level design lesson: in the pyramid

from super mario land, world 1-3

(now in french and russian.)

last month i met with charles pratt of game design advance to record a conversation that should hopefully be available for to insterested eavesdroppers sometime soon. one of the subjects we discussed is the stupifying lack of discussion, decades after people started piecing together digital games, on the art and craft of level design. a week ago, playing through satoru okada’s super mario land for the gameboy, i realized i could probably teach level design just using scenes from that game.

here’s one. this area (click here for a bigger image) appears halfway through super mario land’s stage 1-3, the third stage of the game[1]. numbers in brackets indicate footnotes, by the way.

crucial to this example are two rules of super mario land that a player who approached the game in 1989 after having played the original super mario bros. would have immediately understood.

the first is that mario has two states[2]: “little” mario and “big” mario, the former of which can transform into the latter by finding a magic mushroom. little mario is as tall as a single game block; big mario is as tall as two, making him slightly more susceptible to danger: if big mario comes into contact with an enemy, he’ll revert to little mario. unlike little mario, big mario has the ability to break certain blocks (the light gray, rounded ones in the above picture) by jumping at them from below.

the other relevant rule is that the screen only scrolls to the right. while mario can move freely towards the right, where his goal is, he’s only allowed to retrace his steps as far as the player can see: the left edge of the screen acts as an immovable wall that follows mario through each stage.

halfway through world 1-3, the first “indoor” stage of the game, the player is given a choice. the path rightwards splits into three routes – up, middle and down – though the choice is in fact between only two of them: either of mario’s states makes one of the routes inaccessible. the upper route is blocked by bricks that only big mario can break by hitting them from below. the lower route’s entrance is only one block tall – little mario alone is short enough to enter.

this is interesting because little mario is, most of the time, an undesirable state to be in. little mario can’t break blocks, is only one hit away from death, and must find two power-ups to be able to wield the “superball” weapon[3]. here, mario is given access to a special place as a kind of compensation for this otherwise weaker state – and the bottom route is in fact the most lucrative of the three.

but note that it’s not obvious to a first time player that the bottom route has the best outcome, because at the moment of the player’s choice the screen hasn’t scrolled far enough to the right to reveal the horde of coins. because the entire height of the stage fits the height of the screen, though, a player on the upper route will see the treasure she’s missed. this helps to mitigate the frustration of losing a life against the tough enemies to come: starting over halfway through the stage – and mario returns as little mario, regardless of what state he died in – means an opportunity to nab the coins i saw earlier.

also note that there are no enemies right before the junction: you don’t get to choose which state mario’s in when the path splits. there’s a single enemy on the upper route, behind the bricks, so an experienced player – or clumsy one – can smash her way up to the top, be shrunken to little mario by the enemy, and then double back to the lower route. there’s just enough room between the edge of the screen and the entrance if the player knows what she’s doing.

let’s talk about the way the treasure room is put together. that ? block contains a magic mushroom: in addition to snagging up to thirty-eight coins, the little mario who enters the chamber will exit as big mario. the player must become big mario to leave: those two grey breakable bricks, floor to a mario above but smashable to a mario below, serve as a one-way passage, letting mario out but not in.

so what’s that tiny pit on the right for? it’s for mario if he misses the mushroom. only big mario can leave this chamber – can’t go back left! – and if little mario were to miss the mushroom, he’d be stuck with no escape but to wait for the time limit to tick down and kill him. or to jump in the pit; note that the entrance is only one block tall.

it’s incredibly unlikely that the player should miss the mushroom, of course. see that little block sticking up out of the floor on the left? that block serves two purposes. when the mushroom pops out of the ? block and flies through the air, it’ll hit the wall and turn around; when it reaches the floor, it’ll start moving to the left. that little block knocks it back toward the right, both preventing it from disappearing off the left edge of the screen (a wall to mario, but not to the mushroom) and making its journey to the pit longer, giving the player more time to nab it.

so why have a pit at all? why not just have the mushroom bounce back and forth between two walls until the player catches it? because it’s sloppy. because the threat of potentially losing that mushroom – even if it’s unlikely – makes it far more valuable. because challenge and momentum are both big parts of super mario land, and a time-sensitive situation reminds the player of this.

on the subject of challenge, three of the ceiling bricks above the treasure room fall when mario gets near, potentially hurting him. because they fall the entire height of the screen, they’re a hazard to mario regardless of whether he’s in the treasure chamber or above it. but since they fall from the top of the screen, a mario above the chamber is naturally in greater danger of being hit than one inside the chamber. a mario above, however, is also more likely to be a big mario, while a mario below is definitely a little mario.

the treasure chamber mitigates the frustration of death in a boss level by giving the player a chance to visit a different route (and remember, those are thirty-eight of the hundred coins required for an extra life); it prepares the player for the boss by leading her towards a room that ensures mario’s in his big state; it gives a meaningful play context to the difference in size between mario’s states that is so central to the game and its protagonist. what other purpose does it serve?

it’s a pyramid! this stage is intended to remind the player of an egyptian tomb – you can see pixel hieroglyphs on the walls and big stone bricks[4]. what better way to evoke the idea of a pyramid than by letting the player enter one in the stage itself – one that’s full of treasure and has a secret entrance, no less. you can see that the hollow treasure pyramid is almost immediately prefigured by a solid block pyramid, too; pyramids like these appear throughout the birabuto kingdom, the game’s first three stages. a pyramid the player can enter, then, is the culmination of a recurring level design motif.

all the goals that this “setpiece” accomplishes – and good level design often accomplishes several things simultaneously – it does so using a handful of basic building blocks that are already known to the player: solid blocks, breakable blocks, and ? blocks. concise design doesn’t introduce new game elements needlessly: an element that the player’s already encountered already has meaning to her, and she understands its implications.

this is good level design.

[1] the game’s first world, the birabuto kingdom, most closely resembles the structure of the original super mario bros.’s worlds – a ground stage, a treetop stage, and a castle stage, culminating in a fight with a villain over a pool of fire – to present something familiar to super mario bros. players before the greater digressions of the later stages.

[2] technically three (not counting mario’s submarine and airplane), but superball-throwing mario behaves the same as “big” mario for the purposes of this example.

[3] not just a weapon: mario land’s superball, unlike super mario bros.’s fireball, has the additional ability to collect coins it touches, extending mario’s reach much as the magic mushroom does.

[4] not visible in the image are the stage boss and its minions. they’re sphinxes.

50 thoughts on “level design lesson: in the pyramid”

  1. Wow, thanks for the great, very detailed analysis. I always liked the smartness of this part of the game, but hadn’t considered it in this much depth before.

  2. I like this a lot, great post. I think you are right re: the lack of discussion about level design.

    Also, I haven’t played this game in 15+ years and would surely appreciate it more now. Think I’ll check it out. Thanks!

  3. Good stuff.

    I think it could be quite illuminating if you also took aim at an example of bad level design. I’m sure there are loads of platformers from the 16-bit era that would work.

  4. This is a fantastic essay, thank you. It does an excellent job of deconstructing these screens and seeing what makes them tick, and more importantly what makes them work.

    I agree with Tom, too, I’d love to see you take on a couple examples of bad level design too.

  5. I love when people break down the mechanics of how stuff works. I’d like to see more of these because as it is, when I make levels, I’m pretty much just ab libbing and aside from the very basics like the setting and what I need in there, I don’t work this deliberately at all. Mind you, I’m not working on platformers but, like anything really, the same rules can be applied elsewhere if only in different ways.

  6. I haven’t played this game in years, but just looking at that screenshot made the music for that level spring into my head.

    Super Mario Land felt primitive, but it really did have more complex (and sometimes fiendish) level design than the NES Marios.

  7. mario land is such a wonderful little game, way moreso than its sequel. i jsut wish the momentum wasn’t such that you seem to fall at terminal velocity every time.

    do an entry on ultima: runes of virtue II next! i would have never believed that was such a gem. although i say this while stuck miles into the cavern of pride having apparently gone the wrong way. :\

  8. in particular, mario land explores the ramifications of mario’s changing size a lot more than the nes games do. like the example above, there are lots of situations where whether mario is big or small decides what routes he will or will not have access to.

  9. now that i think about it, it’s funny how back then mario games were allowed to be used as testbeds for exploring alternate design approaches, like with mario land or the fds smb 2, because it’s a great idea that’s pretty much never touched on again after smb3.

    well, i guess there’s yoshi’s island.

  10. Super Mario Land was one of the first games I ever played so this was really neat.

    I always liked how collecting coins felt like a necessity in this game, as opposed to other Mario games where you can ignore them most of the time. You have very limited lives, no saves or warp zones and continues are difficult to earn, but it’s balanced by having huge bunches of coins all over the place, often hidden in pipes or requiring you to be a certain size to collect. A big part of the game is looking for and successfully reaching secrets so you can build up enough lives to get through the tougher levels.

  11. There definitely needs to be more in-depth analysis of what makes games work or not work. The Design Tour videos that the Wolfire guys did were excellent in this regard.

    Really cool. I’d love to see more.

  12. Mario Land was the first game I’ve ever played and is still probably my favorite games of all time. As such I’ve really always been aware of how great the level design is, but I’ve never really known why it works. Now I know.

    It really is a wonderful game. It shows how much people had learned since the first Super Mario Bros- it has almost the same engine, but the things it does with it are so much better.

  13. dess this is seriously the best thing i’ve read about videogames in weeks

    i was just saying the other day on sb that we should be having discussions about the nitty-gritty of game design, instead of arguments about whether whatever AAA game came out this month is culturally whatever that devolve into wankfests over authorial intent and semantics

    keep on keepin’ on

  14. great deconstruction, i hope you keep this on as a running feature.

    it’s really fascinating how integral level design is to the final work. so many words are spent on the development of assets and mechanics and it seems surprisingly rare that implementation on the design level is ever properly examined. looking for literature on design yields frustratingly little that seems useful in practice. i tended to assume that this sort of thing was locked up in academia as is the case in a lot of other fields, but you went to guildhall so i assume you have an idea of what is (or isn’t) happening on that front.

    i definitely disagree with anybody asking for negative examples as it’s way more interesting to have the elements of good design pointed out. certainly finding (and making!) errors is a lot easier, and for every point of good design there can be innumerable bad ones. there are plenty of lists of bad design tropes elsewhere but they are honestly pretty useless.

    thanks again, this is awesome.

  15. 1. you have a weekly allowance now of writing an article about a games level design, be it good or bad. if you miss your allowance i get a free jizz

    2. holy shit you continue to surprise me and impress me i love you THIS ARTICLE MADE ME LOVE YOU MORE

    3. even though ill probably never make a game id hella be in your level design class, you’d probably need to find someone that knows about 3rd person and 1st person games to cover those levels or sjust have it super specialized i dunno

    4. goddamnit this better become a feature SEE NUMBER ONE

  16. no one’s offered me funding.

    i have given a lot of thought to how i’d conduct a class on level design, especially since leaving the gilled hall. none of the teachers in the level design major had any idea how to teach it, while the art and coding majors were off and running. there has to be a way to discuss this stuff.

  17. KENTUCKY NO FUCK THAT IM TRYING TO GET HER HERE unless she can teach it from afar

    -auntiepixelante’s PR rep

  18. There’s one more thing: As big Mario, you can crouch and wedge yourself in the lower path. Then, spamming the jump button while moving forward will make you “crawl” through the narrow passage into the pyramid as big Mario, and then you can get the coins AND the superball!

  19. auntie: you mentioned not introducing new elements needlessly, which is pretty interesting. the current trend is to introduce new shit all the fucking time, until you’re knee deep in complex systems.

    that doesn’t bother me – i can handle it – it may even contribute to the predisposition towards slack design, the fact that a badly designed level can still be pretty fun. reading this, i can imagine the mileage all those game elements could have offered a more thorough designer. you must have played N, that had a good feel about it – i felt like there were more elements than there actually were (although not as economically few as SML).

    i have a question – do you think that procedural level gens might obstruct the good practice of elegant level design? once people like derek yu and steenberg crack it open for everyone else, i’m sure you’ll get a lot of people using it to design levels for them, and regardless of how smartly spelunky knits micro-levels together, will it provide another tool for cutting corners, d’ya think?

  20. procedurally-generated levels work in games that are more about systems: having the world rerolled each time makes the game stronger because what the player is learning isn’t the game stages but the game systems, the understanding of which will allow the player to negotiate whatever scenarios the game cooks up for her.

    spelunky is this type of game, and derek also put a huge amount of work into making his game output levels that are balanced and interesting. some authors use procedural generation as a way to shortcut the challenges of design, and it tends to produce either levels that are far too easy or far too hard (often resulting in the author making the player way too powerful to compensate) or an inconsistent mix of the two. it results in boring levels and a boring game.

    procedural generation isn’t a substitute for design. there’s a video i’ve seen that i’ve never been able to find since of a procedurally-generated super mario bros. the levels are mostly just long, flat plains with pits and lines of ? blocks dispersed arbitrarily. super mario land wouldn’t make a lot of sense with procedurally-generated levels, and neither, unfortunately, do some games that happen to have procedurally-generated levels.

  21. Spelunky was a bit of a proc-gen triumph, really, wasn’t it? A couple of my favourite screenshots are some the crueler configurations that Spelunky arranges itself into – tellingly, they’re all from worlds one and two. One of my editors just completed it, after like 800 deaths. My envy is palpable.

    I guess badly implemented procedural generation is as painfully obvious as good old-fashioned bad level design, then, and as ever, innovation and creative attention are the only proven methods. Thanks for your reply!

  22. I don’t want to shit all over this, but this seems to be an over articulation, of something that is intrinsically apparent to all people, and honestly doesn’t need to be said.

  23. so this is an attempt to articulate what has hitherto gone unspoken in hopes of generating a discussion of design? shit, you’re onto me !

  24. James, I got a lot out of the article, and certainly didn’t “intrinsically know” any of it, excluding obvious Mario stuff. If you mean that this is just a tour of facts extracted and derived from simpler facts, then yep, that’s a lesson. I mean, speed is obviously distance divided by time; you just need to look at the phrase “miles per hour” to know that, intrinsically. Yet, they’ll teach you that in school, because if nobody teaches you, you have to reinvent the wheel.

    So maybe you’ve already reinvented this particular wheel. Good for you! I’m pretty glad of the time-saver, to be honest.

  25. James, have you read much film criticism lately? Lit crit? I’m guessing no, because what you say this piece is (“over articulation”) is exactly what those things are. It sounds like your beef is with the practice of criticism more than anything else. Criticism is dissecting the form, taking it apart to figure out how it works, or doesn’t. This is a true furthering of games discussion, and not, as Glossolalia above put it so eloquently, “arguments about whether whatever AAA game came out this month is culturally whatever that devolve into wankfests over authorial intent and semantics.” In other words, it’s not the article, it’s you.

  26. Wow. I love this analysis. I recall playing this game when I was 6, twelve years ago, and fell in love with it. This articulates exactly why the level design is great (while remaining elegant).

    It is noteworthy that New Super Mario Bros. incorporates that blue “tiny” mushroom with similar design choices to SML. While extra small, Mario gains jumping power and can access areas otherwise unreachable – at the caveat of dying on contact with an enemy. The game even gives you a bonus for beating bosses as tiny Mario. The full implications of this didn’t register with me till I read this analysis. Cool stuff.

    P.S. MOAR!!!!!1

  27. If Flackon hadn’t said it, I would have. One of the reasons I’ve always liked the weirder, glitchier physics of Mario Land is that they open up more opportunities to feel like I’m getting away with something. I can more easily fix jumps. I can force myself under things if I need to — though the trade off is that I have to annoy myself in the process. Some of the strange collisions and hard-falling are a decent price for the positive glitches. If anything, they add further character.

    The Super Ball is, in its way, a workaround for the glitchy, limited engine; they couldn’t easily program the fireball’s motion, so they tried something else — and in the process, they managed to fold the fireball mechanic back into the basic principles of the game.

    It was always a weird appendix; it’s like someone decided Mario had to be able to shoot, and hey, that secondary button isn’t doing a whole lot. Then they never really thought about the implications of this power. Great, you can attack enemies from a distance — which means you don’t have to get close to them unless you want to, which I guess you could consider another layer of protection beyond the mushroom. What’s better than being able to stomp and smash anything? Not having to get your hands dirty at all.

    That’s kind of… I dunno. Thing is, the whole point of Super Mario Bros, the thing that it really does that’s new and different and interesting, is the way all its important mechanisms are immediately tactile. Whereas before this, most of the touch that went on in videogames took place through a barrel sight, you have to physically move Mario up to things and jam him against them in logical ways. I think the reason the game was so successful is that it’s so unusually intimate. So when Mario gets really powerful, he’s rewarded by being able to back off and play this like a seminal Bill Rizer?

    What’s dumber is that the fireball has no effect on the actual physical environment — which is more of the player’s opponent than the monsters that decorate it (though, yeah, they are effectively a part of that environment — that just illustrates the point). But the Super Ball — especially with the more intricate, collection-heavy environments in Mario Land, it becomes a crucial tool. It can reach places Mario can’t (or at least Super Mario can’t, so easily), and when unleashed in close quarters it creates a sort of pleasing chain reaction sensation.

    In mechanical terms, it’s basically a ghost Little Mario that Super Mario can send out to do his dirty work; in some senses like a Gradius Option. The effect is like getting a multiball in pinball (thus the comment about chain reaction, above). It’s not perfect, and further games could well have refined the mechanism. A second level of power ball that could break blocks might be interesting. Perhaps the first level would bounce off coins, and ricochet off enemies without harming them. There are a bunch of directions this could go.

    But yeah. This game is a perfect model of creative response to limitations.

    Yet they didn’t.

  28. the turtles who turn into bombs are another example. my expectation is that they were able to get the richocheting turtle shells, or some kind of reasonable simulacrum, working – but on the smaller screen and more cluttered stages of super mario land (it’s a much more vertical game than the original super mario bros., which has far longer flat spaces), the turtle shell just didn’t make a lot of sense.

    instead, when you bop the enemy, it gives you a few seconds to get away, then explodes. like the koopa, it becomes a different kind of hazard when attacked, but one that makes more sense for this game. time-sensitive situations in a game where you can only go forward are always powerful motivations to keep going forward.

  29. Something else for further development; the bombs might be able to destroy blocks, or have other environmental effects.

    Heck, it would be neat to do a Super Duper Mario Land of sorts. Keep it monochrome and glitchy, even if the pixel count is increased. Maybe even play with that. Add a grain filter at times, or screw with the framerate, to create some cognitive dissonance at just the right moments. Sort of like an old movie, pasted together of scenes of different quality recovered from all over, full of jump frames and camera weave, though using more videogame-relevant artifacts. Pixel blur, something to mimic inadequate lighting conditions (too bright or too dark, or uneven lighting, creating weird glare).

    And that’s just the visuals — though in this way the visuals could more overtly become part of the dialog. But yes, there’s that. Maybe a bit of cognitive dissonance from interesting uses of glitches that don’t seem like they should be there. And more overt mechanical upgrades in the form of level geometry interaction: the turtle bombs, the adjusted super ball, perhaps a few more big/small dilemmas.

    Something else I appreciate about Mario Land is that for the most part, every world has its own zoology. You don’t see the level two monsters in level three, or the level four monsters in level one. As consistent as it is systemically, the game really doesn’t reuse concepts more than it needs to. This creates much more of a sense of a journey than in other Mario games, which often just seem like they’re distinguishing one Sudoku puzzle from another by drawing it in a different color.

  30. one of the neat things about super mario land is that every stage – not every world, but every individual STAGE – is visually distinct from every other. the low stage count helped with this, i’m sure, but it nonetheless gives the game a stronger sense of progression from place to place than in other mario games.

  31. Still, if you duck with big Mario and press jump several time just in front of the path entrance, you CAN enter the lower path with big Mario.
    Consider it a small bug that is not ruining the gameplay.
    Coins and flower!

  32. You write a lot, but all elaboration is functional and informative and you know what you’re talking about.

    It’s like reading a good book, but it’s about computer games! Educate the masses! =)

  33. And all of that is just for normal mode, too! One of my favorite things about Super Mario Land is a legitimately more difficult mode that you get to play after beating the first quest.

    Like in hard mode, there’s a Goomba-equivilent enemy that starts out on the middle path, but if not disposed of quickly, will fall down to the bottom path, which essentially forces small Mario to take the middle path instead of the bottom one, since there’s no room to jump over the enemy down there. Also in hard mode, there is a sphinx enemy right outside the treasure room. It’s relatively easy to deal with if Mario takes the top route, but its fireballs shoot inside the treasure room, and can easily hit Mario. Mario can’t defeat the sphinx until he exits the treasure room, and if a fireball hits him before he breaks the bricks, he won’t be able to escape.

Leave a Reply

Your email address will not be published. Required fields are marked *

Parse error: syntax error, unexpected 'class' (T_CLASS) in /home/ccecce/auntiepixelante.com/wp-content/plugins/sk2/sk2_plugins/sk2_pjw_daily_digest_plugin.php on line 25