|
|
By
Stephen Clarke-Willson, Ph.D.
Gamasutra
January 1, 1998
Originally
published in Digital Illusion, by ACM Press.
|
|
|
 Features
Applying
Game Design to Virtual Environments
Virtual
reality technology-3D navigation plus novel interfaces-is being used to
bring improved entertainment experiences to greater and greater numbers
of people. This chapter examines how principles of good game design dating
back to the best 2D sprite-based games can enhance the design of newer,
3D entertainment environments. In addition, some problems found exclusively
in 3D are described and solved. Everything described here has been implemented
and tested in products under development.
The current excitement over creating 3D virtual environments has, for
the most part, overemphasized the technology and lacked focus on the quality
of the actual experience. The original Super Mario Bros. and Zelda
series of games developed over the last ten years and published by Nintendo
provide a greater virtual experience than 95 percent of the 3D games being
published today. Virtual environment designers must carefully consider
what made those early games so appealing and apply the same principles
today, albeit in a different presentation style. This chapter examines
the following list of key game design principles:
- Third-person
presentation
- Discovery
and exploration
- Movement
versus animation
- Player
control
- The use
of maps
- The use
of "weenies"
- Closed
environments
- Constant
positive feedback with sporadic negative feedback
- Complexity
management and slow bullets
Then,
the chapter addresses three problems inherent in 3D environments that
must be solved in any good entertainment application:
Third-Person
Presentation
The
classic games (except some driving games) all have an amazing attribute
in common-that of third-person point of view (POV). Many driving games,
even today, that strive for a straight-through-the-windshield view still
allow a third-person POV. By contrast, the vast majority of 3D VR games
or experiences adopt a first-person point-of-view wherein your own character
is not visible.
There are two good reasons for adopting the first-person POV:
- It's
easier to implement.
- It's
easier to sell.
The
first-person POV is easier to implement in a 3D world (and nearly impossible
in a 2D world) because several issues, as detailed shortly, are instantly
sidestepped. Most importantly, management of depth perception is extremely
easy: Game players or viewers simply draw straight lines from their view
to anything they see: If it's big, it's close; if it's small, it's far
away.
The first-person POV is easier to sell (in other words, to get someone
excited about funding) because it provides a more visceral experience,
and that's what the majority of game and interactive development today
is about. Each game tries to top the previous top game's "rush" of graphics,
which is easiest to do with first-person POV. However, this places enormous
limitations on the depth of the experience that is possible.
First-person POV is a good first step for the industry, but it must evolve
to support third-person POV in order to grow and to attract an audience
beyond those looking for a rush. Admittedly, a VR skiing attraction might
not be as much fun in third person, where one can see the skier one is
controlling, because most of the fun of skiing is in the rush. Skiing
(downhill skiing at established resorts) is not about exploring, learning,
and discovering.
By contrast, the best 2D games are about these very things. While nearly
all popular games, it seems, have an action component, the enduring games
also create an environment-even in two dimensions-that invites the player
to explore, learn, and discover.
Imagine we want to develop a 3D VR Web site for some commercial purpose.
Our first thought will be to provide some kind of visceral experience
to attract people. A 3D environment where the world whirls about will
attract a certain number of people-mostly males, ages 18 to 35. But now
that we have people visiting our Web site, it's probably more important
that we encourage them to explore, learn, and discover. We want them to
explore so that they'll stay active in our site. We want them to learn
our product; otherwise, there's probably no commercial reason to create
the site. We want them to discover because it's the internal feeling of
"Aha!" that gives intense pleasure and rewards our visitors for coming
to our site. This critically important Aha! experience can only be created
through careful design; it's not something that comes via a purely visceral
experience.
In a Mario Bros. game, the first thing you have to do is jump on
a Koopa (that's an enemy sprite for the Mario-impaired). But the second
thing you have to do is to navigate a series of blocks; this forces you
to explore how your character moves and how the environment reacts to
your character.
In a first-person POV game, your character's movement is very limited.
In fact, your character doesn't move at all; instead, the world swirls
around it. But there is a linear relationship between you, your character,
and the world (Figure 13.1).
 |
|
Figure
13.1: Relationship of user to character in first-person POV.
|
In
a third-person POV game, the experience is more complex (Figure 13.2).
This triangular relationship between you, your character, and the world
provides much greater opportunities for interaction.
Why? Because of your ability to see your character in the world, your
mind can see more complex relationships. Instead of a more intense "in-body"
experience as with first-person POV, you have a transcendent "out-of-body"
experience. That is the real potential of VR-to give you experiences not
available in normal life-and third-person POV provides an increased context
in which to act and react.
Discovery
and Exploration
The
act of discovery is what creates the Aha! experience and can be something
as simple as realizing that a certain shape in the landscape means you
can get help there. Discovery is what cruising the Web is about. Discovering
new information is fine, but, in the design of a 3D environment, it's
also important that visitors learn to discover "how to" and not just "what."
In a Mario game, with a little experience, you learn that you can double-jump
and whack two Koopas in one motion. In a 3D VR Web site, for example,
where you've created an online store, you would want visitors to learn
"how to" by exploring and discovering.
You may need to manipulate your visitors into having this Aha! experience.
For instance, you might program your store-clerk character to purposefully
get in the way of a new visitor so that your visitor learns to ask the
clerk questions.
 |
|
Figure
13.22: Relationship of user to character in third-person
POV.
|
One final note on third-person presentation: Very few movies are made
in the first-person mode, yet moviemakers obviously want you to identify
with the main character to develop an empathetic link. If you watch
a young child (6 to 12 years old) play Mario Bros., you'll see
that the child identifies so completely with Mario that the child "projects"
into the 2D environment. You can create this same sense of identification
in 3D without resorting to a first-person through-the-helmet view.
Movement Versus Animation
In
a substantial VR world where objects may be extremely distant, the movement
of an object in the world is more important than any fancy animated details.
Up close and personal, animated nuances can provide all sorts of visual
clues about the character you are interacting with. But more attention
must be paid to how these characters or objects move within the environment
than is currently being spent because, in a truly rich 3D VR world, there
will be a huge number of objects but only a few that the viewer is attending
to actively at any one time.
These objects may exist in the environment to attract the viewer's attention
and encourage him or her to explore in a particular direction. Extremely
detailed, complex character animation is wasted at such a time, when the
object may be only a few pixels on the screen. But an amazing thing about
people is that we can tell whether that's a person or a robot off in the
distance by the motion of the object in the environment. If it bobs up
and down like a person walks, then it's probably a person. If it glides
a little too smoothly, it's probably not a person, at least as we know
them. The simple motion of a few pixels bobbing up and down is enough
to clue the viewer that "Hey, there's another person over there!" Particularly
if the motion is oversampled and the motion is displayed in a subpixel,
anti-aliased environment, very few pixels can express a great deal.
Player
Control
Ask
any player what he or she likes about Mario Bros. games, and the
answer is always the same: control. It's certainly not the graphics. My
early experience with VR Web browsers is that they leave me feeling horribly
out-of-control of my experience. I'm never quite sure where I'm going
to end up or why. There are a lot of technical reasons for this, including
latency problems and erratic frame rates, but the problem can be solved
even under these conditions by switching to a third-person POV. If I can
see myself in the scene, and how I move in relationship to other objects
in the scene, then I feel much more in control. If something is bumping
me from behind, I can see it. I'm not left in a semiparanoid state where
I have no idea what's affecting my character and therefore me.
Even if your user interface is dedicated to first-person POV, you can
still increase the sense of player control in complex environments by
allowing your viewer to move an icon, perhaps shaped like a camera, through
the scene, position it in a safe place, and then switch to that view.
I know from personal experience setting up architectural walk-through
animations that it's easier to create an animation by manipulating a camera
icon in the scene than by trying to animate by looking through the camera.
The final goal is the same-a first-person experience of walking through
a building-but the control necessary to navigate the environment is first
provided through the third-person interface.
The
Use of Maps
One
way to overcome limitations of first-person presentation is by providing
a map mode. Maps, in general, are pretty handy things to have when you
have a goal but you don't know how to get there. Maps are also handy to
create when you're exploring and want to remember what you've seen so
that you can get back to it (again, the map is necessary to achieve a
goal).
An incorrect use of maps, in my opinion, is to compensate for the lack
of a sense of context that first-person presentation omits. If your game
player or 3D VR user is constantly switching to the map to figure out
what's going on, then you've failed in the design of your environment.
So, when you want to encourage exploration, you want to make sure that
maps are unnecessary, and that's done through the use of "wieners."
"Weenies"
This
somewhat bizarre term was coined by Walt Disney, who suggested that when
designing massive 3D environments (theme parks), it was necessary to lead
visitors through the environment the same way one trains a dog-by holding
a wiener and leading the dog by the nose.
Obvious weenies at Disneyland are Sleeping Beauty's Castle, which encourages
guests to travel from the main entrance to the central hub; the former
Rocket Jets, which encourage guests to explore Tomorrowland; the Mark
Twain Steamship and dock, which encourage guests to explore Frontierland;
and the King Arthur Carousel, which encourages guests to walk over the
castle moat and into Fantasyland.
One of the biggest failures in weenie design ever is The Space Place.
The Space Place is, in my opinion, an extremely cool, high-tech environment
filled with plants and huge cathedral ceilings. It's a restaurant, and
it's only open on days that Disneyland is very full. The reason is that
nobody knows it's there. It's hidden between Space Mountain and the site
of the former Mission to Mars. Even if you walk right up to it, you're
hesitant to go inside because the lighting does not invite you in. It's
so poorly placed that it's not clear how to improve it, without putting
big arrows out in front that say "Come in here!"
Your 3D VR environment needs to have standout landmarks so that it's easy
to navigate without a map. The best games, which have typically been designed
with very limited graphics, always save a few graphics to denote special
and interesting things that should be investigated.
[Gamasutra
Editor's Note: For more about weenies, check out "Environmental
Storytelling:
Creating Immersive 3D Worlds Using Lessons Learned from the Theme Park
Industry" and "Environmental
Storytelling, Part II: Bringing Theme Park Environment Design Techniques
to the Virtual World".]
Closed
Environments
The
best games take place within a limited constrained space. In Myst,
you're stuck on a series of islands. In The Seventh Guest, you're
trapped in a mansion. In Clue, you're limited to the locations of the
house. In chess, checkers, go, and Monopoly - in fact, in every board
game - you are limited to the board. In Dungeons and Dragons, where you
sometimes create your own environments, the Dungeon Master sets the limits.
One of the biggest problems in most flight simulators is the sense that
you can fly anywhere. The trouble is, in most of "anywhere," there's nothing
to look at. If you are creating a virtual environment, you should be sure
that anywhere you can go is interesting and that the places you shouldn't
go are off limits.
It's also best to disguise the edges of your virtual world. Your viewer
shouldn't simply run out of data while cruising your environment; he or
she should come to the end of it and want to turn around and go back.
At the least, put a fence around the legal area for cruising. Or a force
field. Or anything. But don't let your poor viewer walk out into the void.
It's not polite.
Constant
Positive Feedback with Sporadic Negative Feedback
If
you want to exercise editorial control over the way viewers explore your
VR environment, you'll want to use weenies to attract them to specific
places. Once they arrive, you'll always want to reward them with a treat
of some kind. It can be something as simple as a sound bite ("Great!"),
or a quick, fun animation, or increasing the score counter, or anything
that's special and easily associated with success (perhaps a flashing
blue light). For this to work properly, you must always reward progress
toward the goal you desire your visitor to achieve.
Now, from time to time, you'll want to dissuade your visitor from heading
in a specific direction. You might decide to always display a discouraging
sign or play a sound like a game show buzzer when someone goes the wrong
way. Unfortunately, it's far easier to identify the good things you want
someone to focus on than to enumerate every bad or irrelevant thing to
avoid. So, it's best to provide sporadic, almost random negative feedback.
The idea behind this is that your visitors will become slightly paranoid
about straying off the beaten path. Combined with positive feedback, your
visitors will learn to always search for those positive vibes that encourage
them to go where you want. They'll be a little paranoid if they don't
get some positive feedback soon, and they'll start looking for it because
it's the only reliable way to avoid negative feedback.
This technique is really helpful when you're creating your own world for
someone to explore because, in spite of all the technology, it's simply
impossible to delineate every possible path or option. Since your site,
if it's on a network, will likely be constantly under construction, you'll
never have time to make sure every possible location and approach from
every possible angle is complete and bug-free. So, you want your visitors
to explore the places that are known to be safe and avoid the places where
you might not have completed the job. To get them really paranoid, if
they stray way off the beaten track, just terminate their connection from
time to time. That'll get them to behave.
If you provide constant negative feedback that is reliably produced, then
you'll find a lot of people will spend a lot of time exploring the edges
of your universe, looking for this negative feedback. So, keep it sporadic
so that they never know if it's coming or not.
Complexity
Management and Slow Bullets
In
well-designed, seemingly fast arcade games, where player reaction time
is very important, the action is carefully orchestrated and, in fact,
not really happening as fast as it seems.
One huge and common error in game design is to create a game where the
moment you see something appear on the screen you have to react and shoot
it. This is OK once in a while but overall provides for boring game play.
At first glance, you might think that all games are made this way: You
see it, you shoot it. But, in fact, going all the way back to Space
Invaders, the objects actually tend to move quite slowly; the difficulty
comes from the number of objects marching toward you. If you look at the
bombs that get dropped by the aliens at the top of the screen, you'll
see they travel quite slowly. If those bombs were the only thing you had
to deal with, the game would be pretty simple. But, in the context of
all the other action, even one more slow-moving bullet can be overwhelming.
The majority of VR Web sites for the next few years will be fairly static
environments where you mostly just wander around. But, as the ability
of computers to download and manipulate individual objects increases,
the amount of activity within sites, including seeing other visitors,
will increase. It's at this point in time where careful attention to the
complexity of the scene will become important.
You might think that a site where there are a lot of slow-moving objects
(people) milling about talking to one another can't get too complicated.
Not so. Just imagine the last overcrowded party you went to. Notice that
most chat rooms available on the online services carefully limit the number
of visitors per chat room.
Many computer and video games limit the number of active objects on the
screen for technical reasons, but as these technical limitations are removed
by fast hardware, it will be smart to limit the number of objects on the
screen for psychological reasons. If you put too many active objects on
the screen at once, no matter how slow and innocent they seem, you can
quickly overwhelm your viewer.
Lack
of Depth Perception (Even with Stereographic Glasses)
Stereo
glasses seem like the next great thing. Stereo glasses that connect to
your PC are now available for less than $100, making them a mass consumer
item. But here's the problem: People only use their binocular capability
for close-in work (say, up to 15 feet or less). After that, they depend
on other visual cues, like atmospheric haze, to compute the distance to
something.
Perhaps the most important visual clue is differential scrolling of the
environment, used most notably and effectively in animated backgrounds
(formerly created with multiple-plane camera animation stands). For instance,
where I live in Washington state, there are forests on top of rolling
hills. If I stand still, I really can't tell how far away one tree or
another is, but if I am walking and bobbing up and down slightly, then
I quickly get a sense of the distance to different trees.
In the game Doom, the player character bobs up and down while walking,
and this does magnificent things in terms of providing a sense of depth.
It's too bad there isn't an option to turn it off so that you could compare
the effect both with and without.
A company called Vision III in Virginia has developed a 3D effect that
doesn't require any special glasses. In their technique, the lens rotates
around the axis of view at about 4.5 to 5 cycles per second. The amount
of motion is small enough that you don't notice it directly, but the increased
information transmitted to your brain from the subtle differential scrolling
that occurs gives everything you see much more depth. The effect is subtle,
and some people are more attuned to it than others. But it's a more sophisticated
version of the bobbing up and down in Doom.
Management
of Player Viewpoint
When
you develop your 3D environment, you will be faced with a problem that
moviemakers never have to face, namely, that your visitor might be able
to see the backside of your sets. In the movies, the camera location is
strictly controlled, so it's easy to paint only the front side of your
sets. At Disneyland, where you can roam at will, everything needs to be
painted, including parts of the backstage area that you might come across
accidentally from time to time. The same will be true of your VR environment.
If you adopt a third-person presentation, you might consider locking down
the camera, but visitors will quickly want to move it. Still, by constructing
your "sets" so that it only makes sense to place the camera in friendly
places and adding a small restriction (say, the camera can only be placed
on the ground), you can avoid a lot of extra work dressing up the back
sides of your buildings and walls.
In one VR viewer currently available on the Net, there is a Walk mode
and a Fly mode for moving about in the environment. Some environment builders
may want to disable flying in order to constrain the range of movement
in their environment to something reasonable. A typical techweenie approach
is to respond, "But this is VR! You should be able to do anything!" Not
true. It's also wrong from a show business standpoint. You don't want
a visitor to your VR site to do "just anything" anymore than Disneyland
wants you poking around backstage.
Navigation and Targeting Support
Navigating
in a 3D environment is tricky, from either a first-person or a third-person
POV. In Doom, it's pretty easy because there are actually several constraints
on what you can do-the designers know that you're spending most of your
time in buildings and that you're not going to turn things upside down.
In many VR viewers, there are not any restrictions, and you can turn yourself
upside down very quickly.
Even if you can't turn yourself upside down, how do you specify "I want
to go there"? If you point at an object and say, "Go there," should it
take you there and leave you facing the thing? Most of the time you'll
be staring at a wall. Should you have to walk everywhere as in Doom? That
could be quite tedious if you're not busy blowing people away.
One approach I've been using is a targeting system for navigating. You
point at an object, and your player character points at it. Then, as long
as you only move forward or backward, slide right or left, or fly up or
down, your player character stays pointed at this object. If you turn
left or right, then the targeting lock is lost. This has the benefit of
eliminating some of the degrees of freedom available, making it easier
to move through the environment, without permanently restricting your
movement. You can still see the sights as you cruise, but you're not constantly
worried about getting lost because your player character is locked on
the goal. If you come across something interesting on your way to your
goal, you can break the target lock and look around.
If this targeting approach is part of your navigation system, then you
can provide "way-points" for visitors. Just put up signs at interesting
places; if your visitors click on the sign, that becomes their destination
and they can quickly navigate there.
Summary
The
best research investment you can make for virtual environment interface
design is to buy a game machine and to sit down and play it for several
hours a week. In discussing look-and-feel, it's always easier to discuss
look because we have terminology (developed over many centuries). But
for now, while interactive design is emerging from the garage into the
mainstream of culture, it is still best to "feel" the interface directly,
and the best way to do that is to play games.
Stephen
Clarke-Willson [email] received
his Ph.D. in Information and Computer Science from the University of California,
Irvine, in 1986. As a research scientist at the Northrop Corporation,
he developed computer graphics algorithms for the biggest (certainly the
most expensive) video game ever created, the B-2 Stealth bomber. He was
chief architect of the Patran 3 MCAE system from PDA Engineering. From
1990 until 1994, he was vice president of worldwide product development
for Virgin Interactive Entertainment, where he supervised the development
of all computer and video games released by the company. Today he is the
owner and creative director of Above the Garage Productions [visit
site], where he is developing a series of 3D games.
This article was originally published in Digital Illusion: Entertaining
the Future with High Technology. Copyright 1998 by the ACM Press,
a division of the Association for Computing Machinery, Inc. (ACM).
|
|
|
|
|
