[MUD-Dev2] [Design] Avoiding the uncanny valley: cartoons and animals

[Mike Rozak]

I've been creating racial models for my virtual world, at http://www.mxac.com.au/mif . The game only displays heads. For a long time, I have planned to only have static images of the characters to avoid the manequin-like uncanny-valley effect. (And to lessen the amount of work for authors.)

EQII and Oblivion are prime examples of the uncanny valley. For all the millions spent, characters look like schizophrenic manequins, especially when lip synced; the mouth says one thing, but the eyes' expression is blank or contradictory.

However, for the fun of it, I've tried some automatic lip-syncs of my animal races, along with some really cruddy hand-done animations to blink and move the head/ears. You can find them at:

http://www.mxac.com.au/ThylacineAnim.wmv
http://www.mxac.com.au/RatAnim.wmv (also using text-to-speech)

A year ago I did the same thing with a human head, and it was just as scary as anything produced by oblivion/EQII. I don't have a movie around to upload, by I can generate one if anyone wants. (Or just look at Oblivion/EQII.)

The results with the animal heads are so much LESS scary than the human animation that I'm seriously thinking of including some lip-sync in my game, since the uncanny valley appears to be shallower for animals.


What does this mean in general?

For the next 10-20 years, I suspect many VWs will either use humanoid animals and/or cartoons (WoW) in order to avoid the uncanny valley.