I would say that the most important thing is to really work on tge schema, trying to keep a very broad view of things you may want to support. I'm sure this will draw heavily on musicbrainz (you have artists, works, etc).
A big part of that would be defining scope. "Films and TV" is fine; but would that include every moving image posted on youtube/facebook/...? If not, what notability/length/... requirements would there be? Would adult content be allowed, or should that be set aside for a potential future PornBrainz? If disallowed, that also needs to have a clear guideline to determine what is and isn't adult. And so on.
It would certainly be good to have people from multiple cultural backgrounds looking at those things, to avoid an excessively western viewpoint.
Naming is always hard, and will depend on the scope choices made; but I do like VideoBrainz.
Once the schema is there, the rest can flow from there, given enough motivated developers (who can agree on an implementation language; MB is perl/js/java, BB is python iirc (a language i hate with a passion); so I guess VB should be in asp/vb/vbscript ).
The web service could be dual xml/json similar to what mb has, or just json. I would personally like it if attribute/type names were chosen such that serialization is as trivial as possible in as many languages as possible (i.e. no hyphens, no names that might also be keywords, etc).