Can be done, although high LOD takes a while and the current implementation might eat up all memory. You can run it in parallel though. Shard by octants :)