I need a good free NURBS modeler!

So… I am a Blender user of many years
was modeling a piece of jewelry yesterday and came upon a shocking realization:
I can’t build what I want to build in Blender with vertex-based geometry.
(I need to subtractive boolean / chamfer-blend into the subtracted area)

The first thing I tried was to study everything about Blender’s NURBS modeling implementation.

Then the second shock came: Blender’s NURBS implementation is not good enough.
The NURBS features of Blender have somehow become overlooked over the years, and the result is,
well if you want to simply sculpt a fancy tube it will do the trick, but it certainly can’t handle anything like this: http://www.youtube.com/watch?v=XRkcrdwuUO8

So, I need some recommendations for a free NURBS modeler.
Open Source is preferred over just free, but I can’t afford to be picky.
Mac is my platform of choice but I am willing to install Linux or this.

I have an alternative Nurbs-Modeller, www.moi3d.com. It is 295$. I know that`s not free, but a very good and intuitive usable program for nurbs. The developer was a programmer of Rhino too.

A quick search for free nurbs modeller brought this up:

But I haven’t tried yet.

http://opensourceecology.org/wiki/CAD_tools

mmm…

FreeCAD ?
http://www.freecadweb.org/

“there are currently no good open source parametric 3D CAD programs”
Thanks for the informative, yet depressing news. lolz

I did find an application called Nurbana which is open source but is Linux-only, only has 14 people who downloaded it ever, and hasn’t been developed since 2005.

Rhino has a free demo that will let you save 25 times. (No other time limit.)

http://www.rhino3d.com/

Steve S

You did not say what problem you were having with vertex based geometry in modeling what you described. Seems a pretty simple proposition. Boolean is never a necessary modeling method, only a shortcut. Chamfer blend is another description for bevel AFAIK for which Blender offers many options including bevel weights which seems like what the video showed at first. (It was long so I skipped through it.)

I don’t mean to dismiss your question and you have had a few good responses anyway, but I would really have to see this problem that can’t be modeled with meshes give Subdiv modeling, creasing and bevel etc. I saw nothing in the video that could not be done quickly and easily with Blender’s tools, incorporating NURBS or not.

@DruBan: I suspect what he meant was “I can’t use the tools I know and my current work flow to build what I want to build in Blender with vertex-based geometry.” If someone knows and uses Nurbs and Booleans, making the shift to subsurf modeling would take some time.

@Quantum Anomaly: I was also curious to see this thing that could not be modeled in Blender, since I have found it quiet capable. AFAIK, you are right about Blender Nurbs. There doesn’t seem to be much interest in the developer community to push nurbs past what they could do several years ago, and most of the blender artist community has responded with a general shrug and a “meh, don’t use em anyway” attitude.

The things I find missing from NURBS are the ability to Alt-select rows and columns and rotate and scaling around the active knot. Also the space, smooth and relax functions might be very nice to add to the NURBS toolkit. NURBS has the great advantage of being able to change he resolution independently in U and V unlike SS which of course does both directions to the same depth.
Cyclic closes a surface but it seems that it might be nicer to have an explicitly closed surface giving a NURBS solid; however that would be a problem with the U,V description so is unlikely.
All it would take is some coder to hop in and spruce things up a bit.:smiley:

Yup. And that’s what they were saying back in 2006 and 2007. :wink: Don’t hold your breath while you wait…

i heard there is a more advance SVN somewhere
with some new functions added
but never been able to find where it is !
and i think it needs to be built so not an SVN !

would be nice if somone would make a new SVN for windows 32 and other os

happy bl

Just about ready to happen then! My patience is always rewarded!

It looked promising at first, but they haven’t released a Mac build in ages.
I was about to compile one myself but then I realized I need to update XCode and it will require upgrading from Mountain Lion to Mavericks. I don’t know if that may cause any of the software I currently have installed to break, so I want to research that a bit before pulling the trigger. Not ruling it out entirely, just saying it’s going to be a project.

Thanks for suggesting.

[Edit: I should add that I did download the most recent build - August 2009 - and tried to follow some video tutorials, but mine was so old I was missing half the menu items]

Rhino has a Mac beta actually!
I downloaded it and opened it up. It expires in a month, not sure if there’s a way around that.
I also don’t know that I feel like investing my time learning a closed source piece of software I can’t afford, when I could stay on Blender.
But then Blender’s NURBS is so primitive, the only thing you can model with it is a sock with a hole in the toe… lol

DruBan, you brought up some good points.

I disagree. Here is an example of an object you could only create with a boolean operation:


Where a complex surface must carve away another complex surface then a boolean is often the only way.

Chamfer blend is similar, but I think the reason it’s called a “blend” is that it is not a matter of simply specifying a fixed radius. Instead two edge loops on opposing faces of a corner are blended together to create the bevel over that corner.
The result is that the radius changes along the curve like so:

(Rhino)

The closest thing to this technique in Blender is bevel weight. It can be used to adjust the radii of individual edges, but will not yield a smooth gradient because it’s based on a polygonal mesh instead of parametric curves.

The reason parametric modeling is usually better suited for CAD than polygonal modeling is that in order to add detail to a poly-based mesh more polys are required. But in creating more polys the mesh becomes harder to work with.
Furthermore, adding more polygonal data affects how a subsurf modifier modifies the mesh, sometimes pinching places you don’t want pinched. Heh…

Parametric modeling is better for CAD because curves and surfaces are interpolated from the control points which define the shape. Polygonal modeling is approximation. Subdivision surface uses approximation scheme (Catmull-Clark in Blender).

Adding more polygons adds more work but that’s why you plan ahead. Same for pinching. It’s not the faulty modeling paradigm why you need to switch to another, it’s because they are for different purposes and you seek to use what is originally designed for the task.

I think you have gotten all the answers you need… If you have a particular design problem that you want to model in Blender using curves and/or Nsurfaces feel free to ask but I don’t see the point in taking the time to model a random useless shape without using Booleans just to show that it is easy enough.

…in order to add detail to a poly-based mesh more polys are required. But in creating more polys the mesh becomes harder to work with.
Furthermore, adding more polygonal data affects how a subsurf modifier modifies the mesh, sometimes pinching places you don’t want pinched.

A little practice with the Subsurf modifier will probably help your problems although I know that you said you have a lot of experience with Blender. For instance the SS mod adds geometry so therefore it’s not good practice to add more polys and then throw the mod on top of that. A better practice is to have a few well placed polys to which you apply Subsurf, and then adjust the original geometry’s placement to adjust the final shape, much like you adjust the control points of mathematically described surfaces.