Question about the paper

[yogwoggf]

Hi, I've been reading your paper for an interactive fluid simulation mod I've been implementing.

Questions

For this equation, is there a specific method to sample pixels j? In some implementations of this paper, some people sample 4 corners of each fragment in a grid centered at the origin pixel (i).

I'm not exactly sure how this equation works. Is there a specific reason why it relies on the world radius, vertical resolution, depth and camera fov angle? Also, what are the bounds of this equation? Should there be a maximum limit?

Is this filter intended to be used with a nonlinear Z-buffer?