IAnim – Valve
Mar 8th, 2009 by Ninja Dodo
For the next couple weeks I’ll be doing a series of posts on the use of interactive animation in a number of notable games…
A bit of Half Life 2 and Team Fortress 2 this time…
One thing Valve does very well is make it feel like you’re present in the world by having characters react to you. Subtle things make all the difference…
In Half Life 2, people look you in the eyes. They don’t stare… they look and look away, their eyes darting from place to place. If you shine your flashlight in Alyx’ face she squints and covers her eyes.
The most elaborate animation is reserved for scripted sequences: intricate, often spectacular scenes that break up the action. You are free to move around within certain boundaries and characters respond to your location by glancing sideways or turning to face you, but… they remain essentially linear cutscenes. The overall sequence of events does not change based on your actions.
The line though, between the scripted and dynamic, is slowly fading and it will be interesting to see where this goes…
There’s a clever mechanic that began life with Alyx in Episode One and was further refined in The Orange Box and Left 4 Dead: Based on context, characters will say short lines of dialogue in reaction to what is going on around them, sometimes very specific things…
It would not be a stretch to imagine using the same AI that picks appropriate dialogue to pick gestures and other actions to make NPCs respond to the world not just in words, but in movement…
A different beast, Team Fortress 2 is interesting in its use of silhouette to distinguish characters. Each member of its ensemble is instantly recognizable at great distance through shape and style of movement. Just as well, as you’ll want to adjust tactics accordingly, or suffer an untimely demise…
The characters’ appearance matches their gameplay function.
Personality is all too rarely explored in game animation. Clearly it can serve both as icing on the entertainment cake and as core gameplay information.
Worth a mention too are the dynamic facial expressions in TF2. They seem to reflect player health and success with excited and fearful expressions, though, perhaps more significantly, they make screenshots completely hilarious.
Next up: Cecropia’s The Act
I’ve had a half-baker about L4D’s barks myself for a while. Those barks are an old trick going back well before Valve. A word though on these – animation is really really expensive. While we can do a lot with dynamic sound, animation is a lot harder, both to create and to fake. You can do things in the quiet moments (such as Alyx did in HL2), but when there’s action going on moving the face is usually all you’ve got bones for. Where would it happen in L4D, for example? AI will get better at picking potential idle behaviors, but there’s much less of a player gameplay need for it then there is for sound, so it’s unlikely to make big strides. The key will be the player need – developers start new genres where it really matters, like murder mysteries where you are trying to figure out whose lying.
(oops, this got caught in the spam filter)
I agree figuring out the *when* is a really big question that is not so easily answered. Even if you can do it technically there’s no guarantee the player will have the opportunity to see it. I think much can be layered on top of core actions, but it may just get lost in the bigger action. If you ask me, some kind of dynamic staging based on the principles of traditional animation except with flexible prioritizing might do the trick. So you don’t run a subtle action while a big action is going on or if the player isn’t looking, but you also don’t try to do a big (full body) idle action that interferes with a more important (gameplay) action… that kind of thing.
And I think you’re right, the points where animation matters directly to gameplay is where we’re most likely to make some headway.