This technique uses multiple-channel shadow maps to allow shadow casters and receivers to be masked in screen space coordinates.
Since there are no additional passes, the performance impact is pretty minimal.
I implemented it using Unity Engine's Scriptable Render Pipeline.