In general, experts in the field of CFS tend to apply their own approach to the problem, depending on the technology available in their laboratory. For example, when scaling images of the face and skull, some researchers focus on specific sets of homologous landmarks, others rely more on a fit of skull and facial contours or seek morphological consistency.

CFS has evolved according to the technology available to professionals, based on previously established foundations [1, 2]. The variety of supporting technological advancements has involved a large number of highly diverse approaches, i.e., photographic superimposition, video superimposition, computer-aided photographic superimposition, computer-aided video superimposition, computer-aided 3D-2D superimposition, manual, semi-automatic and automatic.

In addition, there are different anatomical criteria that are used to evaluate overlap: contours, lines, proportions, landmarks and soft tissue studies, morphology, asymmetries, positional relationships, etc. The differences are not only found in the set of criteria used, but also in the weight assigned to each one of them, while the skull-face overlap is evaluated. Finally, each expert has their own decision scale, with different names and meanings, and different criteria for moving along the scale.

It can be considered that the CFS technique is currently in a new stage of standardization, which would begin in 2015, as a result of the European project MEPROCS (New Methodologies and Protocols of Forensic Identification by Craniofacial Superimposition), with the first publication by Damas et al. [3]. Surveys and two validation studies were included in the development of this project, which allowed witnessing a serious lack of consensus in the protocols, methodologies and materials used to carry out the overlays and their evaluation.

These studies made it possible to diagnose the main sources of error and uncertainty in CFS and to propose a list of 17 recommendations and 4 practices to avoid in order to eliminate or minimize possible sources of error as much as possible.