Questions about reconstruction process in ContextCapture

Hello, I am wondering if it is possible to know about the reconstruction process used in ContextCapture program.

With my background knowledge, I guess template matching is used for dense reconstruction after the bundle adjustment.

I am careful to ask  as the questions may be treated as a secret but what I want to figure out are mainly 2 :

- Template matching algorithm or template size used in reconstruction process

- How does reconstruction process deals with a pixel which is shown in multiple scenes? Is its location decided where having least squared error like bundle adjustment or some other techniques are used?

Thanks!