Orthophoto reconstruction with accurate pose (from INS) and lidar point cloud but limited side lap

Does anyone know a workflow using ContextCapture to generate an orthophoto reconstruction by skipping the AT, using accurate pose from an INS and a lidar derived mesh/DSM rather than an internally generated DSM from matched points? 

Basically, I have a lidar/imagery dataset in which the side lap of the images is pretty low (25%) but where the images have accurate position and rotation from an AP50 INS, and a dense lidar point cloud from which I can derived a quality DSM for the final rectification.  The standard workflow requires me to solve the AT, which results in a substandard set of tie points, and a poor DSM the reconstruction.  It would be great if you could skip the AT, and proceed directly to the reconstruction, loading an indepent mesh from my lidar survey.  

Any thoughts gratefully received!