• It is well known that an image undergoing uniform translational motion occupies a tilted plane in the spatiotemporal-frequency domain (compressing the the 3-D image sequence into 2-D). The case of uniformly translational motion is a special one. However, if an image sequence is examined over a suitably local region in space and time (as made possible when st/stf representations are used), locally translational motion becomes a reasonable approximation.

• The use of st/stf representations provides a more convenient framework than motion compensation for the incorporation of models of spatiotemporal perception.

• By explicitly enforcing localization in the temporal dimension, st/stf representations allow improved exploitation of temporal masking.