application to 3d reconstruction

[fariba87]

thank you for your great work.
I want to apply your model, for refining the depthmaps predicted from a deepmvs model. i have change the way source view and reference view depthmaps should be called based on pair.txt and intrinsic and extrinsic computation based on cam.txt files. but the ourtput depthmaps become worse. can your model be used to refine depthmaps to create more accurate 3d reconstruction?