Robert Rohde wrote:
That wouldn't get you file descriptions or copyright status, etc. If your goal is something like mirroring a wiki, you really need access to page descriptions as well.
At present, the main solution is to copy all of Commons, which is overkill for many applications. It would be nice if the dump generator had a way of parsing out only the relevant Commons content.
-Robert Rohde
I'd expect a "commons selected dump" to be pretty similar to pages-articles. What you can do is to request just the images used with Special:Export or the API (depending of how small those wiki really are, it could be feasible or not).