How do enemy attacks work with the CAMERA?
Enemy Combat Design VI
This is the first essay series for Signals and Light, and it provides an overview of enemy combat design in action games. It is written from the perspective of a AAA game designer who works with enemies in action games and aims to pay it forward with what he’s learned.
The sixth part of this series explores how different kinds of action games handle the positions of attackers in conjunction with the player’s camera. It covers…
An overview of the most basic elements of how enemies position in combat.
A spectrum between camera-sensitive and camera-insensitive camera styles and some of the basic principles that tend to underlie these approaches.
A brief explanation of how level design can intersect with these choices.
And an analysis of whether the player “should” see what’s going to hit them before it happens (spoilers: it’s complicated).
As established in Part 1, enemy attack cadence is divided into four pillars. The fourth and final pillar is DIRECTION, which refers to where the enemy is relative to the player and the camera when their attacks activate.
(See Part 2 for a full explanation of the enemy attack breakdown.)
Like concurrency (Part 5), Direction is ALSO determined before an individual attack actually activates (it’s basically part of the Assignment of Attacker Role step, but it’s broken out here for clarity). However, it strongly impacts the timing, proximity, and concurrency elements of the attack and the player’s understanding of those elements.
Attack direction is a confluence of the forward vector and location of the attacker relative to the forward vector and location of the player character as well as to the forward vector and view of the game camera.1 The words might sound mathy, but they boil down to: “Where in the player’s field of view is the thing that is hitting them?”
VI-A - enemy positioning
How enemies position themselves around the player during combat is the first thing to consider when it comes to the direction of enemy attacks. Some types of enemies can have fixed positions, but more often, enemies stand and move relative to the current player position. As the player moves, the enemies follow. Some games don’t care if enemies crowd or stumble into each other, but most games use a variety of AI location management tools to handle how enemies move relative to the player and to each other.
In ranged combat games, enemy positioning often involves claiming areas in the arena, such as perches or cover locations. In melee combat games, sometimes developers use a “combat circle,” a series of points around the player that the closest enemies will gravitate towards in order to evenly distribute themselves in a way that looks smart. These systems can have extra layers of rules on top of them. For instance, some games have enemies prioritize moving onto an on-screen position before they attack as the player rotates their camera.
In this talk from Sony Santa Monica regarding God of War (2018), the developers explain a complex AI location manager that the team created to handle the placement of enemies around the arena relative to Kratos and the specific camera style that they went with. It was complex enough that it used a variety of regions and shapes to help the player understand where enemies are without having to always look at them, including splitting the world around Kratos into quadrants so that enemies would remain in relatively the same area of the arena even as the player runs around.
There are a myriad of approaches to this concept, but many of them are not obvious unless the developers explain them. (Everything would be so much easier if they’d just publish their development documentation.)
VI-B - action game camera types
Regardless of where enemies stand, usually, their attack direction is “forward” and “towards the player.” In more complex action games, sometimes enemies move erratically to create unique experiences, but, for the most part, there’s little to say about the literal direction of enemy attacks. What matters most for attack directionality is actually the player’s camera—where it is and what it is able to see.
Without dwelling on them, here are the main types of game camera.
Among each type, there are many approaches to field of view, rotation rate, distance from the characters, and many more factors that one could dedicate entire essays to, but that’s not relevant here. In action games, the camera follows the character, OR it gives a nearly complete view of the currently relevant gameplay space.
VI-C - camera-sensitive combat
When it comes to how cameras serve the player’s understanding of combat with enemies, the most obvious first candidates are games where the player can see all enemies at once or see everything they NEED to see in combat at once. In these games, an attack that matters to the player is always going to be visible when it activates, allowing a near-perfect understanding of the threats that the player interacts with. Usually, this is achieved by having a top-down or 2D camera that, by default, shows all threatening enemies to the player at once.
There are, however, some games in this category that might be surprising for those who haven’t noticed it. In games where the camera cannot see all enemies at once, developers are obligated to consider what their approach to off-screen attackers is going to be, and some developers decide to provide a simple answer: enemies who are off-screen cannot attack.
This approach is in some third-person character action games. In these games, the player is meant to take the spotlight in combat and feel highly empowered, with the skill ceiling being a place where the player endlessly creates attack combos and effortlessly defends against groups of enemies. They’re fast-paced, and the ability to see all enemy actions at once helps to facilitate that pace and keep the player in the spotlight.
VI-D - camera-partially-sensitive combat
Following that principle, it’s easy to assume that a variety of third-person and first-person action games probably have a priority system for assigning attackers based on the player’s current camera orientation or have systems to encourage the enemies to try to stay on screen. In both these cases, the enemies are still ALLOWED to attack from off-screen, but it’s just less likely to happen. This can be achieved by making off-screen attacks cost more attack tickets, or there may be a high-level attacker assignment system, as discussed with the enemy attack concurrency management tools in Part 5.
This approach is in a few first-person games where the developers designed enemies, arenas, and player toolkits around the idea that the front of the player is the most important gameplay area. These games often have self-contained arenas that the player traverses around in loops or corridor arenas that widen and tighten as the player constantly presses forward without having to think about enemies spawning behind them.
It’s almost impossible to say which games prioritize on-screen attackers without the developers releasing detailed information about their AI systems. One can often intuit whether or not it’s happening by paying attention to how often off-screen attacks occur, but it’s hard to highlight a lot of specific games here because any assumptions of how they work could be wrong.
VI-E - camera-insensitive combat
The remaining approach is…everything else. It doesn’t appear as though these games give much, if any, prioritization to on-screen attackers, and combat becomes a free-for-all with regard to the current direction that enemies are attacking from. These games may have plenty of other enemy attack cadence systems up their sleeves, but they set forth a challenge for the player to remain aware of threats remaining in the arena and keep track of those threats even if they can’t see what the threats are doing.
The player’s ability to pull this off depends on a few things. Obviously, enemy positioning plays a huge role in where off-screen attacks can come from. In games where the player can easily manipulate enemy positions and enemies are less mobile, it’s easy to anticipate the actions of off-screen enemies or manipulate the group so that enemies rarely end up off-screen. In much more dynamic and frenetic games that use ALL of the arena in complex ways, it’s impossible for the player to control the enemy positions, and there will often be some or even a majority of enemies who are not on screen at any given moment. Also of note, some horror games actually highlight the unpredictability of off-screen attackers at the heart of their design to create suspense and fear, elevating the importance of audio cues to direct the player’s attention.
In the past decade, more frenetic third-person games have started to lean on one very specific solution to the player being able to handle off-screen attacks, and that is with an attack warning system. Attack warning systems come in a few forms that tie to the game’s camera style and general enemy attack design. Some games center the warnings on individual attackers or directly on the player character, keeping attention largely on-screen for a visceral focus. More complex attack warning systems, on the other hand, involve a directional indicator that tells the player where an attack is coming from when it's off-screen, even if the enemy is far away. Additionally, any type of these indicators may have multiple visual stages for the player to identify where the attack is between its initial anticipation and its actual activation.
How important enemy attack indicator systems are is up for debate and depends on the game. On the one hand, attack indicators can just be a form of accessibility for players who want less of a challenge, and skilled players should be able to control their camera and listen for audio cues to avoid off-screen attacks. On the other hand, a game might be so chaotic that attack indicators feel like a necessary feedback mechanism because no amount of skill or understanding would allow the player to infer when off-screen attacks are occurring.
VI-F - direction and level design
Attack direction links to level design in several ways. This part of the essay cheated somewhat in tying enemy attack direction to camera style because, of course, a game’s camera style dictates literally everything about a game genre. But after the developers decide on a camera (probably as part of the pitch for the style of game they want to make), they do have to consider what kinds of arenas they intend to design. The high-level choice here arguably falls into a spectrum between two extremes—arena-focused combat and frontal-focused combat. Some games fall at either edge, but many action games use at least some of each.
At one end of the spectrum is the “corridor style” of level design. In this type of layout, there are constrained spaces and fewer open arenas, the structure informs enemy positions in the level, and the player has more autonomy in engaging one or more enemies in an encounter. This all puts some power in the player’s hands to influence the directionality of incoming enemy attacks. This approach is in many genres, including cover shooters, stealth games, intricately designed Soulslikes, or 2D Metroidvanias.
At the other end of the spectrum is the “arena style” of level design. Of course, a game can have both literal corridors and arenas, but the distinction lies in what types of spaces the player spends most of their time. Arenas leave the gameplay experience open to the flow of combat. They are inherently more chaotic and unpredictable, and so accounting for enemy positioning and directionality in the game design comes through behind-the-scenes systems and enemy design rather than in hand-placed authorship. Chaos isn’t inherently bad depending on the game design, but for games that want to favor the player more, they might include some of the more refined methods described here to manage enemy positioning in order to make the flow of combat more coherent.
VI-G - discussion: should the player see what’s hitting them?
THIS ESSAY PROPOSES…that…answering this question is hard.
In discussing this, the essay finally wraps back around to the initial complaints that led the essay in the first place—why are players getting hit in the back all the time?
How punishing it is to get hit from off-screen depends on how the game defines what it means to be hit. It would take an essay of a similar length to this one to fully explain damage systems and hit reaction systems in action games, as that is a huge can of worms with many possible solutions. Setting the full topic aside, to put it simply, off-screen attacks are more or less punishing depending on how the game applies a player penalty to each enemy hit.
For shooters, not being able to see all attackers at once is inevitable. Even if the player sees the attacker at the moment the enemy fires, there’s a large chance the player will look away by the time all of the bullets of the attack land. Therefore, in most first-person and third-person shooters, individual hits are not particularly punishing. The player has a large health reserve or regenerating health, and hits don’t often disable or slow them down. On top of that, these games will usually show a directional indication of where the hits came from as part of the HUD, and these flash DURING the hit because the goal isn’t for the game to warn of each individual hit in advance but for it to convey the direction of the offending enemy shooter.
In games focused on hit stuns or hit reactions (usually third-person or top-down games), an off-screen hit might go so far as to entirely interrupt the player’s current action. Hit reactions include things like flinches, stumbles, or even knockdowns, disabling the player for anywhere from half a second to three or more seconds, depending on the severity of the attack. Even if the attack doesn’t do a lot of damage, the interruption to the player’s state means majorly disrupting their flow and ability to deal damage and defend from subsequent attacks. It sounds mean, but these interruption-focused games are fun, as the proportionate punishment of an enemy hit is what makes the enemies appear as immensely powerful opponents.
The best games that rely on level design to inform enemy attack directionality make off-screen attacks something the player learns how to avoid, and the enemy placements help to train them to gain this awareness. There are then directional arrow indicators in some open-arena games with frantic combat because the designers understand that it is their obligation to help the player avoid attacks. However, an arrow is nowhere near as informative as seeing the actual attacking entity, as an arrow doesn’t tell the player what kind of attack it is or even how far away the attacker is. The arrow might have a few different visual states to differentiate when the attacker moves into range versus when the attacker activates the attack, but without rotating their camera, the player has no understanding whatsoever of what is actually going to happen.6
Out of everything presented so far, this is the first moment that may sound like complaining or criticizing, but that’s not the point of this essay. There is no EASY solution to this problem. The games that rely on directional attack indicators or that use open combat arenas where enemies attack from a lot of angles are not invalid, these are immensely fun games with many passionate fans.
The reason these games are the way that they are is a result of a confluence of a lot of design decisions, and in the most frenetic games, one of those decisions is an ambitious striving to make it so that the player is not TOO POWERFUL with the tools and abilities that the game offers. It’s like a combat design scope creep between enemy attack cadence and player design, all to keep up with the absurd spectacle of powers granted to the player.
Gamers WANT to have those absurd spectacular powers, but they also want some form of a challenge from the enemies, and these games have basically backed themselves into a corner that they can’t extricate themselves from without radical shifts to their combat design, camera design, or other sacrifices to their design intent. The enemies become more powerful, including attacking often from off-screen, because if they didn’t, the game would potentially feel too easy.
So…it’s hard to say what the “right” approach here is when it’s clear that even the best approach can lead to some amount of player frustration or at least uncertainty. Sometimes, off-screen attacks are just something that the player must accept and work around, and they’re definitely not going away. Perhaps developers could sometimes step back to think about how their cameras interact with their enemy attack design a bit more, but fully expanding on that is really up to designers and game teams, and this essay isn’t going to try to break down what approaches have worked best since all the games referenced here have tons of fans who enjoy them regardless of their approach to this aspect of enemy attack design.
[NOTE: Big TBA on when Part 7 will be released. Working on another project, and this took about a year, so it could be as late as sometime in 2025. Sorry! The author only has so much spare time outside of game dev. >_<.]
NOTE: The explanation uses the word “forward vector” over and over again not to sound smart, but because vector math is probably >=75% of the math that matters in gameplay development and rotation math can fully go to hell.
They represent the game as though the player is seeing it through the eyes of the character that they control.
They give an over-the-shoulder or above-the-head view of the world that the player can rotate or otherwise control, letting them see their player character or their party easily while controlling them.
They are typically called 2D cameras, even when the game isn’t literally 2D, but generally, these are used when the character is constrained to move on a two-dimensional axis through the game world, though there are a few exceptions where there is an axis of depth.
They give some form of bird’s eye view of the game and vary significantly in their implementation. Isometric games tilt the camera, truly top-down games look straight down and may be two-dimensional or even represent characters in a way completely detached from reality, and some top-down games let the player rotate the camera and function like a very zoomed-out third-person game.
And yes, God of War: Ragnarok has a button that lets the player do an instant 180 rotation, but this is a strange concession that expects players to execute a gameplay verb that has no analog within the genre and that players rarely or never expect to do, and analytics probably show that the majority of players do not press this button.