When using recap to process photos into a point cloud, it does not create the model with an accurate scale or coordinate system. If there is metadata with the jpeg, it seems to get the scale close, but ordinary pictures do not have altitude info with them. Recap doesn't appear to have any way of changing this either. If 2 known points were in the images, we should be able to create a distance, and then scale the point cloud to the known distance between the two points. I think you should also be able to change the xyz axis orientation. You could select 3 points as a plane, and set the z to the normal of the plane, and then select one point for the origin, and then one more point to rotate to. All of these points should be adjustable numerically as well.