I would say they are vital. Unless you are able to produce rtk data and process that straight into your data.
My workflow is normally
1 take geo located images via inspire 1 pro
2 readjust the - sea level to + sea level in exif data
3 take into Photoscan Pro
4 align images
5 add gps control points
6 build point cloud
7 build dense cloud
8 build mesh
obvisouly tweaking the options to your fav. settings.
In the meantime, I would go round the site using common sense to take reference ground control points, using fixed furniture and features. Using something like the Geo 6000 with CM accuracy and Terracync and external antenna at 2m, locate your point - add feature, collect data for a good minute or so per point to get a good 60-100 data points (remember the 600 is 1hz) - I also take an image of the point as well which is geo referenced. Its important to note i am not using a realtime RTK data feed here, just the kit which can use rtk real time. In normal mode, this will get anything from 20-60cm accuracy.
I then get the data back to the office, and post process it with Rinex data for the time period and Pathfinder office to create my CSV import file. This will then give accuracy of normally 1-3cm. It is this data I then use in processing.
I cannot see how you can advise accurate analysis for anything geospatial without this process.
I went for post processing as it saved me the £2k per year for data feed but if you don't have financial restraints and good mobile service, then you can get 1-3cm accuracy in the field using realtime.
Hope that helps!
