When Kamembert looks for duplicates it works in two passes, and the second one is the one that keeps your files safe.
The two passes
- Quick pass. Group files by name and size. This is fast and catches the obvious copies without reading file contents.
- Verify pass. Before anything is deleted, the copies are read byte for byte to confirm they really are identical.
Matching by name and size alone is never enough to remove a file: two different renders can share a size, and two builds can share a name. So the extra copy only goes to the Recycle Bin once both copies have been read in full and matched.
It never freezes
Large folders are verified in the background, so the window keeps responding while the check runs. The same care applies to whole duplicate folders, which are detected as a unit and not counted twice.
There is more on this in the Duplicates docs.