The space estimation currently just asks the server about the size of the individual tiles and and them counts a sum of all the individual sizes. That's would be quite precise in an ideal world, but as already mentioned, there are unfortunately filesystem clusters in the real world So, what can be done:I can add the cluster size to the size estimation feature, so it is closer to the actual result pack the tiles in some transparent single-file filesystem store the tiles in a sqlite database ? BTW, every non-standard method will break the current tile sharing between the various tiles based map aps that use .maps IMO, a transparent single-file-fs proxy linked to the .maps folder would be probably the best.