When confronted with facial dysmorphologies, three-dimensional (3D) facial harmony is preferably assessed with regard to normality. This, however, presents two major challenges. The first challenge is to define normality. What makes non-dysmorphic faces appear normal? The second challenge is to situate the craniofacial dysmorphology with respect to normality. What makes the dysmorphic face not appear to be normal? To tackle these challenges, a novel approach based on a Face Space is proposed. In essence, faces are represented as points in a High-dimensional space, the dimensions of which capture important sources of allowed facial variation within a normal population. To establish an aesthetic objective of a given dysmorphic face, the novel concept of the "normal equivalent" of that face via robust projection into the Face Space is proposed. This technique is demonstrated on an artificial example in which smiling whilst showing teeth is considered a surrogate for facial dysmorphology.