This video showcases the basic quadtree based level of detail scheme that I currently have in my project.
The quadtree is based on the distance from the camera to the object cube (which is converted into a sphere).
Because of this, the LOD doesn't work uniformly across the surface as when the camera reaches the "corners" of the sphere, the distance is incorrect.
Each patch is created using the DirectX 11 tessellation stages and is a 64 x 64 grid, therefore each patch contains 8192 triangles.
When the camera reaches the surface in the video, there are around 10,100,736 triangles (1233 patches) rendering.