(Not) Game Genres, pt. 7: Wolf's Elementary Spatial Structures of Video Games (part two)

Continued from the previous post (these post titles are getting too long).

Wolf's spatial structures of video games, from "Inventing Space: Toward a Taxonomy of On- and Off-Screen Space in Video Games" (Film Quarterly (Fall 1997, vol. 51, no. 1) and republished in The Medium of the Video Game (2001)), was an inspiration for my own "A Taxonomy of Virtual Spaces" (Rowe, 2021, unpublished). Before I delve into the differences between my own taxonomy and Wolf's, I'll review Wolf's work through the eyes of two other scholars who influenced my work: Clara Fernández-Vara and Dominic Arsenault:

Clara Fernández-Vara and her team at Georgia Tech took issues with Wolf’s methods, stating that, “his analysis lacks a historical perspective, and the strict comparison to film misses what the intrinsic properties of the digital medium bring to videogames” ("Evolution of Spatial Configurations in Videogames," Fernández-Vara, Zagal, & Mateas, 2005). The paper rightly ignores Wolf's structures 9 (split screen) and 11 (mini maps) as these are augmentations to spatial representation, not a specific structure of playable game space. They also exclude structure 1 (text description) and full-motion recorded video as the team was not interested in games that "import spaces from other media" (2005).

The Georgia Tech team's paper declares that "the screen is the basic unit of space in videogames, since it frames the interface" (2005). I've adapted this into my own research as it provides a measuring stick usable in so many games, especially 2-D games. However, I can't agree with excluding games with text descriptions of spaces and full motion video from my research. I am working toward a method of understanding the aesthetics of games that can also be applied to other forms of digital media. Excluding the entire swath of text adventure games (interactive fiction), full-motion video games, and countless other games with "imported spaces" from other media would leave a massive gap in the system's usability.

Dominic Arsenault and his team at the University of Montreal took issue with Wolf’s spatial structures, citing that, “Wolf’s focus on historic, early 2-D game spaces led to him creating fine details between different types of scrolling spaces, but little detail for later 3-D game spaces” ("Game FAVR: A Framework for the Analysis of Visual Representation in Video Games," Arsenault, Côté, & Larochelle, 2015).

The University of Montreal team notes that much of Wolf’s spatial structures are focused on “scrolling spaces,” or what they call the Framing Mechanism Mobility. This deals with the player’s ability or inability to access off-screen spaces (if there are any) within the digital game, which as I discussed in the previous post, was the focus of Wolf’s essay. This means that Wolf’s categorizations give mostly superficial definitions of the spatiality presented within the screen’s frame. Like Arsenault’s example, Wolf lumps all 3-D spaces into the same category, no matter what method is used to project a 3-D space onto the screen. In Wolf's defense, he does have a detailed section of "Ways of Representing Three-Dimensional Space" (2001, pp 70-75) that concludes the article. However, the section doesn't differentiate between, say, 1-point (Maze Wars), 2-point (Battlezone), and 3-point (Quake) perspectival methods or pre-recorded film and video clips (Dragon's Lair, Myst) or the different ways they may affect the player's sense of space.

The four image planes of the Game FAVR system (2015)

The Game FAVR system developed by Arsenault’s team is specifically designed to define graphics, not spatiality. It uses a multidimensional approach to define by categories of ocularization (through what “eyes” the game world is presented), frame mechanism (how the frame moves across the game space), and plane analysis (how the agents, in-game environment, off-game environment, and intangible interface are projected to the screen). I took a similar multidimensional approach in the structure of my own taxonomy of virtual spaces, focusing on spatial construction as conveyed through graphics. A sense of space is mostly conveyed through graphics, so much of the methodology (like plane analysis) works equally well for my purposes.

As example, here is a direct comparison between my taxonomy and Wolf's classification:

1. No visual space; all text-based - I classify this spatial structure's image planes (to borrow a concept from the Game FAVR) of agents, environment, and background/foreground elements are all "rendered" in text description (something that neither the Georgia Tech nor the University of Montreal systems account for, by their designs).
2. One screen, contained - This is a fixed framing device with no mobility (in other words, the game "camera" doesn't move).
3. One screen, contained, with wraparound - Wraparound screens imply one of two types of topology, either cylindrical (wraparound two edges of the screen, like Pac-Man) or toroidal (wraparound all four edges of the screen, like Asteroids). Rarely, there are other topologies (like the cubic topology of E.T.).
4. Scrolling on one axis - This is a framing device with smooth scrolling 2-D mobility in one direction along either the vertical axis (Xevious), the horizontal axis (Super Mario Bros.), or a diagonal axis (Zaxxon). Xevious and Zaxxon are examples of auto-scrolling framing devices (what Game FAVR calls authoritarian).
5. Scrolling on two axes - This is a framing device with smooth scrolling 2-D mobility in any direction (Gauntlet). I have other specifications for games like Metroid where, due to technical restrictions of the Famicom/NES hardware, the game space may be vertical scrolling or horizontal scrolling, but not both at the same time.
6. Adjacent spaces displayed one at a time - This is a framing device with discrete 2-D mobility (sometimes called "page flip"). Some have single axis mobility (Pitfall!), others have two axis mobility (Atari's Adventure).
7. Layers of independently moving planes (multiple scrolling backgrounds) - Wolf describes two different cases here. The first is a background layer or layers (often in orthogonal projection) scrolling at a different rate than the player avatar in parallax motion to create an illusion of depth (he wrongly gives examples of Zaxxon and Super Mario Bros. - neither of which use parallax motion. Moon Patrol and Sonic the Hedgehog would be good examples). The second case is and example of a layered space, where there are multiple 2-D planes of gameplay that the player avatar may move between. Either there is a jump layer above the typical gameplay plane (Bump n' Jump, Pac-Mania) or the avatar may move into a background layer (Super Mario Bros. 3, Warioland).
8. Spaces allowing z-axis movement into and out of the frame - Wolf gives examples like Tempest and Night Driver. I call this a framing device with 3-D mobility in the Z-axis only.
9. Multiple, nonadjacent spaces displayed on-screen simultaneously - This is just split-screen multiplayer (Spy vs. Spy, Super Mario Kart, Final Lap). Multiple frames on a single screen do not make a new spatial structure, in my thinking.
10. Interactive three-dimensional environment - This is a catch-all category that I discussed above.
11. Represented or “mapped” spaces - For the most part, Wolf refers to a mini-map, which I note as a spatial modifier in my taxonomy (Rally-X, Bosconian). It can enhance the player's understanding of the game world in its entirety, but isn't really a navigable, playable space unto itself. The map usually appears at the edge of the screen (Defender) or replaces the gameplay frame (Spy vs. Spy) or the map is diegetically displayed in the game world (Myst). In some cases, the game is still playable with the overlay (Doom), but that is not the intended way to play.

Wolf also includes "god's eye" map games like SimCity and Caesar II in this category. I consider these to be games with an environment image plane that may have use plan orthographic (SimCity (1989)), dimetric (SimCity 2000 (1994)), or trimetric projection (SimCity 4 (2003)), or three-point perspective (SimCity (2013)), or numerous other projection options.

Video game imagery is hybrid in nature. Images are montages constructed from various visual elements, overlaid upon one another, transforming and animating, and often projected to the screen in different techniques for reasons of clarity, technical limitation, or stylistic choice. Such a complex and varied visual form for displaying space cannot be described with only eleven categories. It requires a multidimensional approach that accounts for the various conceptual planes (to borrow from Game FAVR) of an image in order to document, classify, and create a framework for studying its expression of a virtual space.