I do not fully understand the complexity of SC's scrolling capture option, which is tied to capturing an object as well. it appears one can try to let SC guess how to do the window, and if it fails, one can set various parameters I do not understand.
Would it be more simple, perhaps to have a `manual scrolling window' options? That is: have a screen capture that lets the user take control of the scrollbars, scroll the window himself and then click a button, or some such to let SC capture the results?
Sometimes, adaptive contextual human intelligence can make tasks easier, that otherwise require complex conditional rules.