Games with party-based or multipresent interaction models need a camera model that allows the player to see a large part of the game world and several different characters or units at once. Normally such games use an aerial perspective, which gives priority to the game world in general rather than to one particular character.
In games with multipresent interaction models, you must provide a way for the player to scroll the main game view around to see any part of the world that he wants (although parts of it may be hidden by the fog of war; see Chapter 14, "Strategy Games"). With party-based interaction models, you may reasonably restrict the player's ability to move the camera so that it cannot move away from the region of the game world where the party is.
The top-down perspective shows the game world from directly overhead with the camera pointing straight down. In this respect, it resembles a map, so players find the display familiar. It's easy to implement using 2D graphics, which keeps its use common on smaller devices, but its many disadvantages have led designers to use other methods on more powerful machines.
For one thing, this perspective enables the player to see only the roofs of buildings and the tops of people's heads. To give a slightly better sense of what a building looks like, artists often draw them cheated—that is, at a slight angle even though that isn't how buildings appear from directly above (see Figure 18.1).
The top-down perspective also distances the player from the events below. He feels remote from the action and less attached to its outcomes. It makes a game world feel like a simulation rather than a place that could be real.
Designers of computer and console games now usually reserve this perspective for showing maps, although it is still common on smaller devices and some web-based games. Flight Control, for example, is a hugely popular top-down game for the iPhone.
The isometric perspective is normally used to display 2D outdoor scenes. While the top-down perspective looks straight down at the landscape from an elevated position, the isometric perspective looks across the landscape from a somewhat lower elevation, with the camera tilted down about 30 degrees from the horizontal. If the game world is rectilinear, as they usually are in games that use the isometric perspective, the camera is normally positioned at a 45-degree angle from the north-south axis of the landscape. This permits players to see the sides of buildings in the landscape, as well as the roof. See Figure 4.5 for a typical example. In the main view, a mixed troop of soldiers marches out through a gap in a city wall. You can see two sides and the roofs of various buildings around the soldiers.
Normally, a 2D display engine draws the isometric perspective using interchangeable tiles of a fixed size. As a result, the isometric perspective distorts reality somewhat because objects that are farther from the camera are not smaller on the screen. However, the camera does not display much of the landscape at one time, so players don't mind the slight distortion. The player can truck or dolly the camera above the landscape but cannot pan, tilt, or roll it. You can also allow the player to shift the camera orientation to one of the other ordinal points of the compass to see other sides of objects in the game world. If you want to provide this feature, the artists will have to draw four sets of tiles, one for each possible camera orientation. You can also let the player choose an altitude from which to view the world, but the artists will have to draw multiple sets of tiles at different scales.
The isometric perspective brings the player closer to the action than the top-down perspective and allows him to see the sides of buildings as well as the roofs, so the player feels more involved with the world. It also enables him to see the bodies of people more clearly. Real-time strategy games and construction and management simulations, both of which normally use multipresent interaction models, routinely display the isometric perspective or its modern 3D equivalent, the free-roaming camera. Some role-playing games that use a party-based interaction model still employ the isometric perspective (see Figure 15.3).
For aerial perspectives today, designers favor the free-roaming camera, a 3D camera model that evolved from the isometric perspective and is made possible by modern 3D graphics engines. It allows the player considerably more control over the camera; she can crane it to choose a wide or a close-in view; and she can tilt and pan in any direction at any angle, unlike the fixed camera angle of the isometric perspective. The free-roaming camera also displays the world in true perspective: Objects farther away seem smaller. The biggest disadvantage of the free-roaming camera is that you have to implement all the controls for moving the camera and teach the player how to use them.
Context-sensitive camera models require 3D graphics and are normally used with avatar-based or party-based interaction models. In a context-sensitive model, the camera moves intelligently to follow the action, displaying it from whatever angle best suits the action at any time. You must define the behavior of the camera for each location in the game world and for each possible situation in which the avatar or party may find themselves.
Ico, an action-adventure game, implements a context-sensitive model, using different camera positions in different regions of the world to show off the landscape and the action to the best advantage. This makes Ico an unusually beautiful game (see Figure 13.7). Context-sensitive models allow the designer to act as a
cinematographer to create a rich visual experience for the player. Seeing game events this way feels a bit like watching a movie because the designer intentionally composed the view for each location.
This approach brings with it two disadvantages. First, composing a view for each location in the game world requires you and your programmers to do a lot more work than is needed to implement other camera models. Second, a camera that moves of its own accord can be disorienting in high-speed action situations. When the player tries to control events at speed, he needs a predictable viewpoint from which to do so. The context-sensitive perspective suits slower-moving games quite well, and frenetic ones less well. Some games, such as those in the survival horror series Silent Hill, use a context-sensitive perspective when the avatar explores but switch to a third-person or other more fixed perspective when she gets into fights.
In the early Resident Evil games the camera sometimes jumped to a different point of view without warning, right when the player was in the middle of a fight. This disoriented the player and upset his understanding of the relationship between the controls and the screen, often causing him to lose the fight. Don't move the camera in unexpected ways during high-speed action. Use a fixed, or at least a predictable, perspective.