Abstract
dc:description.abstractCombat feel, the moment-to-moment subjective character of real-time melee combat in ac- tion games, is a central concern of game design and a recurring subject in design literature, but practitioners currently navigate it through intuition and reference to admired prior work, with no shared formal vocabulary for the design trade-o!s being made. This thesis presents a decision-theoretic framework that formalizes combat feel as a Bayesian network in which designer decisions act as interventions on measurable system variables, those variables drive latent perceptual states whose conditional distributions are grounded in the psychophysics literature on input-lag detection, duration discrimination, and audiovisual temporal binding, and the perceptual states aggregate into an expected utility representing overall combat feel. The framework’s descriptive value is demonstrated through a development study: the author built a melee combat prototype, iterating on animation timing, cancel structure, and impact- emphasis e!ects (hitstop, screen shake, camera response). The framework was constructed afterward as a formal description of the design tradeo!s encountered during iteration, with four published action games —Dark Souls III, Sekiro: Shadows Die Twice, Devil May Cry 5, and Hollow Knight— serving as design-literature reference points that informed specific tuning decisions. Two complementary forms of validation are reported: descriptive coverage of the development study, demonstrating that the framework names and structurally relate the variables actually tuned during iteration; and alignment with design-literature recep- tion of the four reference games, demonstrating that the framework’s vocabulary maps each game’s commonly described feel to a structurally distinct region of the framework’s percep- tual space. The principal contribution is a structured vocabulary, traceable to the empirical literature on human perception, for the kind of practitioner reasoning about combat feel that currently happens informally. Numerical inference, parameter fitting against player-survey data, and prospective use of the framework during development are identified as the most important directions for future empirical work.
Degree
thesis:*- Name thesis:degree_name
- Master of Science
- Level thesis:degree_level
- Thesis
- Discipline thesis:degree_discipline
- Computer Science
Author and committee
dc:creator, dc:contributor.*- Author dc:creator
-
- von Goetz und Schwanenfliess, Grayson Julian
- Contributors dc:contributor
-
- Ray Toal
- John David Dionisio
- Jason Douglas
Subjects
dc:subject × 4Identifiers
dc:identifier.*- Repository record dc:identifier
- https://digitalcommons.lmu.edu/etd/1380