[vtk-developers] vtkExplicitCell

[David Thompson]

> This could work, although I'm not quite seeing how trim edges would fit in.

All the trimming I've seen done in CAD BReps is in world coordinates with splined curves that represent the bounding edges of the surface, not in parameter space; meaning that the bounding curves would also exist in the multiblock of structured grids and could thus be referenced in a surface's field data (by block ID, for instance).

I agree that we would have to figure something out if given a trim polygon in parameter-space coordinates. That doesn't seem like an insurmountable obstacle, though. And having a NURBS representation using standard VTK datasets with adaptor classes that provide access and convenience methods seems really desirable so that filters can be used on the control polygon point coordinates and point data (e.g., vtkArrayCalculator, vtkWarpScalars/Vectors).

> Also sometimes there are supplemental structures like polygonal tessellations to help speed up geometric operations, etc. 

What Lin is working on now is generating those tessellations for rendering. Assuming we generate a multiblock of unstructured grid tessellations in the same tree structure, it could be used for both rendering and accelerating geometric operations.

	David

> 
> On Fri, Jul 3, 2015 at 9:15 AM, David Thompson <david.thompson at kitware.com> wrote:
> > Long term I'd like to think how we could represent a CAD model with a large number of trimmed splines.
> 
> What problem do you foresee with a multiblock of structured grids? The memory overhead of vtkStructuredGrid? It seems pretty minimal to me compared to the b-reps CAD packages use and the multiblock dataset need not be flat... it could be a bounding box hierarchy for fast culling of irrelevant patches.
> 
>         David
> 
> 
> 
> -- 
> William J. Schroeder, PhD
> Kitware, Inc.
> 28 Corporate Drive
> Clifton Park, NY 12065
> will.schroeder at kitware.com
> http://www.kitware.com
> (518) 881-4902