Hello Steve,
You can merge chunks at any point after the cameras are aligned:
- build depth maps in chunks separately, merge chunks, build mesh in the merged chunks from preserved depth maps,
- build meshes in separate chunks, merge chunks with merge models option enabled,
- merge chunks with the alignment only, run build depth maps and build mesh operations in the merged chunk.
There is no requirement to have depth maps, dense cloud or mesh pre-generated in the chunks prior to the merging operation, but if it makes manual chunk alignment more convenient, you can have them.