Blue/red 3D works because the objects closer to the viewer have a greater blue/red shift then objects farther away. (...or the otherway around...not sure) So to have any effect approximating reality you would have to selectively apply these color shifts to the images. You would need some process which could figure out where images in frame are in relation to the camera, then apply the color shift accordingly. I think the 3d processes used in the past either used two cameras to capture or somehow shifted one color by a special lense.

http://science.howstuffworks.com/3-d-glasses.htm
http://www.3dglassesonline.com/how-do-3d-glasses-work/

Manually shifting the blue or red spectrum a 3-7 pixels might work for some of the space sequences where objects fly by camera quickly but overall it won't be 3D-ish.
none