Why in modern video games are all the enemies still clones?

June 9, 2009 9:49 AM

Why, in video games, am I moving through an endless variety of environments but killing the same enemies over and over?

I think this may be changing in the current generation of games, but if you go back just a few years to something like Half-Life 2, you have these games with enormous amounts of work put into creating level after level of unique environments, but then you have just a handful of models and skins for enemies and NPCs. You never see the same place twice, but you see the same characters over and over.

Why the disparity? Does it really take more work to make a single enemy than to make an entire level? Is this just an accepted convention carried over from the days when computers and consoles only had a tiny amount of room for character sprites?

I'd think that even if making functionally distinct enemies was too hard, they could at least create some (random?) variation in the size/shape of the models or in the skins to make the characters as unique as the environments.

And then there's games like Oblivion and Assassin's Creed where they do have lots of unique individual people but they keep saying the same handful of dialogue over and over. Wouldn't hiring a writer to make more dialogue cost about 1% of what they paid the artists to make all the different people? AC is particularly weird because you have different actors reading the same dialogue (when you save a woman from the guards, she'll give you one of three speeches, read by one of a dozen actors) so it's not like they were working within limits on disk space for the audio.

posted by straight to technology (17 comments total)

Does it really take more work to make a single enemy than to make an entire level?

Not usually, but consider the number of enemies in a typical game vs. the number of levels. There's your basic answer.

Also, a level you see over and over. An enemy might show up for a second before he's dead and gone from your mind forever.
posted by ignignokt at 9:56 AM on June 9

3D characters represent a lot of work in concepting, modeling, rigging, animation, effects, AI programming, scripting, etc.

Animation is really the killer time-suck here. Want the enemy to react appropriately to a shotgun blast? Animation and animation scripting. Want to do a judo throw on an enemy? That's a link-up animation, which increases the overall cost significantly.

Unless there's a lot of destructible, animated or interactive elements, environmental modeling and effects just aren't not as costly.
posted by Cool Papa Bell at 10:03 AM on June 9

Levels are static. There are many more dependencies for an animated character with AI.
posted by gnutron at 10:04 AM on June 9

These are mooks. They're disposable. Just like 90% of the guys Arnold kills in Commando, their only purpose is to provide a vague flavor of danger and die on command.

Plus, I fully expect a new matrix to arrive for game reviews soon, something combining the log of polygons rendered per second with number of head explosions, perhaps factoring in the number of weapons you can kill someone with and the various shades of blood. Dialogue? Not so much.
posted by adipocere at 10:04 AM on June 9

Level design is relatively cheap, easy, and can be rapidly duplicated; voice characterization and individual character animation is hard, expensive, and unique. A level can last for several minutes, with multiple perspectives and rooms; a character may only last a few seconds before you blast his head off, so there's a strong temptation to use an army of clones. Realistic environments are relatively unchanging and static, and any animated components tend to be repetitive and rhythmic; realistic individual actions and features tend to be, well, individualistic and quirky, and so must be custom-tailored, absorbing the time of artists and developers and raising budgets.

Even the free, open source games community tends to reflect this: look at the popularity and relative speed of level creation versus the years taken to create complete do-overs like Black Mesa.

Finally, creating unique "goals" for NPCs is difficult: it's a lot easier to put them on rails and send them towards the player to be blasted. (Call of Duty: World At War was notorious for this, to use a recent example). Having NPC's navigate their way through obstacles and react to both allies and the actions of the enemy requires clever programming and some substantial CPU cycles.

As always, software and technology filters from the high-end to the consumer level, and you will find technology like Massive used in games more and more to create more individualized opponents. We've come a long way from Space Invaders, but the mentality behind that game - create lots of duplicate opponents; increase difficulty with speed of attack, rather than intelligence; throw with the occasional unique boss for variety - still pervades the industry, for reasons of culture and economics.
posted by Bora Horza Gobuchul at 10:21 AM on June 9

This isn't exactly a new concept. For ancient examples, take a look at your common chessboard. 50% of your "men" are pairs with unique abilities and looks. The other half are boring pawns that only go (for the most part) in one direction.
posted by sideshow at 10:37 AM on June 9 [2 favorites has favorites]

Enemies are only really seen as obstacles, whereas the level is content. If you want a different kind of obstacle, you make a different enemy. Cosmetic variations are always left behind in the dev process. On top of that, the player gets very very good at recognizing the enemies and variation can make it harder for him/her.

Samey level design makes a game boring quickly. It really is one of the best indications of progress in a game, whereas increased enemy variations adds so little it's never included.

Maybe in a shmup it would be important to have a large number of enemy types, but they are really the content in shmups, rather than the level design itself and they're always designed around what the enemy *does* rather than duplicating an old enemy and painting it green.
posted by Submiqent at 10:38 AM on June 9

What everyone else said.

Also, repetitive elements are a Good Thing, as they allow the gamer to develop rythm and, in a good game, flow, which is the state of mind that will have you hooked to your console all night, forgetting to eat and sleep.

The environment is mostly passive, so the variation is just eye-candy and not too distracting. Having to assess each and every enemy and decide if it is something new, requiring a different strategy, or just a new skin on an old model, would make a game slow, distracting and also very hard.
posted by Dr Dracator at 10:44 AM on June 9

I'd think that even if making functionally distinct enemies was too hard, they could at least create some (random?) variation in the size/shape of the models or in the skins to make the characters as unique as the environments.

This is already happening, particularly for games in which large numbers of random NPCs appear together. In the latest Total War games, for example, soldiers are individuated procedurally (screenshot). Generation of this sort requires a bit of technical overhead, so I think we'll have to wait out another round or two of Moore's Law before the technique becomes widespread.

And then there's games like Oblivion and Assassin's Creed where they do have lots of unique individual people but they keep saying the same handful of dialogue over and over. Wouldn't hiring a writer to make more dialogue cost about 1% of what they paid the artists to make all the different people?

Text is absolutely cheap to create, edit, and deploy. Fully voiced dialogue, which has become a given for modern commercial games, is not.
posted by Iridic at 11:05 AM on June 9

RE: Dialog, you are correct that it is inexpensive to write more dialog. The cost comes with hiring voice actors to record dialog, both in numbers of actors and time each actor spends in the studio.

However, in the particular case you mention, where different actors are reading the same lines, that's just sloppy, and it happens a lot, typically because the writer is a contractor. They write the lines early on, then it's later thought the lines sound repetitive, and the developers use different voice actors to provide variety, without re-hiring the writers to come up with new lines.

Once you have a unique audio file, there's really no reason to have a different voice actor read a throwaway line the exact same way.
posted by Durhey at 11:14 AM on June 9

In some ways, this is no different from other games and sports, which often combine repetition with novelty. In basketball, you do the same thing (bounce a ball on the floor and try to get it through the hoop), against the same "enemies" (the five people on the other team), through a series of different "levels" (each team's possession of the ball). Ditto for football, baseball, hockey, tennis, etc. etc. As others have already pointed out, chess follows a similar format: you do the same thing (move your pieces to capture the opponent's queen) against the same enemies, through a series of different "levels" (board configurations). Ditto for Monopoly, Risk, etc.

I don't know if this is an explicit part of video game design, but the format of perfecting a particular set of skills against known enemies/opponents within changing contexts seems to be a relative constant in games of all sorts.
posted by googly at 11:16 AM on June 9 [1 favorite has favorites]

What an awesome question. I'm sure you can tell from the variety of answers here already that there are a lot of different reasons for it, and it can vary quite a bit from game to game, but I can break down a few different things to help you understand why.

1.) From an implementation perspective

Setting the game design reasons aside for a moment, there are several purely technical reasons why this is often the case.

- Underlying World Systems: In most games, there is essentially what one might refer to as a "world system". This is something that is, in general (although different studios will handle this differently), designed by the game's designers in tandem with the game's engineering leads or team -- sometimes these teams are fluid (i.e. the lead engineer is also a designer from this standpoint), and sometimes they are not (i.e. everybody plays their distinct role). It really depends on the makeup of the team and strengths of its individual members. In simplistic terms, once the system is implemented, worlds are essentially always the same. They may LOOK wildly different to you, but they have a ground you can stand on. Walls, or buildings in them that obscure your view to a certain extent. A sky, or a ceiling. Objects (or props, like chairs, desks, cars, etc) that behave according to certain rules, and effects (like smog, fog, fire or water). Making one level appear in a forest, with trees everywhere and another level appear in the desert, with rolling sand dunes, is essentially "easy" from an overarching design standpoint. Your designers and artists are hard at work to make this variety, but once the system is set, they are essentially working in a vacuum that does not have a major impact on the game's design, or the other departments working on other systems or features.

While this is a major oversimplication when considering the complexity of modern video games, for all intents and purposes, once a world system is designed, the level designers and artists can set out and do pretty incredible things with the items available to them in their "toolbox". Obviously, this "world system" can grow and change depending on whether or not there is time to implement additional changes (for example, "we know that we designed this entire game to happen "outside", but we really need one level in a building and therefore, we need to create a system to handle doors, and all of the behavior that will come with opening them, for the player and for our AI").

Consider also the fact that the majority of games that make it to the shelf are developed by studios that have been perfecting and tweaking their world systems for years and years, simply building on an existing foundation -- or they've licensed an engine (such as the Half Life engine, or Unreal engine, or Quake engine) which is consistently being developed simply for this purpose. While worlds will always look visually different, they will almost always be expected to do the same basic things, so this is, in the scheme of things, quite achievable. So say, in version one, you created a world system with a sky and clouds and a ground, etc, and in version two, you added a system for nighttime and daytime, and in version three, you added a system for cities. Etc, etc.

Underlying Character Systems: Again, I am hugely undersimplifying here, but games also have underlying character systems that are designed much in the same way, but there's an entirely different implementation pipeline and skillset and group of people working to create these creatures, design their behavior, and make them come alive. This is going to vary much more intensely depending on the style of gameplay, the setting of the world, the skillsets of the team, and the budget of the game.

It is, generally speaking, a MUCH bigger deal to suddenly decide that you need not just human enemies (that run on two legs, shoot guns with their arms, and say things with their mouths, for example), but quadraped aliens with crazy tentacles and giant biting teeth.

For the most part, the system that you design up front will need to be the system that carries you throughout the rest of the project (this is not the case for all development studios, but it's largely true for many). Adding something new to the plan can cause a huge ripple effect across every discipline of the development team: The character modelers, the art direction, the animation team, the AI programmers, the feature programmers, the game designers, etc.

This is one of the reasons that in many video games you play, you'll see that the variety in enemies comes through often in how their dressed, the shape of the weapon they carry, the amount of damage they deal, or their size. It's also one of the reasons why so many boss fights will feel so "different" -- because they are often uniquely designed one-time encounters that are created outside of the basic character system.

The Size and Makeup of Development Teams: Here's another "I'm generalizing bigtime here" caveat, but essentially, the place where the games come together is in the implementation of the levels. You often have a pretty large group of game designers/world builders who put everything together based on a structured design and a set of rules. This is their skillset and it's the major focus of the development process from a review and planning standpoint (once you're deep in production, most everything gets reviewed in the context of a level).

The same is not true for the team that actually creates the enemies. In a sense the enemies are almost like a micro-level in themselves, as there is a group of folks devoted to making them work -- modelers, animators, designers and programmers -- but this is usually a smaller subset of the team. It takes a lot of people to create the number of complex and detailed characters you see in video games now, and a lot of other people to animate them in various ways that look real and do not seem repetitive or boring, and a lot of other people still to tweak their behavior and make them play against you in a compelling, challenging way.

Looking back into the way I described what is essentially a variation on a single concept ("level") - forest vs. desert, the variation of an enemy ("special forces soldier" vs. "infantry grunt") is not going to appear nearly as distinct to you, but from an implementation standpoint, they are essentially the same thing.

2.) From a Design Standpoint

Even assuming infinite resources and time, huge variety in enemies would cause not just a major design challenge, but could likely dilute the purity and focus of the game.

There is a lot of effort put forth in creating a stable, enjoyable and achievable difficulty curve. This is in my opinion, hands down, the hardest part of making a video game. There is a huge divide in the skillsets of players, so making something that will satisfy your entire audience is very difficult from the get go.

Generally speaking, game designers have a specific set of rules they use with which to design gameplay experiences. These would be, in the simplest terms, the environment you're in (world system) and the enemies you're fighting against (character system). Throwing in too many variables can make designing a level incredibly difficult. Working with five or six character types is quite difficult. Working with 20 could take an incredible amount of time -- to ensure that you actually know how to fight them and will not get frustrated, confused or lost.

Regarding your question about the dialogue in Oblivion, the first thing I can tell you is that the biggest reason you see so many different types of people is because they had a deep, rich character creation system (which you yourself used when you began playing the game!). I didn't work on Oblivion and so I can't say for sure, but I'm certain that while they must have had a huuuuuuuuuge character development team, the people were working to make different types of skins, clothing, hair, and animations, and not in creating the characters themselves. Once the system was developed, there were probably millions of potential permutations, so it was "easy" (if anyone who worked on Oblivion is reading this, I apologize in advance for using the word "easy" as I'm certain it was far, far from it) to have a jaw-droppingly wide variety of characters. In the development scheme of things, it was achievable is what I mean.

Also, there must be one bahazillionblillion lines of dialogue in that game. I was blown away by the amount of it. Imagine all of the logic trees, all of the different responses, and the sheer mind-boggling number of NPCs in that game. It truly blows my mind!

I'm certain I could natter on and on for pages more here, but I think this might give you a bit of insight into why you don't see the variety in enemies like you do in environments. Also, I am really looking forward to the advances in the industry that will make it all the more impressive as we do continue to see huge leaps in the way games are made and the possibilities for wild permutations in character appearance and behavior, all the while creating a satisfying experience within a reasonable amount of time!
posted by pazazygeek at 11:26 AM on June 9 [6 favorites has favorites]

From a player's persepective, it might be difficult to remember the particulars of how to defeat each enemy character if there's a really wide range.
posted by electroboy at 11:41 AM on June 9

It's part of the language of video games. Enemies aren't individual entities, they are classes of beings, and when their appearance changes significantly it's a signal that something about the class has changed in its abilities. If the leader of an enemy squad looks different than his teammates the player intuitively knows to watch out for that one in particular since he might have a stronger weapon or more health. Modern games have been attempting to randomize enemy appearance somewhat, but it always has to be balanced against giving the player the information they need to distinguish classes at a glance and prioritize threats effectively. Left 4 Dead is a good example of a game where the cannon fodder zombies are varied in their appearance to a degree but the special enemies stand out physically and always look the same. From a resource standpoint it also doesn't make sense to craft a one time use enemy unless it is meant to be a memorable boss encounter or a puzzle. Yet another problem is that a lot of the time you simply don't get close enough to the enemy while fighting to appreciate subtle variations in appearance.

Another reason things are done this way is that part of the fun is learning how to adapt to facing the same enemy in different environments where there might be different hazards or tactics at play. It's a theme and variation thing, where the game throws on more complexity once it has taught you how to deal with a specific threat.

For an interesting counter point, check out some playthrough videos of I Wanna Be The Guy. There aren't actually that many enemies in it, but the environment acts in pretty much the same capacity. The reason I suggest it is that almost every challenge in it has its own set of arbitrary rules, which makes it an incredibly frustrating game. For example, in one of the opening rooms apples fall down from trees to kill you, until you mange to jump above them at which point they start falling up. In some ways this is the opposite of your question, where identical elements behave differently instead of vice versa, but it still demonstrates how difficult things can become if you disrupt the normal rhythm of gameplay.
posted by CheshireCat at 11:43 AM on June 9

One issue I didn't see mentioned above is memory and streaming constraints. The comparative memory needed to run an environment texture, a character texture, an animation, an audio file like a line of dialogue, or lighting information is not close to equal across the board. Consequently, each team gets a slice of the memory budget determined by what's most important. Depending on what the main emphasis of the game is, you're going to weight your memory budgets accordingly. Open world games like GTA usually allot a big chunk of memory for environments and vehicles, which is what the player will largely focus on. At any given time, the player can go pretty much anywhere in the world, making it harder to manage memory and streaming. They might allot less for minor NPCs and VO since the player isn't going to interact with them as much. First person shooters might allot more memory for characters and animations, since the environments will be fairly contained and the interactions with those assets will be more "close up." RPGs will have a lot of memory set aside for characters, VO, and audio files, since you spend so much time interacting with NPCs. In other words, depending on what kind of game they're making, the developers will strategize where to cut corners in order to keep the cost of memory under budget.

Another way of directly answering your question is that an individual environment is fairly cheap, since it's usually just the environment textures, the lighting info, and maybe FX and UI. Characters are proportionately more expensive, since in addition to the character textures, they also demand animations, audio, lighting, attachment systems for weapons, FX, possibly UI elements, etc.

Over a few years in the game industry, I've definitely learned the memory management is like the accounting of game dev: the important thing that everybody wants to ignore until we're over budget.
posted by ga$money at 12:01 PM on June 9 [1 favorite has favorites]

In addition to all of the monetary, time, and computing limits that everyone has mentioned quite thoroughly above, there's also a thematic reason why there are only a few different kind of enemies.

Simply put, how many types of enemies are realistic to have? Sure, random variations in facial shape and body size are realistic, and these have been used to some degree in games, but random computation on the fly is probably a bit processing-power heavy.

Having more different kinds of enemies is unrealistic because generally heroes in things only fight a few basic species/model of creature or robot. It's how life is, so it makes sense to reproduce this, even if you might find it a bit tedious.
posted by JauntyFedora at 3:13 PM on June 9

In addition to what has already been said about the complexities of animating and AI, levels have a lot more duplication than you think - many textures used are repeated throughout a number of the levels, it's just the shapes they're painted on that varies. If you take HL2, there's only 3 or 4 really different environments, really (city 17/nova prospekt, the beach and the citadel)

You also approach them in a very linear way, so the 'look' can be tailored to look good from a particular path and angle. Mobile enemies need to work in many uncontrolled situations.
posted by ArkhanJG at 4:54 PM on June 9