We Have Moved

Pixel-Love is now at: pixellove.wordpress.com

A copy of the content remains here so links don't break, but everything is ported, so if you want to look through the archives we suggest you head over to the new site.

Wednesday, March 07, 2007

GDC07:Ubisoft's Adam Thiery Talks Camera Theory

GDC: Ubisoft's Adam Thiery Talks Camera Theory Adam Thiery, a designer for Ubisoft Montreal, gave a short talk today on interactive cinematography. His basic point was that game cinematography is player-driven. Simple it may sound; real application is always trickier. One of the big sticking points is that camerawork, being player-driven, is limited by current understanding of game design and player psychology.



A modern camera knows when to change state, explained Thiery. In Far Cry, when the player is pressed against a wall, the standard tracking camera shifts from a behind-the-character perspective to show the player character left-of-center, and focus the player’s attention to the right, around the given barrier.

Thiery said that a good game camera is a matter of functionality, rather than cinematography – yet given that, it pays to consider the visual composition within each camera state. The reason is that any action a player takes is generally guided by what he has been shown to do.

The original Half-Life takes places in a disorienting sci-fi setting; to drive the player forward, it uses huge stripes painted on the walls, like a trail of breadcrumbs or an arrow. Though this is an artificial and somewhat clumsy application, that same principle applies to any 2D screen composition.

When level design and camera work together, guiding lines like the edge of a table that has been pulled away from a wall – or areas of contrast, like light within dark, different levels of detail, or the difference between animation and an immobile object – tend to subconsciously guide the eye, therefore the player, in a certain direction. Thiery showed a dark scene from Splinter Cell, in which a small, white light was visible in the distance. Again, where the eye goes, so goes the player.

All this said, not every object in a given composition has meaning. A chair that happens to be sitting along the wall may just be a chair. Still, between the camera and level design, every scene should be possible to compose toward a guaranteed point of view on the player’s end.

A player will base decisions upon what is perceivable, and a given scene may well present a variety of options to the player. As those decisions should be predictable in a controlled environment, it may be possible to design particular consequences depending on the actions taken.

Thiery’s third major issue involves camera transitions. Most games avoid changing views, so as not to disorient the player. The most obvious problem here is the danger of overextending a shot beyond the point where it is dramatically interesting. The simplest way of avoiding confusion is to use “controller patches,” where as long as the player keeps holding the same direction, the character will keep traveling the same direction.

There are a couple of rules to follow here, though. One is to avoid drastic changes of view during action sequences, such as fighting or platforming. Another that Ubi is trying to incorporate is borrowed from cinema: the “180-degree shorthand”. A camera transition must not change movement direction entirely. “Left” should remain vaguely left, and “right” should remain right. Any change of 2D direction should remain within around 90 degrees, the narrower the better.

When it comes to camerawork, Resident Evil 4 is to Thiery’s judgment probably the most advanced game to date, about which he admitted having to resist a temptation to base the entire lecture around that one work. He explained how camera cuts can, in fact, add rhythm and intensity to a sequence, and can accelerate play by removing unnecessary transitions and skipping directly to the strongest perspectives.

In one scene, Leon approaches a window, and the view immediately cuts from a behind-the-back view to a somewhat overhead perspective, showing a ladder propped up against the side of the house and an enemy climbing upward. Leon tosses the ladder aside and steps back, and the camera cuts back to normal.

“I want the action,” Thiery said. “Show me the action, as quickly, as cleanly as possible.” A moment later, Leon is fighting a group of enemies. When the player performs a jump-kick to an enemy’s head, the camera quickly cuts to a side view, showing the length of Leon’s leg to illustrate exactly what Leon is doing, and then cuts back. The cut is directly related to player action, so it does not seem confusing.

Trailing off, Thiery speculated about where game cinematography might go in the future. With the added power of current and future hardware, perhaps depth-of-field or screen-in-screen will grow more important.

The important thing to remember is that, as game cinematography is player-driven, what works in cinema is not necessarily functional in game terms; cinematography is tied to game and level design. “Whenever you design a new control, think about how it affects the camera,” and how that in turn affects the controls.

No comments: