Ryan

Here you go. I put comments in your file. grain_test_collision_stuck_fix.hip

Here is my attempt. For sure you need unique names and ids. RBDs_SOP_solver_testing_01_fix.hip Hope this helps.

// Relative Neighbourhood Graph // https://houdininote.tumblr.com/post/162427874300/connecting-points // This wrangle runs thru all points. // Current point = point being ran thru // 'nearpoints' gives you a list of point numbers near the current point foreach(int pt; nearpoints(0, @P, 1.0, 10)) // For every point 'nearpoints' finds in the given radius.. { if(pt == @ptnum) continue; // Skip itself, go to the next iteration vector p = point(0, 'P', pt); // Store position of the near point float dist = distance(p, @P); // Store distance (mydist) from current point to near point // Building 2 new arrays... // We will search for any points within the radii (mydist) of both the current and near points // to check if they are closer to each other. int ptsA[] = nearpoints(0, @P, dist); // Look for any more points around the current point within mydist... int ptsB[] = nearpoints(0, p, dist); // Then, look for any more points around its near point within mydist. int chk = 0; // Initialize a flag foreach(int cpt; ptsB) // For every point (cpt) found in our 'ptsB' array.. { if(cpt == @ptnum || cpt == pt) continue; // if cpt is the current point or near point, skip it. chk = removevalue(ptsA, cpt); // Remove cpt from our 'ptsB' array. If successful, set our flag to TRUE. if(chk) break; // If we successfuly removed cpt, break out of this nested loop. } // If our list 'ptsA' is empty, lets draw a line between the current point and the near point. // If there were points left in the list, no line will be drawn. if(!chk) // If we failed to remove a value previously... { int pr = addprim(0, 'polyline'); // create a new empty polyline addvertex(0, pr, @ptnum); // Add a new vertex to that polyline at the position of the current point addvertex(0, pr, pt); // Add another vertex to that polyline at the position of the near point } } Im not the original author of this code. Its an example I pointed to in that link.

Here is a solution using some stacked grids and intersect analysis: isosurfacetest_fix.hiplc

Here is one solution. Sort Points_fix.hiplc Here is a really nice blog on connecting points: https://houdininote.tumblr.com/post/162427874300/connecting-points There is an example file with it too.

Yea. That's what it is. Even voronoi fracture is different. Good to know.

When you negate your Normals, your color values are going from 0 to -1. You need to check the sign ( + or - ) of each vector component of the Normal. If it is negative, you need to take the compliment of the Position component [ ex. (1-P.x) ] then do absolute value.

Send a screen shot. Is this what you mean?

It might be because the RaySOP doesn't wok well with curves. Here is a VEX idea. I'm not sure if this is what you were going for.

You might need to reset the cluster numbers and names. rbd_cluster_constraint_name_fix4.hip Hopefully that bug will get fixed in the next release.

Here are 3 ways to emit packed prims. Hope this helps. emitting_packedprims.hip

Here is my attempt. I am using a TransformPiecesSOP to transform the glass shards. Then using those transformed shards in a new dopnet. Hope this helps. layered_destruction_fix.hiplc

Small fix... rbd_cluster_constraint_name_fix3.hip

I think this might be what you want. Try this. rbd_cluster_constraint_name_fix2.hip

Your problem was that the RBD Cluster node was generating pieces with the same cluster number that were too far away from each other. I switched the Cluster Type to Group Constraints. That seemed to work. But I dont know if its what you want. rbd_cluster_constraint_name_fix.hip