Hi Michael,

I'm wondering if you've plans for sub-object manipulation like Rhino added in their last iteration? Seems like a cool feature.

Hi Chipp - in the future I would like to make it possible to do transformations on faces like move, rotate, or scale a face - but probably not quite in the exact same way that has been done in Rhino because the way it was done there is somewhat problematic with surfaces tending to mutate in somewhat strange ways.

My plan is to follow more of the "direct modeling" type mechanism in some other CAD programs like SpaceClaim. I'm not sure when that will happen though since it will involve a lot of work.

There is some limited amount of that in place with the new feature for MoI v3 in extrude where if you extrude a face sub-object the result is automatically booleaned onto the main object.

- Michael

Just looked at the SpaceClaim video and it looks pretty cool how they are able to provide some parametric modeling. I think this is a great feature for enabling the refinement and 'what-if' types of edits to an existing model. From my brief research, SC doesn't seem to have better 'editing' than 'creating' types of tools.

Just wondering, does MoI and SC share the same underlying kernel technology? I know a little (what Google tells me) about ACIS, Parasolid and CGM as solid geometrics kernels, but really not much more about what kinds of 3d kernels are available. I assume MoI and Rhino3D have the same kernel technology. Is it a NURBS type of kernel, or a solids kernel (or are they one and the same?).

I *assume* surface modelers, like Modo, Lightwave, SketchUp don't use a 3rd party kernel, but maybe they use some open source libraries?

If I'm overstepping proprietary bounds, then please let me know-- I'm just interested in understanding how things work. :)

Hi Chipp,

> Just wondering, does MoI and SC share the same underlying kernel technology?

No, SpaceClaim uses the ACIS library and MoI currently uses the Solids++ library. In the future I will probably be doing some exploratory work to see if it's possible to switch MoI to use ACIS for some operations, but that's a ways off.


> I assume MoI and Rhino3D have the same kernel technology.

No, they use different kernels. MoI does use the OpenNURBS library which deals with reading and writing the contents of 3DM files, but that's the only part that's in common and it's specific just to file I/O. When MoI reads a 3DM file it translates the data into Solids++ data structures and then it's all Solids++ stuff that it deals with after that.

Rhino uses their own internally developed geometry kernel.


> Is it a NURBS type of kernel, or a solids kernel (or are they one and the same?).

They are usually part of the same kernel, usually in solids modeling kernels the way a solid is defined is by a set of surfaces (sometimes with some various possible types of surfaces including NURBS surfaces) that are joined together at common edges to define a watertight skin. There are usually a combination of NURBS functions and topology functions within the same library.


> I *assume* surface modelers, like Modo, Lightwave, SketchUp don't use a 3rd
> party kernel, but maybe they use some open source libraries?

Usually polygon mesh modeling programs use their own proprietary internally developed libraries for their geometry. Typically there isn't as much fancy analytic mathematical work involved in these and so it tends to be easier for it to all be developed internally rather than coming from some other library. There are some open source libraries for polygon mesh geometry as well but as far as I know none of those particular programs that you mention are based on one.

- Michael

Fascinating. Thanks for the explanation. The fact that your program does not use the same underlying technology as Rhino yet you can read/write the same file format (which I assume you had a big part in defining) is impressive in itself. My guess is translating data files from one kernel to the other is not a trivial task.

Hi Chipp, yes it can be challenging converting data between different kernels. The tricky part is that each kernel can have somewhat different ideas about what is valid and invalid data, like how close 2 values have to be to each other to be considered equal and various things like that.

- Michael