I think the last time I posted an article here was back in 2003 or 2004. I generally don't write this kind of thing because I don't feel like I know enough to lecture about it. In fact, I didn't write the following article to be read by anyone else. These are my own notes (which is why some phrases might seem strange or unpolished) from observing/playing other games (Super Metroid, Castlevania SOTN, Zelda, Megaman), reading other articles and a couple of books regarding story-structure.
I wrote this when I was building my game's world and structuring the story beats for it, so it's influenced by Campbell's monomyth analysis (with which I don't agree very much anymore). Some examples I offer are pretty dumb—again, I wasn't expecting anyone to read it. Apologies. The reason I'm uploading it is so that someone else might benefit from what I couldn't find online when I was doing research on this subject. I invite critiques and corrections if you have any. I'm sure there's room for tonnes of improvement.
EDIT: It would be nice if other developers could also write down how they make their own worlds. It would be a great way to share ideas or experiences.
Notes: Castlevania in this article is almost exclusively Castlevania SOTN. Also, I just noticed sometimes I make reference to the items in my own game. Sorry.
--
The ellipse principle
When we're determining the shape of our world, we have to block-out where the main areas will be. We can use circles and lines to connect these areas and place them more or less where we want them to be. Then we imagine how the player will go through these places while the story is taking place. The shape of the world is important because it will ultimately determine how the game will be played and how tedious the backtracking will be (this is a metroidvania after all!). Linear worlds can be fairly tedious to travel, unless of course the content of each room is so good that it makes you forget you're just going in a straight line back and forth. If we don't have the luck to be such good designers, we can use the elliptical approach and it will almost instantly make our worlds more friendly and accessible.
Ellipses are the feminine approach to designing worlds. Most medieval cities are built in ellipses because they stay close to the centre; if they risk to stray away, since there were constant sieges back then, they might be destroyed. Masculine cities, which are more recent, are made of stretches of construction, each branching out to dominate the terrain they denominate part of the city. Although the maps in super metroid and castlevania are composed basically of halls or shafts, they generally compound together as circles or loops. This saves space and keeps things close together so that there's little boring time between the different beats of the story/game. Super metroid does this better than castlevania sotn, which is probably why it has aged better. Even the rooms themselves have a circular or elliptical approach to their design. Loops are good because they let you enclose areas, so once the player returns to where he began, he'll know he has finished said place after getting his reward.
The map itself should also be an ellipse through which the player either moves elliptically or in zig-zag. In cradle-like maps, it's easier to create zig-zags compared to circular maps. The advantage to circular maps is that no place is really too far from another; there's always a short-cut from one place to the other. Cradle-like maps don't share this advantage, but have the advantage of changing the direction in which the player makes progress through the game, letting it feel like separate acts.
▪ Consider that the game world must be built to be travelled from one place to the next (according to the plot or game-objectives). Thus, you need to understand how passing through all its areas will feel like to the player.
⁃ The best method I found was to build the structure of the world (according to the ellipse principles, perhaps?).
⁃ Then, I placed the areas around the map (still without rooms, room shapes, or terrain design (the level design inside the rooms)).
⁃ Then I specified the shape of each area and began making paths/routes that the player could take (in terms of movement through the map).
⁃ Then I sat down to turn the paths into rooms and room-shapes. The room-shapes are not very clear at this point, and obviously much less the terrain design. This is kind of what blocking is for 3d games. It is important to make your world big enough so that you can later shift rooms around, add or subtract ones at other places, etc. I made the mistake of making it too crowded at first, and it looked completely confusing and it was very hard to build.
⁃ Then I finally started making terrain-design. There was still much to be changed, but I figured the structure should be strong enough so that it can withstand such changes. After this, i saw it was very hard to make changes to the structure without having to go through all the areas around it. This last is a pain. It's like a jigsaw for which you decide to remove a piece and make the rest use that empty space. I learned I have to be slower in the blocking aspect next time I make a map to avoid this.
⁃ I also learned that each area must be clumped together with itself, at least more than the paths to other areas. There should be a clear distinction from one area to the other. If this is not the case, the player will not identify his position very well on the map and might get confused or lost rather easily. The map shape is probably as important as the level design of each room, although secondary to the latter—that is, I found it's more important to prioritise the level design than the room shapes or their positioning. There should be a balance, but when sacrifices must be made, I almost always choose to favour the level design.
Symmetry
When the player moves about the world, he will encounter two types of terrain formation or room design: symmetric and asymmetric. The asymmetric design is generally the more common one: Mario, MegaMan and Sonic use this. It lets you build levels only to be travelled forward. Castlevania doesn't generally notice this, but Metroid does: if you want the player to return the same way, you should build him symmetric rooms. Why? Well, because he's already gone through it once, the way forward; the way backward will be the same thing, so it will be obvious that he must go back that way--but as a last resource! (Edit
If we suddenly are stumped and there's no way to proceed, unless we are being chased or we're running away, we always assume the exit of a place will be at where we came in. In other words, the end of the adventure is at the beginning. The beginning, however, is always a base or a place to rest (again, if we're not being chased); we always wish to return there unless we find another safe spot.
To give an example, if we're going into an abandoned house, the safe spot is always outside. We will go about inside it until we've visited every room, and then come out the same way we came in. What generally happens in games such as Metroid or Zelda is that, when we're at the inmost part of the house, we will realise someone has entered the house or something has happened that will make it impossible to leave the way we came in. So, the way out will be the way back or out the roof, etc.
I found that it might be true that, if the gamedevs want a player to keep on going forward, then the room shapes will often be asymmetrical. You can also place soft locks, as they call them, to prevent the player from returning this way. However, if the player notices this last, he will feel the design of the game being imposed on him. If built correctly, though, the player won't notice it too much or at all and will assume the world is open for him forward--especially if he finds new things and replenishables that will give him confidence to go forward without returning to the savespot.
If you want the player to eventually return the way he came, build symmetry until the room that will begin the return trip. Said room must initially present the player with a centre that branches in two or more dead-ends. The player will explore them and then return the way they came. This is obvious.
However, what if said room leads to a new area or is just at the middle of the whole map? Well, symmetry might suggest to the player that he will eventually need to come back the same way. How? It will leave a resolution pending. Of course he'll forget about it but, when he comes back, such a thought might pop into his head: nothing happened in this area—why? Did I miss something? But then symmetry must be made apparent: maybe it should be fun or surprising to find such symmetry. This can be a tunnel that slides in a U-shape, or it could be a room with a different background (like Castlevania's 3d chapel). It must be noticeable. This will create a landmark in the map.
Secrets
When you add secrets to a room, be it secret items, power-ups or paths, you must always have a curve of difficulty when placing them. If you hide stuff from the player from plain view and there is nothing to suggest it is hidden there, if the player has just started the game, he will never chance to find it. But if you teach the player to always look for things in suspicious places, then by the end you will be able to get away with hiding things in very obscure places. Metroid did this very well. It also added spaces from which you could see the path, but not the entrance. The entrance is hidden rather well, but this makes the player curious and aware of the presence of said entrance and its secret item/path.
Dead-ends
You should almost never have dead-ends. They are very regular in real-life, but in games like these, it only frustrates the player. Dead-ends should be places where there are secrets to be found. If this last is not the case, then some plot advancement must be seen, or some back-story of the game-world (even with a simple sight, you can get an idea of what happened there without narrative). There are, of course, exceptions.
▪ If you want your player to feel lost, add dead-ends. If you want him to suffer through a maze, then have a very good reward at the end (not a power-up, for it is not enough. A new item, weapon, etc. makes for a good reward.).
▪ The other kind of dead ends are those that aren't dead-ends, but seem so because the player doesn't have the required equipment yet. These are game-locks, as some people describe them. But then again, they are locks if the player is aware that it is not a dead-end. If he cannot see the point of a room and criticises the design of the game, then it is a dead-end (in his mind!). When he goes back and manages to get through with a newly-acquired ability, he will feel ashamed for having blamed you for his blunders, and perhaps his respect for the game might grow a little.
Branch routes
Routes that branch from the main path (that lead to a power up or something similar) usually have to offer a smart way to loop around so that the player doesn't have to retrace his steps. The player likes going out and far into adventure, but retracing is a drag unless it's through a different enough path. It can be the same path, of course, but something should've changed the rules so that it's either 1) easier to return or 2) more difficult, but entirely different the way back.
It's generally easier to simply do another path through which the player can return. The first path can have collectibles if the game allows it, or maybe it can be an easy room in which to grind for stuff.
Routes to access parts of the same room
If part of a room is inaccessible ONLY from one side to the other [if the player is supposed to encounter an obstacle he can already pass without the need to acquire a skill power up or spell], make a route that goes around the room (to the bottom or top, left or right--depending on the direction of the obstacle in said room).
The detour can also be made as an alternative to a barrier that can be destroyed with power-ups not yet obtained. This will become a shortcut the second time around.
• Said detour must have in its way a few branches with power ups, recharge rooms (where the player can replenish his strength). Rooms that are themselves not branches must have obstacles (enemies). Remember to reinforce the idea that the player must follow the line of enemies like breadcrumbs. If there are no enemies in some areas, the player will assume he took the wrong path since it will feel empty. This is of course no replacement for real obstacles, but it will guide the player with an invisible hand.
• If the detour holds itself an entrance to another area of the level (but is inaccessible yet to the player that lacks a special skill--like the double jump, a spell, etc.), make it rememberable by adding something unusual by the entrance (such as a special kind of enemy, some new decoration, different floor tiles, or simply that the design of the room seems different or strange).
▪ Also be sure to add a room in that detour that is especially different from the rest of the rooms in the game (super metroid had a branch like this in norfair when the screen would start shaking too much and the room filled with lava). This makes the player remember that place because it was uncomfortable to play or because it made them rush. If no such thing is added, the player won't remember this particular route, even if it's pointed out on the map.
▪ This is a better alternative than having there be text telling the player this room is important.
▪ Keep in mind that detours will seem as detours the first time around. The second time around, one of the branches must be able to be opened so that it reveals further parts of the same area (and so that the detour stops being a detour and is encircled by more rooms.
• If this last doesn't happen, then the branch must connect another area to this one.
• If you don't want further areas to branch out, then add an upgrade power-up that can be gathered the second time around to justify the uselessness of this dead end.
Trap areas
When the player must learn to use a new skill, be sure to trap him until he manages to figure out how to use said skills. The smaller the area is, the more clear to the player that he must be missing something important. If the area is too big, the player will assume he is lost instead of challenged.
How to trap
Trapping is only good when you do it to teach the player a new ability without telling him. If you trap him to delay him from getting somewhere, unless he's trying to get in or out of the inmost cave (in which the player expects to encounter traps), he will grow tired of it.
You can trap the player in said areas in two ways. You can have the game trap the player:
▪ Silently: it doesn't tell the player he is trapped. The player will be unable to proceed through the apparent way and will backtrack -- only to find out he is trapped. If the trapped area is not too big, he will figure out he is being tested and this trapping was intentional. Not only will he not feel lost, but he will feel he is missing something very silly and will be gratified when he discovers it.
▪ Evidently: sometimes it's good to add a hidden trap (such as crumbling floor, collapsing walls, etc.) to challenge the player to find his way out. It is very important that the player understands that he was trapped by the game because of his own carelessness. Metroid does this the second time the player goes to Brinstar: he gives the player a long hallway which he can run across (seeing that the game has given him enough hallways to speedboost his way through them). However, there's an energy tank at the end, and the player will with more reason try to get quickly to it. Alas, four blocks before the energy tank can crumble and the player, unable to jump because of the low ceiling, will inevitably fall into a loop of rooms.
⁃ Important: The player knows it was his fault, though, for running carelessly to the power up instead of checking to see if there was brittle flooring. If the trap were a wall that collapses on an exit, it must collapse because the player interacted with the terrain around it. If the wall just collapses because the player came close to it, he will feel cheated and will think badly of the game level design. Be careful with this.
⁃ Note: The difficulty of beating a trap should determine what the reward should be. If the trap is not too elaborate, a dead end with an extra container, experience diamond or whip seed is okay. If the trap is very elaborate, it should lead to a new area entirely, or to a boss that gives a new special ability (with which the player can backtrack).
Small Short Cuts
Sometimes it's good to add a very apparent shortcut, such as sailing a current, a tunnel through which the player can run, an elevator, etc. However, sometimes it's more gratifying to let the player find the shortcuts by himself.
▪ This is done very cleverly in super metroid when the player can skip bombing several blocks: if he holds left or right, he will bombjump into an invisible tunnel above the other tunnel filled with blocks. The same game mechanic led you to find the shortcut.
▪ Another way to do this is by having enemies use the hidden shortcut; the player will notice them and will try to go there himself.
Big Shortcuts
This kind of shortcut is basically a 1 or 2 cell long room that connects rooms that are seemingly far away from the other (the player must go through a long circular backtrack to get from point A to point B). These shortcuts can be activated or opened during the second revisiting with new powerups that the player didn't have the first time around. Example: super metroid—you can get from Brinstar to Old Tourian in the second revisit.
⁃ Note: remember that letting the player skip areas by the use of their skill will give them gratification for beating the game's sequence, but will also add uncertainty as to how much their game will mess up. Thus, try to never leave a sequence-break open unless it requires very specific skills (that the player wouldn't have the first time playing the game).
Deviations to new areas
Such deviations must not distract the player from his original road through that area. If you offer the player two roads, you risk losing control over his progress. The easy way to avoid this is to add barriers that are only removable with future upgrades. Thus, this is not really a bifurcation: he can only go one way this time around. If done too frequently, the player will notice he's following a linear path, which is why you must resort to a few tricks:
▪ Have the two roads be accessible, but one road will have high level enemies, telling the player it's not time for him to go through this way. This is a workaround to adding a physical barrier, but it's still clear that the game doesn't want you to go this way.
▪ Another way, which is inherently more elegant, is to make the player think he chose one of two roads (the one you want him to take). He indeed chose the road you wanted thinking it was his decision, but it was because you suggested it without saying it. Such ways are very common with Super Metroid. This last does it by having option 1 be a new road entirely, and option 2 be a road that begins with a room that you already visited. Metroid generally wants you to take the new road, so such visited room will be one that made you very frustrated--it hopes you won't want to go back that way. It also furnishes the new road with nice backgrounds, a new enemy or a ledge that you couldn't reach before.
▪ Another way to make the player choose a road first (even if it's just a small branching, such as getting to a save room) is by adding dangerous-looking terrain (bright colours, sharp shapes (spikes), vertical terrain formation, etc.)
Before boss rooms
Be sure to tell the player somehow that a boss battle approaches. This could be by adding a background decoration (like the monster's head, or by special tiling, or maybe even a sound). Metroid did this by using the eye doors; mini bosses weren't predictable, but they were easier. Castlevania uses special doors, as does Megaman.
When to give upgrade power-ups (stat expanding items)
Power ups should have somewhat of a reasonable motive for being there. Whip seeds grow from special plants in certain areas, so the background around it must always be the same plant from which the seed grows. HP containers and expanders are special jewels or craft scattered around in the world map, so there should be other smaller jewel backgrounds or crafted pottery around these items.
• They should be placed in small branches off the main roads. Castlevania didn't quite do this, but Super Metroid did it almost religiously (except for the power ups --energy tanks, missiles, etc.-- that the player found at the beginning of the game).
⁃ Generally it's best to have them be at small rooms or pockets in said branches.
⁃ Make those rooms longer or those pockets bigger if the power-up is to be found later in the game.
• Having them be at branches distracts the player from the main quest a bit and gives him a little satisfaction.
▪ Only place them in the main paths if the player needs to be more powerful for a specific area.
▪ Placing them in an area will level the player out so that he grows stronger by boosts instead of linearly.
▪ Remember that the placement of items must obey the sequence of objectives (the paths the player will likely take through the map). Things should appear as the player requires them.
When to give ability power-ups or spells
Often, it's best to reserve these to special points in the game. There aren't as many abilities or skills compared to upgrades (health, weapon, etc). The tradition is that a boss has such skill or ability and, once beaten, the player acquires it. This is traditional because the hero in the monomyth absorbs the powers of the threshold guardians. Bosses are generally threshold guardians because the game plot takes a turn once you defeat them (it could even be a change of objective).
• Zelda gives the special item after the player defeats the mini boss; Super metroid gives it right away if it was a boss or makes the player go through a small area if it was after a mini boss; Castlevania gives it right away if it was a boss--mini bosses don't generally give anything except passage to a further part of the level; Mega man gives the power up right after the boss is defeated.
• The most content-filled solution is Super metroid's, in my view. If it's a boss, the player requires instant gratification. If it was a mini boss (the battle won't be as hard), it's best to let the player go through a small set of rooms (generally a circle). The power up is at the middle of the area. Getting into it is like knowingly falling into a trap, and getting out of the second half requires the power up (and you must learn to use it!).
----
It seems back then I was thinking of writing more, but never got around to it. I do recommend this article, which is a fantastic analysis of the entire super metroid game: http://www.gamasutra.com/blogs/HugoBille/20120114/9236/
Pretty good article about classic metroidvania design.
I would personally try to create something different and less predictable though. When I was still working on my project The Bubble (now pretty much abandoned), the idea was that the game would randomize the location of key items, and also randomizes where the locks are that require those items. There would've been a main route that snakes around the map, allowing player to travel to the final boss with very few locks in-between. Being able to open 'detour' locks would've allowed access to shortcuts and powerful equipment that make defeating the final boss easier. Without any such equipment, the final boss would've been pretty much hardcore mode.
I don't know if that would've worked in the long run, since I abandoned the project for lack of experience with the coding part at the time, but it seemed like an interesting twist to the usual design.
Normally metroidvania games seem to lack replay value, requiring things to be done in precise order, but Castlevania SOTN is a pretty good example of the genre that does have a lot of it. Lots of extra stuff to find, enemies may drop rare weapons/items, and then there's the whole fake ending thing, which genuinely left me baffled the first time, and I was convinced I was missing something. Optional and randomized things like these are a good way to add replay value, without affecting the core game flow. Even if you know exactly what to do to play through the game, there's still stuff left you didn't experience last time.
Very nice article - particularly the second half
I must confess I haven't actually played Metroid or Castlevania. The only Metroidvania game I have played a lot is one called "Zeliard" (a very good and extremely underrated game btw), and it does tick virtually every box on the list:
doors that lead to a boss fight always make a thumping noise when you get near; you can always see objects long before you can reach them (and in some cases you get clues from talking to people as well) ; there are big shortcuts once you have certain items; there are caverns that you technically can get through without the right items (eg. the asbestos cape to survive heat, or special shoes to survive slime), but it's so tough that you know you're on the wrong track; dead-ends always have a nice hidden bonus (but it's often protected by some moderately tough baddies) ; new magic is gained at each new town (which requires beating a boss).
I'm guessing it's a fairly formulaic genre...
HitmanN, it would seem as though that is a very different way of building worlds. It would be interesting to know how the story would fare with the random placement, at least excluding the problem of having to develop content multiple times to accomodate for the variations. But I agree, metroidvania games lack replayability, except only if you're trying to speedrun them. I studied film, so I have a linear understanding of story-telling.
Sketchy, I agree with the formulaic nature of metroidvanias today. However, back in 1994, or even 1986 with the original, Super Metroid was probably one of the first games to do this kind of things. It was a kind of experiment in a new direction. Unfortunately, either we only copy these things and don't offer much innovation, or maybe the genre is already too explored to find anything new. I didn't know about that Zeilard game—I looked for a LP of it and it seems kind of cool.
It would be nice if you guys could write up something describing how you build your game worlds (maybe including story, which I realise I didn't mention here, nor difficulty curves, sidequests, puzzles, etc.). I think it would be a nice place from where to get ideas and learn from if we all shared our different views on this. Or if you're not interested, maybe you know someone that is?
Alonso Martin, the story progression in The Bubble was indeed not entirely decided yet, but the main idea was that the main character has a definite goal/direction from the start. To find a way home. But she's in a place she doesn't recognize, and she's not sure how she got there. Hints, clues and answers to the latter questions are found around the map, some hidden behind locks, others at the ends of dead ends, functioning as sort of side quest enabling items. You can even find a small friendly 'village' that acts as a hub for many side quests. The game flows past these things, but doesn't impose them on the player. Ultimately, you could, if you really wanted to, ignore all such elements, travel through the game to the final boss, beat it, and get home. But picking up the correct pieces of answers will allow, at the very least, different endings. Better endings. Saving people. Preventing bad things. Reversing bad things. Maybe even rewrite history as the ultimate ending.
So, basically, the storytelling is as rich as the player's interest in exploring is. The main quest is pretty linear, to get from A to B. but the story itself is not, if you keep digging.
Anyways, I'll probably forget to check this article again, so no need to reply. Just thought I'd drop some info on what my plans were for the randomized item placement approach.
Personally-speaking while developing Hasslevania (and moreso with Hasslevania 2) what I do is make a big list of all the items in the game that allow for progression. Ex the pyro mace iac's fire attack burns bushes, iron armor allows for walking on spikes and etc.
In Hasslevania for example, things are semi-linear up to certain points and then they open up into maybe 3 or 4 different possible paths. Grabbing the above-mentioned mace, you can now go back and burn bushes or melt giant candles to advance somewhere. Or you can use it to destroy enemies that would otherwise be impervious and attempt to go down that path.
Or better still, you can use it to solve some puzzles which open up new treasures that can be used for entirely different paths.
The entire map of Hasslevania 1 is pretty amazing, in a put-together kind of way. It's amazing in that I was able to make something like that and have it actually WORK. I truly believe having a good Metroidvania game that's not generic and shitty is one of the harder things to pull off by any game maker.
For Hasslevania 2, I'm attempting to do the same type of thing but on a scope 3 times as big. Yeesh, and no wonder it's still not done yet!
EDIT: Here's the entire map for Hasslevania 1. It might be hard to follow on what items get you to what places but you'll get the gist of it: