With each new generation of games, virtual worlds and animations the bar is raised in terms of visual “realism” or so says the popular consensus. But is it?
That depends on what you call “realism”. The current definition seems to be stuck in the view that “realism” is how a scene looks in a snap shot, like a photograph. What this fails to take into account is the most important aspect of realism, movement.
Our focus here is on character movement, and while great strides have been made in terms of physics, lighting, modelling and textures character behavior has so far lagged far behind. Why? I think it is primarily because it is the single most difficult challenge to approach and for that reason has only seen recent incremental changes.
Of course developers and artists love challenges and in the industries we deal with they are amongst the smartest and most talented you will find (looking at some recent games and a number of animated movies can attest) but the time and staggering costs involved in producing convincing character movement puts it beyond the reach of most studios. Much of that content creation could be automated leaving artists and developers to concentrate on the “special touches”.
What we see today is in many ways little changed from the days of side scrolling games, a small number of pre built animations with a small number of moving parts per animation that are triggered by a custom built system that can only handle a small number of states. What that results in is repetitive, stiff, wooden character movement with little or no emotional expression, little or no automatic interaction with the world and most importantly, disappointed and frustrated users.
Does any user really care if your highly detailed character has beads of sweat on its brow, if its face does not move?
Does any user really care if you have 500 characters with highly varied stylised looks if they all behave in the same way with the same animations?
For us, the most important aspects of a character are the emotional and social gestures and movements. These are the elements that we believe separate animate from inanimate, believable from pixels. Crossing the “Uncanny Valley” is not about looking more realistic, it is about moving more realistically. So what are the elements of movement that are important?
- Attention – That is focusing on objects in the world that are interesting at the current point in time. This transmits “intent” and “thinking” (event if it is being driven in other ways it is what the user perceives when watching).
- Orientation – How the character turns to look at something, involving the whole body and face and not just rotating eyeballs (unless perhaps your character is a lizard). Having eyes focused correctly and not staring “zombie like” into infinity. Having the head rotate like a primate not like a robot or a barbie doll and having the torso move with it.
- Reaction – How the character reacts to events in the environment and thoughts in their heads, in real time and in ways that are complex and unique. In any reasonable world the possible number of events and combinations of events can be almost limitless however so being able to manage how to react to events is critical.
- Complexity – We are very attuned to pattern matching which means we spot repeating patterns, or animations, immediately, and start gradually ignoring them. Movement needs to be complex enough that patterns cannot be seen, repetition is either not there or not obvious.
- Consistency – Movement should be consistent with the current context or environment, this implies, for any reasonably large environment, a very large “palette” of behaviors and movements to drive the character, not “shoe horning” the same 8 animations into any scenario.
- Uniqueness – Characters need to be able to behave, move and emotionally express themselves in a way that is consistent with their personality and their mood and any given time in order for users to create an emotional bond with characters and believe in their presence as an animate object
Right now you might be thinking “What the hell, we have a 10,000 characters in our world, if each of them was able to move convincingly to every possible event in a way that matched their personality we would have to spend $100 million and the next five years building content!”. Well, using traditional methods, yes you would. Thats why you would need help. In this case we would like to introduce you to the brave new world of emotional and social animation synthesis.
Looking at the bigger picture, content heavy industries, like games, virtual worlds and animation are being squeezed from both ends, content creation costs are going through the roof while sale prices are only going down. This means lower and lower profits. Obviously this cannot go on for much longer and ways to automate the production of content (in all its forms) are the key. This is the brave new world of procedural content and I will discuss that in a later post.
September 23, 2009. We exhibited our technology today at the Cambridge Enterprise Conference amongst some of Cambridge’s hottest new technology start ups. We also pitched the gathered audience of business leaders with our vision of powering a billion digital characters with Emotion AI products.
