What is the cue for stereopsis, and how is it generated by points in space at different depths? be certain to define crossed and uncrossed binocular disparities for objects along the optical axis. why is it a relative measure of depth? the work of julesz with random dot stereograms shows that (1) form need not be present in the monocular stimulus for stereopsis to take place and (2) stereopsis is impossible on the basis of point-by-point comparisons. how can both of these statements be true? what is the “matching noise problem” encountered in deriving depth information from random dot stereograms? how do cooperative algorithms potentially solve this problem?