Voronoi fracture issues

[catchyid]

Hi,

I've used solid modeling (i.e. cookie) to model a wall with thickness, then I applied Voronoi fracture. I am getting bad results (see attached). I did the mesh myself (find attached), it's very primitive, just a box (wall) intersecting with other boxes (windows).

-I don't see why such a simple geometry is problematic to Voronoi fracture node? I checked Voronoi docs and I could not find specific requirements for the input geometry?!

-I've inserted "clean", "facet" nodes before Voronoi but it did not fix the problem...

-I converted the mesh into VDB then convert to polygon and applied Voronoi fracture --> that did work, but it's not practical! it's very slow, needed high resolution volume which produced very detailed geometry and it consumed all my computer memory, and this happened in one wall only, not the entire building...

-I've read once about "Volumetric Shattering" on this forum and its superiority, is it the above method or different?

-If anyone has a video about this Voronoi fracturing that discusses possible problems, pls share

Thanks

wall.bgeo

[Atom]

Here is a fracture setup that I generally start off with.

It uses an ISOOffset to create more generalized sized chunks and not spears. Uniform Sampling should be set high for best detail.

On long low poly imported flat objects I often add a Divide with Convex Polygons turned off.

Enable Bricker on the Divide and chose a lower size until grid lines start to appear. Dial in the grid to match scene resolution.

Scatter points but use a low count.

Use a PointReplicate to scatter more points from the original scatter points. This is why we use a low count above. Drop the count to something low like 10. Then use the Uniform Scale to dial in the scatter around each scatter point. This creates a pocket of smaller fracture areas around a scatter point.

Follow up with a group set to a bound box or object. Any points inside the bounding area will be connected to the VoronoiFracture. Any points outside the bounding area will be deleted and thus produce no fractures. Use the bounding box to decide where you want fractures to occur. Try removing only a lower corner and view the sim. Each new bounding box position will generate a new sim. There is also a Switch to turn on all points and not use group.

In this animation the group fracture points are used instead all the scatter points. This allows you to create a fracture focus area where major destruction will occur. The side effect of using a small group focus area is that you will be surrounded by larger parts. But you can drop down more bounding groups to break up those larger pieces as well with a new group node. Make sure to set any new nodes to contribute, not overwrite, points to the master fracture_points group.

ap_prepare_geo_fracture.hiplc

Edited October 9, 2016 by Atom

[catchyid]

Thanks Atom for your reply...I've replied to your post earlier but it's gone !? so, in summary:

-First, thanks for your detailed example and destruction pattern

-I've realized that Voronoi Fracture node does not work all the time, meaning, the input mesh must have some conditions in order for Voronoi to work correctly. For example, the same wall (see attached) is fed to:

>Voronoi node directly --> fails

>Divide with bricker distance = 1 then Vornoi --> works

>Divide with bricker distance = 2 then Vornoi --> fails

The problem here is I don't find anywhere in the documentation, the conditions that the input mesh must satisfy in order for Voronoi to work correctly? To make it worse, even when I use "divide" node, sometimes it fixes the problem and sometimes it does not depending on the bricker distance...

simple_wall.bgeo

divide_and_voronoi.hiplc