« February 2006 | Main | April 2006 »

March 20, 2006

A nod's as good as a wink to a blind bat!

The facial expression of avatars at the Black Sun nightclub, social hotspot in Neal Stephenson's Snow Crash (1992), was so natural and nuanced, that it was "just as good as a face-to-face." Unfortunately today's metaverses (i.e., MMOGs) have a loooooong way to go before they approach that kind of sophistication.

There does not currently seem to be any consensus among game developers regarding the best way to implement facial expression. Different games employ different approaches. Yet each approach seems to grapple with the same basic problem: avatars' facial expressions are usually hard to see. (Another basic problem is how to control them easily, but I'll save that for another post.) The reason for this is that although avatar features and proportions tend to reflect those of real human bodies, the player's visual perspective usually does not. Players tend to play from a perspective that is at least several feet behind and slightly above their avatar's head. In other words, the player's viewpoint is disembodied. This unnatural distance makes it hard to see the expressions on avatars with realistically proportioned heads and faces.

Now one solution might simply be to force first-person perspective, at least in certain circumstances. Nick's data suggests that some players, especially women, actually prefer first-person perspective. However this solution creates different interactional problems. First, in first-person perspective, it's often difficult to notice other players and mobs unless you're looking right at them. The reason for this is that standard computer monitors don't allow for peripheral vision. A player's view of the game world is much narrower than it is in the physical world (see Hindmarsh et al. 2001). So the "disembodied," and usually zoomable, viewpoint in most games helps mitigate this lack of peripheral vision by increasing the player's field of view.

A second problem with first-person perspective is that when in it, you don't really know what your avatar is doing. If you can't see your avatar, you can't be sure, for example, that it waved, bowed, shrugged, pointed, etc. as you intended. The reason for this is that games don't simulate proprioception, the sense of the position of one's limbs that is independent of sight and touch. Indeed we are all like the unfortunate "Disembodied Lady" documented by Oliver Sacks (1986) when it comes to controlling our avatars. So forcing first-person perspective could hinder players' ability to use avatar gesture in sophisticated ways (e.g., for RP).

So given the low visibility of avatar faces in third-person (or first-person-zoomed) perspective, how are game developers handling facial expression? Here are a few approaches:

No facial animation - some games, like World of Warcraft and EverQuest Online Adventures (PS2), don't animate facial expressions at all. WoW has slash commands for things like /smile, /frown, and /wink, but they merely produce text emotes, e.g., "Bob smiles at Eric." EQoA doesn't even provide text emotes. So orcs always look pissed off, even when they're happy.

Facial animation + text emote - other games, like Star Wars Galaxies and Second Life, provide some nice facial animations that are accompanied by corresponding text emotes. So for example in SWG, when you type /smile your avatar smiles and you get the message, "Bob smiles at Eric." Typing /wink makes your avatar wink and you also get the message, "Bob winks suggestively at Eric" (apparently you can't wink non-suggestively). Similarly in SL, a simple text emoticon, such as :-) :-D :-( ;-) or :-P, accompanies the facial animation. Now although facial animations in SWG and SL are rather detailed, from my own experience, I find that I usually orient to the text emotes instead. Again, when my viewpoint is zoomed out, avatar faces just are hard to see.

Body animation + text emote - now in EverQuest II, the devs took a creative approach. They gave up on trying to animate the face (in most cases) and instead tried to translate common facial expressions into gross bodily gestures. So for example, /smile produces the text emote, "Bob smiles at Eric," plus an odd swing of a curled arm with a fist, but no smile. Or /wink produces, "Bob winks at Eric," plus a couple of nudges with the elbow, but no wink. Now these body animations are indeed much more noticeable than facial animations, and I tend to notice them rather than their corresponding text emotes. However, while I applaud the devs' effort, the animations in EQ2 just don't feel like an adequate substitute for facial expressions. Perhaps if they used more appropriate body animations, ones that people actually do when they smile or wink in an exaggerated way (and also included the facial animations), it might work better.

Automatic positioning of avatars - another creative approach to the problem of seeing facial expressions can be found in There. There solves the problem by automatically rearranging avatars in the supposedly optimal configuration: a semi-circle. When an avatar steps into a chat group, all the avatars are automatically shuffled to make room in the semi-circle. With this configuration, the player can see every avatar easily, including faces. (In addition, the text commands - 'wink, 'smile, 'frown - appear in the chat bubbles in case you miss the animations themselves.) Another benefit of this approach is that you can see your avatar from the front and thus can see your own avatar's facial animations. None of the other approaches enable this. Now while this is indeed a clever and effective solution, I personally don't like the system pushing my avatar around. If I had to choose between control over my avatar's position and orientation and a clear view of facial expressions, I might pick the former.

Other possible approaches...

Close-up view - one simple, although perhaps not so elegant solution, would be to create a small close-up window of the face of your target. That way you could easily see the facial animations of the avatar you're looking at but also the gross embodied gestures of your own and others' avatars in the main view. While two disconnected views of the other's avatar may not be ideal, it might be worth experimenting with.

Amplified animations - another approach might be to amplify avatars' facial expressions making them bigger in size. I once heard Raph Koster suggest borrowing the style from anime in which characters are drawn with disproportionately large heads and eyes. This certainly make facial expressions more noticeable from a distance. I think another interesting technique in anime is that of super deformed ("super-D") emotes. When characters express an extreme emotion, their face and bodies become radically morphed for a few moments. For example, in Teen Titans, when characters get angry, their heads grow huge and menacing for a moment; when scared, they shrink to a tiny, baby-like form; or when overtaken by love, their heads balloon and their eyes turn into hearts. This use of super-D really feels a lot like an emote in an MMORPG to me since it is abrupt and lasts for only a few moments. I can easily imagine super-D versions of /rofl, /OMG, /cheer, /mad, /scared, or /goggleeyes. However, I'm not sure how well it would work for the more subtle, basic expressions like winks, smiles, or frowns. Also, while these kinds of techniques would feel "natural" in an anime-themed world, I'm not sure how well they would work in Norrath or a galaxy far far away (they might work better in the ever-humorous Azeroth).

So which approaches did I leave out? Which is the best?

Post by Bob

Posted by bobmoore at 03:34 PM | Comments (12) | TrackBack

March 13, 2006

Raid Content Use

In the month of January, we tracked 223043 characters. Of these, 11098 (5%) spent time in high-level raid content (BWL, MC, or ZG). The majority of these were level 60 (as expected) - 99.4%. The remainder were level 56-59 (0.06%).

Of all the level 60s, 30% have spent time in raid content. On average, characters who spent time in raid content spent 310 minutes (about 5 hours) over the month of January in raid content.

Of those who spent any time in raid content, 28% spent less than an hour in raid content. In other words, 72% of these characters spent more than an hour in raid content. Thus, 3.6% of all observed characters spent more than an hour in raid content over the month of January.

Server Sample: RP (High), PvE (High), PvE (High), PvP (High), PvP (High)
Sampling Period: Month of January
Sampling Resolution: ~12 minutes
Parsing Method: All unique characters were tracked.
Data Filter: None
Sample Size: 223,043 characters

Posted by nickyee at 11:45 AM | Comments (26) | TrackBack

March 03, 2006

Guild Churn

It's easy to talk about guilds as somewhat stable entities over a one month period, and by and large, most guilds with more than 10 members do survive from one month to the next. But we were interested in exploring the amount of guild member churn that occurs. For example, given the guilds with 30 members, how many characters were in that guild at some point during the month but are no longer in that guild?

To do this analysis, we tabulated two guild rosters:

Full Guild Roster: For each guild, note down all characters who have been observed to bear this guild tag at any point during the logging period.

Current Guild Roster: For each guild, note down only those characters who actually still bear this guild tag.

A character who is in the full guild roster but not the current guild roster is not simply a character who was not observed towards the end of the month. For this difference to occur, they must have deguilded (not bearing any guild tag) or joined another guild (bearing a new guild tag).

Thus for each guild, the difference between those two roster sizes is the member churn - the number of characters who were at one point in the guild but aren't there any longer. Below is the average churn for guilds of different sizes. The churn percentage was around 25% and was fairly stable across guilds of all sizes. In other words, if we see a guild that currently has 20 members, then over the past month, there were 5 members who have left the guild.

Server Sample: RP (High), PvE (High), PvE (High), PvP (High), PvP (High)
Sampling Period: Month of January
Sampling Resolution: ~12 minutes
Parsing Method: The sample unit is each unique guild with a guild size greater than 1.
Data Filter: None
Sample Size: 5,285 guilds

Posted by nickyee at 12:20 PM | Comments (4) | TrackBack