Chasm: Creating an Infinite Metroidvania
Kickstarter has become the go-to resource for everyone who hopes to create the video game of their childhood dreams: A way to gather cash from the masses in exchange for the promise of realizing labors of love that no major publisher would ever bankroll.
Not all Kickstarter game ventures are created equal, though, and some are more compelling than others. Most recently, a project called Chasm caught my attention by promising two of my favorite things in a single, faux-16-bit package: Open-ended action-RPG gameplay and randomly generated environments. In short, a marriage of metroidvania and roguelike games. How could I resist, especially when the two concepts being wed stand at such odds with one another? I reached out to Chasm’s developer, Discord Games, to talk producer James Petruzzi about the inspirations and challenges faced by his team’s project (for which a Windows and Mac OS demo currently exist).
IGN: What can you tell me about the game at the general level — namely, its design, the ambition behind it, and your influences?
James Petruzzi: Chasm is a 2D action-RPG platformer, with inspirations from games like Castlevania, Metroid, Zelda, Diablo, etc. We’re designing it to be a replayable experience, which was what sparked the whole idea to begin with.
A year ago, I had a simple idea for how to create Metroid-style maps procedurally. Later in the year, I began working on the original version of Chasm, which was to be a mash-up of Zelda and Terraria. I really wanted that open-world exploration, but mixed with the simpler progression systems of Zelda. I was never able to get the balance right, and after months of work just decided to scrap it.
To me, the problem with replaying a game like Symphony of the Night is that you know exactly where everything is after your first playthrough. The rooms are always in the same order, the enemies are in the same places, items and secrets once known are always known. So I began thinking of ways to mix it up.
Basically, what we’re doing is hand-designing tons of room templates. The templates are loaded into memory when the game starts, then when you enter a dungeon for the first time, the generation algorithm runs for that floor. We pick templates off the pile, build a dungeon with them according to a bunch of rules, test it, and populate it with enemies, treasure, traps, etc. The current plan is to have three floors per area, and six areas total, along with some special side stuff to do. Each area would have a big boss fight before you proceed to the next.
So the next question is, will there be backtracking like most Metroidvania games since there are only entrances and exits on each floor? Our (pending) solution to that problem, is to have floors branch and have doors that are not immediately reachable. For instance, when you get to floor 3, it could branch and lead to two other areas, one immediately available, and one not. Perhaps you go route A, obtain double jump, then backtrack to route B that’s now traversable. This could of course get way more complicated once branches start branching. For the special abilities that allow access to new areas, we want to focus more on acrobatics than weapons like Metroid or Shadow Complex. I’d like to do double jumping, wall jumping, ledge grabs, grapple hooks, sprinting, etc. and really flesh out the move set so it’s incredibly fun to move through these dungeons.
I grew up on NES, so I have memories of games like Castlevania, Mega Man, Zelda, etc. ingrained in my subconscious. Looking back, the thing I loved most about those games was that they didn’t over-explain anything. Zelda was probably the most alluring to me, just how mysterious the whole world is, and how you’re left on your own to explore it and solve problems. To me, you can’t beat an interesting world in a game — I’d take it over traditional narrative any day. To that effect, we’ll avoid telling you the story through character dialogue. Instead, you’ll be piecing together the history the of the world by exploring it, and finding and studying artifacts, discovering journals and notes, etc. These things will be collectibles, and you won’t see all of them on your first playthrough. Unfortunately, the story side really got the shaft in the demo, but we’ll be fixing that for the full game.
IGN: I’d like to talk more about the procedural aspect of Chasm. To me, the terms “metroidvania” and “randomly generated” are very much at odds with one another. I’ve played tons of free-roaming, exploration-based action games, but very few of those are genuinely good; the structure of the world makes or breaks a game like that.
I know a lot of procedurally generated game use a library of predefined spaces — I see that often in roguelikes. But even that makes for a tricky balance, especially with this style of game where the details of individual rooms has less impact on the quality of the experience than how those rooms relate to one another. So how are you approaching this, on both a philosophical and practical level?
JP: I would have to disagree with you there. To me, the contents of the room and the gameplay itself is much more important than the order of the rooms you run through. If you swapped some rooms around in Symphony for instance, would it really make the game less enjoyable?
The fun part of the game for me is exploring every possible nook and cranny, platforming through awesome set pieces, battling enemies and bosses, and improving your character. To that extent we’re also developing a concept of set piece rooms that are very unique, and guaranteed to always be on a certain floor. These rooms will help control the overall flow of the floor, leaving the procedural generation to make interesting new ways of connecting them. Maybe “metroidvania” and procedural are antithetical, as someone on NeoGAF so eloquently put it, but personally I don’t think so.
IGN: See, I’d make the case that the most important (yet most frequently overlooked) aspect of a good exploration-based platformer is the overall structure, the relationship of rooms to one another and the flow of the adventure. You don’t want things to be too obvious and straightforward or else you feel like there’s no freedom to move about, but if you have to do too much backtracking it becomes a boring slog.
You can see the difference in the Castlevania games: Symphony of the Night’s castle is pleasantly short on padding, with a logical flow from room to room. There are points where you can choose from different paths knowing that you’ll ultimately have to go back and explore the other route as well. On the other hand, Harmony of Dissonance’s castle is overly convoluted and confusing, with lots of superfluous rooms linked in an illogical way. I feel the biggest difference between these games’ reception isn’t their graphics or music or skill systems, but rather the overall design of their castles.
Look at Shadow Complex and Super Metroid. They’re largely linear with some freedom to poke around here and there, but then you hit a point where suddenly you have to completely change your course. They each only do this a one or two times apiece, so each instance has meaning, but the timing of these situations (around the midpoint of the adventure) is carefully considered and has a huge impact on the experience.
That’s why I ask about how you’re handling this aspect of the game. It’s a subtle thing, but important. Not to say that the design of individual rooms isn’t also essential, but how they all fit together plays a big part in making this style of game work, I think.
JP: I agree with you on the overall flow of the adventure, that’s more of the level we’re concerned at. If you can nail the tight gameplay and fun combat, and keep the overall flow nice and mysterious (you shouldn’t feel like you’re always on the “only” path), then I think the experience will come together. When people hear procedural generation, they automatically think that were dumping a bunch of blocks out of bucket onto the floor and that’s that.
I believe the overall flow of the adventure can be controlled, while using these pieces to build slightly different routes to checkpoints everyone is going to have to get to. For instance, you start the game, head to objective A, then go to B but see you can’t get there, so you go to C, etc. That kind of flow can still be preserved. Our goal isn’t to change that flow, it’s to make sure the 25th room you walk through isn’t the same exact one as last time, with the same enemies in it, and the same bag of money that drops from the third candle.
Anyway, we can argue about this forever, but its my job to prove it can work!
IGN: Ideally, how big will the game be? Metroid-like games usually span several hours, but randomized games (say, Spelunky) tend to be pretty compact unless we’re talking about some kind of massive ASCII-based roguelike. Do you have a target time in mind for an average single playthrough? How much of the play value do you expect to come from scale, and how much will arise from the simple novelty of constant change between sessions?
JP: I’m really shooting for the game to be about 10 hours in length. I can’t guarantee that yet, but that’s the length of the experience I’m really striving for. Now, remember, in Normal mode, you just reincarnate back on the surface, and can continue playing. In the demo there’s not really any penalty for death, other than the inconvenience of backtracking to where you were. We may change that for the final game, but we’ll have to figure out what works well for us. Maybe dropping some EQ, losing gold, or nothing at all.
We’re definitely going for a more classic action-RPG experience, it’s just some roguelike elements mixed in. To that effect, we do have the Hardcore mode which will be debuting with the next demo update. In that mode there’s permadeath, and you’re scored when you either die or finish the demo. Someone else can use the seed you used to play the exact same dungeon layout and try for a better score. We plan on the full game having online leaderboards and such to complement. We also added a Time Trial mode to the demo that just times you on your run to the end. We’ll probably allow a single save for permadeath in the full game, but it will be deleted if you die.
IGN: One other question I have about the randomness, and this might be getting a bit into the weeds, is precisely how many different “key” components the game will have. The best games in this style turn the player’s skills into keys to open new areas: Double jumps, more powerful weapons, grappling hooks, etc. It makes the process of growing stronger allow for organic exploration, which is a very compelling approach to design.
That being said, I worry about Chasm becoming repetitive with repeat play if you’re acquiring the exact same tools in the same order — first, say, the super missiles, then the double jump, then the bomb, etc. Will the appearance and order of these elements be randomized as well, and if so, how do you balance that with the core play mechanics?
JP: The problem with mixing up the power-ups is that we can’t tailor each area for highlighting the uses progressively. For instance, if you swapped the grapple hook for double jump, perhaps in the next area you couldn’t focus on making it grapple heavy, or tailoring the boss and other aspects for its use.
I think even with the major destinations being the same, we’ll still be able to do plenty of variation on the way there. Just think of a mini-boss that’s randomly thrown in your path, or perhaps you fall through to a mini dungeon, etc.
IGN: Based on the demos you’ve released, it seems Chasm is pretty far along. What role does your Kickstarter campaign play in the game’s development, and what happens if you don’t meet your goal?
JP: We’ve been working on it for about 5 months now, but that includes starting the engine and tools from scratch. I’m estimating it’ll be a year’s worth of content work, which is what the Kickstarter budget is based on. There will be 5 of us working full-time on it, plus operational, legal, and promo costs.
After we took the game to GDC, we had interest from several publishers. When we got back we were contacted by a couple, and we quickly found out why they have a bad name. I think it’s especially bad for a niche game like this that’s trying to preserve some of the old style systems. We were immediately asked to make changes and to give up IP rights. So Kickstarter seemed like the perfect solution for a niche game like this one. The fans want us to make the game we have envisioned, and in return they secure its development with our full creative control.
If the Kickstarter doesn’t work out, our only option is to seek out a private investor, or perhaps crawl back to a publisher. The risk there though, is that the game that gets made won’t necessarily be the one we want to make. I quit my corporate job to pursue what I wanted creatively, so I’m trying my hardest not to get back into the same situation.