Professional Documents
Culture Documents
Stealth Games
Stealth Games
stealth games
Paul-Ionut Arsenie
University of Derby, Computer Games Programming
Course: Games Behaviour
100369891@unimail.derby.ac.uk
Contents
1. Introduction:..................................................................................................................................2
2. Explanation:...................................................................................................................................2
2.1 Sound:..........................................................................................................................................3
2.2 Visual:.........................................................................................................................................3
2.3 Awareness level:..........................................................................................................................3
3. State of the Art:.............................................................................................................................3
4. Recommendations and Conclusion................................................................................................4
5. Bibliography..................................................................................................................................5
1. Introduction:
Artificial intelligence and its implementation in games makes NPCs looks more realistic and
it cover decision making process of them. In stealth games, AI is used so intense that it gives us the
impression that the NPC have senses (sight and sound), each sense giving them a state of awareness
as the challenge to the player is to move from a location to another undetected.
In a game with stealth mode, breaking the reality can be the fact that determinates if you will
play or not anymore (Rabin, Game AI PRO 2, 2015). Being one of the most realistic genre of games,
using a high implementation of AI, some issues comes along with it, as the details can easily break the
fragile reality this genre of game achieve. Playing stealth games or games with stealth mode (Metal
Gear Solid, Hitman, Dishonoured, Assassin’s Creed, etc.) the player will notice that the quality of
NPC senses is different from a game to another (as in one game sound sense is better the sight sense
and vice versa) and the awareness of the player can be for a very short time. Is it possible to make a
stealth game with a high implementation of senses? Can the NPCs character act like a human once
their awareness level increase?
The challenge was always to create the most realistic game, giving to the NPC all senses to
make him exactly like a human being. AI should act as intelligent as a human, but it means that it
should do mistakes as well. (Schwab, 2004) In the document below, we will discuss about the issues
that does not let the stealth games to be 100% realistic and about the limitations that could occur if all
the algorithms will be implemented in order to create a perfect game.
2. Explanation:
As the states before example the unsatisfactory uses of the algorithms for sense and how the
state of awareness is insufficient to achieve a perfect realistic game, we will discuss about each of
them apart and try to find a solution.
As it can be seen in the image below (Fig 1.0), in order to create a realistic NPCs, it needs
things to see and things to hear, so the game needs sounds source and visual source. Having these two
important things, the NPCs achieve visual sense and sound sense (Rabin, Game AI PRO, 2013), fact
that makes our NPCs act like a real human.
(Fig 1.0 from: Rabin, 2013)
2.1 Sound:
To create a sound sense, a pathfinding from the noise source to the position of the agent is created
(Rabin, Game AI PRO, 2013). In this way the NPCs detect the position of the player, and even track
him by footsteps sound. The distance between the NPC and source point must be calculated
accurately, otherwise, bugs like hearing through walls can occur (Rabin, Game AI PRO, 2013), fact
that makes the sound sense algorithm even complex.
Furthermore, with a sound radius implemented, NPCs should detect any object that make noise
around him, not only the footsteps, fact that take us to the first issue. Even if many games have
implemented the idea of throwing object to distract the NPCs, it is distracted just by thrown object,
not by dropped one as well (Ubisoft, 2015). Based on the experience form the game it can be noticed
that NPCs should hear any object that make noise but most of them just pass unnoticed, fact that
breaks some important aspects of the reality created in the game.
2.2 Visual:
Based on Blacklist experience, when a player enters the visual cone of an NPC, a timer starts,
scaled with the distance between them, when it reach 0, the NPC immediately act (Rabin, Game AI
PRO 2, 2015), it represent a very good example of how vision cone works, and how the visual sense
is implemented for NPCs.
On the other hand, a critical issue is represented by shadows as it is a part of the graphics settings,
it can be turned on and off. The algorithms implemented for visual cone interact directly with objects
or characters, so as the shadow is a part of the graphics settings, it does not represent a physic element
in the game, resulting an inexistence of the interaction between the visual cone and the shadow, fact
that makes the NPCs to let the shadow pass unnoticed.
5. Bibliography
Rabin, S. (2013). Game AI PRO.