I was watching a tutorial on youtube for Rhino, and the teacher there said something like this when he was making a surface "don't worry about the strange surface, it's Rhino it's always going to be bad". That got me thinking about wether moi geometry or building tools were better than rhino's, my gut feeling says yes, because a far fewer booleans and fillet work in Rhino than in Moi for me at least.

Could this be right? is the underlying technology better or different somehow?

And as a side-note, I was reading through old posts here and I saw something about, that you Michael were thinking of adding SubD modeling in Moi, is that right? That would be really awesome, for those organic shapes!!!!

Hi Anna, well it's difficult to make just one blanket statement that "x is better than y" - it's more of a mixture. In some particular things MoI is better and in some particular things Rhino is better.

I'd say that MoI is especially strong in the area of booleans though, I frequently see people post examples of boolean problems in Rhino that are able to be done in MoI. But even that can vary in a particular case.

MoI's filleter is not up to the same level as something like SolidWorks, but it does seem to handle mechanical type parts somewhat better than Rhino but again that can vary in particular cases.

Rhino has advantages in its deformation toolset (the UDT tools), and more things set up for maintaining continuity in surfacing operations.


> Could this be right? is the underlying technology better
> or different somehow?

Actually there are some differences in the underlying technology - I think the geometry engine that MoI uses is able to represent non-manifold shapes more fully than Rhino's system.

The system that MoI uses is a full implementation of Kevin Weiler's thesis, including the "faceuse" structure where it is possible to set up a non-manifold face (one face bordering on 2 adjacent solid regions kind of like a cell wall) rather than only non-manifold edges. This can be useful as an intermediate stage in boolean processing and I think it is a big factor that helps MoI's booleans work so well.

But on the other hand it is also a more complex structure.


Also MoI's geometry engine is able to represent solids with void pockets inside of them, such as a smaller interior sphere booleaned away from a larger outer sphere solid which Rhino is not able to do.

 

> that you Michael were thinking of adding SubD
> modeling in Moi, is that right?

Yup, it is definitely something that I'm interested in.

But it is going to take quite a bit of work to make it happen, so I'm not sure when it will happen, it may be a while.

- Michael

"Yup, it is definitely something that I'm interested in.

But it is going to take quite a bit of work to make it happen, so I'm not sure when it will happen, it may be a while."


This is definately something you have to elaborate on:)????? would it be some kind of a new mix of nurbs and subd...box modeling techniques???

Thanks for a good answer on the other question:)

Can we have an image of something "Manifold" and something "No Manifold" ?

One representation:

each edge should be used by exactly two faces (not more, not less).

Non:



Non:



Manifold:




I'll look for other examples. It is a mathematical concept.

Hi Burr, normally the case where the edge is solitary and only on one face is not considered non-manifold (like in your second image).

But yes the first example you have there is a good one, where having those 2 boxes combined together into one structure would require that one edge to be shared between 4 faces instead of only 2 faces.

Internally the geometry library that MoI uses allows for such things and it tends to help to make booleans work better since things can be combined all together into one unified structure before it focuses on eliminating different parts of it.

There is a kind of normalization process that MoI does to any non-manifold results after the calculations are done, to make sure they are split up into different manifold chunks.

The equivalent for faces is having it possible to have one structure that has a face in common between 2 different solid regions, like this:



- Michael

Hi Anna,

> This is definately something you have to elaborate on:)?????
> would it be some kind of a new mix of nurbs and subd...box
> modeling techniques???

Well, it is hard to elaborate on it when it does not exist yet! :)

I just won't know a lot of particular details until I get a chunk of time to work on it in the future at some point.

But I think ideally it would be great to have a sub-d method of creating a base surface, as just another option to construct surfaces. Like in addition to sweep, network, extrude, etc... it would be good to be able to have a "sub-d" cage surface that generated a NURBS surface.

One of the difficulties though is that sub-d / box modeling has a fairly different kind of toolset than NURBS modeling, so it would mean introducing quite a number of brand new tools to work on the polygon cage. That means it will take quite a bit of work and that's why I don't know when it will happen.

It still has yet to be determined if it is even possible to introduce all those additional specialized tools into MoI without making MoI become overly bloated.

- Michael

This technology was explored in another software called EIM, the ElectricImage Modeler. It was quite popular, though the company ended up canceling the software due to licensing problems with Spatial. They're currently planning to resurrect it.

What's nice about subD's within a NURBS framework is that you can sculpt using your familiar Polygon subD cage techniques, but also make use of NURBS boolean and trim functions (depending on how it's implemented). Also the final polygon mesh can be exceptionally clean, with density set by the meshing function.

Malcolm Thain was the acknowledged master of this 'sculpting' tool when using EIM.

http://transputer.org/3dgallery/main.php?g2_itemId=964

If this can make it into MoI it will really set the software head and shoulders above the crowd.

Anna wrote: "I was watching a tutorial on youtube for Rhino, and the teacher there said something like this when he was making a surface "don't worry about the strange surface, it's Rhino it's always going to be bad". "

Or he mean that bad look caused by the antique Rhino mesher. Here I'm sure - the MoI mesher is much more better and faster. :)

@Bur and other thx for the Manifold precision

<Malcolm Thain was the acknowledged master of this 'sculpting' tool when using EIM.
indeed!

Kurt that sounds very promising BUT that tech meant to be released years ago... no wonder why Dassault/Spatial gives them trouble...They want to sell Imagine and Shape after all :)

I also like the Autocad 2010 approach, The Maya & Autocad Team have been working together to combine Sub & solid modeling operations...

http://www.youtube.com/watch?v=Xy9ZIqVE-jQ

http://www.youtube.com/watch?v=rK58M_y43jY

http://www.youtube.com/watch?v=0hlLIkXxIxk

Hi Neo...

What is the result from the model in Autocad 2010, mesh or nurbs ?

you can save as .sat then import to the app of your choice...autodesk makes a big time move on .sat...it works well with Rhino, spaceclaim... (not always with moi (?) still in search for the magic bullet :)

Nice. Looks like others have moved ahead and the technology is now fairly common. At least in AutoCAD land.

yep Is very promising BUT is nothing new as I said the code comes from Maya and is been there for years. One of the reasons why the T-Splines connection to Maya was not a success IMO.

here is an example from maya, modify>convert>subdiv to NURBS.

>>>here is an example from maya, modify>convert>subdiv to NURBS.>>>>


And hows the resulting NURBS model from that conversion??

Burr wrote : "And hows the resulting NURBS model from that conversion??"

Now that's the real question, isn't it? Amapi also had a sub-Nurbs converter but the result was not on par with T-Splines...

Hi Michael,

<< One of the difficulties though is that sub-d / box modeling has a fairly different kind of toolset than NURBS modeling, so it would mean introducing quite a number of brand new tools to work on the polygon cage. That means it will take quite a bit of work and that's why I don't know when it will happen.>>

Maybe a way to import the sds cage would be a great start :) there are allready so many poly modeler around.
Now, if you come with a new poly modeler, with the same 'MoI Touch', it would be amazing !

"Maybe a way to import the sds cage would be a great start"

WOW, would that be possible?

> One of the difficulties though is that sub-d / box modeling has a fairly
> different kind of toolset than NURBS modeling, so it would mean introducing
> quite a number of brand new tools to work on the polygon cage. That means
> it will take quite a bit of work and that's why I don't know when it will happen.

Well, you could just make it more generalized then. MoI already has the ability to push and pull points, curves, and surfaces, with and without the precise transform tools. Thus editing a sub-d cage would be very much like editing the hull cage of a surface now. If the purpose is to use this cage to generate NURBS surfaces, then the only real additional tools I could see as being needed would be a fast way to add edge loops and some loop and ring selection. And even these tools would be of use for NURBS work; adding a loop to a surface would just be a very fast trim and such.

The workflow I envision would be two ways of entering into this sort of cage edit. One would be to create a box (that would be making a sphere). The next would be to do something akin to the Network tool, only you end up with a cage controlling the surface instead of the surface directly. From there you could pull the points and edges around, use the new tool to add loops, and box model your way around. Tools like Extrude or Chamfer would work off a face or an edge much the same as it does on NURBS surfaces. The tools that can't work on a surface or solid would likewise not work on a cage. And I don't really think it'd be safe or wise to let Booleans work on them (way too messy to make it work consistently).

Once you like it, you could then drop the cage down to a surface and have a nice, smooth NURBS surface made from the more gooey and sculpture sub-d methodology. It wouldn't really be a full-fledged subdivision modeling solution, but that isn't a big deal I don't think.

Ideally, one could also take a surface and push the same Cage Create button and get a lower-res cage around that surface (or even solid). One of the main issues I have with trying to push or pull surfaces generated in MoI from lofts, sweeps, or networks is that the control point density is often huge. Without a soft-selection or free-form deformation cage, modifying these surfaces naturally is a pain. There's also no way to deform things (like a bend or twist) without doing it the very tedious and unnatural way (moving point by point). If I could just toss on a cage, such things would be cake; add in a soft-selection falloff and I think I'd be in heaven.