I dont claim this to be a "good" tutorial, and its been pieced on an hour or so, but to whoever it can be interested the basic workflow to GI+FG comes down to this:
_____________________________________________
QD Mentalray GI+FG Cornell box workflow example
1 make a scene like that: a box with a hole on top and some balls inside
put a spotligh pointing to the hole, on its atributes turn on emit photons
on the camera used to render put its environment color to white
2 assign lamberts with various colors to the box and balls go down to the mentalray tab and asign a 50%gray to irradiance of the lamberts (and put a blinn without specular for the chromeballs)
3 render with mr prewiev settings...
4 just changed the settings to globalillumination preview
5-6 in the light´s atribute editor slowly iterate and finetune trough the values at
a)photon intensity
b)#gi photons
c)intensity
d)exponent
(this is a good moment to see what changing kust one of them actually does... ) keep going until the brightest and darkest areas look more or less right(not too overexposed-underexposed) and the spots(photons) cover more or less uniformly the scene... (not seen when reduced quality to fit on thread)
7 emit more photons from the light pump up photon volume accuracy
8 Turn on final gather
9 pump up render settings and put even more photons, iterate steps 5-6-7 and play with the irradiances and ambients of the lamberts until youre done...(ie you either like it or youre so frustrated you declare it "done")

another variation is putting "windows" to the box or deleting the wall behind the camera
And numerically pumping up beyond 1 the v value of the v (HSVmode) in the white of the environment on the camera. Or placing an image(hdr or whatever) on a big scaled normal inverted ball surrounding the box
_______________________________________________