People have an amazing ability to recognize other people. You will be surprised if you try using a 3D scanner to recreate a head -- the result will be different from what you expect. If you try to capture the entirety of a head with your naked eyes, even though our eyes are only three inches apart, it's difficult to do accurately.
Figure 8: A nose that appears straight actually has a lot of complexity.
So with a scanned head, you can be sure there's going to be some mass and volume you didn't expect, which will result in an unexpected silhouette when you move your in-game camera. A straight nose when viewed from the side may be slightly hooked (see Figure 8).
A cheek may actually be puffier when you see it from certain angles. Great sculpture changes its silhouette elegantly as you physically move around it, and every moment is a surprise. The concept applies to CG and you should be prepared for it. The illusion should betray your sense of volume every time the angle changes.
A good character should have a good signature pose. It's easy to recognize Spider-Man by his gymnastic flying silhouette, for instance. But again, that is visual language, which belongs in the realm of concept design.
What I mean here is the posing that defines the character's personality. Good film actors change their body mannerisms for the roles they play. The neck angle and spine curve are typical targets. Simply pulling the spine in and out for default poses gives the character a really different personality.
You also rarely see a profile shot in movies, because actors rarely turn on the axis that's perpendicular to the camera. Good animation doesn't do this either, though it still happens if the director isn't good.
See the image above for an example. The story behind this image, created for a serious game, is that this girl has HIV, and she prostitutes knowingly. A guy (we do not know who is -- this could be a customer, could be police) is approaching her. I needed complex emotion on her face to indicate guilt and ignorance, but since I did not have time to add real facial expression, I solved it with simple posing. The girl successfully conveys some of the complex emotion I was looking for.
It's a fun process to decipher how your favorite actor creates his role through poses and movement, setting aside typical acting aspects like voice tone or eye movement.
Human beings live from moment to moment -- we are not really conscious of what we do between describable actions. For example, when you wash your hands, your focus is on your hands. Next, you want to wipe your hands, so your focus jumps to the towel.
People are not exactly aware of what they're doing while moving between these focus objects. In these in-between moments you can redirect focus to create good acting. The same idea applies to the face. A smile, a frown, anger, or even lip syncing, all are examples of visual language.
Between these states, that's where you can create real character. If you step-pause a movie at home, you'll find actors making funny poses and faces between the conscious points. That's a good subject to study. Boring acting like simple smiling or frowning may turn into some attractive expression by injecting this type of essence.
There is no guarantee that you can find a perfect cast. Even if you do, there are often many legal reasons that we cannot use the particular actors we want for our animation. That's when you have to design a human, and the result needs to be more attractive than a regular human. It must be as attractive as an actor.
I think this is one of the most difficult things to accomplish. It's easier to take a specific human target and try to model it, but if you can't use the actors, you can't use them! Figure 9 shows an attempt at using a real actor (Gary Oldman) to evoke a certain mood.
Figure 9: An early test for the GoldenEye: Rogue Agent game (recreated for this article) using Gary Oldman to create a more complex character.
The typical way to design a human is to pick stamps of features from many people and combine them like a police composite drawing. It usually takes a hell of an effort to blend them so that all parts and skull shapes work in harmony, and even then there is no guarantee that your results will look strong.
You might need to go back and start choosing feature stamps again if the result isn't strong enough. Designing a human is a repeating sequence of this trial and error process. When I do this, I need to be ready to get completely exhausted, and still there is no guarantee of getting a good result.
Sometimes character descriptions are difficult to visualize. For instance, consider this one: "35 years old, but still looks somewhere in middle her 20s, a mix of Arabian and North African. Her eyes are solemn yet very determined. Although she is not a classic beauty her beguiling serenity makes everybody mesmerized."
It is cruel if this complex requirement only falls on a character artist. As CG and game fidelity goes up, more actual references are required. I think it's worth having auditions or working with domestic and oversea casting agencies to help artists when modeling.
Various elements need to be considered before entering full production. As the technology improves and budgets go up, It's becoming very challenging to tie each component into a single direction. But great impact cannot be achieved without all elements working in harmony. Masterpieces are borne when everything works in sync.
[All images created by Takayoshi Sato with the exception of Figures 2 and 7.]