I am rendering content over photographs screen mapped to geometry with the ShadowCatcherMtl and they all have DOF which I'm trying to match in the rendering. Unfortunately, Corona's DOF will add defocus blurring on the shadow catcher geometry and it's too extreme. I can composite the rendered content over the original backplate in post, but in places with more extreme DOF it will not blend correctly as I still have doubled DOF in places where the alpha is not completely black or white...
I am trying to get a solution for this but it seems the only working way would be to exclude that screen mapped geometry from DOF calculations. Would something like this be possible at all?