It would be really great if there was a way to use the Blendsrf command to blend more than two edges. Say I had 3 or 4 surfaces that did not touch and I wanted to create a surface that blended between all of them. I assume the resulting surface would have to be a trimmed surface of some sort based on tangency of adjacent edges. I could fake it now by blending the adjacent edges and then using the patch command, but the patch command usually is too inaccurate to be able to join the results afterwards. It would be great to have a tool that combines the accuracy of Blendsrf or Networksrf with the versatility of patch.
Similarly, if there was an option within Edgesrf to maintain tangency, curvature, etc. when selected edges are surface edges, that would be very helpful.
On a different but related note, there are certain instances where it would be really useful to be able to have untrimmed surfaces with more than 4 sides. It doesn't happen often, but when it does, it's really hard to work around.
Hi Jacob- I guess what you're asking for in the first bit is a G2 patch, correct? Can you post an example case? To your last point, one of the givens of NURBS surfaces is that they are bi-directional/four sided... no way around that, so at the moment, an untrimmed surfaces with more than four sides is not a possibility.
I don't necessarily want a patch that can match G2, so much as a patch that will actually match the edges given. If you look at the example file attached, the patch doesn't actually match the surface edges. That is the biggest issue I have with patch. It's sort of a last resort command when nothing else will work because it does not offer very much control and it doesn't really match the constraints given.
I really like the interface of the blendsrf command because it gives a lot of control over how it connects the surfaces. If it was possible to use that interface to blend between more than 2 surfaces, that would be fantastic.
The edge surface command is easier to fake by creating a 4th edge using the blendcrv command and then networksrf as shown in the example attached, however if it was possible to have more control similar to the blend srf command, that would also be nice.
Hi Jacob- is this an example of something you really want to do, or 'exaggerated for the sake of clarity'? It is pretty extreme - not really well suited to any clean surfacing. If it is not a realistic example, can you please post one that is? I don't see how to be of much help, to be honest, with these surfaces.
The last file was an example to try to illustrate in the simplest terms what I am looking for. I've attached a portion of an actual file I'm working on that is giving me trouble. As you can see, at the corners where there is the discontinuity, patch really isn't able to match the edges. However, I can achieve a pretty good surface using the blend surface command between any two of the edges, so long as I am careful to set the control points logically. I'm looking for that sort of accuracy between all 4 of the edges, knowing that this will actually create an 8 sided surface with extra edges to connect the discontinuities at the corners.
On an unrelated note, is there a reason that network surface won't accept closed polysurface edges as boundaries?
Thanks for your help,
Hi Jacob- if you trace the edges around the shape you have patched, you'll see this is not a realistic thing to surface to the edges but you can get something , maybe what you are looking for, if you turn off Adjust tangency in the patch controls. You'll have to do some work to trim that surface since the input curves do not form a closed loop but it may get you something.
Unfortunately, the tangency is pretty important for the project. I'm attaching a revised file that includes some surfaces I made with the blend surface command. They do a good job of matching G2 of the surface edges, but because I can only do 2 edges at a time, they don't align in the middle. If I could input all of the edges at once so that it could create a surface based on all of them (the same way that matchsrf will take into account multiple edges to find a surface that fits all of them), that would be great.
If all of the edges formed a closed loop I could use networksrf to fill them in, but because they are separate there aren't any tools currently capable of it (at least that I know of).
You need to resolve what's supposed to be going on in those corners better, which will presumably lead to breaking the problem down into some more manageable steps that will give you a clean result. Those "extra edges" you're looking for it to create, you need to define them.
Thank you for the feedback. The edges actually are defined, although they are very small. I've tried to break down the process into smaller steps. Unfortunately it always seems to have a hole at the surface creation step. There just doesn't seem to be a tool capable of creating that surface.
I'm not a programmer, so I don't really know what such a tool would entail, but it seems like the complexity should be at those new edges in the corners. The blend surface command resolves those new edges exactly as I'd like. It's the transition along the long edge boundaries that blendsrf doesn't resolve. I would have thought that would be the easier part since all of the other tools (networksrf, patch, etc.) seem to do okay at the long edges and fail at the corners.
Maybe what I'm asking for is impossible, but it seemed like it was worth wishing for.
When Pascal's reaction to your input geometry is Are you serious? then something is missing.
You're trying to "network" or "patch" a surface with edges that don't form an actual perimeter of a shape, let alone a single NURBS surface, so there's no surfacing command that's going to magically make this in one step. I have kind of a vague sense of what you might be trying to go for, but it would take a while to figure out how to get there from what you have.
What might help to figure out a structure would be to take the tangency out of the equation, either ignore that you want it for now, or rebuild things(this would require changes to your initial surfaces) assuming that you will put that "tangency" back in via fillet or blend operations.
Sorry this thread is out of order; it wouldn't let me reply to your last post.
I managed to create a polysrf of what I was looking for using blendsrf, sweep, matchsrf, & networksrf. I attached it so that you could see what I was aiming for.
There is a closed boundary for the surface. It is made up of surface edges and curves at the corners (creating an 8 sided surface, which would have to be trimmed obviously). That boundary has been there since the beginning since I realized early on that it would be necessary.
I tried creating the surface without tangency, but that didn't actually help at all. It didn't really give me a starting surface that was any more accurate than the ones I got including tangency, but it was worth a try.
So I guess what I'm wishing for on this wishlist of our wildest dreams is that there could be a command that is capable of matching those edges accurately. While I wasn't able to make it out of a single surface, the form does not seem too complex. I would think that a surface could be made to fit it, but maybe not.
Thanks so much for everyone's help,
Well, seeing what you've come up with I can see how this could be simplified a great deal, though I just don't have time to give it a shot right now maybe Pascal can. The problem is not a new surfacing tool needed, it's just understanding how to arrange surfaces to get the shape.
Given the tools we have, I guess I'd start with a patch and then decide how much area to devote to damage control at the corners, trim, and make those surfaces separately.