Help with GCP tie-ins after initial survey

[SuperNewb]

Hello again everyone,

I am working on a survey where some of the GCPs placed on-site had moved just before photos were taken...  I have isolated these erroneous markers and processed the data to not include these GCPs.

However, the resulting DSM is flawed in the areas missing the GCPs as expected and I have been trying to correct this without reflying the site. 

A surveyor has collected a series of points on-site now in areas where no changes to topology have occurred. I want to use these to drive the elevations of the incorrect regions.  My thinking was that I could simply drop these phantom GCP points into my data, go along and tag them to their corresponding photos where photoscan estimates that they should be, and optimize the points. 

It looked as though it was working well, too well in fact, and it is now clear that it did not work at all.  After optimization, Photoscan tells me that there is error in the new points between .2 and 3cm...  but after building a new dense cloud, these points are still stranded in the air and not actually helping to drive the new elevations....  The Resulting DSM is exactly the same as the DSM without these points added.

Any ideas from the gurus out there?

Thanks