I've noticed a distinct lack of variety amongst the evil we encounter during our many adventures. During my travels I encounter a great many enemies, but if there is any discernable difference between them and those I have already dispatched I am often unable to find it. It would seem that whatever being or force brought these creatures into existence simply made one of each race of enemy and then cloned it to create the armies that now infest the land; I see no evidence that there are any young or any elderly, or that any are shorter/taller or fatter/thinner than others, or even that any seem to have a different sense of fashion. They move in the same way and predictably make the same decisions as thier brethren when placed in the same situation. In some of my more recent adventures I have even noticed the odd occurance of seemingly smarter, stronger or more powerful enemies who seem be still be clones of the same enemy but with different coloured skin and/or clothing.

---

My question to you, good forum members is if you see any merit in attempting to provide more variety to our game worlds by solving this problem. Obviously back when Super Mario Brothers was new there were some pretty extreme limitations on memory, but this is a problem that is significantly reduced for us in the current day-and-age. There's also the argument to be made that producing variations on our enemies creates a significant amount of additional work for the art department -- and this is a very valid point which may well apply to some games, particularly hobbyist and indie efforts more so than industry titles -- but there are ways to work around this. With 3d models we're able to programmatically alter small details of our characters without causing a loss of quality; it would be relatively easy to produce slight variations in height and weight of our enemies for example. There are also games for which this wouldn't be appropriate, and having all enemies of a certain type appear identical does have it's uses. In an RTS for example the player needs to quickly and correctly discern the type (and in some cases condition) of his/her own and enemy units without having to think about it. Varying the units in this type of game would just make this more difficult and would likely result in mistakes identifying the type of units. Your thoughts?

Ultimately, the issue will have to be resolved. At some point, a lack of some sort of procedural methodology will make even the development of that first koopa disproportionately time consuming. Artists will have to use procedural tools to automate the menial tasks associated with content production. The logical extension is that procedural programming involved in the use of the tools can then be applied to finished art products - manipulable models based on an archetype.

Right now, games are developed in a manner akin to movies. You list out the assets you need developed for your static product, produce them, and the package and ship your game. Makes sense for a movie - you work for a few years on what will become two hours of footage, you send it out the door, and then you pretty much worry about PR.

Except that it works differently for games. You're not just working on a story about a handful of people that will provide two hours of non interactive content. Your product is not static: there will be patches and fixes and changes over time. The entire setup of a game inclines itself to procedural production. Budget and time constraints will hopefully cause developers to realize this. Movies are a painting: write once, film it, finished product. Games are more akin to a machine - you build it to do most of the work for you, you maintain it, you extend it.

The phenomenon of the "different enemy, same model" comes from the flaws in the production paradigm. You shouldn't be trying to build all the characters that will be in your game. That'd be like trying to grow all the extras that will be in your movie from embryos. You should develop a system for generating those extras in a game: "programmtically alter small details without causing a loss of quality;." In a movie you just pull them "off the shelf" and clothe them mass production style. In a game, it's as if you can grow them from embryos. That ability needs to be capitalized on, for the sake of better games and more manageable production.

I don't think that the issue is on the production side of things since it shouldn't take a skilled artist too long to make slight deformation changes to a model or create bitmap variations if their layers are setup correctly in Photoshop. Most often, it is probably an issue of performance and expense. Characters can require multiple materials, each requiring several bitmaps (diffuse, specular, normal, ambient occlusion, self-illumination, etc.) at resolutions such as 512x512 or 1024x1024. To make just one character variation will require a new skinned model that the engine has to keep track of plus several unique high-resolution textures. In some cases the bitmap issue can be circumvented through the use of decals, but decals still demand extra image map space and they have to be skinned to the character somehow.

As an artist I can say that creating a character variation for a high res character might only take 2-8 hours depending on the degree of change, which doesn't seem like a big deterrent for designers/project managers not to consider allocating time to create variations. It seems that the computational expense is the primary reason not to pursue this approach.

I do however agree though that the lack of distinction between NPC’s in games breaks the suspension of disbelief, and it seems like there should be economical ways to allow for greater variety among characters.

Back in renderware 3.5, they had a concept called 'RWCrowd', which was basically just a way of getting the developer to pay for interchangable body-parts :)

Assuming you had a a fairly decent set of body-parts for a 'paper-doll' system, you could assemble a unique looking character trivially... Most RPGs already do something like that, of course, but by taking inspiration from something like Sims2's face creator, you could generate large sets of NPCs.

There might be a space out there for that as a 3rd party tool, combining artist controlled assets with dynamic details.

Allan

I've only occasionally found the identical nature of most game enemies to be irritating, and that's usually when it's so blatant it's hard not to notice - for example in Wizardry 8 where you can be attacked by a group of eight bandits who all look identical.

Even in RPGs where you'd think it'd be jarring to repeat NPC models I find it more amusing then annoying; sometimes I like to think of the models as a cast of theatre actors so it's okay that the mayor of Betaberg has the same face and voice as the beggar in Alphaville.

That's not to say that I wouldn't mind seeing more variety in the characters in games, but I'm not sure how much of a problem it is.

My main concern is with the point you made about the easy identification of enemies. While I'm in favour of aesthetics it can't come at the expense of functionality, and being able to tell at a glance what a character is is important for a wide variety of games. Note that this is an issue for the general choice in art for characters not just for changes in each individual model; I like seeing solid choices in colour and shape for characters that make it very easy to tell its function. For military units however you've got the benefit that everyone is in uniform so this problem is less of an issue.

I've been wanting to experiment with customisable avatars in games although it's very far down on my checklist of things to do (building a basic 2D engine is more of priority [smile]). I'm hoping that if I ever properly integrate 2D vector graphics into a game engine so that the engine itself draws the vector images (rather than having to go through the intermediatory of a graphics program translating vectors to raster) then I could experiment with computer modification to those vector images leading to more customisable 2D avatars.

Large similarities between enemies is often a positive and realistic thing. When you hunt, don't expect to be able to identify specific instances of the animal (well, unless you're skilled enough to do so). If you storm an FBI security center, expect all suits. But if you wander onto a beach, there's your variety. Micro bikini, short shorts, etc. If we're talking freckles and 1 inch height differences, I believe that only those with a very sharp eye for detail would appreciate it. If you're Mario stalking koopas, they would all look the same during the 4 seconds that pass before you stomp them into the ground and mug their coinage.

If you're constructing a beach, then I would rather have motorbikes riding past and large buildings reaching into the skyline next to me than different color shorts on passerbys. That goes toward artists and designers. But if the shorts are cheap or free, then of course. It's better, but not significantly for me.

PS. Goomba are not evil. They're like Jaffa. A slave army.

It actually goes into production wise just like movies. If you know something about movies you know that many times in a mass off people it will be a group of extras dubbed over and multiplied to give the illusion of a large crowd. I'm pretty sure clones of the same enemy are to convey a sense of inhabitants and not individuals. I don't think the trend will stop because thats more than enough to convey to the player. It saves time, money, and helps you get your point across.

None of these arguments really act in defense of what MMORPGs suffer. I.e. look at WoW or EQ, the same kind of model is reused again and again and again without ANY variation. At least UO would COLOR their models (even if it was done poorly :-/ )

I say this because WoW doesn't need to preload ALL of its models in to memory, but none the less, when I go to one side of the world there is a Creeping Crumster that is tall black and represents death. Then I go to the other side, and there is the Swamp Master of Doom, and its tall and black and represents the bottom of the swamp. Same model, same texture, different names and identities. Poor.

Of course there is merit to this problem, but is it a good one to solve? is another question. Taking this to the extreme say you have a game which has 100% totally different and unique enemies you encounter the whole way, would this game be any fun? Probably not because everytime you battle an enemy you will die as you have no experience, no way to know how to beat it, it would be very frustrating.

Remembering how to beat an enemy and then defeating it rewards the player with a little cookie for doing good, this is what Super Mario enemies are, a bunch of cookies all lined up ready to eat, and when the player is done, they feel good.

So creating unique enemies could be fun, or it might really suck, it highly depends on the game and implementation. I would guess having some variability with a common thread throughout the game would be the best move, as is already done in almost every game.

I don't think it's a big deal, but I do think it's something that will gradually get addressed. Oblivion has the procedural face thing which is good, and the game I'm working on (and probably countless others) has varying height and girth values for characters on top of other customisable aspects. It's non-trivial though, because people quickly notice if a model is just geometrically scaled up or down or shaded in a different colour, so you need to put a fair bit of thought into how the components are made and how they each vary.