I’ve just uploaded a screenshot of the automatic level-of-detail algorithm I’m implementing.
It’s taken me quite a while to get the algorithm completely stable and handling all the nuances of the meshes that are thrown at it, but I think it’s ready for the prime time now. I’m now integrating it into the rendering pipeline with automatic reductions with distance, LOD biasing (for small viewports and particularly important entities) and getting the exporters to precompute LOD data at export time. Watch this space.