Thanks!
We took a hybrid approach using ground based terrestrial laser scanning from ground level and some at roof level, in conjunction with photoscan.
Photos of the facade were taken by a person in a boom lift with a nikon d800, and the bell tower was reached by someone dangling from a crane in a basket - not me!
For the main wings of the elevation we only used the laser scan data for the surface reconstruction.
For the bell tower and portico where the laser scan data was patchy we combined it with the dense cloud from photoscan, literally just copying and pasting one pointcloud into the other.
Surface reconstruction was done in an external application, and imported into photoscan for texturing.
We took 1,343 images in total. About 500 around the bell tower, 250 on the portico, and 300 on each wing.
Photoscan data was registered to the laser scan data using 95 ground control points, total error around 15mm.