Experienced Points: The Big Cost of Small Places

By Shamus
on Jan 13, 2012
Filed under:


During the recent thread on Campster’s Half-Life commentary, ps238principal asked about the cost of making games larger. My response was a bit rambly and unfocused, so I expanded it into a column that simply discusses how art production pipelines have become more complex in the last 11 years.

Enjoyed this post? Please share!

A Hundred!17117 comments. Quick! Add another to see if this message changes!

From the Archives:

  1. noahpocalypse says:

    I ++ && <3 this post. Awesome.

  2. Fang says:

    I’m just surprised that you made an article. You haven’t done that in a while.

  3. Exetera says:

    You mentioned in your article that it matters more in modern games that you be able to see reflections… actually, Deus Ex: Human Revolution’s engine can’t do real reflections. In places where there are reflections, they’re very, very bogus. For example, the cars in Detroit all appear to have reflections off of their shiny surfaces. However, they’re actually wrong – they all show a reflection of an information display near Sarif HQ, which would be correct for exactly one car in the level. The rest just reuse that texture.

    I personally find this humorous, because the original Deus Ex could do reflections, and they were used rather often. Mirror-polished floors and bathroom mirrors were fairly commonplace in the universe of the original Deus Ex, whereas in Human Revolution there are no reflective floors and the bathroom mirrors are either broken or covered in ads.

    • Shamus says:

      Mirrors seemed to vanish in the early part of the last decade. Those old mirrors used to work by rendering the mirror rectangle itself. Then literally mirror the entire scene around the plane of the mirror, draw the scene where it intersects with the rectangle, then un-mirror it and draw everything AGAIN. This was actually kind of cheap to do, because the CPU had done a ton of work that you wouldn’t need to repeat for that second pass.

      This worked with those old single-pass rendering pipelines, but it’s probably madness to do this with today’s multi-stage pipelines. There are just so many steps to rendering (drawing transparent stuff, multiple layers of bump mapping, alpha-sorted bits for hair and clothing edges, plus different shader combinations for plastic, metal, or stone – type surfaces. Doing all of that twice in a frame would, without fail, double the GPU load. In most cases, this ends up cutting the framerate in half.

      It’s not impossible, it’s just that if you want a game that can do reflections then you need a game that only uses half of the available power so that when a mirror shows up there’s enough power left over to draw it.

      What’s odd is that Half-Life 2 had full water reflections (with ripple effect!) and Crysis 2 had cheapo cube map reflections like the kind you mentioned in DX:HR.

      • Alan says:

        I’d actually been wondering about that, and that makes sense.

        It would be nice to put mirrors in controlled situations where your scene is half as complex as usual. Your average video game bathroom isn’t the most graphically complex scene. You’d need to play some games to ensure that outside the bathroom and the mirror couldn’t be seen at the same time, but it seems feasible. And mirrors in bathrooms would do wonders for my suspension of disbelief.

        • Rosseloh says:

          I don’t consider myself a bathroom-designer, but I think that would be really easy, too. Lots of bathrooms I’ve been in have the door in a short hallway of its own, often at right-angles to the walkway between the stalls and sinks.

      • Adam P says:

        I can name two games on the N64 that had reflections: Super Mario 64, and Perfect Dark.

        In SM64, there’s a room with a giant mirror. You can see the Lakitu that’s holding the camera! I’m positive that Nintendo pulled this off by just doubling the geometry, inserting a semi-transparent plane between them, and then having two extra characters (Mario and Lakitu reflections) respond to player input.

        In Perfect Dark, there were a few shiny, reflective floors and a number of puddles scattered about. The game had an x-ray scope (which did some sort of fancy effect to make everything look shades of green and red), which revealed that these floor reflections were indeed just mirrored geometry revealed through semi-transparent floor textures. And Perfect Dark required the Expansion Pack (an extra 4MB of memory!), an accessory which wasn’t around when SM64 was released… which is why I’m certain SM64 used the same trick.

        It blew my mind when I figured this out on my own a few years ago. Of course, this trick wouldn’t work in modern games, because transparent surfaces are often very thin and have stuff on the other side.

        From what I’ve seen in modern games, “reflective” surfaces are often kinda dirty. So it seems to me that the engine guys could just do a pass on geometry and textures. With all the grime on the “shiny” I doubt anyone will be able to say that some detail is missing.

      • MichaelG says:

        HL2 water reflections were kind of bogus, actually. It would reflect parts of the landscape, but not others. There would be docks, buildings, rocks, etc. missing. Very strange once you looked for it.

      • JPH says:

        I always figured the real reason why they stopped using mirrors was because seeing your character move around through the mirror in a first person game looks horribly artificial (see Deus Ex Vanilla).

      • Phoenix says:

        The water in half life 2 is so beautiful…

      • ivoras says:

        I obviously don’t know what I’m talking about but aren’t small mirrors (like those in bathrooms) special enough that you could apply any number of tricks to effectively compose them from pre-rendered textures? Especially since todays games still have this huge separation of what is scenary and what is a gameplay object. I imagine you could pre-render (at level load time) a single bitmap containing the whole possible static scenary as reflected in the mirror, then use common panning / scrolling to approximately correct for the player’s angle to the mirror, and let common perspective correction on this rectangle do the rest. If you wanted to be fancy, maybe you could even pre-render the scenary in a few passes, depending on the object’s distance to the mirror, and then do some parallax scrolling (of course, this would work only approximatly, but bathrooms are small and not very complex). On top of all that you could then render the player (if visible in the reflection). Of course this wouldn’t work for scenes with many dynamic objects or for large rooms.

    • X2Eliah says:

      Yeah, the game devs discussed this in one of their audio mini-podcasts, in response as to why there really aren’t any mirrors around. According to them, it was directly because of the limited amount of RAM on consoles, and near the end of development they were forced to choose between ‘more detail in the world’ or ‘mirror reflections’. They went with the detail, and were left with not enough RAM to have the mirrors.

      Since the PC-port was done out of the original game’s resources without changing anything regarding look/art style/functionality, they left the no-mirrors concept.

    • Rigonzo says:

      I did think it was rather odd that there was such a big deal made of the fact that Jensen’s apartment mirror was smashed. I’m not very far in to DX:HR, as it keeps crashing on me, so I don’t know if it gets further attention down the line. I also found the “mirror function disabled” screen in the Sarif Labs toilet amusing but somewhat jarring.

      To be honest, I think if they had just papered over their inability to render mirrors and designed the bathrooms without mirrors, I wouldn’t have even noticed that they weren’t there.

  4. zob says:

    AI has come a long way in the last ten years, but it hasn’t come so far that the enemies can figure this stuff out for themselves.

    Actually FEAR was doing this. Then plague of consolitis spread around the land.

    edit: Also this just occured to me. You do remember Carmack’s speech about Rage and Consoles, right? There are some serious bottlenecks in console architecture and I am guessing thats also a limiting factor in AI design.

    • Shamus says:

      As I understand it, FEAR had a LOT of environmental hints. It’s actually a worst-case scenario for this sort of thing. The level designer had to set up tags for where foes could use ladders, crawl under things, and vault over walls.

      The fact that the AI used these elements wisely is to the game’s credit, but those hints still needed to be there.

      Maybe this is why the AI seems so much dumber in subsequent titles.

      • ps238principal says:

        That explains a lot about how FEAR played. It also used a military opponent to make all the units use the same models.

        While I thought the combat was fun, the environment seemed a bit of a step backwards when it came to interactivity, at least when compared to Half-Life 2 and just about any other game where you could destroy nearly every box and every pane of glass.

        • Tizzy says:

          Hmpf! I have yet to see a shooter that isn’t overrun with bulletproof glass everywhere.

          • ps238principal says:

            Very true, but I see more shooters have an all-or-nothing approach: All windows are breakable, or they’re not. All boxes might as well be made out of adamantium held in place by a teaspoon of dwarf star, or you can smashy-smashy to your heart’s content.

            It’s when there’s some that are and some that aren’t, it’s harder for me to enjoy the game, especially if they’re similar in color, size, etc. I think FEAR had a bunch of ammo/gun crates that I just missed because I didn’t try to destroy them or (more likely) they weren’t near a place where explosions were occurring.

            One example of this kind of thing was in one of the Penumbra games. In nearly every room, light switches were just a wall decal and you couldn’t make them do anything… except in one particular room, but the switch looked like all the others that the game had pretty much been training you to ignore.

          • Daemian Lucifer says:

            Ugh,why did you write that?I hate you now!You have no idea how much I hate you now!!

            Modern warfare has glass that not only shatters,but can be penetrated with multiple bullets before it breaks completely.

            Damn you!Damn you to hell for making me praise that foul series!!

  5. JPH says:

    Sometimes I feel like I’m the only person on Earth who’s happy with the length of time games occupy nowadays. Well, with most games I play, anyway. I imagine if I’d bought Modern Warfare 3 I probably wouldn’t be happy with the size of its single player campaign, but Human Revolution more than satisfied me.

    To me it’s a quality > quantity deal. Yeah, old games tended to be longer, but they were far, far less refined and polished than the games we have today, and I’m not just talking about graphical quality.

    • Daemian Lucifer says:

      I wouldnt be sure about that.Id say that we did a bit of a trade between what was polished back then and what is now.See when people say how original deus ex was huge,they arent talking just about the levels being bigger than in human revolution.They are also talking about what you could do in all those levels,and how npcs would react to much more of those things than they do now.We lost that freedom and responsiveness,but we gained a tighter gameplay.Is any of those better to have?Depends on which you prefer.

      • Tizzy says:

        I do miss the good all days when I feel like games were experimwnting a lot more with mechanics. It often failed in dreadful ways, but you spent a lot more time experimenting: “Gee, I wonder what will happen if I do *this*…”

    • Zagzag says:

      I really really enjoyed Human Revolution, but I was quite happy for it to end when it did. I was thinking as I played it that (with the exception of the boss fights of course) the quality was high enough throughout that making the game longer would surely have impacted on this. As a wise man once said, the trade off between quality and quantity is a perilous one.

    • Sumanai says:

      I don’t think that games need to be longer than they are. In fact, I hold the insane belief that Portal 1 felt too stretched out near the end and wish it were shorter.

      Some games would need larger map sizes in order to feel less claustrophobic, but I can’t really say how many since it’s usually a problem only in first or third person perspective shooters and I don’t play a lot of them.

      To me a far bigger problem than game length is badly done filler in an effort to make the game longer.

  6. LegendaryTeeth says:

    This is why I like indy games and procedurally generated content. I want more stuff that is awesome. It doesn’t have to necessarily look awesome.

  7. Eldiran says:

    Coincidentally I just started re-reading through Experienced Points today… good to see a new article! It’s only a matter of time before someone asks, so I may as well: is this going to become a somewhat regular-ish thing again?

  8. Naota says:

    As I’ve said in the past, one of the biggest casualties of the push for ever more minute detail in video game levels has been entertaining level design. On almost every front, the quality of game levels has either remained exactly where it was 10 years ago or sharply declined as everything was simplified and shrunk down to accommodate the all-powerful deity that is shiny new graphics with painstaking detail.

    Games have all but stopped featuring design that reflects real-world spaces, too. We’re given either linear pathways dressed up and decorated to suit the locale, or open expanses of terrain with the occasional isolated structure plopped down for good measure. In both cases it seems there’s minimal thought given towards level flow or generating an interesting space for gameplay (eg. Fallout 3 and New Vegas have only two gameplay environments – narrow internal hallways which are all the same size and huge open spaces with big structures in the way).

    It’s rare to find buildings designed like they were in the old Thief games, where every room had a believable purpose in the game’s setting and you were free to choose how you traversed the place without also having an entire outside world cluttering up the design goals. Focused design on a big level rather than a sandbox or a linear theme park ride seems to be a lost art, and I really wish it would make a comeback. Human Revolution seems to have understood my woes, but I wonder if it would have ever happened if the original game hadn’t been so expansive in the first place.

    • Paul Spooner says:

      Read your article; good stuff!
      So, basically you’d like designed spaces that are both not linear (one entrance, one exit) and not open (enter or exit anywhere) and which are also conceivably usable in everyday life.

      • Naota says:

        Open is fine so long as there’s enough complexity for an enjoyable area (you need stuff to work with to make a good design – particularly vertical space), but the level itself has to be self-contained. Sandbox worlds like Fallout 3/NV/Skyrim are too big and too open for the designer to micromanage the experience. Players can approach from anywhere and there’s far too much empty space to ensure that all of it is well-designed or makes sense.

        In an isolated level you can put the gigantic labyrinthine city all around the player and miles off into the distance using a few locked doors and a low wall, and so long as they have a big playable space that reflects the city they can see in the backdrop they’ll be happy and you can craft a diverse, multifaceted experience. In a sandbox game you have to create every street and building top of the entire metropolis and allow the player to explore all of it or you’re missing the point of being a sandbox game entirely.

        Case in point: in a game like Thief or Deus Ex you can enter nearly every building in a city block and find something important inside. Each one is a part of a planned and tested gameplay area. In GTA IV or Fallout New Vegas you’re lucky to find more than a single open door or building interior in the same amount of space.

        The verisimilitude of the buildings is more for flavour and immersion, but it does help the player too. There’s no feeling neater than spotting the security room on a game map, going there, and finding dozens of interactable camera feeds and door controls.

  9. Blake says:

    Always a pleasure to read your column Shamus. Of course, those of us who found it through TwentySided, rather than the Escapist, know exactly what prompted your particular choice of comparison games.

    What kills me though, is the same thought everyone has when this discussion comes up. Why is no one making a massive and expansive game of PS2 quality? Give me a game that looks like Morrowind, but plays on a scale we’ve never even seen before.

    Skyrim took six years to make, and its gorgeous AND expansive. But not everyone can do that. Not everyone is Bethesda or Zenimax or whatever. So build out of old cheaper bits and flood the market with games similar in scope to Skyrim, or heck, maybe even longer. The math (which might be off the top of your head) says that games are 20 times more expensive to make than they once were. Assuming that the rate of increase in cost was reasonably paced, that means you could produce a Morrowind or an open world title of similar fidelity for about half of what 6 hour shooter would cost today, and in a similar time frame. Don’t tell me there’s no market!

    • Gamer says:

      I wouldn’t say there’s no market, but a lot of people nowadays are severe graphics whores. While people like you or me would buy it, many more would say “That’s so UGLY!!!”.

      It’s a sad fact of gaming.

      • Naota says:

        A sane way to mitigate this would simply be to do what Shamus suggested and run with a creative art style that isn’t so harshly judged against the way things look in real life. Nobody complains about the lack of detail in games like Grim Fandango or World of Warcraft because they’ve chosen an aesthetic style which goes well with an abstracted video game reality. It’s not as hard as people think to make a game which has good aesthetics but less minute detail and less graphical tech hindering the design.

        Also, there are many graphical enhancements these days like ambient occlusion and various lighting systems that only need to be added to the engine once and require no real effort from the art team to make the game look better. A great example of how easy it is to make a cheap indie game look great on a small budget with a bit of graphical wizardry is AAaaaAAaa For The Awesome, especially when compared to its prequel A Reckless Disregard For Gravity.

        • Tizzy says:

          Yes! Back when photorealism wasn’t even an option, people made do by carefully tailoring the art style to the capabilities of their tools. Same goes with CG animated films: Pixar and the likes were usually very good at matching their characters and environments with the technology of the time (toys & bedroom and no humans much easier han fish and sea and so on).

        • JPH says:

          I agree with this. I do not want to play a game that looks like Morrowind, but I’d love to play a game that looks like Wind Waker.

      • JPH says:

        If it looked like Morrowind, those people would be justified in criticizing it. Morrowind looked ugly even for 2002.

  10. rofltehcat says:

    Nice to see another post about game design after quite while :)

    What do you think about crowdsourcing or something like that? Game producers could do a contest: Create the best *insert room description* (or the best X for kinda-repetitive hotel rooms) and get it into the game as well as a mention in the ending credits (long enough already, so who cares). Give out a basic ground-map for the rooms and let the community go at it. Would probably also make generic rooms (like hotel rooms) look more organic than making one guy build 20 hotel rooms.
    I think this could be a good way to increase hype for a game. Of course one employee would still have to check them, game-relevant stuff would have to be placed by hand… but it would probably be a good way to save some man-hours and especially for mod-friendly games would build a community even before release.

    However, many games today don’t even give out mod or map kits even after release… so this won’t work for a studio working for a big-ass publisher.
    It is a bit sad that publishers seem to prefer the money they can make off DLC to the increased sales from having a thriving community that supplies itself with new content. Hell, there could even be DLC that supplies a modding community with new objects and mod/map options.

    • Ben says:

      Relying on the community to do work for a games company is usually fraught with enough legal issues to scare away all but the most determined publishers. Plus you have to then hire people to vet all that content, and presumably deal with a bunch of irate modders when they realise someone else is making money off their work.

      I’ve read a lot of posts on forums with people complaining when developers don’t release editing tools. It gets somewhat disheartening to read all of the associated bile that spews forth on forums when you work hard to release a game and it doesn’t have tools to edit with. It always reads as if people think it is a right they have, or that it is just laziness on the devs part, or greed, not wanting to give up DLC profits.

      Making tools for games is as much work, if not more, as making the actual game. Making tools suitable for public release is often exponentially more work.

    • X2Eliah says:

      It would be impossible to maintain a singular art style or design ethic in such a case.. People would notice that the rooms woud vary wildly in terms of looks, layouts, quality.

      And building a community that has the illusion that they are contributing a lot to the game *will* backfire, when it comes to selling, dlc, and any other “hot button” issues that people on the Internet like to get all pissy about.. Some of that can be seen at this very moment on the “X3: Albion Prelude” forums (company-Egosoft).. The community has been very involved in the development, and now every time something is not done exactly as they want 100%, there is much whining and moaning on the forums. Generally, such methods build you a community, yes, but it is not a community you would actually like to have.

  11. Kimagure says:

    Great article. Sadly, I feel like you only briefly touched on one of the key reasons why art has become more expensive and why it ends up constraining design: complexity of the interactive space. If it were just a question of more couches and people and animations, you might be able to outsource that to animation houses in India or Taiwan. The real problem, however, is that modern graphics requires multiple layers that all need to come together and work just right for the level to look good. Not only do you have more complex models, you have normal maps, texture maps, light maps, bump maps, spectral maps, questions of ambient occlusion, etc…

    This leads to a number of problems, most notably:
    1. As the system’s memory can only hold so much at any given time, this leads to real constraints on the size or complexity of any given level/room (though there are workarounds and fakes that can expand that to a degree).
    2. Because it’s so complex to set up a space, and because it takes so much time, levels need to be locked down rather early and there’s far less opportunity for the designers to playtest and change the experience than with earlier games. This is even moreso when a room/level/area involves a large central (and often original) setpiece.

  12. Abnaxis says:

    That article was very…untechnical, from what I have come to expect when you write about these sorts of things. To me, the problem isn’t that players expect more little fiddly bits in their scenes, it’s a combination of more fiddly bits * more layers added to each fiddly bit. It’s not just ‘all the windows need mouldings,’ it’s ‘all the windows need moldings and everything needs higher resolion textures/bump maps/etc.’ Similar to the problem of too many players on an MMORPG server–the problem is nonlinear because you not only need to to tell the player the results of their actions, you have to tell every one else as well, for every player.

    This multiplication effect is huge, but I don’t really see it in the article. This is unfortunate because it is a highly technical issue that is difficult to explain (at least for me) which is usually where you are strongest. Do you run into a word count wall or something?

    • Shamus says:

      My word count is pretty wide, and didn’t really impact this article. This mix just felt right. To get into technical specifics I would have needed more recent development experience with these current-gen engines, and I don’t have that.

    • Naota says:

      Not to invalidate your point (because it’s an exceptionally sound one), but higher resolution texture maps shouldn’t be a serious impediment to anything but the overall memory budget. It pains me to play the OCD nitpicker like this, but sometimes a technical explanation is just too tempting to resist :/.

      All it takes for developers to make higher resolution normal maps is to type in a higher resolution when they render them out to flat textures in a 3D modeling program. When painting colour maps, specular, or any similar textures it just requires the artist to start off with a bigger canvas in Photoshop, which they should be doing already if there’s even the smallest doubt. It’s trivial to resize a wall texture from 2048×2048 to 1024×1024 with no major loss of detail, but you can never scale a texture back up and create detail where there was none.

      The real problem is that artists now have to spend time creating dozens of these layered textures and models for every single object in the first place, which is the worst offender of the multiplication effect you mentioned.

      • X2Eliah says:

        but you can never scale a texture back up and create detail where there was none.

        Heh, a lot of the authors behind Skyrim’s “highres texture” mods should be informed of this ;) – most of those are pretty much no more than cheap upscales + 3 brush strokes….

        • Naota says:

          Good god, have they never heard of detail textures? (Edit: I knew Shamus wrote about them somewhere :P) You could literally achieve the same effect as those misbegotten upscales without quadrupling your texture memory just by tossing in a few detail textures at a higher scale than the defaults.

          • Alan says:

            I miss detail textures. Seems like their use died out in the last 5 years or so. Sure, the walls are hired resolution, but at some point I’ll still by smashing my nose into it, and detail textures remain an effective technique for creating the illusion of more detail than is actually there. Even a subtle white noise would do wonders.

            Shamus, any idea why games stopped using them? I was under the impression that multitexturing became very cheap more than a decade ago.

            • Irridium says:

              Actually, the Halo games make use of this. From a distance it looks nice, and when you zoom into something (character, environment, vehicle, weapon, whatever), you can see a bunch of scratches, marks, and lots of other small details.

              At least, the Halo games on the 360 do this.

      • Abnaxis says:

        But that’s kinda of the point I was making. Every single problem I see brought up, when taken by itself, is really not that big of a deal. I’m teaching myself Blender right now, and I can say it is trivial to add a moulding to all the windows. I took me, a complete newbie, ten minutes to do. I can’t imagine it taking longer than a minute when I become practiced. The same goes for scripting, and textures. But the multiplicative nature of these things make the cost of game time go up exponentially, and the devilish thing about exponential relationships is that even when the multiplication factor is something small like 1.02, your cost STILL explodes after a while.

  13. Adam P says:

    “If you look at the original, you’ll see that rooms are often very square, as if they were made from elongated cubes. This is because they were made from elongated cubes.”

    Woah, slow down. You just blew my mind!

    I don’t know why I found that second sentence to be as funny as I did. Something about stating the obvious, maybe?

  14. Graham says:

    Just as an FYI, so you know who you’re writing the article for, I believe ps238principal is Aaron Williams.

    Don’t recognize the name?

    He’s the author of such comics as Nodwick, Full Frontal Nerdity, and (appropriately) PS238.

    If you aren’t familiar with his work, read it. It’s excellent.

    • I actually found dm of the rings (and hence this blog) by a link posted by Mr. Williams many many moons ago. Great guy, very talented, and his Nodwick computer spoof thing still hangs (with strips numbered in binary) in one of the computer science labs at my college.

  15. RCN says:

    It is just disheartening that this needs to be explained to people. On the other hand, yeah, it seems the current console is also holding back and incurring huge costs with the smoke and mirrors necessary to make their limited processing power cheat a prettier game out of it.

    If only instead of getting a new console generation they launched the current ones with improved hardware. Increasing RAM and CPU cycles would be trivial, and it would’t even need to be optimized to be a significant step up. Maybe leave those hardware spots open to upgrade so that developers are never in this situation again.

    Oh, right, those are called PCs… it is not for nothing developers love to give praise to PCs… when it comes to creating their games in them, not FOR them.

    Anyway, good to see you back to the escapist. I was beginning to wonder if you left out of respect for the Extra Credits crowd.

    • Pete says:

      But thats one of the major advantages that consoles have in the first place. When making a game for, say, the PS3, you know EXACTLY what hardware youre dealing with, so you can tailor it exactly. Compared to that, making PC games work is an absolute nightmare.

      • Eruanno says:

        I was just thinking that. It seems like it would be a huge pain to find out what kind of hardware combination generates that particular bug that batch of users is having, if everyone just has the exact same hardware.

        I don’t think we need new consoles with a crap-ton of RAM or a 16-core CPU or what-have-you hardware to get better games, but rather I think games need to:

        1. Have better writers. Or maybe write more diverse stories. I’m tired of playing the story of the heroic soldier-dudes running through foxholes, shooting enemies while shouting clichés. I’d rather play the game about a boy in a summer camp for psychic children solving the mystery of who is stealing the childrens’ brains, or the game where you are a test subject making wormholes to solve puzzles while a homicidal AI is taunting you.

        2. Stop using so many hallways and areas with obviously placed chest-high walls. Maybe I want to be surprised when there’s an ambush for once, instead of just going “Oh, enemies here, what a surprise. Pew, pew. Yawn.” Maybe I want to be AMBUSHED, for reals!

        3. Consider the fact that games are interactive. I’m tired of games that won’t let me open doors, where have to wait for my partner with a big “FOLLOW” strapped to his head to kick in the door for me. My characters foot is fully functional too! I’m playing a game, not watching a movie!

    • Tizzy says:

      But then, could you release games that would work decently on both old and new versions of the console? So long as you have to maintain compatibility with the older consoles, I don’t see hiw having the new ones around could help. And I can’t picture Microsoft and Sony just saying: buy a souped up version of what you already have.

  16. Big Steve says:

    I liked your article, but as I read it my most common thought was, “yes, but why don’t they just automate that?”

    I mean, once you’ve made a window for a given architectural style, can’t you just keep cutting and pasting all the little submodels whenever you need a new one? Something like an elevator is more or less the same: you might want to retexture the interior from building to building, but the actual function of one personnel elevator is identical to another whether you’re in an apartment complex or the Pentagon, which means you can pre-fab most of it and just paste it in as needed with minor modifications.

    FWIW I seem to recall that facial animations for speech already are automated. Back when Valve was still doing tech demos for Half Life 2 I saw a video where they had the G-man model lip syncing some Korean speech, allegedly all done on the fly.

    Unless I’m missing something important, it seems to me that a lot of the increased time is being spent doing complex tasks over and over again when they could be automated just because those tasks were trivial to do on an individual basis ten years ago. Of course there’s still quite a lot of overhead in producing higher quality models and textures and so on, but the difficulty in adding a few extra levels to make the game longer seems like it could be greatly reduced if the editing tools were able to handle repetitive tasks on their own.

    • rofltehcat says:

      This is also one thing I was thinking about and I can only hope they are already doing it.

      I work with CAD software for my studies and when designing parts I can either just build a part (depending on complexity 10-XYZ minutes) in a way that it works for my case or I can invest 3-15 minutes more to add some parameters and formulas.

      With those, I can change a part on the fly. Without changing every single measurement, for which I’d have to go several layers/levels down into the construction tree, and would in the process probably break some other dependencies of other construction elements.
      I can also make many different versions of the same part/assembly by simply linking a whole excel sheet to those parameters. Plus, it takes much less disk space and calculates a lot faster, especially if the part is used many times in the same assembly.
      Furthermore, if I give a part with parameters to someone else, this person can easily adjust it if needed (within the boundaries of the parameters) without having to look into how exactly the part is designed. Actually, I’ve often preferred rebuilding a part myself to fixing the parameter-less (and often not very well designed, as in very chaotic/following no system/ignoring all rules of design and rational thought) mess of others.
      It even works for whole assemblies. Need an engine that has 10% more cylinder capacity? Bam, the whole assembly is adjusted by just changing a few parameters. Sure, it is often not as easy as this and the engine surely isn’t optimized for weight, clean combustion and optimum power (otherwise you’d have to recalculate that whole thing).
      It is not perfect but last time I checked, game designers didn’t have to go for optimum and 100% physically correct behaviour.
      Oh, and it can do colors, too.

      I can only assume/hope that map editors have a similar functionality and that it is actually used.
      Now you’d expect those parts to look exactly the same, just a bit smaller or bigger, etc… If built right, they can look very different and even have different features enabled etc.

      For Shamus’ pillar example this would mean that one guy would build a pillar type (or maybe 3, round, rectangle and octagon) and some different bases. Those options are already enough to create a lot of different combinations. And those combinations can be easily varied in size and even shape by the level designer who only needs to enter a few variables.

      • Eric says:

        Of course, this is what is done in 99% of games. Usually it’s referred to as using static meshes, due to the conventions set by the Unreal Engine. Static meshes are basically pre-defined objects created and placed within levels. They’re actually more efficient to render than traditional constructive solid geometry, it’s far faster to create detailed levels using them, they can be scaled, rotated, etc. freely, and for the most part it makes level design easy.

        Most games use a mix of static meshes, a terrain system and CSG, with CSG as the base for, say, buildings, and static meshes used for the vast majority of detailing in the world. Terrain is a whole other story but put simply, it also doesn’t require any sort of modeling, and can be automated to a degree by terrain generators, importing heightmaps, etc.

        However, it’s also worth pointing out that the average number of static meshes that make up a game level has skyrocketed. Whereas previously you’d have maybe a dozen objects in a given room (chairs, tables, coffee mugs, lights, etc.), now you require far more… just think of the average office in a game like Human Revolution versus the original Deus Ex and you’ll see that there’s nearly a fivefold increase in the number of objects. Those art assets have to come from somewhere, and that’s what takes so much more time and money to do – especially as a series of cubes and a 128×128 wood texture no longer passes for a bookshelf these days.

        That’s usually why the actual number of level artists/environment artists/”people who use the SDK to make game levels”/whatever you want to call them has remained relatively constant over the years, as have level design teams (which are often differentiated from level artists), while the art departments themselves have got so much bigger. Hopefully down the road, stock libraries of 3D models and textures will become more widely used and viable than they are now – we already have them (see CGTextures.com, which supplies dozens of AAA games with ready-made textures), but rarely are they production-quality or usable without serious modifications.

      • Shamus says:

        Boom. You’ve exactly explained what I’ve been on about with more “procedural tools”. (As opposed to procedural generation of “random” worlds, you design a specific world but leave a lot of the particulars to the automated tools.) The level designer says where a window goes, but some script works out the frame, type of glass, sill, etc.

        This type of tool would be most useful in a game like DX where you know a lot of your game spaces will have shared elements. Like, most of the game will take place in an urban setting. This becomes less useful if you’ve got levels in an apartment building, an amusement park, an oil rig, a forest, a dojo, and a stadium. Then again, that sort of eclectic design work is crazy expensive no matter WHAT you do.

        • Simon Buchan says:

          I’ve been trying to figure out exactly this for a while – I imagine it could get to the point that a level designer could drag a line to make a corridor, drag another line to make a branch off it, drag a square to make a room, etc…, while artists are filling out templates/generators to say how a corridor, junction, wall *looks*. Obviously this style only gets you 80% of the way in the look, and you’d have to spend at least the same amount of time figuring out the unique features.

    • Piflik says:

      You basically described the workflow for modular design. People are doing this, especially with environments (characters and vehicles not so much). When you create, for example, a house for a game, you would cut up the facade in separate small chunks that you reuse for the whole facade. You would end up with one corner piece, one window piece, one door piece, one basic wall piece and then some additional detail pieces. With this fairly small number of pieces (<10) you can then build the complete facade and then use them to create additional houses with the same architecture.

  17. Daemian Lucifer says:

    I think its high time we developed procedural generation a bit more.Making just a few assets that can be used on a bunch of couches that populate your level would cut the time required to fill it sharply.And for the cost of creating just one apartment,you could now build an entire building.True,the first excursions into this field would lead to some samey looking levels,but I think it would be worth it.

  18. Eric says:

    To be perfectly honest, I do think that Shamus overstates the amount of work that goes into this sort of thing. Yes, it takes more time to make game levels, that much is true, but a lot of that comes down to the art pipeline and not the actual level design and creation pipeline. Scripting takes just as long as it ever did (and often is far more approachable, with flow charts instead of actual code), setting up cover nodes, interaction points and other bits isn’t as time-consuming as you’d think (or more so than it used to be), etc. You might have to place specific nodes for the AI to understand, but you also don’t need to place waypoints yourself due to auto-generated navigation meshes, lights are increasingly being done all in real-time, collision is automatic, etc.

    What I would say has changed is design itself. Game levels are now created with far more thought, planning and purpose behind them. You could get away with “a series of rooms” back in the day because the standards were so much lower, but now each and every game event, the placements and behaviours of every enemy, the layout, flow and pacing of the level at large, etc. all need to be considered in greater detail. Part of this is because production itself is bigger and more compartmentalized (you can’t really just have a single designer go at a level for a few days and have it production-ready), and part of it is also just because there is far more art to things than there used to be. Design, implementation and testing are all far more thorough simply by virtue of the fact that creating AAA games is no longer the sort of thing that a few college students can do in a dorm room.

    Granted, I’m not speaking from a professional capacity (though that may change soon, and I do have many years of my own game and level design experience), but I do think that this article oversimplifies the situation a little bit more than it should. It’s not all down to better graphics, and not everything actually takes significantly more time to do (much of it is actually much quicker).

  19. SatansBestBuddy says:

    I have been saying the exact same thing for at least 3 years now, because it’s so very true yet so many people STILL don’t get it!

    Some common rebuttals to this argument include “well, if it takes so much time for one person to do it, just hire more people!”, which is another great example of how people have no concept of correlation. Hiring more people = spending more money. Then there’s “if it’s done once, it can be just copy pasted, so why don’t they copy paste more?!” Or how about the old “well, they’ve already done one game with this engine, so why is the sequel taking so long!?”

    I thank you for trying to explain a little bit of what people are missing, but the fact remains that people who are more familiar with playing games than making games have, honest to god, no idea whatsoever the amount of work that goes into these games.

    I bet most people have no idea the amount of work that went into making Deus Ex, nevermind the two orders of magnitude more work that went into Human Revolution! If people don’t understand the baseline as taking a lot of work, then trying to explain the concept behind something taking MORE work than that baseline is like saying “There is more water in the ocean than there is in a lake”

    They’ll respond with “Okay, but that still doesn’t explain why the ocean is so much bigger than the lake!”

    People just don’t get it.

    Also, cross posting this from Escapist cause I realized I never post there so I never check up on discussions there.

    • Bryan says:

      Heh. Not only does hiring more people end up costing more money, you *still* end up getting finished later.

      “Adding more programmers to a late project makes it later” — Fred Brooks

      The time something takes is a function of the number of communication channels required between the people working on it. As more people get added, the number of communication channels goes up quadratically.

      (Assuming a full mesh. Simplifying the mesh is how open-source tends to work out, or at least the Linux kernel. It might be possible to do something similar for games, I suppose, but it’d depend on the complexity of the interactions between the various teams, and whether a single maintainer could handle a single subsystem. It’d probably work for some but not all.)

    • MatthewH says:

      If programmer man-hours are 100% interchangable, then simply adding programmers will not increase the cost. If they are even mostly interchangable, and the interest rate is greater than zero, then it is possible that adding programmers could actually be cheaper.

      Though my understanding is that game studios are more like a medieval or renaissance art studio than Henry Ford’s assembly line -so programmers may be negative interchangable (by which I mean, adding a single programmer necessitates adding addition QA people to make sure the styles match).

      So, really, the industrial technology we want them to invent is Mass Production of Creative Talent -which may be helped by the development of school programs (by teaching a standard system it eliminates all the ideosyncracies) or by the long age of a console generation (since people can reuse what worked earlier).

      • Simon Buchan says:

        Programming is VERY VERY VERY non-interchangable, in the extreme case, old ’70s era IBM estimates found their best programmers to be around 100 times better than the average. Due to several factors you are unlikely to see that sort of disparity within the same company (largely good programmers don’t hang around with bad ones for long), but that gives you an idea – but that’s not why hiring N times more programmers doesn’t get you done N times faster, with effort and good HR you can probably maintain high quality average on your programmers (especially for games). The real reason is that programming speed is closely related to how well you understand both what’s needed (far harder than you’d think!) and how what already exists and will exist in the future works – obviously the second is hard enough with one person working on a 1-10 million line codebase, but with 50+ other people continuously changing the rest of it it becomes a nightmare! And that’s only the start of the trouble with coordinating such tightly interdependent work.

        Fortunately, artwork has no such restrictions, meaning that without hardware limitations your game detail is linear in how much you spend on it. Unfortunately, linear isn’t good enough for the amount of detail we want. Also, we do have hardware limitation, and so you need lots of programmer time to pack that much more detail into the same memory, shader cycles and fillrate – and since you can’t easily increase programmer counts that means less relative time spent on everything else in a game!

        In short – linear costs are bad, but exponential costs are killers.

        • Naota says:

          I’d also say that the effective payoff of spending more on realistic graphics is absolutely an exponential function where each new level of detail costs an order of magnitude more in cash/time/manpower to achieve than the previous standard. You can throw more artists at the work to complete it faster at a linear rate, but actually raising the level of fidelity higher requires exponential effort.

          Going from a Doom enemy sprite to a textured model from Quake is a difference of maybe half an hour for one artist, but a huge leap forward in visual quality. Going from Half-Life 2’s character models to those in Crysis is a difference of half a dozen people and entire days, maybe even weeks of work, yet the payoff is so much less. At any significant distance I doubt you could even tell the difference between a Crysis character and one from HL2.

          • Simon Buchan says:

            Well, yeah I guess I was thinking more in terms of “work done ~= artists * time” rather than how effective that work is. I don’t even know how to subjectively compare graphical detail – for example I’d probably say from a modern perspective Doom looks better than Quake (heck, I’d be OK with someone saying Half-Life 2 looks better than Crysis – in particular their animation and character design is more appealing) although by any objective measure in both comparisons the second game is clearly better.

            PS: At a sufficient distance you can’t tell the difference between a doom-style sprite and a Crysis model, so I don’t think that’s fair :P. You know better than me that a model should be judged by it’s worst (likely) case, not the average.

            PPS: Out of interest, I checked out the credits for the id games:
            Doom II (1994): 2 artists
            Quake 1 (1996): 2 artists
            Quake 2 (1997): 3 artists
            Quake 3 (1999): 4 artists
            Doom 3 (2004): 7 artists + 3 animators
            Rage (2011): 26 artists + 7 animators
            Quake 4 (2005): 19 artists + 8 animators + “Creat Studio”

            I’m not even sure what this says (other than holy crap did Quake 4 get a fast turnaround). Maybe there was some game design shift that made art real hard ~ 2005?

          • MatthewH says:

            It has been exponential -but I think, while costs have continued to be exponential or linear, the argument here is that the actual benefits lately are at best linear if not logarithmic. We’re hitting diminishing returns.

        • MatthewH says:

          Just to make sure I understand, you are saying that programs do note break down into discrete parts -all the parts work together. Therefore, a programmer has to know every part of the program -which is a daunting task, even when the programmer has coded every single line of code. Add another programmer who can make changes that the first programmer doesn’t know about and suddenly the entire program falls apart. Have I understood correctly?

          It would seem here is the potential place to innovate -find a way to have multiple programmers work on a project in such a way that their work can’t affect other people’s work. Maybe some type of super engine or generalized AI. (I’m not a programmer, so please, feel free to tell me this wouldn’t work.)

    • Hitch says:

      They should just review the games honestly and state that after the first couple graphics configurations the game refused to run at all and that’s why they’re giving it such low scores.

      Ubisoft would probably blow their top at that. But I can’t imagine many gamers siding with Ubisoft after reading an explanation of why that happened.

    • Tse says:

      Too bad they can’t use a cracked version to test it, because Ubisoft games are getting cracked faster than ever these days.

    • Bubble181 says:

      Ubisoft has been on my no-buy list for quite a while now, along with THQ. Them and EA are probably the “worst” companies about this sort of thing. I’d put Blizzard on there as well with their always-online crap in SC 2 and DIII, but since I love both franchises, well-meraning people keep gifting me these games :-P

  20. Unseer says:

    not really on topic, but i saw you back at the escapist and i was thinking: couldn’t you try to get Spoiler Warning picked up by the escapist?
    you contribute for them already and it’s a consitently made/ consistently funny show. besides, the escapist don’t have an let’s play contributor yet…

    have you and the others ever discussed this?

    • Daemian Lucifer says:

      I think theyve already tried that,but it didnt work out.

      • giantraven says:

        I think it’s probably too long overall for The Escapist. I could see them perhaps doing a show featuring LPs of short sections of a game, but not an entire game.

        They also don’t really do videos over 10 minutes long and with only one video a week that means a hell of a long time to remain on playing a single game.

        • Perhaps a Spoiler Warning Abridged would work better for the escapist. hmmm *ponders*

          Hey Shamus & Josh, I’m sure it would be editing hell, and no idea if you got all the original files (doubt it) nor even all the masters laying around, but if you do have all the “masters” (or enough) then a season could be edited down quite a lot I guess, almost like a highlights reel or abridged version of a season.

          Say if there are 30 eps of 20 min each, (600 min) those could be edited down to 150 min (30 eps 5 min each), to a 1/4 reduction.

          Of course those viewers will miss out on a lot of comedy gold (fastforward through the n’th time trying something again due to a silly mistake), or inside humor that only those who see the full season eps would actually get etc.
          Also this means that the escapist would lag a season behind but in return get more tightly edited episodes (shorter & fewer).

          The escapist could always need more quality content, and Spoiler Warning could always enjoy a larger fanbase, so… Interesting idea or…?

  21. Hitch says:

    I had to go back and start over after reading the first paragraph of this article because it didn’t make any sense to what I thought the topic was. Influenced by the illustration, I read the title as “The Big Coats of Small Places” and was expecting an analysis of the wardrobe designs for JC Denton and Adam Jensen.

  22. MatthewH says:

    Related to the game market, though not exactly to the cost of graphics -I was in GameStop earlier this week and noticed that many of the new games are down in price $4-$7 from post-Christmas prices. A handful were still at their usual prices (MW3, I noticed). It was not, to my knowledge, MLK-Day sales -the prices on the entire stock just seemed to have dropped.

    Has anyone else noticed this? I had been observing prices going up. I am curious if this is an indicator for the Games Market, or just the American Economy (or maybe just my region’s economy). Or do game prices usually spike near Christmas and New Years?

    • Gamer says:

      Odds are that’s just the usual post holiday thing. Christmas is when stores get the most sales. They have to discount what they didn’t sell in order to make room for more shipments.

  23. Irridium says:

    Bit off topic, but I got an ad for ya.

    Didn’t get the url for it, though. Saw it right when I opened the site, took a picture, and clicked the comments to write this and didn’t think to do it until I started writing this sentence.

    • Adam P says:

      URL is http://waywardsons.keenspot.com/, and it’s a SFW webcomic (unless they go heavy nude after 50 pages).

    • Shamus says:

      Hm. Not impressed by Project Wonderful ads so far. I don’t want to upset the metrics by talking about it in a post just yet. I’ll probably have a post later this week where I discuss what I’ve learned with regards to PW vs. Google Ads.

      In the meantime, please bear with me while I let stuff like this slide. I’m going to clean house in a few days and this problem will be solved then.

      Thanks for the heads up.

  24. decius says:

    Is there a way to create a program which will handle the details (baseboards, door panels, windowsills automagically/procedurally within the tool? Once the tool is made that does all that without direct input for it, the labor cost of adding rooms and areas within an existing style would be reduced, at the cost of a significant investment in setting the stylesheet for the area.

    That also might provide a way to break up the sameness between areas; change the style, literally.

  25. uberfail says:

    I’m just waiting for us to achieve photo-realism so we can get bored of it an move onto something more stylised. Like modern art after the renaissance.

  26. Rick says:

    Is it just me, or does The Escapist’s redesign look very “MS Metro UI”-ish? I’m not saying the design is bad, I’m just implying it’s not theirs.

Leave a Reply

Comments are moderated and may not be posted immediately. Required fields are marked *


Thanks for joining the discussion. Be nice, don't post angry, and enjoy yourself. This is supposed to be fun.

You can enclose spoilers in <strike> tags like so:
<strike>Darth Vader is Luke's father!</strike>

You can make things italics like this:
Can you imagine having Darth Vader as your <i>father</i>?

You can make things bold like this:
I'm <b>very</b> glad Darth Vader isn't my father.

You can make links like this:
I'm reading about <a href="http://en.wikipedia.org/wiki/Darth_Vader">Darth Vader</a> on Wikipedia!

You can quote someone like this:
Darth Vader said <blockquote>Luke, I am your father.</blockquote>