I'd be interested in camera locations, but for another reason:
It would be really really nice to be able to get the point cloud in its current format (x,y,z,r,g,b) but also get camera location and orientation inside that same framework with a tag for the image name associated with that location (imagename,x,y,z,rho,theta,phi).
From the standpoint of doing the kinds of things sir_ivar is doing with turning point clouds into GIS products, this would offer a lot of utility. A fairly dense set of photos with a sparse set of ground control points with known coordinates would essentially tie all the images into a fixed frame of reference. At that point all sorts of possibilities open up.