Hi Max,
The new deringing algorithms are (in theory) independent of scale, so there should be no difference whether you apply Deconvolution to the whole image or to a partial preview.
Could you put an example, i.e. upload an image where this is happening?