Below is a simplified set of data that can be used to recreate a cube.

Code:

allPlanes = #() /*Add all the known points where each defines a plane: */ append allPlanes #([-64,64,64],[64,64,64],[64,-64,64]) append allPlanes #([-64,-64,-64],[64,-64,-64],[64,64,-64]) append allPlanes #([-64,64,64],[-64,-64,64],[-64,-64,-64]) append allPlanes #([64,64,-64],[64,-64,-64],[64,-64,64]) append allPlanes #([64,64,64],[-64,64,64],[-64,64,-64]) append allPlanes #([64,-64,-64],[-64,-64,-64],[-64,-64,64]) allPlanes.count /* How many faces in the resulting mesh */

If you were to create a plane for each item in the array, they would all intersect and the innermost section would be a cube. You could then select a face and, in this example, see that each has a fourth vertex. More complex geometry would have more verts added in faces .

Can anyone share some expertise on a way to convert this into a full working piece of geometry? The various methods I've come up just don't work... and the easiest cheat I've tried (

**nvpx.CreateConvexFromPoints()**) always ends up with verts that are not exactly where I fed in the positions (always moving a tiny fraction off... but unusable because of this).