In this case, Multi-Blend gave me a good result. The other two options were not satisfactory, unlike xNURBS, which performs better in this case.
@Quan_Li have you tried this yourself in Plasticity or do you base this of an instagram video only? The patch generation is only part of the total time it takes to create a patch. I would argue that selection of constraints and being able to go back into the patch to change settings after creation makes up for the generation time. Also take into account that we’re still in WIP phase.
It is your file. I failed to get G0 with the new patch. I also tried surface edges, with the same result. Since I only want G0 at those positions, curves and edges are just the same, I guess.
What are the tolerance settings in that Instagram video? Is it 0,001 mm for “Distance” and 0,05 for “Curvature”, like what you use in your Rhino?
By the way, what is the “Flat” setting and why it uses 1E+03 instead of 1000? Perhaps the idea is to change the last number to add another 0 (1E+04 is 10 000)?
@Quan_Li see the attached file. Technically not fully G1 / G2 at the edges, but the majority of the edges are. Inspect with GlobalEdgeContinuity and by double clicking the Patch surface.
patch-test tubi_sg.3dm (972.4 KB)
@menno this is abit OT but I remember the original announcement for the new Patch command saying it would let us to swap between Nurbs, SubD and Mesh but I can’t see this in the latest implementation
See this: Rhino 9 Feature : Multi-sided surface patches
Is it something that is lost or it will be added in one of the next releases?
I think it’s a shame to not have this capability.
The topic you link to is obsolete. Recently, mesh and point cloud input was added. SubD has never been on the table.
Yes the link is at the old post.
I probably misunderstood because I tought the Patch to be able to BUILD meshes and SubD as well and not only accept it as input.
Becasue nurbs is quadrangular and your algorithm is making nurbs it look possible to use it also to generate quad mesh and then subd. Of course the trim would not be possibile but it’s something that may worth to think of… just after the bug fix.
Thanks for the attention.
patch test.3dm (203.2 KB)
I wasn’t able to patch either of the two…
The only valid result is by using multiblend option on the right one.
Attached is a way to do both. I don’t understand why the five-sided patch was giving you problems. The quad-pipe one is, I think, better done in multiple stages.
patch-test-MDvR.3dm (704.5 KB)
(60 seconds video)
patch-test-3.3dm (201.7 KB)
In the video at 11 seconds and 22 seconds i try to quickly change the U and V spans by entering a number , TAB , entering a number, this cause the solver to fail. I have to push the G2 button to trigger recalculation.
Instead at 33 seconds I first change U spans, wait for recalculation, then change the V spans, and wait, everything is fine this way.
At the end, similar problem by setting the internal curve from “G0” to “I”: quickly clicking both curves make the solution ignore the second change (and the trim of the surface still fails), and I have to press the G2 button again.
Those problems can be solved by pressing ESC key before the preview calculation is finished.
(not showin in the video)
Quick sequence:
Uspan change > TAB > Vspan change = fail
Uspan change > TAB > ESC > Vspan change = success
or
Quick sequence:
G0 to I > G0 to I = fail
G0 to I > ESC > G0 to I = success
Proposal:
any current change in the settings or input data should abort current preview calculation, and trigger a new calculation.
In the U/V spans maybe consider having the TAB key not triggering a recalculation at all, but wait for an Enter keypress instead. (user often change both U and V .. ?)
I only learned it for one hour, I don’t know how to flatten the surface yet but xNurbs in Plasticity has very impressive speed, almost real time.
@Quan_Li For me, it’s fine.
@brvdln setting spans to 20 should not be necessary, especially not for multiblend, and significantly slows down things.
side note: did you know you can enter 0 for 0,0,0?
@Gijs No, thanks, I didn’t know that. I’ve always been used to entering 0,0,0. As for the spans, yes, I tried several settings; I didn’t record it in the video, but the results are good.
What happens if you set the angle to 0,1 degrees? In my opinion, a value of 1 degree is way too loose for a product design.
Also, if you’re going up to G2, note that the supporting surfaces between the pipes are only G1. If you upgrade those to G2 prior to matching the central patch, the matching will be possible down to angles of 0.02 degrees.