ZIP famously doesn't specify any character encoding in its file names (a later version of APPNOTE introduced an additional bit to signal that UTF-8 is in use which is to my knowledge not really taken off). Many archivers therefore assumed that they are in the active code page, which meant you can experience mojibake even in Windows and in fact was a major pain when you deal with ZIP files originated from other East Asian countries. Later archivers generally have an option to set or guess the character encoding---if you still have those files, try them.