Almost T-Joint/Star-Joint how would you model this?



Hej Hannes,

you could maybe try this tutorial for Y-branches, it’s the same principle, no matter if the tubes have the same or a different diameter. Just model the primary domain and then reflect.

Happy branching!


I think that subD is the efficient way for not spending 40 min on Y branch…
Nurbs is used for controlling curve for mechanical constraints and aesthetical purpose in the same time but for aesthetics only, for me subDtoNurbs is the key.
You just need to extract boundary control polygon for starting your sub-D topology


When you know what you’re doing, you can model a Y-branch in a few minutes.

Instead of futzing the model with polygons, I believe the OP will learn many good things about proper curves, good patch layouts and continuity. After all, Rhino is primarily a NURBS modeller.

Polygons are ok for visualisation and prototypical 3D printing, but not for passing on the data to structural engineering and toolmaking software for production, depending on how the product will be moulded or cast.


90% agree for the learning purpose but for machining G2 are enough and at the end you get a nurbs from your subDtonurbs so you can evaluate everything… finally you just have more flexibility for the form finding in your design creation and it’s the most important ! everything beyond is just projection and rebuild…

But if you aim class A for commercializing your nurbs model and not the final product I’m 100% alright with you

Ps i love nurbs but vertex location is prime.


We’ve spend a lot of money/effort for reverse-engineering polygons into surface or solid models, but with no good result, comparing to the time/budget it would have taken to develop the design in surface or solid modelling in the first place.

But probably, the OP just wants a model for rendering or 3D printing at scale, who knows?


Nurbs and subD are just paradigms nurbs can produce bad surface if they are bad modeled too and once you rebuild a projected curv no matter on what kind of support you are: mesh, subD(it’s different) or nurbs, you enter in the one span world.

(Lukasz Domagala) #12

Just watched full thing…wow! I think I should pursue a career in class A modelling.
So relaxing.


The SubD implementations inside CAD apps are usually not polygon based. Hence you do not get a facetted model. What you see e.g. in Gijs’s screenshot is the control-polygon (cage) which also exists in any super duper surface model.

I still see a lot of issues with Nurbs-compatible SubD for precise, complex models with cutouts and exact dimensions – but for isolated shape-transitions limit surface based SubD should work quite nicely. Did I already say that the result is not polygons? Ah yes, here :o)

(Gustavo Fontana) #14

Do we need another thread about Nurbs Vs. SubD for production?

I think these 242 responses on this thread cover the topic well:

@lagom, @hifred, stop it! :joy:


Class XYZ… if it looks good, it is good ; ) When a client does not care (very sad) or does not have budget for proper 3D (equally sad), then I would not bother. At the end of the day, you want to be in the pub at 5pm!


Well, the OP came with an interesting NURBS modelling question; then the subject of polygons was introduced ; )

(Lukasz Domagala) #17

Yeah these can wait!

Sign out…see you next door! (pub)

(Lukasz Domagala) #18

Have a great one everyone!

(Gustavo Fontana) #19

Here you go Lagom,

All Nurbs, Class-Zy


You got the right spirit - chasing a Guinness with a Stella… stellar ; )



As already have been suggested a double blend seem to work fine and is superfast to produce.

A small clip I made here:




Probably not, however, as a hot button topic, such is likely one of the most interesting, evocative, and prescient professional debates, now and into the foreseeable future .

Indeed! And NURBS still represents the gold standard, and there are people in this world with exceptional knowledge and skills at such (that’s right @Lagom I’m referring to you :wink: ) who can bang-out top-notch solutions in short-order, and are willing to give back and share the art. Bravo!

Correctamundo! A place for everything. Native addition of such inside Rhino should be win-win. It is time, if feasible, for Rhino to focus on new (or match existing/lost) modeling tools, IMO. Such can and will add value to the Swiss Army tool of - Modleing Tools for Designers. (Plugins of any value having exhibited their inherent vagrant transgressions.)

NURBS, absolutely, long live ‘em… but there should be a future place/option for the below with respect to the OPs question, which was - how to achieve in Rhino 6, with NURBS. Just sayin…


My friend who is an architectural student and asked me to model this beam was only about 3D-printing a lookalike version so no precision demanded at all.
What I was instantly trying out and very happy not to invest too much time (as I was checking a lot of approaches) was to use the Patch. This was doing the trick. thank you for participating in this discussion as I was always stumbling and avoiding this problem with different pipes coming together in one joint.


Would have been professional conduct to include such vital information in the first place.


@Lagom too bad. Got this decision when thinking about a solution to tackle this. Wasn’t aware, just always build things in Rhino and try to result everything in perfect surfaces.

The whole topic is kind of interesting and I’m very surprised so many people being involved.

I had no problems with this, until TSplines was discontinued in R3D v6 So I’m waiting for the SubD feature in the Rhino Beta to be introduced in the full package.