Please, Subordinate T-SubD's

If we had Subd’s which we could match a vehicle’s interior panels–with an optimized exterior. Even if we had even one level of hierarchy, SubD’s would be much more useful.

In this example, the Subd hull, is supposed to represent a car vehicle side in Box Mode, (and not Lord Varder’s ride.)

This is the surface I might want in smooth mode:

This shows the differences with soft edges:

This is the difference if hard edges were used.

Granted, this would be somewhat unconventional implementation of SubD, but the smoothing rule would be the same, so it is a natural extension.

So, the GUI might work like this. You pick the superordinate object, and the subordinate, by virtue of being coincident vertices follows the superordinate.

And then, this result is a step in right direction:

In other words, I do not want the vehicle interior panels to interfere with the exterior.

One of the only problems with NURBS is I cannot reproportion them.
One of the only problems with SubD is: If I subdivide the mesh, I would not be able to retain the outer shape–without making one that that is no longer ruled by the fewest nodes, which gives me the smoothest shape. If I subdivide them mesh–it will also become uneditable.

A node-based solution would also work, by specifying which nodes are subordinate.

A solution whereas SubD faces could be hidden from the rendering solution, would work to some degree, but that would increase the memory usage, and it would not quite solve not-mangling-door-panel problem.

hey Brenda,
I’m trying to understand the exact problem you are trying to solve, and I’m a bit confused.

are you trying to get the edges of this area closer to the outside surface?

can you clarify your problem a bit more?

Hi Kyle,

Thank you for the interest.

I am trying to create an opening without affecting the curvature of the outer panel. Sorry, about my quickie example.

I’ve been working on an aerodynamic vehicle body using Rhino SubD, having done eleven generations of OpenFoam study, and work on aesthetics, which was needed.

As similar technique as NURB’s, there are very few control nodes in the shell, as few as possible, which allows the shape to optimized–without extraneous intermediate nodes which would only degrade the the shape.

In parallel, I also studying the vehicle frame/chassis. It’s not as far along because of the question of process and materials, but also, it’s not as far along because my computer model cannot complain, so a real-world mock-up is necessary.

Also, in parallel, I have been roughing out the interior.

So…

I want to be able to cut door and window openings and into the vehicle body–without affecting the larger sense of its shape.

I also like the technique of matching the dash to the hoodline, which is not always done.

I have a sense of how to make the body out of NURB’s, but once they are trimmed and joined, that hull could not be easily re-proportioned, as to say, make the window opening a bit smaller. I would need to maintain several stages of body surfaces and primitives, also.

Presently, it’s in SubD, if I add nodes to add window openings, the face splits they would cause would make it difficult to keep the exterior shape. Rhino is very forgiving by adding and maintaining hidden splits, but for a big hole, more car body surface nodes would need to be added, and because of that, changing the the surface would be less accurate. Smoothing might then help, but presently, we cannot smooth to a guide spline, which I have been doing manually ; )

[ Basically, weather I have a SubD or NURB surface/curve with say 5 nodes, I use a 3-node curve as a guide to attempt to fair the more complicated surface/curve the same as the the more simple one.]

The problem for either SubD or NURB is: once cut, the more complex resultant object cannot be re-proportioned as well as a lesser node variety.

~

If we could do this, it would have two advantages.

1.) You could more easily make something like a Porsche 911, Americana, and Cayenne from the same geometry.

2.) It would allow body/interior/chassis to be developed in parallel more easily. (My application)

3.) It would allow styling updates and engineering changes to happen with fewer problems, such as the vehicle’s hood now has to come down lower to meet anti-pedestrian rules, or American ground clearance heights, or whatever fanciful things they have planned for the Euro 6,7,8.

~

It would seem that two things would be required/helpful to make this happen.

1.) The ability to hide SubD faces, so they can be part of the shape solution, such as NURB trims retain the original shape.

2.) The ability to intersect SubD’s using a hierarchy, and let us go back to the original shape. This would also be handy for NURBs, too. For this purpose, the thing is that somehow, the later subtractions cannot affect the original shape.

[I usually keep an Home-Depot orange Tool layer just for tools, but I also keep layers of each boolean-step. Managing this would be more fun, with just a right-click menu. I like the way 2d programs make compound objects. The metaphor of Rhino’s multiple undo would be good thing to start from, but let us do booleans like this, too.]

Using another metaphor, we would be making T-SubD’s.

interesting.

my thoughts would be to over build things and then keep them as a “tool” that you can repurpose later.

then as you hone in on a solution make a break point and convert to nurbs, then trim and refine etc…
I do this quite often when I want to iterate on solutions. I keep the subd, copy to nurbs, then edit, then copy to nurbs etc..

currently we do not have a live link between the nurbs and the subd that created it. I’m not sure if that is even technically possible, @dalelear would be the person to answer that question. and your others as well.

I’ll hang tight and see what he has to say before I offer more ideas.

1 Like

Thank you for looking into it. : )

Rereading your post. I do think that it is possible, but there must be a firm set of rules, so that the programmers can do their job.

The challenge is: it’s a new thing.

@dalelear any ideas here?

There are three routes to do this:

1.) Many 2D programs such as Inkscape and Coreldraw. (I hate using Illustrator, and own it, perpetual.) allow compound objects. Well, these would be great for even NURB’ Objects. The drawback is, for instance, that raddi could not be applied on the result.

2.) T-SubD’s, which oddly–has almost been done, because many engineering applications such as Finite Element Analysis and Computational Fluid Dynamics use 3D filled meshes, such as this.

A subdivided mesh is created using the surface mesh. The difference here is: the mesh is a Sub-D friendlier mesh.

So, it would have a dynamic 3D mesher, or depending on how you look at it, a semi-automatic mesher : )

At an intersection, edges would delineate between subordinate and the parent face.

Mesh faces would be flagged as visible or invisible. Some mesh holes may appear to be there, but really, they would be load/tension bearing allowing the parent surface to keep its shape.

3.) Wait for Autodesk to patent it out from under you, again.