Rhino 7 Subdivision Surface Project

@kelt0004, that is the Raytraced mode I’ve been working on. It is shipped with the WIP.

1 Like

I guess it is not in the Rhino4Mac WIP huh?

Hi guys, I have a corner that converts oddly, can you take a look?

Bug- Odd SubD corner.3dm (59.9 KB)

I see that this also makes an odd corner:

It behaves “ok” when mirrored and joined, but the “star” isn’t perfect so a slight remodel improved that.

There isn’t anything wrong with the subD, see how the control points for the outer curve are:

1 Like

Ok, thanks, that makes sense!

These are the simplest subdivision algorithms that anyone can imagine.
introduction to subdivision surfaces: http://www.cs.dartmouth.edu/~cs77/slides/10_subdiv.pdf

I am skeptical. The main issue, from the user point of view, is getting rid of the polysurfaces because it is nearly impossible to edit their edges.

And that’s bad because …? The link you provided also only talks about Loop and Catmull-Clark. If there’s a problem with these schemes can you enlighten us as to what it is and what alternative scheme fix it?

Do note that we do not subdivide a mesh recursively. The mathematically exact limit surface is computed and any additional mesh vertices are placed directly on it. Any additional scheme we could implement must have the mathematical property that an analytic limit surface can be computed.

It is not bad - all good ideas as simple. When I talk about new ideas I have a habit of explaining every new technical term so that I do not confuse anyone. (Most scientists have the opposite habit - they use technical slang to “prove” that they are smart.)

I may be biased against NURBS and its extensions (subdivision surfaces) because I invented very compact geometric modeling kernel which has nothing in common with NURBS.

We want arbitrary topology and no stitching.

Hi Andrew,
You have claimed this before and I am very, very curious where I can read and see more about this invention. Can you elaborate please or at least tell more about it?


1 Like

The math of my invention is simpler than the math of the NURBS. If the CAD model is made of curves, my invention reduces computer file size about 3 times compared to the NURBS and T-splines. If the CAD model is made of complex organic shapes, my invention and T-splines make files of similar size, while NURBS doubles the file size.

I was thinking about simply posting this invention on the internet and giving it away, but it is not good idea. If I post it on the internet, nobody will invest a penny in this invention, so it will be wasted and forgotten. Patent application makes sense only if I hire several programmers and supervise them for several years. I have more important things to do now. Nearly all CAD programmers are NURBS programmers. The programmers and their employers have invested decades of work in the NURBS, so they are not eager to abandon NURBS. They are more likely to embrace tweaks of NURBS, such as subdivision surfaces and T-splines.

You can patent and then grant a license for the application to a software company in return for a fee.

Filesize != maths.

You mentioned your approach requires T-splines to make surfaces? T-Splines are patented and owned by AutoDESK¹, which would mean only they would be able to develop this technology. However until any specifics are made public you are the only person who knows anything about this new approach and any further discussion is pretty pointless.

¹ And the maths involved is not simple.

…[quote=“Andrew_Nowicki, post:175, topic:16562”]


Looking forward to a full version of a Rhino native SubD plugin soon! One of the main problems with T-Splines is its hungry share of processing resources in smooth subD mode. Complex models become unworkable or crash. The topology of T-Splines is very unforgiving. I hope Rhino subD will be more robust.

If you need a beta tester, I’m in.

Not soon.
What we have will be in the WIP after V6 ships.
There i a lot of development work to do to get a fully functioning, ready for release tool.

Hi There since T-splines is no longer developped, we need this kind of tool to add some skills to Rhino.
I’m looking forward too this.
(Image made with fusion360 and rendered in Rhino 6 WIP)

There are still some odd display issues with SubD objects in rendered mode.
SubD’s are sometimes shown glossy when they should not be and visa versa. And when I select objects, both subd and mesh then other objects change their appearance:

Neither the white nor the red material is glossy.

And as you can see metal is not shown right on SubD.

When I select the metal ring it turns black:

And when I mouse over the vidget it turns to silver:

In case you need some inspiration on subd modelling:


Hi guys,
Just downloaded Dale’s example with SubD object and I see a problem with the display in the latest WIP.
The second side of the SubD is completely dark.

See enclosed:


1 Like