It used to be that this required post processing. You would take the distance from
camera to primary focus and divide by twelve to
get the camera separation. Then, use a two-frame animation to render left and right
eye views and finally piece them together with
DTA, which had a specific red/blue anaglyph switch.
Grim
