How it works

PathEngine's 3D content processing works by combining a set of solid obstacles into a single set of 3D boundaries, essentially like the kind of boolean combination operation you will find in many content applications, but with an added extrude operation to remove any areas for which an agent of a given height would be obstructed.

Refer to this section of the documentation for more detailed information about this process.

Tiled 3D content processing

Note that the demo uses the tiled 3D content processing setup to enable fast updates when only part of the scene has been modified.

This works by splitting the scene up into a set of horizontal tiles, with each part of the scene processed independantly, and the resulting set of ground pieces then combined into a single ground mesh for run-time pathfinding.

The first time you switch from edit to test mode (and run the 3D content process) a set of tiled result pieces will be saved to the SDK bin directory, together with information about the tiling used and the source content elements used to build each tile piece.
Next time you run the process, the demo checks the source data for each tile against the corresponding saved snapshot, and if these are identical then this part of the process can be skipped, with the existing ground result reused.

Building a mesh federation from the tiled pieces

In addition to generating a single run-time ground mesh, it's also possible to generate a set of run-time tiled meshes for use with PathEngine's Mesh Federations functionality.
(Enabling the 3D content process to be applied for seamless pathfinding across worlds of effectively unlimited size.)
The mesh federation example project includes (commented) code paths to demonstrate this.