I’m trying to understand this as well.
There are at least two roles for these points: manually overriding the alignment process, and imposing scale/a coordinate system.
I think tie points and GCPs can both link components that the alignment algorithms haven’t connected. (My experience is that this is a powerful tool, but there will usually be a step in the mesh between the two components.) So I suppose both behave the same in this respect?
For completeness - will ground test points do the same?
Tie points can be used as end points of scale constraints. GCPs are given coordinates. A question here - if I do both, are both taken into account? Or do GCPs override scale constraints?
One thing I’ve been wondering about though is whether GCPs can influence alignment other than by linking compoents. If alignment is (re)run after GCPs are defined, might model geometry be changed by the GCPs? Or will they simply be used to set the coordinate system from a “best fit” between the points defined on the model and the coordinates specified? I’m thinking about a long, narrow model where accumulated errors might result in a false curvature, would GCPs suppress this? Or would they simply highlight the error, by showing the varying fit along the length?
Thanks.