Satellite images of the earth are currently commercially available in resolutions from 14.25m (Landsat 7) up to 0.61m (QuickBird). Visualizing large terrains or whole planets at these resolutions requires highly efficient level-of-detail techniques to reduce the amount of information to the capabilities of the display device. The same applies to simulations that drive a visualization pipeline: Although the data could be computed at very high resolutions, it is often not desirable to generate more information than the visualization part can display. We describe a conceptually simple terrain rendering pipeline that handles on-thefly data decompression and synthesis in a unified process. Possible data sources range from static satellite imagery over per-texel-processing such as image blending routines to light-weight simulations and synthesizers such as noise and filter based texture generators. These sources are evaluated in multiple threads and the generated data is fed to a geomet...