There's a great (but technical-oriented) article on NaturalMotion's euphoria physics engine over at AI GameDev. You might notice that two of the games using that physics simulator, Backbreaker and GTA: IV, are on my wanted list. I'm anxious to see Star Wars: The Force Unleashed as well, but that one could more easily be a short and shallow game (and Star Wars is easier to spoil than football).
Anyway, the article got me wondering how animations could be made better. I have zero experience in animation or programming, so odds are that my speculation will mean very little (which is why I post this here, and not over on the pros' forum). But sometimes an ignorant outsider can raise the right questions... in the same way young children sometimes amaze educated adults with profound insights. So why not throw some stuff out there on the off-chance that it will help?
Besides, I need a Friday post. ;)
A body as a community of symbiotic member-parts
Perhaps one possibility is to divide up a body into functionally separate, but joined and cooperative parts. Just because a player collectively perceives two arms, two legs, a torso, and a head as one object (a human body) does not mean the game's simulators and animators must focus on them in the same collective way.
As millions of diverse living cells in a body act without consideration of their fellows (self-focused programming) while simultaneously maintaining cooperative/collective value (much like thousands of self-focused, individual persons inadvertently cooperate in a stock market), a dozen or half-dozen parts of an animator's model might operate independently of one another and yet not disrupt the player's focus from the whole model.
There are two downsides that I see. First, it means extra processing, but that is becoming less of an issue as an increasing number of games are shifting resources to physics and A.I (away from the tyrannical graphics). Second, more interactive elements means less predictability for programmers. But this, too, woud become less of a problem over time, as one designer's experimentation becomes the next designer's foundational knowledge.
This is probably not a good example, but let's look at walking. The human body when walking is like a series of connected machines, each one reacting to the movements of another. The ankles allow the feet to pivot forward, back, and (to a lesser extent) sideways. The legs then make up for what balance deficits the feet and ankles can't counter. The back then twists in reaction to the legs and waist, also occasionally bending (vertebrae and ribs stretching or collapsing into each other) to absorb shocks. The arms shift, not just forward and back, but in all directions to act as counter-balances (your arms swing outward when you turn). Finally, the head attempts to remain upright in most circumstances. Each section of the body has its own goal which generally requires attention to only one or two other sections.
Joints: more is better
How often do you see models of humans in games with flat feet? Almost all the time, I'd guess, if not all the time. Now stand up and walk around the room, paying close attention to how your toes are involved in walking. The mere fact that the front of your foot (i.e., your toes, collectively) can bend enables a much smoother motion.
Now look at your hands. Each finger has three joints. Try to pick up a coke can or water bottle while only using your back knuckles, the joints that attach your fingers to your hand. There are two differences between that and normal gripping of the container. First, you obviously have a better grip when allowing all of your knuckles to bend and your fingers to encompass more of the container. The second consequence is that removing knuckles from the equation concentrates pressure on two specific points; the pressure is not as widely and evenly distributed. Especially as we move into a gaming era of more depthful and intuitive interfaces (such as virtual gloves), this considerations like these might prove useful.
Lastly, straigthen your arm back as far as it will go. If it was forced much further, your arm would break. Now imagine if injuries in Madden were actually related to events such as that and were not almost entirely random, as they always seem to have been. I know a lot of players who would be much more accepting of injuries if they actually got to see how the injuries happened (from a T-rated distance, at least).
There's a couple of ideas, anyway.