Quick and dirty method would be to set up 2 cameras one for each "eye" and then render to separate bitmaps and then combine in the 3d image creator/compositor of your choice ? ( You may have already done this, hence the request
) Or are you thinking it could be a preset in TG2 ? Then you get into what format of 3d do you want supported...etc...
Whichever way you do it...it's going to be a double render time for each stereoscopic frame/image, I would imagine.
Nice idea though...