Read on Twitter
Fantastic illustration of the types of things animation systems people (like myself), think about. Quality of character motion isn& #39;t just about the animations themselves, but which anims you& #39;re deciding to make, how they& #39;re connected into a system, how the character behaves, etc. https://twitter.com/KaoruGans0/status/1290967203278290944">https://twitter.com/KaoruGans...
I think it& #39;s worth talking about why game animation systems are like this. There are a few major constraints we& #39;re working under: 1. Player& #39;s expectations for responsiveness, 2. Production cost; as in the amount of work that it takes to create, 3. Technical limitations.
Player& #39;s expectations for responsiveness aren& #39;t a good reflection of how people really move. Characters in games generally accelerate and decelerate much faster than people do in real life, because it feels better to control.
We have to find ways to make this seem realistic, for example sometimes characters will step over their stop point and then step back, so that we give them more time to decelerate.
We can also implement both left foot forward and right foot forward Idle animations so that the travel time to hitting Idle can be reduced. We do lots of tricks like this to try and keep things realistic looking when it& #39;s anything but.
Some games have tried to do more realistic movement models, which can definitely make for better looking motion, but is often met with resistance from players. RDR2 is a good example. They took a lot of flack for responsiveness, but the trade-off is very robust animation systems.
Cost is a huge aspect of what we do. There& #39;s a concept in games called the "possibility space". Say you have an empty room and you add a character in there. What are all of the possible things that character can do...
... Now add an object into that room, e.g. a ball. How did adding that ball change the number of possibilities that could occur?
Of course we can& #39;t simulate everything, so our game genres usually focus on a limited set of actions. But when we design any new feature we have to think about all the ways that new feature is going to need to interact with the other elements of our simulation.
For animation systems that can mean multipliers to the amount of animations that you need to build: If you want to add a new weapon type to a game, you have to think about how that weapon type will work with locomotion, crouching, jumping, opening a door, etc.
Some new systems can be isolated. So for example, the interaction with the statue in the video seems stilted, but that& #39;s to keep that interaction isolated from other core systems. This is why in some games, characters will holster weapons before doing these types of interactions.
Animation systems quality is directly related to the number of animations a team can produce. More anims means more possibilities that can be accounted for, but more anims is more work, and can mean that individual motions don& #39;t get as much attention. You have to find a balance.
This leads to tech limitations: With better technical approaches, there are ways in which we can produce higher quality interactions, without dramatic increases to cost. For example, I& #39;m very passionate about automation of animator tasks.
And new technologies like motion matching and neural network based animation systems, have the potential to create highly interactive systems direct from captured motion, allowing animators to focus on the things that can& #39;t be captured.
We& #39;re a way off of being able to create completely realistic, context appropriate character motion in games, but it& #39;s also an exciting time: There are a lot of new approaches on the horizon that should hopefully make a big difference.
Another quick thing on cost: Status effects, like being injured, tired, cold, etc. are usually expensive as they require an alt version of every animation in your system. You can lean on layering to solve this, but it can be challenging to get good results with layering alone.
Some teams might just apply the status change to loco, and when the character does another move, e.g. climbing a ladder, they switch back to the base set. It& #39;s a case of seeing how weird that looks to have an injured character suddenly be fine when they get on a ladder.
