next »

[Tumetsu]

I while ago I was pondering about how to design levels (platfomers specifically) which encourage speedruns from players? What one should keep in mind when trying to design levels which could give the feeling for player that they always could do it a bit faster?

So far I haven't had any clear ideas for this since they haven't been my thing ever really but I had these thoughs about making levels so that there is constant possibility to fail or lose second or two which makes reaching the perfect time harder. Could be bad idea though...?

Any thoughs or experience in this area for designing or speedrunning levels in games?

[SirNiko]

The best way to do this is to create multiple paths and many different actions / items / choices for the player to use. Depending on what sort of a speed run you want, you can either include an obvious choice that is harder but faster (such as a hard path through the level, or a powerup that is difficult to use but defeats foes quickly) or a bunch of choices that seem to all be the same to the beginning player but a clever player can work through and narrow down which one actually provides the greatest time savings for each situation.

What sort of speed runs are you looking at? Do you see the player rushing to beat a static time goal to get a Gold Medal on a level, or do you see players playing the game over and over, making 3 second optimizations to squeak up a notch on a leaderboard?

-SirNiko

[Lifeisagame]

The best way to figure this out I think is to start doing some research in to the speedrun community and look at a lot of speedrun videos then analysing those. It can be quite hard if you haven't played the original game though to see what there doing different:
http://en.wikipedia.org/wiki/Speedrun
http://speeddemosarchive.com/

Here's a few things that I think you can keep in mind based on experience with friends that do speedruns:
1. Speed: Sounds simple but speedrunners will always look for the fastest way of movement. If they find a set of moves ( no matter how hard) that results in the fastest way to move forward they will use it.
2. Shortcuts: if they can reach the end using 2 out of 5 platforms they will do it. If they find a way to get higher then usual and skip stuff that way they will be happy. So provide multiple options to explore the game mechanics (jumps they can combine with kicks for example for just that extra height). Part of the fun for speedrunners is finding new strategy's to beat the level faster.
3. Glitches in the level collision: Speedrunners are very keen on finding glitches in wall so they can skip parts. They like to find stuff that is hidden.

There's a lot more probably but this is all I can think of right now. Good luck.

[Gnarf]

Oh I dunno. But I'd start by doing the thing that the beginning of Half-Life doesn't do...

[Ego_Shiner]

don't have any super precision jumping that makes you think twice, make it so you can make each jump without too much difficulty at a high speed, but still with a challenge involved.

[Brother Android]

I've always found this interesting and cool, even though I have little interest in speed runs myself. Jumper 3 has a lot of good design like this - every level has a time trial and some of them are quite ridiculous and fun to try for, often requiring you to take the riskiest path through the level.

I also remember watching

long ago and wondering if the game were actually specifically designed to make this possible. Some of the arrangements of enemies seem to suggest as much. Even if not, though, I think you can learn a lot about how to design speed runnable levels simply by watching speed runs, and I think it'd be pretty interesting if there were a game that used little things like enemy formations that specifically allowed you to make ridiculous jump chains, etc. Secret moves or moves with hidden applications are also a good idea (see Hero Core), as well as secret paths (obviously).

Aside: what is this, Google Chrome spellcheck? "runnable" is a word while "velociraptor" is not?

[Derakon]

Here's a few more checklists:

 * Deterministic levels, or at least levels without large random factors. You don't want your players to have to check 5 different locations to figure out which one you put the plot coupon in. That just makes the speedrunner do the same level over and over again until you randomly decide to spawn it in the fastest of those 5 locations (see also: Facility level in Goldeneye 007).

 * Non-triviality of speed. If the fastest way to move in your game is just to hold right, and holding right with the occasional jump is all it takes to complete the game quickly, then there isn't much depth to speedrunning (even if the placement of the jumps requires precision). Set up your game to require a variety of movements so the runner has to focus on creating smooth transitions from one set of obstacles to the next.

 * Taking damage as a speedrunning technique. Many games push you around when you take damage; this can be used as a speed boost. Some games knock you up into the air when you're hit; if you're already in the air at the time, this knock can let you access higher areas. Many games make you temporarily invincible after getting hit, which can be used for all sorts of things: running through enemies, walljumping off of instant-kill spikes, getting inside of enemies to spam attacks at 60Hz, maybe even avoiding event triggers. The more forgiving your damage mechanic is, the more useful it will be for speedrunning.

[baconman]

1. Variable abilities. For example: Sonic's spin dashing, MegaMan's Rush Coil, etc. If all your character can do is move and jump, there's not too much choice in what you can accomplish. This is even better if there's an ability or two you can acquire, that also contains an unexpected, unexplained use that players may not originally consider using it for. (Rush Coil/Search can actually block a shot for you in some later MM games!)

2. Risk/Reward. Contrary to another post, I say HAVE the tactical, precise jumping - this is a KEY factor in it, in fact. Don't make it actually impossible to do quickly - just make sure that the margin of error in doing so is fairly small, and that there's always a way to keep the keen player active and moving. For instance, rarely in a Sonic game do you have to wait for proper platform positions - most of the time, a tracklift will have multiple ledges to stand/alternate on, and they're all within jumping range of one another, albeit with a risk in doing so. The entire point is that it seperates the pros from the Joes.

3. Take "the high road." Put some safer "detour" directions in your levels, and direct the player towards them with arrows, coin trails, and so forth; and consider masking/hiding some of the more direct pathways, to give players something unexpected to discover. Lead them to one, later in the game (like a maze level), just so they discover that these alternative paths DO exist, but then don't tell them where!

4. Enemy positioning and reactions. Don't JUST post enemies where they're most obstructive - post them in appropriate positions to be CONSTRUCTIVE to players, as well. Perhaps a shortcut jump across an otherwise-impossible gap by bounding off of the correct sequence of enemies PLACED WELL BEFORE THE JUMP. So if you finish off the first one or two before you anticipate using them that way, the gap is no longer crossable.

5. Hidden Invincibility-or-similar powerups; and the necessity of using such to cross some too-trecherous terrain. Or as forementioned, a place or two where hitting an enemy produces a beneficial result.

6. Don't worry about intentionally glitching or breaking your game. People will do that for you, trust me. No matter how good you think you are.

[Tumetsu]

@Brother android
Holy shit that Mario video That's almost holy for guy who couldn't get past the first actual level of the last world in SMB3.

On topic, thanks. These are very interesting to read. Part of something which I had halfway figured out like hidden passages and such but mostly new stuff for a gamer/developer never gotten into speedruns.

[Derakon]

That Mario run is tool-assisted: it was played on an emulator with savestates and slowdown to create a movie of "perfect play", and the fact that the YouTube video doesn't call this out is annoying because it makes viewers think that this is a player with amazing reflexes, when in fact it's "just" a player with amazing planning. Other good tool-assisted runs to watch include Super Metroid 100% and Gimmick. There's tons more, but those two in particular show off alternate routes, creative use of abilities, abuse of damage boosting, and so on.

[psy_wombats]

* Deterministic levels, or at least levels without large random factors. You don't want your players to have to check 5 different locations to figure out which one you put the plot coupon in. That just makes the speedrunner do the same level over and over again until you randomly decide to spawn it in the fastest of those 5 locations (see also: Facility level in Goldeneye 007).

Though 'large' random factors are obviously not a good thing, I'd think levels with seemingly random yet deterministic elements would work well. For instance, moving platforms. For optimal times the player must reach the platform at the start of its, say, 5-second pattern. Otherwise the player would have to wait for up to 5 seconds for the platform to come back. It might open up the possibility of expending time earlier in the level for some reward later, as experienced players know they still have some time to kill that comes later in the level. Just a random thought.

[Derakon]

Though 'large' random factors are obviously not a good thing, I'd think levels with seemingly random yet deterministic elements would work well. For instance, moving platforms. For optimal times the player must reach the platform at the start of its, say, 5-second pattern. Otherwise the player would have to wait for up to 5 seconds for the platform to come back. It might open up the possibility of expending time earlier in the level for some reward later, as experienced players know they still have some time to kill that comes later in the level. Just a random thought.

I don't know about human speedrunning, but in the tool-assisted realm these are known as "frame rules" and are generally disliked. The reason is precisely because they introduce some slack into the game: improvements that you make that are not enough to make it to the next frame rule are basically pointless.

To make an extreme example, say you have a door that opens for one second every 20 seconds, and that it takes your average player about 50 seconds to reach the door. That means the average player will wait 10 seconds, and take 60 seconds total for that segment. Good speedrunners might be able to shave off enough time to make it to the previous cycle, which means they only spend 40 seconds for that segment. But it's physically impossible to make it two cycles earlier, so the best you can hope to do is 40 seconds. You might be able to do the segment in 25 seconds, but then you have to wait 15 for the door to open, so your gains are lost.

That said, I'm fairly certain that the moving platforms in the Genesis Sonic games are set up so that you can proceed smoothly onto them if you get to them at the fastest "intended" time.

[psy_wombats]

Though 'large' random factors are obviously not a good thing, I'd think levels with seemingly random yet deterministic elements would work well. For instance, moving platforms. For optimal times the player must reach the platform at the start of its, say, 5-second pattern. Otherwise the player would have to wait for up to 5 seconds for the platform to come back. It might open up the possibility of expending time earlier in the level for some reward later, as experienced players know they still have some time to kill that comes later in the level. Just a random thought.

I don't know about human speedrunning, but in the tool-assisted realm these are known as "frame rules" and are generally disliked. The reason is precisely because they introduce some slack into the game: improvements that you make that are not enough to make it to the next frame rule are basically pointless.

Oh hey, guess I learn something. Yeah, that would irritating, but any system like that could be made interesting with the addition of some way to sink time into score. (Well I guess that's not a pure speedrun anyway, but oh well) For instance, using your example, if there was a longer route earlier in the level that cost 15 seconds or so but rewarded the player with a small-ish point reward. A player knowing that they had those seconds to spare would be able to take advantage. Of course, it would be best if there was a way to "move" that spare 15 seconds from one side of the platform to the other, but that probably introduces ugly mechanical elements anyway. Probably best without that kind of stuff, you're correct.

[Gainsworthy]

Holy cow, people here really know their speedrunning. I'd like to add one little extra thing, though - quick reload times. Don't make the game go "OH HEY YOU DIED... wanna continue or go back to the main menu or what? Oh okay you wanna retry hold on a second loading respawning and yeah you can go now."

Have game over text you can skip. Put in a fast restart button. In the case of a game like, say, Spelunky, keep the action moving.

[SirNiko]

I'm also reminded of one game that I hated: Metal Gear Solid 4. You're rewarded for going fast, right? But there are all those obnoxious cutscenes that count against your time, and aren't interesting. Considering making a speedrun mode that eliminates the text and scenes so the player can smoothly complete their runs.

Metal Gear Solid for the Game Boy made it so that the speed run challenges were separate versions of the levels that had most of the objectives streamlined. That was good design, I think.

Mario Galaxy 2 made the speedrun levels shorter at times, cutting out bosses and other parts when they would get in the way.

I wouldn't really bother designing a game for a tool assisted speedrun, since it requires thinking on a whole different level, but if you did one thing to watch for is movement types that are boring and repetitious. Nobody wants to watch a 30 minute speedrun where the players is jumpkicking thirty times per second because it's the fastest movement method.

-SirNiko

[BorisTheBrave]

No one has mentioned sequence breaking!

Most typically of metroidvanias, this is when you complete levels out of order, or simply skip chunks of the game in order to save time. Typically this requires outside the box thinking to achieve, and you have to make though the rest of the game missing whatever power ups you might otherwise have got. I've seen game designer's do this on purpose.

[baconman]

No one has mentioned sequence breaking!

Most typically of metroidvanias, this is when you complete levels out of order, or simply skip chunks of the game in order to save time. Typically this requires outside the box thinking to achieve, and you have to make though the rest of the game missing whatever power ups you might otherwise have got. I've seen game designer's do this on purpose.

First one I discovered was called "Warp Zone." It let you finish SMB 1 and 2 in a mere 8-10 levels apiece. That actually doesn't seem so hard to do, outside of MetVania range. But you do bring up a good point.

[SirNiko]

Metroid Zero Mission went so far as to make sequence breaking a game in itself. There were purposely hidden alternate paths that let you avoid areas that required powerups you wanted to skip, and even a few places where powerups were placed along the path to force you to take a different route. Then, to top it all off, there was an extra ending if you beat it with a really low item percentage.

Of course, when you design it into the game, it's not really sequence breaking anymore because the sequence was designed to be broken.

-SirNiko

[Derakon]

From what I've seen, Zero Mission gets less attention from fans than Super Metroid does, largely because since the sequence breaks were built into the game, they're less fun. You get a nice little thrill of "I'm doing something the developers didn't intend!" when you break sequence in Super Metroid. Now, Super Metroid had mechanics that were obviously capable of producing sequence breaks -- most notably walljumping, bomb jumping, and the speed booster -- but it didn't go out of its way to put in secret side passages that let you bypass otherwise-powerup-requiring obstacles. It just designed those obstacles in such a way that you could navigate them with some cleverness instead of with the required powerup.

For example, there's a path leading to the Wrecked Ship sector that requires you to use the grappling beam to swing from one platform to another. In fact, the grappling beam is not required. The Zero Mission way to handle this would be to have a morph ball tunnel that lets you roll past the entire segment. Super Metroid instead has the platforms spaced carefully so that you can jump from one to the next if you know exactly what you're doing (with some leeway if you're good at walljumping). Or you can build up a speed charge on the shoreline in a space that looks too short for it, and then fly past the entire thing.

In other words, if you're going to design a game with the intent that its sequence be breakable, be subtle about it. If you hand the breaks to your audience on a silver platter, they're less interesting. Sequence breaks are meant to be hunted for.

[BorisTheBrave]

Right, good sequence breaks should make you *know* the game, and exploit its mechanics ruthlessly.

But herein lies the problem I've found with level designing around that. You see it in quite a few games, that, say, because there is a secondary route without double jump, there is always a way to complete the game without double jumps. But that means that the double jump route is always very easy, because you have huge margins on anything. There are no segments that really test a player who isn't doing the alternate route, but is still reasonably good. It also becomes blindingly obvious that that it is possible, after seeing one too many conveniently but unnecessary platforms.