In general, I’m more interested in mapping the objects in and interaction with a place than the places themselves.  For my project, I intend to use multiple photographs featuring a common, recognizable object to reconstruct a composite 3d environment centered around that object.  I’ll probably use reconstruction tools like VisualSFM and a lot of manual tweaking.

I’m still undecided as to what the object will actually be.  There are a couple of choices:

1. A particular form of street sign like a stop sign or 25 mph speed limit sign, which has a very recognizable shape and high contrast with the surrounding environment.  This also has the advantage that there’s no shortage of images featuring street signs and there’s possibly an interesting story to tell through such a composition.
2. A common movie prop and the movies that feature them.  An example is the bag of Let’s Potato Chips that appears in almost everything.

http://www.buzzfeed.com/jordanzakarin/lets-fake-potato-chip-brand-tv-shows

3. A recognizable landmark or place that people tend to take close up shots of like the Eiffel Tower.