Your content is safe.
As I understand it, git doesn’t actually have any concept of a first-class rename operation (only bzr does, of the big 3 DVCSs): the mv is sugar on top of the underlying machinery, which is basically an add and a rm. Since git can track the content that changes during such operations, though, it can use heuristics to guess when an add and a rm are actually a mv. Since that takes way more work than just displaying what git actually recorded—the docs for git-diffexplain that it “…require O(n^2) processing time where n is the number of potential rename/copy targets”—git won’t try it when too many files are involved. The setting you mention just controls that threshold.