Prosopology
The Scientific Study of the Human Face
“What would I do if I had no more external obligations from day to day, such as attending a job or writing a thesis?”
Thinking of this in the larger context of self-actualisation through building expertise and creating new knowledge, and the redemptive qualities of work… one idea was to teach the computer to read and understand millions and millions of portrait photos, then be able to generate convincing virtual characters. This is one application. What is the generalised field of study?
It’s not a new field, it’s also straightforward to describe: the study of the human face. It’s just not obvious if anyone knows what it’s called. So I decided to call it prosopology, from Ancient Greek πρόσωπον (prósōpon, “face”) + -λογία (-logía, “study”), IPA: /ˌpɹɒsəˈpɒlədʒi/, rhymes with “anthropology”.
When we go and start studying biology, there are textbooks, there are lectures, there are animation videos narrated by Hank Green. This setting is very different to what was faced by early naturalists, which was eclectic and largely unstructured.
There are at least two challenges entering a field without its own curated academic tradition:
- How to construct a curriculum from a vast and diverse range of sources, from textbooks on anatomy, to journal articles published by plastic surgeons or computer graphics researchers
- How to find specialised areas, powerful techniques, worthwhile research questions, and develop fruitful applications
It’s also quite exciting, think about LaValle, who did go on to write a whole textbook on virtual reality, and might even start a new university track or program called “perception engineering”.
The medical discipline is one example where there’s a term for almost every specialisation, cardiology, ophthalmology, etc. Even then, it’s not clear there’s a specialisation dedicated to the exterior features of the head (save for, again, a subset of plastic surgery). On a funny note, if you search for studies related to the cranium, you soon land in the museum of wacky pseudo-science, from phrenology to something much more amusing: physiognomy — misguided theoretical framework, but exquisite in artistic detail.
Perhaps many interesting questions and useful application also relate to perception, for example, why are caricatures easier to recognise even though they are distorted representations (with idiosyncratic features exaggerated)? What causes the “uncanny valley” effect, and how to overcome it?
Then there’s a whole area of commercial applications related to facial detection, recognition, authentication, etc. — though these techniques may rely more on statistical methods (e.g. abstract “landmarks”) detached from biology and inter-connection and structure.
To start, the study draws on a number of more established fields:
- Anatomy, human biology, physiology
- Emotion studies, facial expressions
- Aesthetics, anthropometry, perception
- Social cognition, neural basis of facial recognition
- Animation, computer graphics, special effects
- Portraiture, life drawing, medical illustrations
- Deep learning, computer vision
It appears many researchers doing work in this area mostly publish and socialise through platforms (e.g. conferences) for computer graphics…
[To Be Updated]
Who know about relevant topics?
Portart artists, anatomists, forensic anthropologists, computer graphics researchers, plastic surgeons, 3D animators…
Some starting topics:
Face dynamics, facial construction, emotion library, Facial Action Coding System, the bizarre world of “face synthesis” in ISO14496, facial rigging in animation, volumetric capturing, photometric capturing, photo-realistic avatars, emergent (unscripted) facial expressions, likeness and remix…
