Great analysis.

After some thinking on the topic, I have to wonder how much of the migration away from facial emotes to body gesture is about making communication more visible, and how much is about streamlining art production.

Many of the recent release games seem to have very little in the form of facial animation. City of Heroes, like WoW, has a largely unchanging face, and although I'm a big user of emotes, I have to confess that I hardly notice it.

By making faces static, how many "bones" are eliminated from an animated form? how many production hours are saved? If I substitute facial emotes with body gestures, as EQ2 did, I'm making use of points of articulation that already exist in the game for other purposes, so the level of effort isn't as significant.

If the issue IS about the development effort, and not about the difficulty in visibility, then I see EQ2's solution (albiet, with less exaggerated animations) to be the most likely path of future development. I enjoy the superdeformed emotions shown in anime and Teen Titans, but they'd still mean additional development time- perhaps more than traditional facial animations, as the morphing variations would need testing for all points of character customization.

---

Another possible element lies in the context-based animations. Rather than just animating on command, the text parser scans for keywords in dialogue and runs animations.

In SWG, if I mentioned the term "cold" in my text, the character shivered. There were many other context-based triggers that I can't remember, and they weren't always accurate, but they were better than the static "neutral" stance.

I had high hopes for the "groundbreaking" animations I saw in SWG. It demonstrated a potential that I assumed would be carried on to other games, with more articulation, more context-animation, and more realistic body gestures. Instead, developers seem to have moved away from this path.

Imagine a rich text parser that animated something when you started typing (SL has the "typing animation"... I'd prefer something more immersion-setting, not immersion breaking). When you hit "enter" and your text balloon appeared, it could do such things as:

- animate body based on your /mood or different /say options (like /angry /sad etc)
- animate the face, doing a quick syllable parsing and associating a lip movement similar to the sound...
- animate based on context, or on specific commands inserted in the text, at the appropriate time (example ":polite: please refrain from such conduct :threatening: you wouldn't like to see me angry")
- animates eye contact realistically, with characters meeting eye contact, looking around, etc, rather than the extremes of "ignoring entirely" to "focused gaze"

Heck, I was dreaming of the far-future day where I could define my own context-based emotes (perhaps even assign a % likelihood of it happening, so "cold" doesn't trigger the same emote every time) much like players can customize their appearance

-------
One more note:

City of Heroes offers animations for "lecture" or "argue" that gave your character entertaining arm gestures... there are probably about 3 or 4 different animation sets available. I've hotkeyed them and use them while chatting to add a little body language, but they have to be used sparingly.

At least for WoW, the solution seems simple: bring back the animated portraits of targeted characters that appeared in Warcraft III to allow for a close-up of facial animations. Throw in a little 3-D imaging and you could even rotate the targeted chracter to automatically reflect your avatar's viewing angle. I imagine it's only a matter of processing power/memory that led them to change back to static portraits for WoW.

I think Dan has a good point. I feel like it might also work to have multiple (albeit smaller) portraits persisting on the screen. Maybe as soon as you type someone's name in an emote, or they type your name, their portrait pops up on your screen to signify that you are "watching" them. It could be a static image of their expression that changes depending on their emotes. You could then, I don't know, walk away to stop listening, or click an X in the corner of the portrait.

I think the idea of a player window that pops up that demonstrates the facial expressions of others is a good idea. The only downside might be that if you are in a group of 5-40 or more there could be a lot of emotes coming your way and that could prove difficult to work out. Maybe a manner to deal with the computer power issue would be to limit to friends on your friends list or some kind of list of people with whom you're concerned about regarding the emotes they may show you.

I would also like the idea of exaggerated emotions, but the difficulty there again is when you start dealing with larger amounts of people (this all based on playing wow, sorry haven't tried the others). Maybe a combination of the two with a window that appears and to grab your attention when someone types /love their eyes become hearts.

BTW: 1st post here, just found the site recenlty. I really like the work you all are doing.

Thanks, Kyle and welcome. Yeah, 5-40 close-ups of character faces is not going to work. You could of course just have close-ups for group members (5-6), but this does not support communication with players outside your group.

I was thinking just one close-up for the character you have targeted. If you want to really see someone's facial expressions, you select him or her. This would probably work decently most of the time and would be very manageable. However, it would miss reactions from other characters whom you don't have targeted.

I've been thinking more and more that 1st-Person Perspective (1PP) might be the way to go. If you want to see characters' facial expressions, you just stand close enough to see them like in real life. Now to solve the problem with lack of proprioception, you could simply create a close-up view of your own avatar (i.e., a "paperdoll" view in the corner of your screen). That way you can see when your avatar waves, shrugs, bows, etc. You could also, depending on the size of the close-up, see your own facial expressions. With this approach, you only need ONE close-up view, your own character, instead of multiple. Lack of peripheral vision would still be a problem, but perhaps it would not be such a problem in the context of a conversation. When you start to travel or do combat, you could pop back into 3rd-Person Perspective.

I've often done the "paperdoll" view via EQ2's "inventory" page to preview the emotes and see what others see- it's a bit big on the scren, but serves the purpose Bob mentions.

In SWG, I could comfortably remain in 3rd person view with my character centered, zoomed out far enough to see my feet, and still make out most facial expressions.

Even the "only seeing the target" or "only seeing the group" portraits doesn't seem too bad either- how often do you, as a person, notice ALL the nonverbal expressions going on around you? You notice your friends (group), the focus of your attention (target) and maybe those immediately around your target. (missed, in this case.)

One method of handling the "portrait mode" would be to use the software to prioritize gestures- first groupmates, then targets, then within x of target, within x of character, etc. A max of 3-5 portraits would be shown at once, the others, missed (or only appearing in the chat log.)

Heck, you could even make it hyperlinked in the chat log, so if someone wants to go back and see it again, they could...

very interesting conversation going here... it seems the approaches could be plotted on a spectrum with realism at one end and gestural on the other. using animations, povs, close ups to get a tighter view of the face, in other words, higher preicision/realism. Or, developing gestural languages and implementing ways of invoking gestural cues that would communicate not through their facial expression but through their use of a cultural coded gesture. Both are valid from the perspective of symbolic interaction/performance/speech analysis.
Interesting would be to have a coded version that could capture and analyse game play based on the instances and frequency of gestures: friendly to unfriendly, responsiveness of others; directionality of gesture (to a person or group). etc.,

Related idea: Silent films used a combination of exaggerated facial gestures, feature enhancing make-up and "cut aways with text" to overcome the limitations of their medium. Perhaps similar strategies would be applicable here?

So far, the only solution I've found to limited field of view in first-person is some form of letterboxing (possible in EVE, possible with viewport in WoW). By resorting to a shorter and wider frame you get more of the stuff off to the side, but at the cost of losing screen real estate as well as any extended below/above perriphery.

I suppose all of that "blank" space in the letterbox view could be used to hold UI clutter. But in factI don't think you will still get enough peripheral vision without a weird fisheye effect. What is the field of view on a human? About 120 degrees or so? (I am asking, I have no idea). How much of a FoV would you need to get over that feeling that you need to constantly turn your head to see things you should be able to see?

With respect to that, I remember a couple of keyboard shortcuts from the old mac-only Marathon games that let you quickly glance to the left and right (Bungie referred to that action as a "vid"). That really helped, as the glance was lightning-quick and did not require you to change the direction you were running/facing/fighting. Perhaps something like that could help to eliminate the FoV problem (once people learned to vid regularly). Or it could just be even more cumbersome... I seem to recall it felt pretty natural after a while.

Eww, I loved Marathon. It was the first FPS I ever played. I think side-glancing is pretty standard in FPSs and other genres (e.g., Grand Theft Auto has it). I've never used it much. I think in most cases, the view just switches abruptly, which I find pretty disorienting. I would think quick panning would be better. Do any games do that?

The silent-film approach is also very interesting, and I think very applicable to virtual worlds. It reminds me of certain tricks in CG movies. I once heard someone from Pixar say that while the human face has 30 points of articulation, for the Incredibles they built the characters' faces with 300! That way they could create facial expressions that were much more exaggerated than real ones. Such displays of emotion nonetheless FEEL very real to the viewer, although obviously they are unrealistic in interesting ways. The fact that the animators even know how many points of articulation are in a human face shows that they've done their homework and that they deviating from reality in very deliberate ways.

"What is the field of view on a human? About 120 degrees or so?"

Approx 180 degrees. In game a 75-90 degree FOV is often used.