The format of the document you retrieve is the format that it went into CMOD as. So, TIFFs come out as TIFFs, Line data as line data, etc.
If you're using the -c option, the documents are concatenated together, resulting in one (or many) large files that contain the documents 'stacked' one after the other in the file. You can use the 'offset' and 'length' values from inside the generic index file to get the 'address' of a single document.
Be cautious about leaving off the -c option though -- depending on your query, it could produce THOUSANDS of files.
-JD.