Tuesday, October 14, 2008
Personality For AI in Video Games
In video game production, more often then not there is an assumption that any attack or motion will do for an AI as long as it is “cool”. A lot of this has to do with the fact that animation is often created before design has finalized the game plan. I find this workflow very frustrating. I understand that games are about the fun play factor and animation is the icing on the cake, but with a little staggering of roles I think both can be achieved. In games there are very few places to show personality for AI that doesn’t rely on cinematics. If possible I try to milk as much personality into every available piece of motion. The attack of a character should compliment its personality as much as the idle or fidget. An example of this can be seen in my TR creature reel.
The second creature showcased in the movie is the Boargar (Boar looking animal). Illustrating his swipe attack. Originally the design doc only called for a melee attack. During the swipe he spins around 90 degrees then has to manually turn the body the rest of the way. What I was trying to show in this move was that he is normally a very happy go lucky piggy but when threatened he strikes hard and strikes fast. So much so that his energy and exuberance can overcome his cute little body structure. Or something like that. This may seem like overkill but for a video game animator it can help keep the fun of animating alive and push the artist to make every animation as unique as possible. I also believe that this mentality helps bring life to a game without compromising design. This idea is not limited to only attacks but should be applied to everything. I have found transitions to be one of the most furtle ground for personality animations. When a creature is passive and then activated it is a great place to show how that character feels about being bothered. Also transition from active to passive states can show frustration or boredom. All of these things are usually supported by most tools such as Havok Behavior or most video game technology. I would love to have further tech that would support more situational or occupational animations. That would definitely go a long way to make creatures really belong into the world. I have not yet been blessed with that type of tech, so till then I will keep sneaking in as much personality as I am allowed.
Use this form to send your friend this post.
I agree, animation is games is not given the same attention or consideration as other art facets.
My biggest clashes with animation in games:
1. Return to Idle pose, zaps so much of the fun and realism out of any animation.
2. Animations (attacks specifically) are meant to be instantous so the player feels in control. This kills the wind up and anticpation that typically go into making the motion read well.
3. Frame rate reduction. In an effort to slam in as much normal maps, texture space and world elements, once the bad guys enter the frame, frames get dropped. So what was once a fluid animation, now is junked and staggered as the frame rate tries to keep up.
4. No Occupational animation sets. You and I talked extensively about this for Tabula Rasa, but it opened a whole new bank of animations, concerns, tech and unknowns. But giving creatures a ‘purpose’ in the world, other then just standing/patroling waiting to be shot, would help tremendously in increasing their believablity.
While working on Tabula Rasa and various other games, I found the most ‘personality’ based animations where Idles, Fidgets, and Emerges (Spawns). Idles are a given, the general stance would define the demenour of the character. But fidgets could actually show what a character is doing while bored. Doing aimations like stamping the ground and sniffing up pollen from the dust, scooping up a pile of rocks or burrowing a hole, showed more life and character. Lastly would be Emerges, one of my favorite animations to do. I believed that an emerge showed more character then a death, since it wasnt being affected by an action. Was the emerge slow and dilberate? looking around or focused? The Stalker ‘rocket to earth’ animation you did is a great example of showing something clever, unexpected, character defining, and technically smooth. Unfolding its legs, flipping upside down and stomping to show weight, all revealed that the character was alien, mechnical and heavy.
As for future technology aiding animation, I think we’re a long way off from anything major. I still don’t know what the equivilent of Normal mapping for modeling, will be for animation.
Rag doll and human simulators used in the Force Unleashed?
Lip syncs and emotional expressions based off audio used in Mass Effect?
Custom animation based of character creation in Spore?
Environmental Dynamics in Little Big Planet?
All are leaps and bounds in their own direction, but the framework is based around that one selling point. And though it helps us achieve more motion, I’m not sure that these facets add to developing the character.
Thanks for the plug Jorma, much appreciated. That reel is definatly the result of working around yourself, bren and mead. That was a crazy team we had going on.
I definatly agree that the spawns could have been greatly increased to make them more effective. Though we had a few that worked, most just felt… out of place.
What worked:
-Thrax teleport/beam in. But combine it with the dropship hovering overhead, made it so much better.
-Tree lurker jumping down from above. Works great until he spawned with no branches over head in the middle of nowhere.
-treemites burrowing out of the dirt. Small, insectoid creatures digging out of dirt is fine.
-Dropship dumping off a Mech, having the mech unfold admist the dust.
What didn’t work:
-transpanecy fade ins. Worse yet is one with no emerge animation. And the grand finally is that the creature is huge. Your Treeback is fine example. Great potential, poor execution (on the design side, not animation : )
What i’ve seen in other games that felt good:
-Business men/women coming out of a tinted doorway from an office building.
-Spiders in a cave descend from a white thread to the ground
-Teleporter pods that the player has used before, keep emitting guys.
-Effects heavy spawns like a wizard casting smoke bombs with a demon emerging.
Like you mentioned, making the creature emerge in the world, not just appear, makes a much smoother transition to increasing the population. However, it more costly to do. And like in your reply, the use of a prop is usually that conduit of believablity. A doorway, teleporter, hive, dropship, anything of that nature that brings a character to the world, drops it off and leaves. Your minds eye allows you believe that to be possible.
As for future tech and the animation balance in games. Personally, I think less is more. Less number of characters, with a larger animation set each instead of pumping out loads and loads of generic sets of swapable creatures. MMO’s will always suffer from this due to the limitiation of the medium. I think first person shooters/3rd person shooters will have a better success rate, however. Looking at a game like Dead Space, DOom 3 or more isolation/thriller games, allows for more initimate interactions with creatures, instead of making them mindless fodder and bullet sponges.
Personally, I lean more in favor of in game cut scenes to introduce new characters (doom 3 did this wonderfully). However, thats not always possible or even partical.
Too true. Though I disagree with the idea of a few creatures with a more robust animation set. Personally I think it is variety. Designers have always wanted to have a horde of creatures coming at the player. This would be really fun for only a short period of time. But fighting 3 creatures at a time even with a well defined animation set will eventually become just as boring though it will just take longer. I think designers can use the rolling out of creatures as a tool to create a rhythm of intensity for the player. This would take a coordinated effort between all departments to control environment, mesh, and particle detail to allow for the eb and flow of creature numbers. I have never worked at a studio that had that much organization. Most games I have worked on rely on the lego methodology so we can usually only have a couple of characters on the screen because we can not predict the playing field. Anyway thanks Kiel for your thoughts.