Cool renders, guys! :-)

-Will

Interesting thread here on polys. Personally I agree with XROK1 that I would like to see MOI stay as clean from polys as possible. Maintaining simplicity is vital. Once too many functions or tools have been added to an application it becomes overwhelmingly complex and that attractive user-friendlyness would soon be lost.

However one thing I have been thinking about is if you guys out there know of any application that is able to translate standard polygon meshes e.g. OBJ or STL into Catmull-Clark subdivision surfaces. Such an application if existent opens the door to translating trimmed NURBS into subdivision surfaces for postprocessing surfaces where more sculpting flexibility is needed.

Is there perhaps something available out there that is able to directly translate trimmed NURBS into SDS?
P.s. I think T-splines has not yet the ability to translate trimmed NURBS into SDS.

All the best,

The Lapstrake Lover

why not just take an .obj/ngon export from Moi into any app (ie Modo) capable of SDS, and apply SDS to it?

it works for me.

XROK1, Maybe that´s not such a bad idea. My apologies for asking silly questions, but I am not familiar with MODO or any of these high end polygon/SDS modeling applications.

However I have specific topological requirements. I want to convert meshes based on triangles into Catmull-Clark subdivision surfaces having a rectangular controlpoint grid. Is modo capable of translating meshes consisting of triangles into SDS divided by a recangular control point grid? Direct translation of NURBS into rectangular SDS topology l is easy, but when the NURBS has been trimmed things become complicated.

which modelling program do u use.

I ask because (someone please correct me if i'm wrong) you should be able to take an .obj with ngon divisions out of Moi and apply cat-mull division in another modeller that supports it (not 100% sure, do some experiments)?

Hi XROK1,

Just like you I have no desire to make monsters.Currently the only SDS modeler I have is Delftship (a great application for boatdesign). It uses natively SDS based on rectangular (Catmull-Clark) topology. It´s also able to import NURBS (via IGES) but only untrimmed NURBS. I was hoping that by tessellating my trimmed NURBS files to OBJ, STL or any other common triangle based mesh and subsequently converting it (if possible) to rectangular Catmull - Clark topology I would be able to bypass the "trimmed NURBS" import problem.

Delftship is not able to import .obj, it however imports VMRL 1 and I was hoping to use that route for getting my trimmed NURBS converted into Catmull Clark into the application.

P.s. attached is a screenshot from Delftship.

All the best,

The Lapstrake Lover

One more thing.

Here is a link to an article on the currently hot and exciting subject of bridging NURBS and SDS.

http://portal.acm.org/citation.cfm?doid=1138450.1138455

Moi's n-gon meshes are not suitable for subd smoothing because the extra vertices which create the n-gons are not matched up to the vertices on the coneccted faces. The exceptions would be more simple objects such as a cylinder or box. But then you might as well have modeled them in subd to begin with...

NPower makes a 3dsMax plugin which translates trimmed nurbs objects into quad meshes which can be smoothed as catmull-clark subd...

jonah

JBShorty,

Thanks for the tip. The tessellation to of trimmed nurbs to quads mesh is exactly the thing I have been looking for. Unfortunatly, the cost of 3DSMAX + NPower + the quads mesh plugin is too great to be justified by just this minor requirement:-(

Are you aware of any other application (and perhaps not as expensive) able to tessellate trimmed nurbs to quads mesh??

By the way, here is another link to info on the interesting topic of bridging NURBS and subdivision surfaces:

http://www.cgv.tugraz.at/CGV/research/NURBSUBS/esubs.html

The Next Engine 3D Scanner comes with software for handling scan data. And in the demo video you can see when they automagically skin a mesh with a whole network of Nurbs patches. Assuming they are all untrimmed 3 and 4 sided nurbs patches, they could easily be reparameterized to some uniform spans in U and V directions, then extract the control polygons from the Nurbs patches. And that should be close to subd ready (at least enough that some minor hand work could fix in a short time). So i think there are solutions out there, just not in obvious places. And not always budget-minded...

jonah


run the "Scan Studio Core" demo video and watch closely at the topic labeled "Surfaces"

https://www.nextengine.com/indexSecure.htm

Boatstalker.

I have been using Blender for rendering and noticed that it does export in VRML 1.0 and VRML 97 format.

Perhaps you can import into Blender ( there is a Python script (Blender Importer) in the Stand Alone Utilities, located on the Scripts page and it works very well) then export to you other program, And of course Blender is as free as a bird.
Maybe this might solve your problem.

http://kyticka.webzdarma.cz/3d/moi/ Custom Commands at the bottom.


Best wishes.
Malc

> Moi's n-gon meshes are not suitable for subd smoothing because the
> extra vertices which create the n-gons are not matched up to the vertices
> on the coneccted faces.

Hi Jonah, that's not correct, there should be a complete alignment between all vertices of adjacent polygons across a joined edge.

If there is any mis-matching, then either you have an unjoined edge in your NURBS model there, or you've run into a bug.

Can you please post an example of the non-matching vertices that you're talking about so I can see what is happening for you?

- Michael

sorry, i phrased my statement incorrectly. It's not the vertices which cause the smoothing problem. It's the edges which don't meet up at the ends. Most of them form a "T" where the edge terminates, which is not ideal for subd...

jonah

Hi Jonah, maybe what you mean here is that the underlying quad mesh (before it becomes n-gons) is not perfectly aligned between all surfaces?

By this I mean the quad "posts" that form polygons internal to each surface.

That quad structure follows the UV layout of each surface, so it isn't typically feasible to get identical quad structure between different surfaces that meet along a trimmed edge instead of along a natural surface edge.

Things like cylinders or spheres with symmerical cuts can often align along a natural surface isoparm, but when you just boolean 2 shapes together it isn't typical for the surfaces to meet at an isoparm.

It's basically the same problem as trying to reconstruct a complex model made up of booleaned parts by using only one single big untrimmed surface.

So certainly MoI's mesher won't do that, but every vertex on every n-gon there will be a matching vertex on the adjacent face though, making a "water-tight" mesh.

In the future there will be a few things I can do to increase quad-structure alignment in some different cases, like for fillets that have a plane on one side and a curved surface on the other side.

Doing all matching quads along every edge is a lot harder problem - it means creating a mesh that is not initially aligned to the natural UV grid of the surface, which doesn't really fit very well with some of the normal processes of surface algorithms such as splitting surfaces into smaller ones which has to happen along UV iso boundaries.

- Michael

Hi Jonah, it is kind of a general problem that arbitrary n-gons and subd do not mix well, regardless of a "perfect" vertex alignment.

Part of the sub-d process involves taking a centroid of the polygon, and if your polygon is not a convex shape, it won't have a nice centroid-ization to it.

Here's an example:



From an n-gon structure standpoint, that is pretty much a perfect mesh structure with a minimal amount of vertices. Yet regardless if you try to sub-d it, it will work but the result will be ugly as pieces radiate toward the centroid of the n-gon and kind of fold over the concave areas.

For nice sub-d-ifiying, you don't really want minimal vertex count n-gons at all, but rather a set of convex polygons even if that requires adding a lot more points to the entire mesh.

It's a completely different thing... Just more vertex alignment does not solve the problem.

- Michael

Hi Malc,

Currently, Blender doesn´t import NURBs .

However, take a look at the German video on ESubs (extended subdivision surfaces). Imagine modeling with MoI featuring an ESubs kernel and the possibility of being able to click a "convert to subdivision surfaces button" which would convert the NURBS model to SDS for further refinement. Imagine the possibility of being able at the click of a button to go to NURBS mode and SDS mode back and forth directed by which modeling method suits you each time, until your model is perfected.

http://www.cgv.tugraz.at:28888/CGV/research/NURBSUBS/_id246b1_

Best Regards,

Boatstalker

Boatstalker.

So what you are saying, is that when a file is imported, it is converted to a poly mesh and does not retain the Nurbs and thats not what you want.
I have very limited knowledge of programing, so I find it hard to grasp all the different formats.

I am basically an amateur artist that finds 3d modeling an extension of my artistic capabilities and quite a challenge sometimes.
Blender is a challege with it's millions of buttons to learn, whereas MOI (Thankyou Michael) is a joy to use, so I tend to use the software and hope that it comes out ok, without having to understand how I got to the end result.


That software certainly looks awsome.

Hope you succeed in your goal.

Regards

Malc

boatstalker, you might want to try this. not sure if it will work but it seems promising. let us know;-)
http://www.freedesign-inc.com/products.html

Hi Michael. I understand it is more complex than what i mentioned, it was just the minimum example of a simple object which could be exported from Moi and be ready for sds (or at least with minimal amounts of mesh refining)...

jonah