Please use this identifier to cite or link to this item: http://hdl.handle.net/10397/84361
Title: Lag compensation in first-person shooter games
Authors: Lee, Wai Kiu
Degree: M.Phil.
Issue Date: 2018
Abstract: Multiplayer games are important constituents of video game sales every year. While all multiplayer games are affected by network latency, or lag, the extent of the effect varies across different genres. Games with a first-person perspective, such as FPS and racing games, have been found to be the most sensitive to lag [1]. Sports and RPG games are less sensitive but not as less as omnipresent games like real-time strategy and simulation games. It is not uncommon to have players with high network latency in a public mul-tiplayer session. In shooter games, lag compensation is used to help mitigate lag for high-latency players. Upon receiving a shot, the server rolls all other players back in time according to the lag of the shooter. We refer to this method as traditional lag compensation (TLC). The drawback of TLC is that it introduces inconsistencies we call "shot behind covers" (SBC) to whomever receiving the shot. A player suffers from an instance of SBC when he is shot shortly after taking a cover, which according to that player, should not have happened because the cover should have broken the line of sight between him and the shooter. Recently a few games try to tackle SBC by imposing limits on lag compensation. Players who violate the restrictions, such as having a latency higher than a certain threshold, are not lag compensated. This approach, however, can be an overkill. As we will show in Chapter 7, most of the time shooters do not cause SBC even when the limits are broken. In this dissertation, we propose a novel lag compensation algorithm which we name advanced lag compensation (ALC). ALC ameliorates SBC while retaining the benefts of lag compensation. The fundamental idea of ALC is that the eligibility for lag compensation depends on each shot but not the shooter. A shot is lag compensated if and only if it does not cause SBC regardless of properties such as the latency of the shooter. A highly lagged player can continue to beneft from lag compensation as long as his shot does not cause SBC to the victim. The core of ALC is the algorithm for detecting SBC, which to the best of our knowledge, is the first of its kind. Not only does this algorithm form the basis for ALC, it also allows us to objectively measure the frequency of SBC in a match regardless of which lag compensation algorithm is used. In the past, the SBC frequencies could be inferred only from subjective opinions of the players. With the SBC detection algorithm, we are now able to perform objective assessment of the effectiveness of ALC in reducing SBC. Our results show that ALC signifcantly reduces the number of actual and perceived SBC (p<.05) while achieving similar hit registration accuracy (p = .158) and responsiveness (p = .18) as TLC. We also study how SBC changes with movement speeds and the amount of lag. These two parameters directly affect lag compensation because it rolls players back in time according to the lag of the shooter. A player is rolled back further in distance if either the shooter's lag or the player's movement speed increases. It may be easier to spot an instance of SBC if it occurs when the player is further behind the cover. In this dissertation, we show how movement speeds and lag individually, and together, affect the perception of SBC, and how they in turn impact the perceived fairness of the match. Our results also suggest that the limits used by the few recent games to tackle SBC are reasonable despite their approaches being an overkill. Other useful insights provided by this particular study include what game modes should be used in similar future experiments. To improve lag compensation, we have to understand what infuences player perception of SBC and how that in turn affects their experience. In this dissertation, we conduct an investigation on the effects of common gameplay mechanics and features on the perception of SBC. Because these elements are irrelevant to lag compensation, those that are found to signifcantly affect how SBC is perceived have to be adjusted before subsequent experiments on the problem of SBC are carried out. Our results show no signifcant difference in the perceived fairness between different settings for all tested mechanics and features. This means that in the rest of our experiments, we can configure those elements to behave as they do in commercial games. In order to perform all the aforementioned experiments, we had spent approximately 10 person-months to build a multiplayer first-person shooter game named LAGCOM from scratch using Unreal Engine 4. This provides us the highest and necessary controls over the game that would not have been possible with commercial games. We can customize the implementation of any mechanism in the game to suit our experiments' needs. In fact, some mechanisms and features are implemented differently from commercial games for experiment purposes. For example, we have implemented ALC in LAGCOM to evaluate its performance. Building our own custom game also allows us to include elements that are designed specifcally to facilitate experiments in LAGCOM. The game is open source and available on GitHub (https://github.com/stevenlwk/LAGCOM). Researchers working on any problem about shooter games can obtain and modify the project to their needs. The problem of SBC prompts the suggestion of abandoning lag compensation entirely. In this dissertation, we show that the lack of lag compensation significantly decreases the mean shooting accuracy. We argue that a better solution should therefore be improving the TLC algorithm instead of eliminating it altogether. Limiting the distance players can be rolled back is a good starting point. Eventually, lag compensation should move away from using indirect indicators such as a player's latency, and start deciding based on whether a player is/will be shot behind covers or not.
Subjects: Hong Kong Polytechnic University -- Dissertations
Video games
Video games -- Programming
Pages: xvii, 100 pages : illustrations
Appears in Collections:Thesis

Show full item record

Page views

106
Last Week
1
Last month
Citations as of Mar 24, 2024

Google ScholarTM

Check


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.