a great epic, and how is one born? Nathan Phail-Liff, 3D artist at Ready at Dawn Studios (Daxter
), was at the 2008 Game Developers' conference to discuss his studio’s challenges and solutions in bringing a hit franchise to the PSP, in the context of their recent work on God of War: Chains of Olympus
Describing his studio’s philosophy concerning the many problems that occur during development, Phail-Liff said that “the difference between a AAA great game and a good one... is being very self-critical and how you react to those mistakes.”
Moreover, Phail-Liff wanted the lessons of his talk to be "in no way a 'PSP thing'. We didn’t want to make any excuses for the platform we happened to be on.” These concepts, he suggested, should be applied to any platform.
In defining “epic” from a visual standpoint, said Phail-Liff, “I want to take the hyperbole out of it and take practical examples of what the art department did -- conceding that while “cinematics can be great,” the team “wanted a seamless experience from beginning to end.” They worked with an overall goal: “See where you’re going and also where you came from. It’s about anticipation.”
For example, a temple level that takes place later on in the game, the team aimed “give a small tease of the temple in the background” in earlier levels, showing just glimpses without a full reveal. That way, Phail-Liff suggested, the player will be aware of the significance of the place by the time he or she arrives.
“A more mundane example would be avoiding jarring transitions between game regions,” he continued, saying how, for a cave level that eventually leads to a underground lake, just adding moss to the cave’s walls builds a bridge between the two visually.
Also important was creating “strong, consistent scale relationships... you can’t sell grandeur without a strong point of reference with size relationships.” Scale was important not just in visual style, but also textures, Phail-Liff said. “We are also very careful to keep consistent resolution. There’s nothing that takes you out of a game more than blurry textures and UVB stretching.”
He illustrated his point by showing a scene from the game in Maya. “Once you know the rules, you can break them all over the place,” he noted.
The background of the level showed an approaching armada, which was achieved in a similar style to movie backlot studios. Just outside of the main foreground action, polygon density and scale are reduced the further in the background they are in what he dubbed “virtual forced perspective.”
In addition to this is creating a convincing atmospheric depth. “A well balanced fore, middle and background conveys an expansive world... you want strong focal points in the foreground, not just for performance but to keep focus on where the action is.” In the same scene, he showed how adding fog and blurs helped distort the line of horizon and make the water not look tiled despite only being two different textures.
The Polish Curve
Added Phail-Liff, "An extra 10 percent of effort yields an additional 40 percent to the final perception of polish.... The curve doesn’t fall off.”
However, he cautioned of "death by polish", which can produce diminishing returns. To avoid this, the team ranked tasks according to: dependencies, importance to story/gameplay, reusability, and bug potential.
“We’ll fix it later” is a great way to ship a bad game, Phail Liff concluded. This is no "it’s done when it’s done." Time taken away from the front of production has to be saved on the back end. For Phail-Liff's team, the way to do it was through content cuts: “Cut often, cut early!” and pipeline optimization.