Our human face recognition skills are pretty impressive. We are adept at it after about 4 months. On average, we can remember about 5,000 and clocking one we know takes about half a second. But how do we make synthetic, digital humans that meet our strict believability requirements?
The “uncanny valley hypothesis” was first put forward by a Japanese robotics professor in 1970. Masahiro Mori’s point was that our emotional response to robots would be more positive as their faces became more human. The trouble is, at the tipping point of believability, he realised that we become concerned that they look too real and they begin to revolt and repulse us.
What is it in the face that makes it most human-like? Two things, mainly; the imperfect texture of our skin and the corneal reflections in our eyes. In terms of digital humans, the absence of these two key factors is particularly apparent with current de-ageing techniques, where the skin invariably looks smooth and waxy and the eyes look dead and glassy.
If you would like to know more about how Mondatum can help your digital humans escape from the uncanny valley, get in touch – contact@mondatum.com.
Source: Psychology Today