As we know the geometry created in Houdini is composed of surfaces.

So, can we do “solid” modeling in Houdini? or is there a way to convert or export the model created in Houdini to “solid” type of model?

I ask this because some of the simulation software such as Computer Fluid Dynamic (CFD) simulation software can only work with “solid” type of model rather than 3D model enclosed by a series of surfaces.

Hope you can kindly advise!

Thanks!

- Ji

VoronoiFracture SOP does create new polygon faces when it break a model. Just not sure whether that's what you want since it's not modeling.

eitht.

Hi,

I'd guess this won't be possible, not even in the future. The only supported file format which is even theoretically capable of containing solid models is IGES, and this is not handled really well. For CG, you don't really need solid models, so Houdini isn't really to blame for that.

You will need some CAD application which can read one of the supported file formats and try to make a solid from the surfaces. This again can be very nasty, as you won't get nice NURBS surfaces, but rather triangles.

The other way around is much easier …

Regards,
Carsten

You can do what is in the illustration with the Cookie SOP.

The Cookie SOP (Boolean Operations) uses CSG logic to do it's work but as mentioned above, there is no knowledge of a “CSG” object. You have to do all the leg work to ensure that the incoming objects and the results are manifold watertight objects with primitive normals facing outward.

Hi, Jeff, thanks!

Can you explain a bit more on what a “CSG” object is?

CSG stands for Constructive Solid Geometry. Google it and you will get quite a few links.

In it's purest form, you start off with basic primitive mathematically defined shapes: spheres, boxes, tubes, tori. Houdini has these primitives but you can't perform CSG boolean operations on them. I believe you can in RenderMan btw. You then take those shapes and intersect them and perform other operations on them to create a final object. This final object along with all of the initial primitives are manifold or water tight and mathematically defined. A nice property to have when you need to drive tooling paths for machining these parts.

At any point you can alter one of the original basic shapes or any of the boolean operations and everything re-evaluates. It is this ability to tweak any of the initial parameters that makes this very powerful and is sometimes referred to as parametric or procedural modelling as well.

As features needed to be added such as fillets and rounding of corners, the CSG approach starts to get real complicated. Every round fillet needs to be created with tubes intersecting edges and spheres for corners. Great for regular objects with no curving surfaces.

I remember using CSG for modelling back in the mid 80's as it was all the rage with engineering and architecture.

Then came NURBs. NURBs pretty much killed CSG for industrial and automotive design that ultimately is used for the machining of parts. NURBs modelling is similar to CSG modelling where you define surfaces and then perform intersections on those surfaces with controlled fillets and more. Precision is required to create final smooth surfaces with what the industry refers to as c2 continuity (continuous in the second derivative I believe…). Change any of the underlying NURBs surfaces and everything that comes after re-evaluates. This, like CSG modelling, is referred to as parametric or procedural modelling. NURBs to this day remains one of the key surfacing techniques for creating tool paths for creating designs and the creation of parts in the manufacturing and auto industry.

Our CG industry gets away with modelling with polygons because it requires little operator experience to model decent geometry and is very flexible and forgiving. The resolution independence comes from converting the poly cage to Sub-Division surfaces (a spline-based surface that is resolution independent). Polygon modelling works well because most CG modelling tasks require the artists to replicate objects quickly, not design them with numerous revisions. Well, ok, redesign in polygon modelling usually means starting from scratch, over and over.
NURBs (and CSG for that matter) greatest strength is in the ability to endlessly tweak the surfaces and see the results pretty quickly in the design phase.

If Houdini's NURBs toolset were to hold together under all modelling operations, it would certainly challenge Polygon modelling for non-deforming objects as it is inherently procedural unlike Polygon modelling which tends to fall apart when you make upstream changes.


And that's it for today's history lesson.

haha, Thank you very much for the history lesson, Jeff! will take some time for me to digest it …