As far as I know, ZIP does not enforce any encoding of filenames, it is just saved as it comes from the operating system as raw bytes.
So how well non-asciis get carried over depends on where you zip and where you unzip, but as one does not use non-ascii in filenames, this in not a problem at all 
If you know the encoding used by the “zipper”, then you should be able to use iconv like tools to transfer in any encoding you like.
Also remember, that you get double trouble if the encoding in the database differs from the encoding of data you write to it.
