Create a rectangular hole with rounded corners within a SubD face

Tried a different topology, this is probably the best I found so far. However, I am not sure that it is topologically correct, creating Ngons in the nearby faces



If you add this line you can minimize some of the distortion in the next picture

SubD.3dm (173.1 KB)

1 Like

Thank you Tommy. You are absolutely right! It looks fairly clean.

2 Likes

I stayed out of this intentionally because I wanted to see what everyone came up with…

Now that a great solution was put forth, I’ll jump in and say… I’d never model this part like that.

I’d model that surface as a flat subd, then trim the hole in it with nurbs later after converting the subd part to surfaces. I’d project curves on the Subd to make sure they are going to land like I want, but I’d trim them later.

I get that subd is great and the temptation is strong to use it for everything, but the very best thing about subd in rhino is it’s ability to work with nurbs so well.

if you think you’ll need to iterate, by all means keep a copy of the subd on a hidden layer for later, but, the time it’ll take to fight the topo to get this modeled cleanly is so much longer than drawing a rounded rectangle and hitting trim on a nurbs part later.

the other option is to simply extrude a sharp rectangle with creases in the corners and then fillet it after converting to nurbs. I do this all the time… I use creases to mark edges that will get filleted or blended later in the pipeline.

Use the best (and fastest) tool for the job- Don’t pound in screws with a hammer unless you have to…

Or Ignore me completely, and carry on, you are all clearly doing great with this stuff!

happy new year all-
-K

4 Likes

This is the most important thing in SubD construction!

Hi Kyle, thanks for coming to this post and for your advice on different techniques to achieve this task. I completely understand your perspective. From what I can see, we all have different needs and goals when it comes to modeling. My experience spans architecture and product design, primarily using NURBS, and now venturing into car design concepts using SubD, which I would call a “difficult”, rather than “challenging” workflow.

In my case, when working on concept designs where the focus is on exploring options that will eventually be rebuilt as a NURBS model for production, using SubD isn’t just tempting—it is often necessary. Since I’m being paid for this explorative work, I need to ensure my client can explore multiple concepts efficiently and with minimum costs, without having to constantly switch between SubD and NURBS or remodel everything from scratch. With SubD, I can quickly adjust a geometry once the topology allows for it, but the “difficulty” is in getting there. With Nurbs I cannot have the level of flexibility I need from my specific case.

As you know better than I do, car design involves tangencies and other surfacing complexities, which many users on this forum perfect with tools like xNURBS (which I bought myself for a later modeling stage)—a whole other world again. In my experience so far, the direct transition from SubD to NURBS hasn’t provided results good enough for anything beyond the conceptual stage.

The issues I encounter during this process, as you rightly described as “hammering,” can’t be passed on to my client. Instead, I absorb the time and cost of finding a decent topology that I can refine later during NURBS modeling. Sometimes, even simple details like the one in this post take hours to get right.

I’ve seen incredible SubD work on this forum, particularly using Blender, but since I am unfamiliar with Blender and its advanced SubD tools, I prefer to stick with Rhino, trying to squeeze as much as I can out of it, given my limited time to learn yet another tool.

I hope this makes sense?

2 Likes

@luca_biselli your design problem looks very much like a majority of my own work, combining a smooth ‘body work’ with a mechanical ‘hard-points’.

These can be very difficult design problems to solve in a single modeling space - SubD or NURBS - however combining these with some very simple Grasshopper can lead to a very easy to update an amend workflow.

I made a video using one of @theoutside models last year that goes over setting up this work flow:

Cheers

DK

Many thanks @kiteboardshaper! I will have a look.

@luca_biselli, here is the setup I would start with:

I hope you find it useful!

My personal thought is this use of Grasshopper is the killer appilcation for GH in product design - it can greatly enhance a designers ability to do rapid changes to a model by retaining the ‘source’ of small, easy to amend parts of the model.

250101_250101 - SubD rectangular hole GH workflow.3dm (423.9 KB)
250101_250101 - SubD rectangular hole GH workflow.gh (3.2 KB)

1 Like

This is how I approached the square holes as well, triangles and ngons. Far from perfect, but close enough for concept visualization. And after the concept is accepted I would rebuild that region and move on to nurbs boolean. SubD is super powerful for sketching up complex organic shapes, but not for production of accurate geometry like squares etc.

Here you can see the patch layout unsmoothed:

As you can see there are lots of vertices that could benefit from more tweaking, but this was the level I got to within the timeframe I had, but the super cool thing is that it is easy to tweak it further if I would like to.

But these models are SubD challenges for my self, models to explore new approaches and getting to know the possibilities and limitations of different topologies.

2 Likes

why using Subd here at all? using Nurbs specifically with such objects which are predestines for Nurbs is just faster and more precise without having to write a doctor on subd topology.

2 profile curves and a rail and off you go with a swift sweep, then trim a hole into that surface and fillet it as you wish. subd is great, but permanently fitting a square peg into a round hole is just not necessary.

i may be little to none diplomatic when i say there are other options to torture yourself :call_me_hand: :vulcan_salute:

@encephalon, thanks. As described above, this is not an isolated geometry that, as you mentioned, could be modeled in Nurbs in just a couple of minutes (I signed an NDA therefore I am not allowed to share extended part of this geometry). Instead, it is part of a concept car, which involves numerous areas requiring this level of detail. These elements must adapt within the context of an evolving concept, and we need to explore several options before finalizing the design, while a steel tubular frame has just been completed. Once we are satisfied with the overall idea, we will transition to final Nurbs modeling for manufacturing.

The point I want to emphasize is that, at least for me, SubD modeling (and increasingly in the car design industry) plays a significant role during the conceptual stage of the design process. A concept I develop often incorporates critical details that inevitably make a substantial difference in the final outcome and aesthetics. If it is too difficult to achieve, it may be necessary to reconsider the role of SubD in the process altogether.

I personally don’t come from a subD modeling background, but from a Nurbs background. I am trying to use and adapt myself to this new technology I have now at hand in a familiar environment that has been with me for a while.

I may be wrong here, but since AI is now capable of producing 3D models (I explored Vizcom and the result is simply astonishing), in a matter of seconds from a sketch, it is my belief that subD will become even more critical shortly, considering that the current output is meshes, and not beautifully modelled Nurbs.

1 Like

Thanks for responding to the call @Holo. As we have discussed before, this SubD work within Rhino is simply outstanding, and the level of detail you’ve achieved is exactly what I aspire to accomplish for my concept visualizations. This before, as you say, moving on to Nurbs. Of course, this was an already-designed car, but if it weren’t, SubD would still be my preferred approach to explore design options—despite it often feels like trying to climb a vertical wall.

1 Like

Link to @Holo’s post regarding his amazing workflow WIP SubD Dino 246

1 Like

Subd is certainly a good method to chomp out quick iterations of certain overall designs, i still would plead for Nurbs in that case or anything closely related, specifically when filets come to play. subdividing subd appropriately to get these filets as you want will not help you accelerate that for sure.

dont get me wrong subd is good stuff for certain stuff, trying to use it for everything else even if you might have an idea of the future which could justify it i am not buying it.

Ii keep experimenting with AI either and it just has a mind on its own, reworking everything is inevitable, believing that you can trust AI for any crucial step without any major reconstruction at all is currently nothing that i believe in.

on the other side if you become a master in subd and can do those fillets faster and as precise as with nurbs then why not stick to subd altogether either. mixing the paradigms is not a good idea since you will lock your way in very fast ending up having to rework it entirely even more.

I think we have found a solution, which, within the context of a concept illustration, seems to be good enough and, above all, has only quads and no distortion.

4 Likes

I thought about above layout.
but it fails as soon as the surrounding is not planar:

my guess: the opening should have at least 2x2 faces - this will allow nicer layouts for the details:

Thanks for your suggestion. So far, the solution found above looks like the one less problematic, requiring minimum edges addition.