int [] primfind(<geometry>geometry, vector min, vector max)
Find all the primitives whose bounding boxes overlap the given box.
int [] primfind(<geometry>geometry, string group, vector min, vector max)
Find all primitives in a group whose bounding boxes overlap the given box.
<geometry>
When running in the context of a node (such as a wrangle SOP), this argument can be an integer representing the input number (starting at 0) to read the geometry from.
Alternatively, the argument can be a string specifying a geometry file (for example, a .bgeo) to read from. When running inside Houdini, this can be an op:/path/to/sop reference.
min, max
These vectors define the minimum and maximum corners of the bounding box to search.
group
If given, only match primitives in this group.
An empty group string will include all primitives.
The string supports Ad-hoc patterns like 0-10 and @Cd.x>0.
Returns
An array of primitive numbers.
Note
These functions are intended to be used as an optimization to finding primitives in a particular area for processing.
A primitive being returned means it MAY intersect, not that it WILL. However, a primitive NOT being returned means it will NOT intersect.
For instance, to find all the curves from one input intersecting
polygons on another input, we may naively iterate over all
polygons for each curve to determine their intersection. To speed
this process, we may find which primitives may intersect a
particular curve using these functions, and iterate solely over
the potentially intersecting primitives. This significantly
improves performance since primfind uses an underlying tree
structure to speed up search.
Examples ¶
Remove primitives that may be intersecting the unit box centered at the origin:
int[] prims = primfind(geometry, {-0.5, -0.5, -0.5}, {0.5, 0.5, 0.5}); foreach ( int prim; prims ) { removeprim("primitives.bgeo", prim, 1); }
Alternatively, we can use a query bounding box from an auxiliary source:
vector min, max; getbbox("bbox.bgeo", min, max); int[] prims = primfind(geometry, min, max); foreach ( int prim; prims ) { removeprim("primitives.bgeo", prim, 1); }
To see the performance benefit of primfind, compare it to the following equivalent
implementation of the function above:
float tol = 1e-5; vector min, max; getbbox("bbox.bgeo",min,max); int n = nprimitives(0); for ( int prim = 0; prim < n; ++prim ) { int[] verts = primvertices("primitives.bgeo", prim); // compute primitive bounding box and store it in prim_min and prim_max vector vert_pos = point("primitives.bgeo", "P", vertexpoint("primitives.bgeo", verts[0])); vector prim_min = vert_pos, prim_max = vert_pos; for ( int v = 1; v < len(verts); ++v ) { vert_pos = point("primitives.bgeo", "P", vertexpoint("primitives.bgeo", verts[v])); prim_min = min(prim_min, vert_pos); prim_max = max(prim_max, vert_pos); } // bounding box intersection test if ( prim_max.x - min.x < -tol ) continue; if ( prim_max.y - min.y < -tol ) continue; if ( prim_max.z - min.z < -tol ) continue; if ( prim_min.x - max.x > tol ) continue; if ( prim_min.y - max.y > tol ) continue; if ( prim_min.z - max.z > tol ) continue; removeprim("primitives.bgeo", prim, 1); }
| See also | |
| intersect | |
| prim |
|
| search |