As a byproduct of the Local Group dwarf galaxy search I have a collection of a few hundred very faint objects. They are distributed all over the sky and include wisps of dust, planetary nebulae and galaxies all jumbled together in one list. The only things they have in common are a (sometimes tenuous) resemblence to dwarf galaxies in the Local Group and a very faint surface brightness.
But in our analysis of the Dwarf Hunt data it became clear that this could be an extremely useful list. In surface brightness it goes two magnitudes, maybe more, below any other catalogue of galaxies. (In fact since it stands alone out there in parameter space it's very difficult to tell just how deep it goes.) This is territory that severely tests theories of galaxy formation and evolution, as well as possibly giving new insights into star formation in small and tenuous systems.
(There are fainter galaxies, a few of them. They've been found by picking out their stars, one by one, in large surveys (like the SDSS). Our catalog is still unique, lying between these extreme objects and the rest of the population in terms of total brightness as well as surface brightness.)
The observational challenges of using the list, however, are pretty daunting. The brightest parts of the brightest objects are outshone by the night sky itself by a factor of two or three; the faintest ones come in at a few percent. Even measuring an accurate brightness under these conditions is difficult, to say nothing of extracting a spectrum. It is a great help, however, that the telescopes I'll need are not necessarily the big (expensive and hard-to-get) ones.
The first chore is to get a good, well-defined sample of faint galaxies. Many of the objects on my list could be galaxies, but might not be; sometimes it's very hard to tell just what a faint wisp of light actually is. So I'll have to go back and get a deeper look at most of them, along the way doing a better job of measuring brightness so I can say just how deep I'm going. Then I can turn to other observations and analysis.
At the moment (October 2009) I lack the resources to attack this project, so it's on hold. It's possible that I'll be able to use LSST data for it, eventually. . .