> If we use a blank robots.txt file, does this override the metatag instructions on pages that we don't want indexed (noindex,follow)?
No, it won't override those tags, since the robot will be allowed to fetch the pages and read them.
Beyond that, it gets a bit complicated...
If you exclude a URL-path using a Disallow in robots.txt, then the search engines won't fetch pages in those paths, and therefore won't see the robots meta-tags on those pages. Since images don't include html meta-tags, this is a separate issue for them and for other non-html filetypes.
However, Google, Ask Jeeves, and quite recently, Yahoo will list a page in results even if they have been disallowed from fetching that page by robots.txt. They do this when they find a link to any page from any other page they have indexed. The unfetched page is listed by URL only, without title or description in Google and AJ; Yahoo uses the link text from the link it found as the page title in search results.
This leads to an interesting sort of paradox: In order to tell Google, AJ, and Yahoo not to list a page, you must allow it to be fetched in robots.txt, but include the "noindex" value in the on-page robots meta-tag.
The crux of the matter is one of semantics; We want to tell robots which pages not to list in search results, whereas the Standard for Robots Exclusion specifies that robots.txt tells search engine spiders which pages not to fetch. Furthermore, Google, AJ, and Yahoo have adopted the stance that they want to list pages they find links to, in order to reveal more of "the hidden Web" -- obscure pages for which no effort has been made to rank in search engines.
Therefore, I recommend a mixed robots.txt and on-page robots meta-tag approach for maximum control and flexibility. You may also find situations where it is necessary to use server-side URL rewrites to (technically) cloak some pages in order to keep them from being listed in the search engine results.
Images are typically not fetched by text-based search engine robots. Instead, they are handled by separate robots which feed the image search or shopping-service functions of search portals. You can handle them with separate robots.txt records targeted at those specific robots. So far, a robots.txt disallow seems to keep them from being listed.
pdf, xls, and several other non-html filetypes currently fall in a grey area, since they are "text" but not html. I haven't been able to get mine out of the search listings, but then again, I haven't spent much time trying, either.
I've tried to pick my words carefully, so I hope this is not too confusing.