This paper introduces a novel Depth-from-Focus (DFF) framework that leverages the focal stack’s unique gradient patterns induced by focus variation, jointly modeling focus changes over spatial and focal dimensions.