Forensic facial reconstruction or facial approximation?
Our body changes from one day to the other, from fat to thin, from young to old, without any significant facial bone surface modification. The same from upright to horizontal position, with gravitational deformations of sub-cutaneous fatty and muscular tissues . Some databases using for such female reconstructions are taken from post-mortem CT-scan or ultrasound data, not evidently applicable to living individuals due to taphonomical ionic and water movements. Some structures such as the ears are, to date, absolutely impossible to reconstruct for technical reasons and a total lack of serious anthropological landmarks. Position of hairs is absolutely impossible from purely bony remaining surfaces. So, when using computer methods and anthropological databases, do we reconstruct a mean face, a special, a plausible or any face? In fact, we reconstruct only the closer to the mean, i.e. the more probable, some general traits that could be recognized by familiars (meaning: subjective visual assessment of resemblance). A proposition of reality, not reality itself, as reality changes from time to time.
 Charlier P, Froesch P, Huynh-Charlier I, Fort A, Hurel A, Jullien F. Use of 3D surface scanning to match facial shapes against altered exhumed remains in a context of forensic individual identification. Forensic Sci Med Pathol 2014;10(4):654-661.