Blender NURBS Development

While working on different projects at the university I started to use Blender about a year and a half ago. Since then I used it more or less recently for different modelling, animation and rendering purposes. As I was and still am attired in the possibilities a open source 3D Application is able to offer, I started to mention it to certain profs.

Thing is, those are interested in using Blender for education purposes and asked me about the possibilities of it. Everything went right to an edge, where Blender could have been used for a whole course at that institute. And then things came to NURBS, and everbody froze. In modern architecture development NURBS are just not to substitute, and Blender is a really poor NURBS handler.

So that project died, at least for the next future. But I dont want to take that as a forever and hope sombody can inform me about the future of NURBS in Blender. In this Forum, and in some others, I found different discussions of that topic, but none was younger than a year. So what happened with the Blender NURBS project and the integration of Nurbana?

They way I see it - Nurbs are DEAD.
Nurbs are also dead with Max and Maya as well… ok, they’re rockingly cool
in Maya and very functional, but dead still since hardly anyone use them.
Nurbs where all the rage 5-8 years ago - but polymodelling seem to rule
our world today. When was the last time you saw anything modelled with
Nurbs today?

I work at a place where Maya is used on a daily basis, but nurbs are non-existant with these people, they’ve heard about it …but doesnt really use it… And neither does ANYONE that I know in any business…game industry, prototype-design or otherwise.

Maybe that’s just the ones I know…maybe you know someone else who
actually use nurbs actively? Or anyone out there?

To me - Nurbs never existed.

half way through LOTR’s gollem was changed from nurbs to polygons (sub-div) since the technology of nurbs was superseaded with subdiv modelling.

it certainly is not taught here, but nurbs and other curve tools are still handy for industrial design models (which need to be made, and need to be accurate).

it would be nice to see better handling, but the integration project seems to not be happening (it got fairly far i think)


NURBS may be dead in terms of “I just want to visualise it”, but in architecture they are used for a lot of semiintelligent self structuring programmed forms. Or for pseudo genetic algorythms for the achievement of similar but different forms (a kind of “different repetition”). And for the Development of selfrelated Proportions of Buildings (something the old Greeks did in an analogue way - each columns diameter is 1/7th or whatever of its height - but in a extremely advanced way). A very impressing example for this is the new mercedes museum planned by ben van berkel

Could all that be achieved without NURBS? As Im not a too advanced 3D Artist I dont know the answer to that Question, but as soon as somebody qualified tells me I can forget about NURBS for doing such stuff i will.

Perhaps NURBS can be useful in some situations, but overall Polygon modeling is way better and faster. The blender devs would hardly give an effort to improving the NURBS tools.

This topic is just one facet of the whole “technical accuracy” vs. “what looks good” balance in 3D modelling. On the one side are the CAD guys and on the other, the DCC masters. NURBS surface modelling has dangled the tantilizing possibility of “the best of both worlds” out there like the proverbial carrot on a stick.

There are industries in which it has realized this dream. As Alltaken said, its anywhere that a 3D digital model must be turned into a real-world product that (typically) must be mass produced. Since NURBS can accomplish both “what looks good” and “exact accuracy” due to its mathematical nature, its heavily utilized in certain ID/CAD related apps, such as Rhino and Alias Studio Tools.

The question as to whether blender should make use of this asset is where the real debate will flare. Since, currently, blender leans very heavily in the direction of DCC, NURBS is likely to not see much support. If blender were to be expanded in the CAD direction, this toolset would be one of the best components to flesh out. Will blender ever exand in this direction? Not until there is significant/dedicated/committed support from coding-capable individuals. So far nothing’s showing on the horizon… but last year’s SoC project gave it a good push. Did any of that work get merged with the main bf tree?

Seems not dead at all! :slight_smile:

I think Blender needs patched nurbs and a Rhino3d nurbs aproach too (for architecture and industrial design), instead a 3dmax nurbs way.

I personally would really like to give NURBS a shot. I just like them. I hope someone will finally finish the Nurbana integration. But we’ll see how this evolves.

poly and subdivs are great but when it comes to object modeling NURBS by far surpass what polygones can offer you when you want smooth curces and correct cuts, trimed surfaces, fillets at edges, and other nice things which will tell you somebody just made a poly car or somebody used NURBS.

In game design you do not need NURBS anyway.

i know studios which still work with nurbs patches even for character animation because you can do shapes you cannot do with polygones with it.

NURBS are not dead, Polygone tools just became powerfull again. try to do any boolean with subdivs and you will see where the limit of object design with polygones are.

In industrial design NURBS will be only used anyway because Polygones just cannot do it.

And polys are not better over NURBS. Both serve a different purpose. However NURBS can model both organic and inorganic objects with precission, while Polygones cannot.

Furthermore are NURBS in maya a perfect tool for extrude,lathe, and other things and offer a toolset impossible with polys and that all together with Mayas construction history :wink:

Can we write a small C program to create a nurbs structure in C.
My point is that create character in Mesh, Use Python and get all the vertex points and put them as Nurbs using C.
Can we do that?

I would love to see a good NURBS toolset in blender… Personally NURBS rock my world. A lot of things just seem quicker to visualize that way rather than messing with sub-ds, at least to get objects up to a certain recognizable level, and you can always switch it over later. If blender could get something like Maya’s NURBS I’d be in heaven.

Yes I agree, NURBS are highly superior to polys when you are doing boolean operations. With polys you always get messy meshes and some rendering artifacts after boolean operations, but with NURBS since it’s all mathematical algorithms anyway, you get a perfect surface.

Personally, I think unless you do product design & manufacturing, you will not have much use for NURBS. Blender is more for artistic multimedia content creation.

Why is it that some people can’t people actually answer this guys question without editorializing about the suitability of NURBS? He made it pretty damn clear in his initial post that they needed NURBS for the work they did at his university. Despite this some people can’t resist telling him that he essentially ‘dosn’t need NURBS’. How typical.

LaPostal: It dosn’t seem like any dev is activley working on NURBS in blender.


Nurbana integration was one of the Google Summer of Code 2005 projects, taken on by Emmanuel Stone as detailed here:

It never made it into the BF tree because there were still some outstanding issues which could not be sorted out within the SoC period. Other people have asked about this before and here are the replies from Emmanuel back in April:

… and earlier this month:

So, unless some interested individual takes over active development of this, it is on hold for the near future at least.

Hope that clarifies things LaPostal.



you cannot translate a polygone point cloud that easily into a nurbs model.
i do not know if you ever worked with NURBS. nurbs are 4 point patches.
for example you can model a head out of many patches stitched together.
each patch can have an individual subdivision level to model out fine details where you need it and to leave it simple where you do not need it.

however Maya has for example such a funtion to turn polygone models into NURBS. but the problem is often that also the mesh should be made out of 4 point faces and also often you get very dense results.

and than you have to make sure that the tangent angles of the nurbs patches will flow together so you will not see the individual nurps patches in rendering. i fear thats not just a small c code program.

I would rather think that nurbs together with the new topo tool would provide you faster results. you could drap nurbs curves over the body where you need them and later loft them together. but again you also need stitch and attach functions. otherwise you will also run into problems because a texture will be applied only to individual patches and will not handle all attached patches as one patch!

the front of the Prodesign magazine here in New Zealand (the design mag, for the design institute of New Zealand DINZ) has a full page front cover render 100% done in Blender…

Lobo_nz (on IRC) did it in his industrial design job.

I will be using blender and have been using it professionally in the design field… we certainly DO need more accuracy. and i support any effort to allow more accuracy.

i do however think productivity of workflow is important to keep, its blenders current #1 asset… i’ll post a link to the article containing the blender renders some time, i was given permission to do-so.

ok here it all is in PDF


So which of the blokes on pg 3 are you? :smiley:

Interesting ATV design! Seems as if they could have made better use of the space under the bonnet, with no engine there…

Side note: There’s a small company near where I live that’s been developing safer ATVs with some interesting front steering/suspension:

Nope not me, i am not part of it at all.

lobo_nz Is an industrial designer who also uses blender, he hangs around IRC, but not the forums. so i was posting it.

He has submitted it to the Blender Gallery so it should show up there.


this was the summer of code projects i was most looking forward to.

too bad it never materialized in the BF tree. has anyone looked into the SoC code tree to verify how much progress has been made?

I think he is using Blender as a backend for better rendering.
Apps like Rhino or Cobalt have only very limited render and material system.

Recently Rhino got Flamingo which adds a GI engine to Rhino.