I had what seems to be a good idea, so I''d like to solicit some opinions on the matter. What I have are two ideas related to the issue of uploading large meshes to clients. The kind of setting I have in mind is a game where most meshes are variants of some other mesh, and where players and mod designers can specify custom meshes. The first notion is that when a client downloads (or uploads) a mesh, it receives the differences between that mesh and a mesh it already has. For example, a mesh of a man in armour is probably not that different from a generic humanoid mesh. Suppose that 20% of the armoured man mesh is new or different from the generic humanoid mesh, your mesh download is now reduced to 20% of the time. The second notion is that many mesh formats incorporate LOD, or different LOD levels can be automatically generated for them. The idea here is to download the lowest LOD levels first, before play, and download the higher LOD levels during play, if there is enough bandwidth spare. I figure a combination of the two would greatly increase loading times for the kinds of games I described. I''m not sure if something like this could work with scenery. I suppose it depends upon how the scenery is stored. A lot of maps in games are obviously collections of prefabs with slight modifications made to each piece to make it all fit properly. If it didn''t break whatever partitioning scheme was in use (or if the scheme was fast enough that it could be done on the fly by the client), then you could send the prefabs and their modifications through, which would probably be faster than sending the completed map. CoV