SketchUp Nurbs

pilou · 26 Sept 2008, 12:36

That yet exist Moment of Inspiration with Skp export
Maybe not with all that you show above

And maybe all that you describe is in the 7?

kwistenbiebel · 26 Sept 2008, 12:44

That yet exist Moment of Inspiration with Skp export
Maybe not with all that you show above

And maybe all that you describe is in the 7?

Hey Frenchy, as you have a direct line to the developer of MOI: wouldn't it be a nice idea to ask him to make a 'mini-MOI' as a commercial Sketchup plugin?

plot-paris · 26 Sept 2008, 12:50

I imagine it as a merge between polygon- and nurbs workflow.

as far as I know at present you allways start with volumes or at least faces in nurbs modelling.

this new way makes use of SketchUps edge-based approach, which in my opinion makes it far easier to start sketching up rough shapes.
you still could use the program exactly the way you do right now.

basically SketchUp Nurbs creates geometry as we are allready used to it, but allways combined with a functionality that in some way the FreeFormDeformator gives us - only that you are able to dynamically add or delete FFD's control vertexes and apply changes directly to the geometry, rather than to a group.

but I definitely have to have a look at MOl, sounds very interesting...

Stinkie · 26 Sept 2008, 12:52

@kwistenbiebel said:

wouldn't it be a nice idea to ask him to make a 'mini-MOI' as a commercial Sketchup plugin?

He'd be undermining the position of his core product if he did, wouldn't he?

pilou · 26 Sept 2008, 12:57

@Kwist : Moi is yet like a mini "Plug" of Rhino and so many other
I don't believe that it can be more small but your suggestion is recorded

plot-paris · 26 Sept 2008, 13:01

another short example for the difference between the current version and SketchUp nurbs. and remember. all the three cubes have been drawn with lines and extruded with push/pull, then the top face rotated 90 degrees arround the centre...

remus · 26 Sept 2008, 16:02

Cool idea plot, and id certainly welcome it in SU, but i think its a bit more complicated than you make out.

One area that could get very sticky is changing between nurbs and polygons, as it is pretty much a one way street, if you want to maintain proper NURBS surfaces, anyway. I can imagine this causing a lot of annoyance, with models that you think are finished, have been converted in to polys but then need editing.

I reckon youd also need some very clear way of distinguishing between NURBS stuff and poly stuff, otherwise your models could get very confusing.

A quick thought: perhaps all nurbs surfaces could act as proxies, kind of like subd and smooth, and the proxy could be kept on a different layer, so basically all the geometry would be polys, but you would have NURBS surfaces controlling the polys.

plot-paris · 26 Sept 2008, 16:33

oh, I am afraid I didn't make myself clear enough there:

all the geometry consists of nurbs - but SketchUp nurbs lets you manipulate it exactly the same way you handle polygons in the normal version.
so even an edge is in fact a simple nurbs-curve.

and unlike all nurbs-modellers I know, SUN will automatically create a face, when several nurbs-curves that lie in one plane, are closed...

solo · 26 Sept 2008, 17:03

I like this thinking.

I remember when .... http://www.sketchucation.com/forums/scf/viewtopic.php?f=15&t=1782&st=0&sk=t&sd=a&hilit=nurbs&start=30

Hazza · 29 Sept 2008, 04:02

Of all the whining and b*tching I have read about SU (poly count, multi core this, 64 bit that... pffft) this suggestion is the only one that has made me look at the current version of SU as limited. This would put SU head and shoulders above where it is now.

I can imagine all sorts of possibilitys if Google implemented this. To the begineer SU would behave exactly as it does now. As they learned and grew they could add Nurbs control points and learn how to manipulate them.

plot-paris · 29 Sept 2008, 07:05

of course now we could do with some detailed information about nurbs models. does anyone know how a nurbs model behaves in terms of filesize? or does it need more graphic card power, more/less ram, better processor?

my guess is that it behaves pretty much the same for rectangular objects but brings a huge improvement when entering the realms of curved (organic) shapes.
I may be completely wrong though

pilou · 29 Sept 2008, 10:01

The native 3dm format of nurbs is very consuming!
But you can compress it with a 30% factor

Here a 3DM file (compressed 900 kb ) original 3 Megas : Test by Steph 3D
And the direct SKP file 20 000 polys (but you can ask more of few polys at the convertor as you want)
(compressed 1.5 mega original 3.7 megas)

Render Podium with Biebels' settings with any regulates except a reflection

@unknownuser said:

or does it need more graphic card power, more/less ram, better processor?

Depending of what nurbs progs you use
Moi can run on an very old comput and video card!!! more 5 years old! (SU works fine too !
So for this one Memory is the more imortant

Test by Steph3D.net.zip

testskp1.zip

RayOchoa · 30 Sept 2008, 01:18

@ Paris, you make this seam so easy. I love the way you do all these presentations which make one believe that that is actually possible.

plot-paris · 30 Sept 2008, 06:28

thanks, Ray
unfortunately my vivid imagination doen't go beyond Photoshop. I have no idea of programming. and this SketchUp Nurbs (in short SUN ) is something not even our ruby masters can achieve for it is a completely different technology.

and I have to admit, I fear there is some huge drawback of nurbs. otherwise, why do polygons exist in the first place? they have to have their own advantages. I should do some reading about nurbs and then report here, what I understood - so the dream of SUN lives on

Bruell · 17 Oct 2008, 22:25

Hi, I think this is a great idea, and should be doable.
My input to this tool is, to keep the models clean, you can start with NURBS then turn it to poly's, since the benefit of NURBS lyes in the creation. And I would like to see 2 rail sweeps and all the good stuff that comes with the NURBS tool set.

And with this could come real circles, that stay round when you zoom closer.

plot-paris · 18 Oct 2008, 12:20

@bruell said:

And with this could come real circles, that stay round when you zoom closer.

and the great thing is the versatility. at the moment if you draw a circle in SketchUp and delete a segment of it, the information of the original curve is gone forever.
with SUN (SketchUp Nurbs) you not only the information of a circle is kept (because the formula for the curve is still that of a circle, except with manipulated start- and end-point), but you can even change a straight line you created with the LineTool turn into a circle by entering the NurbsEditMode:

so if you select any elliptic shape and click the "simplify" button often enough, it will eventually turn into a perfect circle. if you use the simplify too to "paint" on certain parts of a curve, only these parts will smooth out...

of course it is dangerous to loose yourself in the mere number of functions that could be useful - if you are not careful you end up having something like almighty Maya, that can do anything you want and takes years to master.
important is to keep the main interface simple, to reduce the number of tools as much as possible and keep them intuitive, like SketchUp did with the copy function - instead of introducing a seperate tool they integrated it into the move tool. this approach has to be kept in mind when shaping SketchUp Nurbs.

plot-paris · 18 Oct 2008, 19:12

well, NURBS means "Non-uniform rational B-spline" and fredo's bezier curves are splines. I guess, that soapskin bubble works in a similar way.
as far as I know they both use mathematical algorythms like nurbs and later transform it into polygons.

so just imagine what our ruby-geeks can come up with if you give them a SketchUp that is solely based on nurbs!

I have something in mind, where I don't know, if it already exists and I saw it somewhere or if it is my own idea:

Imagine you create a curved surface in SUN, calculated with nurbs technique. it is supposed to be a rocky, uneven ground. you could apply a material with displacement map, that creates the rough surface during the render process. but you want to see the result while modelling, perhaps even interact with the uneven surface. instead of modelling it yourself, you apply a displacement filter preset to the surface, that recalculates the whole surface, because it is a nurbs element, you don't have to translate it to polygons every time you make a change.
thus you can easily change the overall shape by manipulating the original curves of the surface, while the displacement filter will adopt to the changes.

chango70 · 19 Oct 2008, 10:11

Sounds great! I'd imagine whoever can do this and indeed does it should get paid.

chango70 · 19 Oct 2008, 10:15

Isn't Bezier Curve and Bubble Skin script able to approximate Nurb Modelling?

minguinhirigue · 19 Oct 2008, 12:44

This is quite a big idea, but implicit geometry (NURBS, Blobs, and other forms that could be redraw from control points, lines and faces) is really far from what sketchup is doing today. However, there are a lot of soft which are actually doing what you are talking about :

But, I recognize that Sketchup interface is easer than Rhino one... And Rhino or Moi do not have any free version. So this could be a great improvement for SU7 : providing a free version which keep a clear interface and user-friendly workflow, while allowing edition on complex NURBS and spline.

Note that if anybody want to model a NURBS the way it is suggested in the first message, then it could be usefull to implement T-spline instead.
T-spline are more powerfull than spline if you want to edit the object in the way you do with a polymodeler : extruding, pulling, stretching, adding details all the time. Just have a look here.

For those who are searching, there was a quick job on Bezier surface plugins for sketchup : http://groups.google.com/group/Ruby-API/browse_thread/thread/0e044d3b1f7124e5

PS : Chango70, plot-paris, I guess Soap-skin-Bubble don't use Nurbs or spline maths, there is probably just a relaxation calculus of a fixed grid of points, so it's hard maths and physics, but not the same that defines NURBS and Spline.