@KevinCain You are welcome All mapping applications use square tiling for their mosaic manipulations, that is because this is the best way to define areas and use/delete needed tiles in the area. As we define area of interest by giving Top, Left, Bottom & Right coordinates, any offsetting in these coords in order to define rotation will lead to stretching and resampling of the original data output from server. Because the nature of satellite imagery and also heightmap capturing systems are based on square tiles. In addition giving rotation features in Area Selection for your purpose is an overdue of the whole system and something very customized which is not needed so much to be considered as a general purpose implementation. The reason that you won't find any references regarding your question is that the final performance on your system is based on many many factors. Talking about terrains, just to name a few of these factors are: terrain tiles count - each tile will have its own drawcall(s) heightmap resolution pixel error quality (LOD & Tessellation factor of terrain surface) - edgier surfaces are harder to render on GPU texture count painted on terrain(s) textures resolutions textures compression and importer settings shader complexity and features (calculations accuracy, PBR, DX11 features, procedural texturing, POM) does terrain shader have different map types (e.g. normalmap) in it? rendering parameters to define dynamic detail and resolution in different distances do terrains have tree or grass in them? tree and grass density and their animations if available lightmapping & light probes reflection probes shadow casting - if enabled, engine's defined shadow distance & resolution on terrains And guess what! These are only some of the upfront factors on top of my head only for terrains regardless of the other assets siting in your scene or current machine configuration the game is running. Maybe I'll add more items in the list if I get back to this post again And please note that, before rendering heightmap-based terrains, the CPU needs to loop through the 2D array of data in its heightmap to extract heights from it, calculate LOD & Tessellation based on camera distance all in realtime and finally pass it to the GPU to render. Also CPU needs to do realtime physics calculations on millions of generated polygons on terrain surface. So terrains are mostly CPU intensive than GPU at first, and then of course if the terrain shader is getting more complex, the GPU rendering will also be close to its bottleneck. So the word "large" in your question is too general and can't get its exact answer. If this question was asked in stackoverflow it will definitely get a "too-broad" tag As I cannot give any exact directions to the question, the best option is to do error and trials for such purposes and do your own benchmarks in different terrain settings, so that you may find some pre-calculations for your target platform.