Source metadata
- Type: MSc thesis (Master of Science Thesis, Examensarbete för masterexamen)
- Authors: Erik Fagerholt and Magnus Lorentzon
- Title: Beyond the HUD: User Interfaces for Increased Player Immersion in FPS Games. The cover and the Chalmers repository record use this title; the inner title page shortens the subtitle to User Interfaces for Player Immersion in FPS Games.
- Institution: Department of Computer Science and Engineering, Division of Interaction Design, Chalmers University of Technology, Göteborg, Sweden
- Date: 2009 (the authors’ copyright notice is dated September 2009)
- Partner and supervision: written in cooperation with EA DICE (tutor Anna Wendelin); Chalmers tutor and examiner Staffan Björk, co-author of Patterns in Game Design (see source-patterns-in-game-design)
- Length: 124 PDF pages, including references and five appendices
- Availability: freely available from the Chalmers Open Digital Repository, handle
20.500.12380/111921
Page numbers below are the thesis’s printed page numbers.
Key takeaways
- The thesis is about immersion in first-person shooters. Working with EA DICE, the authors set out to find how immersion “affects, and is affected by” the user interface of FPS games, with the aim of producing UI guidelines for greater immersion (abstract).
- Method. An objective study of 19 FPS games, played analytically for 30 to 60 minutes each and chosen across the Metacritic score range (pp. 25–26); a subjective study of eleven games picked because they break FPS conventions, including Mirror’s Edge, Dead Space, Far Cry 2, Killzone 2, Left 4 Dead, Grand Theft Auto IV and BioShock (pp. 34–44); a literature study across ludology, cognitive science and media studies; interviews and a focus group; and three sets of user tests with five tests in each set (p. 28).
- The design space has two dimensions. The authors first tried a diegesis × spatiality grid (Picture 7.1, p. 48). Because diegesis proved ambiguous, they replaced it with a fiction question and kept the space question. In the final model (Picture 7.5, p. 51) the axes read “Is the UI element existing in the fictional game world?” and “Is the UI element visualized as a part of the 3D game space?“. Diegesis then becomes an area of the design space (fictional and spatial) rather than a dimension (pp. 48–49).
- Six categories, not four. Placing the UI elements found in their game studies into this space gave six categories (pp. 51–52; restated as the designer’s “tools”, pp. 73–75):
- Non-diegetic (HUD) elements: non-fictional and non-spatial; overlays, which make up most of a conventional FPS interface. Suited to detailed information.
- Geometric elements: in the 3D geometry but not part of the fiction, for example Left 4 Dead teammate outlines, Mirror’s Edge Runner Vision and checkpoint beacons, and Splinter Cell: Conviction briefings projected on walls. This is the category that later writers call spatial.
- Meta-representations: part of the fiction but not shown spatially, such as the Grand Theft Auto IV phone drawn as an overlay, GPS devices or a terminal shown full-screen.
- Meta-perception: non-spatial elements that make up for the “broken perceptual link” between player and avatar, typically conveying internal state such as damage or health through full-screen tinting, blur or distortion, heartbeat audio or pulsed vibration. The authors say meta-perception is “conventionally” not connected to the fiction, though Killzone 2’s on-screen blood bridges it into the fiction. In Picture 7.5 it spans both non-spatial quadrants.
- Diegetic elements: both fictional and spatial, presented as the player character would see them, as in Dead Space and Far Cry 2.
- Signifiers: a sub-group of diegetic elements that give indirect cues the player must interpret by reasoning, such as smoke from a weapon about to break in Call of Juarez or blood trails in Dead Space.
- Classification is about the sign vehicle. The authors classify the UI element as “an informational vehicle”, so ammunition digits can be non-diegetic even though the bullets they count are diegetic (p. 47).
- Immersion is redefined. After reviewing Salen and Zimmerman’s “immersive fallacy”, Juul and Linderoth (p. 67), the authors define immersion as moments when players use their real-world perception, reasoning or emotions to play, or adopt the game world as primary and reason from the character’s point of view, “rather than having to refer directly to the rules of the game” (p. 69). They split it into immersion through reasoning and through perception, leaving emotion out of scope (pp. 69–71).
- Four guidelines (pp. 72–92): Know Your Design Space (and How to Exploit It), Know Your Game, Establish Player Agency and Strengthen the Player-Avatar Perceptual Link.
Notable claims
- Diegetic is not automatically better. “Contrary to what one might believe, a fully diegetic user interface is not automatically more immersive than a user interface incorporating informational elements represented outside of the diegetic spectrum” (p. 75).
- Players accepted non-diegetic elements. In the navigation test, none of the participants said the non-fictional Runner Vision or the BioShock compass arrow was a problem, and only one felt the Mirror’s Edge checkpoint beacons were out of place (p. 63). In the health test, no participant said non-diegetic or meta-perceptual overlays broke the experience, and genre conventions were given as the reason (p. 64). The authors conclude that such elements “does not necessarily weaken the game experience” (p. 64).
- Players preferred no navigation aid. All five participants in the Mirror’s Edge/BioShock test ranked “no navigational aid” as best for the game experience, citing freedom and discovery. The aids they rated easiest to use were the ones they liked least for experience (p. 62). The authors still observed that players who lost their sense of direction under stress fell back on trial and error (p. 62).
- Match the UI to the level of competition. In narrative-led games “diegetic elements and signifiers are important tools”, and some ambiguity may even help. In competitive, skill-focused games “non-diegetic and geometric elements” give the clear, unambiguous information that is hard to achieve diegetically (p. 77).
- Fiction limits diegesis. Diegetic options depend on the setting: holographic screens like Dead Space’s “would feel awkward in a medieval fantasy-themed game” (p. 75), and full diegesis is much harder in a historically accurate World War II game (p. 77).
- Design objects as prompts. The team should know “a game object can be designed not only as a prop but also as a prompt”, for example a weapon that itself signals it needs reloading (p. 79). Exploring diegetic solutions early is worthwhile, but long diegetic explorations for highly competitive multiplayer games “may often prove unproductive” (pp. 79–80).
- Affordance amplifiers and suppressors. UI should amplify affordances the rules support and suppress ones players might infer but cannot use. Spatial categories (signifiers, geometric and diegetic elements) suit this best (pp. 80–81). Their worked example: swaying curtains would have shown Resident Evil 5 players which window they could jump through, where a close-range button prompt failed (p. 81).
- Tiered guidance on request. A proposed Mirror’s Edge guidance button would give more detail each time it is pressed: first turn Faith’s head towards the goal, then highlight the building, then highlight the objects needed to reach it (pp. 82–84).
- Progression versus optional affordances. Progression affordances must be amplified so players are not left to trial and error. Optional ones need not be, because discovering them is rewarding in itself (pp. 86–87).
- Reify rules that cannot be integrated. Battlefield: Bad Company turns its map boundary into “enemy artillery” rather than an invisible wall (pp. 87–88).
- Meta-perception strengthens the player-avatar link. Screen tinting and tunnel vision are generally appreciated. Audio meta-perception (a simulated cocktail-party effect, exaggerated street noise at roof edges) and haptic meta-perception are described as largely unexploited (pp. 88–90).
- Casual and experienced players differ. In the Resident Evil 5/Dead Space test, the less experienced players preferred persistent button prompts, while the three “hardcore” players accepted them only in an introductory section (p. 66).
- World coherence matters most. The authors note that the factor “maybe most influential on immersion” is Juul’s coherence of the game world, which is a shared responsibility, not the UI designer’s alone (p. 96). Because players interpret fiction subjectively, an element’s place in the design space “can actually differ depending on how the representation is interpreted” (p. 96).
- The guidelines are unverified. The authors say the guidelines and definition are “the unverified result of the last iteration” and should be tested with purpose-built prototypes, since the user tests used existing games (pp. 97–98).
Relevance
- game-interface-types: the origin of the two-question model, and a correction to how its categories are usually named.
- ui-design: the interface taxonomy section, plus evidence on readability versus immersion.
- diegetic: the thesis’s own use of diegesis, and why the authors moved from a diegesis axis to a fiction axis.
- presence-and-immersion and immersion-models: a definition of immersion based on reasoning and perception, built on the “immersive fallacy” critique.
- affordance, signifier and player-agency: affordance amplifiers and suppressors, signifiers as a UI category, and communicating agency.
- player-guidance: navigation aids, tiered guidance on request, and the finding that players preferred working the route out themselves.
- left-4-dead: teammate outlines as a standard example of a geometric (spatial) element.
Open questions raised
- The guidelines rest on small samples (five tests per user-test set) using existing commercial games. Would purpose-built prototypes, as the authors recommend, give the same results?
- Players in the health test were split by their prior genre experience (Halo players ranked Halo 3 highly, others ranked it last). How far do “genre conventions” explain the acceptance of non-diegetic UI, and does that acceptance shift as conventions change?
- The authors expect the design space to generalise beyond FPS games, noting meta-perception in racing and third-person action games (p. 97). Third-person, top-down and 2D cases were not tested.
- The popular four-label version merges meta-perception (usually non-fictional) into “meta” (fictional). Is that simplification harmless in teaching, or does it hide a useful distinction?