Geoclipmap
Renders vast terrain as a set of nested grids around the camera, each outer ring half as dense — level of detail that falls off with distance and leaves no visible seams.
Described by Losasso and Hoppe at Microsoft Research in 2004, the method places a number of concentric, progressively coarser grids around the viewer. As the camera moves, height data is scrolled into the rings' edges like a torus (toroidal addressing), so only a thin edge needs updating per frame. The transitions between rings are morphed so no cracks or sudden pops appear.
Unlike chunked LOD, which splits terrain into fixed blocks with their own detail levels, geoclipmaps are centred on the camera and need no precomputed tree. They suit heightmap-based terrain particularly well.