Doing DoF and MB in post on CG is always a ‘hack’ atm… … can’t use motionvector pass after things have been defocussed and even DoF is missing data on what’s ‘behind’ the pixels it’s moving. Defocussed objects really close to the camera become almost transparant, especially near the edges. So you’d need to know what’s behind. Years ago I worked with a TD to have mental ray (yeah, that long ago) render a ‘behind-pass’ which we could use to do proper DoF… but then he got snatched by Weta and that was that… hah…
I’m with GPM… wing it until it looks good… but then in linear… 