If you can't use any other engine, best is to compose it (take center area and make it smaller; iterate).
Note, when rendering,
ray depth depends on the size of the image - the larger the image, more depth is needed (as goes for Texel density.See attached*)
Finally, IMHO, it'd be of great use for advanced users, if such options as ray depths, volume priorities, photons (+domain), ... were simply exposed.