Dude's intro to basic level design principles

Pretty well-laid out. I dare say I agree with most of what you said. Pretty clear, too.


My main gripe is in the post's layout. I found it hard to read the way it was spaced out.

Now that's been fixed, I'm quite glad with this.

 

EXCELLENT read. Man, there are not enough thumbs up in the world.

I'd love to hear some other prominent people's opinions on level design. I was talking to tweaker earlier in IRC about the use of the random dips and small bumps in the road you see in various levels, and he went into the technical reasons for them being there (engine intricacies related to how high you jump based on your angle and how its tied to the speed limited). It'd be awesome if he came in here and further explained.

Frankly, I feel this is a discussion that's been a long time coming. I'm looking forward to reading the replies in this thread.

Oh, and to contribute something to this thread, I'll repeat what David Jaffe (god of war's creator) had to say about level design in a gamasutra article - when it comes to rewards and puzzles, there is a specific forumla you should follow for success. That is - show the player the question before you show them the answer. The end result is that you give the player a sense of reward and it makes them feel as though they're accomplishing something.

The example he gave was in a zelda sort of game. You don't give your player a key in room A, then have it lead into room B where there is immediately a locked door. There's not puzzle solving in that. Rather, have there be a locked door in room A, and an open passage to room B where the key lies. This gives a sense of reward.

Applying this to sonic, think of marble zone act 1, where you have to push the block onto the switch. You see the switch before you see the block, and you interact with it first. the player gets to play with it and realizes when something is on it, the gate opens.

 

Yeah, sure thing.

Basically, the way collision works in the Sonic games is through a map of a bunch of predefined blocks in various shapes. Associated with these shapes are different types of responses that the physics engine will have to them. One specific reaction you'll notice is that, when going at a decent enough speed and jumping at an incline or slope at roughly below 70-80 degree angles or so, you'll be propelling much higher in the air than you would from a flat surface. I can't speak as to whether or not this is an engine quirk or an intentional effect, but it can be very nifty in reaching higher places in a level.

The cool thing about this trick? Many levels—especially hilly levels with lots of slopes—take advantage of this engine quirk. It can be used to traverse high sections of levels, used to access different and potentially-easier-to-clear paths, and to get certain goodies. Time your jumps and propel yourself right, and you can do cool shit like get on top of an inclined loop, reach high platforms hidden at the top of a level, and so forth. One thing to note is that when these types of angles are present in a level, they're always around open areas, and will generally lead to easier access to an upper path that might otherwise require some massive backtracking.

This trend isn't as present in games past Sonic 1 and 2, but they can still be used in some situations.

Another cool design quirk worth analyzing is the speed cap in Sonic 1, 2, and Sonic CD.

You see, Yasuhara's initial Sonic design principles centered around the idea that speed—and subsequently the proficiency with which a player can clear a level—were things to be earned, not given out liberally. As the player continued to play the game and develop their skill, the levels would become easier to traverse; you'd develop your own shortcuts and methods of beating a level in the most efficient way possible, whilst still working within the restrictions of the game engine.

Sonic 1 has a cap on the speed the player can accumulate speed through normal running. What this does is keeps the player from simply holding right and expecting to go faster throughout the whole game—remember, speed had to be earned by learning to circumvent this restriction. This is why Sonic can go at above-"max" speeds when not holding a direction; however, he also doesn't have a limit on how fast he can go whilst rolling. By learning to control the physics of Sonic whilst rolling, you develop skill at the game, and are subsequently rewarded with—you guessed it—speed!

This is where the speed in the classic games came from—trial and error, and quite literally being a good player. Anyone can go through the game, but only super players could go through the game with Sonic speed. This philosophy largely went out the window with the introduction of the Spin Dash and the removal of the running speed cap, which gave speed out liberally at a player's demand. However, as you'll notice, the game has a clever way of dealing with this.

First off, while the player can keep accumulating speed from running normally, he can't keep it the second he goes in the air whilst holding a direction. What this does is prevents the player from abusing their stored momentum, and forces them to roll into a ball to maintain those physics in the air; however, by rolling into a ball, they trigger a flag which forfeits the player's control over their momentum whilst in the air. In this sense, if you want to skip over large portions of the level by taking advantage of stored momentum that is released through the air, then you need to have the skill of timing and a lot of practice with rolling physics. If you get one bit wrong, you can miss your jump, fall back to the ground, or even potentially run into an obstacle that can kill or otherwise damage you.

While Sonic 1, 2, and CD had these sorts of restrictions, Sonic 3 did away with them. You could run really fast, jump, and not see the ground in a level for a pretty impressive amount of time. Sonic 3's tendency for levels to stack vertically made this less of a problem than it could have been, but imagine if you paired that behavior with Sonic 1-esque level design? You'll literally be running on top of the walls that you normally wouldn't be able to access.

Anyway, this is a fantastic topic! I'm particularly awaiting design lectures from both Jayextee and Jan Abaza—they always did have fantastic takes on the classic level design that really helped bridge together all of the intricacies in the Sonic games.

 

If I may, I'd like to add something to this that I have been meaning to say for quite some time.

This applies mainly to rom hacks of the 16-bit Sonic games but also other games that use a tile based system.
And that is "Make sure the tiles fit".

I can't stress this enough. That amount of times I've played an amateur rom hack and seen misplaced tiles that have obviously don't fit together is ridiculous. There are some exceptions to this rule (I.e. GHZ uses objects to make some tiles fit that you wouldn't expect too), but most of the time if it doesn't fit, use another tile.

To keep on topic though(if I was straying off-topic, that is), general flow of a level is important as well, which is something I'm slowly learning. I know only too well that I have made several levels in the past that are either too linear or too "stop and start" based.

Thanks for the advice Dude and I'll be sure to consult it in the future.

 

Great read, you pretty much nailed it.

One thing that I think you forgot to mention though, is that back then (and now, I guess), each level had varying degrees of this, so you didn't feel like you were doing the same thing over and over. There were levels that you could just blow through like Green and Emerald Hill, and then there were slower levels, like Marble Garden. Then there was at least one agonizing level, one that separated the boys from the men, like Labyrinth and Scrap Brain, where you really needed to be cautious and keep your wits about you. This helped the levels stand out from each other, and unlike games like Mega Man it's really difficult for me to confuse the levels in Sonic.

 

Something to also keep in mind in terms of the very first stages, and this goes along with proper tile lineup, is that they actually had a lot of empty space to them. Most layout edits of EHZ don't keep this in mind and as a result tend not to look or play as good as the original.

I like to jot my layouts on paper and then make them -- this way larger (e.g., 128x128) tiles can be re-arranged to get you what you want rather than another loop you don't need. EHZ's tiles were very much made with that level layout in mind.

 

Chalk this up to something that I always assumed was obvious. You should absolutely design your art around the level, not the other way around. If you design 128x128 chunks first and then try to arrange them into a level design on the fly, your game will either feel flat out broken and disjointed or very cookie cutter and repetitive.

Excellent point to bring up.

 

Absolutely wonderful topic you guys. One big thing about Sonic 1 that I miss is having to pick up your speed in order to go through a loop. Waaaaaay back in the day, when Sonic 3 had just come out, there was a video game contest at the martial arts studio I studied in. We had Sonic 1 and Disney's Fantasia running on the Genesis. I was the last to play, and was laughing at that point because obviously everyone there had picked up with at least Sonic 2: On everyone's first loop, they all spent their time trying to Spin-Dash, and just kept jumping in place. Everyone got frustrated and just gave up because the game "wasn't working right" and we weren't allowed to help our fellow students.

Luckily, even at my young age of eight years, I had played all the Sonics, and new how to fucking play it, so I think I got through the first Act of Marble Zone before my time had run out. My score was faaaar beyond theirs...but I still got second, because everyone else did great at Fantasia, and I didn't: I'd never played it, and back then, even movie games required skill and were actually difficult for "younger gamers". Everyone else who entered the contest was at least twelve. Well, that's what I get for not branching out too much farther than the games we had at home, not to mention at the time my only Genesis playtime happened at friends' houses; and they all just had Sonic 1 and Golden Axe.

I do remember my surprise when I first saw the Spin-Dash though. I was playing Sonic 2 at a friend's house (he had just got it and it was my first time), and I played without it. At one point in Casino Night, I accidentally did the Spin-Dash for the first time. I thought it was the best thing ever, but I never actually took advantage of it. I was always running or rolling, and stopping just to crouch and mash the jump button was kind of a pain. For me, it's always been one of those where I only use when I have to, and to that extent, it's a very nice mechanic.

 

Speaking of the spin dash, I always felt it was an underused concept in Sonic 2. It's a really cool mechanic that could have been used in level design but just... wasn't.

Thankfully, however, Sonic 3 (and knuckles) put the spin dash to excellent use - not via actual level design, but with tons of gimmicks (again, going back to what Dude said). Think of how many gimmicks there were in Sonic 3 and Knuckles that revolved around the spin dash to work - the spinners which raised the ledges in marble zone, the swings in ice cap zone, and obviously the elevators in lava reef zone. Pretty great idea to design gimmicks around sonic's only real move in the game.

With regards to Tweakers notes about rolling vs running in Sonic CD, it makes it seem like there was a far bigger difference between the super peel out and the spindash in sonic CD than I ever realized.

 

Actually, you just reminded me of a very important gameplay difference in Sonic CD—it allows you to maintain control of Sonic after a spinning jump! Could this be some sort of testament to the differences between Ohshima and Yasuhara's different takes on the Sonic formula? Who knows—but thanks for reminding me nonetheless.

 

Thats something I'd actually noticed before. I keep using it as an example, but collision chaos zone act 1 there is an entire jumping puzzle based around this.

Skip to 2:30 to see what I'm talking about

I love that part of the game (probably my favorite part) and I've noticed that for years.

 

You know, to be quite honest, I've always had a love/hate relationship with the Super Peel-Out. I love it because it helped introduce Sonic's "figure eight" running animation, which I love because...well, the "wheel" legs just didn't do it for me. I mean...several cartoon characters at the time would use such an animation when running fast. Even Wile E. Coyote. Because of that, I felt the "figure eight" animation brought Sonic's "speed" to a level higher than other famous cartoon characters of the era.

But I hate the Super Peel-Out due to how it worked. For one, I would've been more satisfied with it (and Sonic CD's Spin-Dash), if I had to mash the buttons repeatedly to do it: Holding up/down and A was kind of lazy to me, especially since I'd already played Sonic 2. It also didn't have much of a purpose in my eyes outside of show/something new to do. Even though running/rolling physics are different, the Peel-Out did (virtually) the same thing as the Spin-Dash, only if you ran into enemies you'd get hurt.

I don't even know if I explained it..."completely", as something like this is just, well, difficult to explain. Obviously you guys have your opinions, and I have mine; it's just that mine dismiss the Peel-Out as a "useful" technique.

That aside, I'd still keep it in the game because it just looks cool.

I'm one complicated, indecicive motha fucka.

 

You know what level was REALLY GOOD at this? Sandopolis Act 2. That level is beginning to surpass Lava Reef as my favorite in S3K having replayed it with that in mind, as though I was doing so for the first time.

Between the light switches, the sand dune ride, and the push-switches to move pillars, the level did a very good job in that regard.

 

That was a great read, and it hit pretty much everything on the nail.

 

I'd like to add something to this. Gimmicks are great. They are fundamental and all, but something I and others haven't really emphasized enough is the concept of gimmicks working in conjunction with one another to form what I'd call an activity.

Example.

Hilltop Zone 2: You enter a cave and a gate (gimmick 1) closes behind you. Lava starts rising (gimmick 2) and a small challenge begins for you to beat the flow of lava upwards with the use of moving platforms (gimmick 3). Along the way, some falling platforms (gimmick 4) can impede your progress and there are even a couple of optional areas where you can score some optional rings and an invincibility monitor. At the top right of the chamber, another door closes finishing the activity.

Another simpler one is when Mystic Cave has two moving platforms and between them, a hanging switch that opens the a gate. Another simple one is when Oil Ocean uses the pressurized popping caps to launch you into a machine that does the same thing as those ball shaped cannons and starts one of those cannon sequences. Complex systems littered Sonic 1's later stages, especially Labyrinth zone.

Good levels will introduce the concepts needed for the more complex systems individually before combining them to create activities.

 

Yea, the great thing about Sonic 1 and 2's levels were that they had their own personality. None of the "activities" were the same in every level. In Sonic games today, it's always light speed dash, boost panel, light speed dash, boost panel, bouncer, boost panel, a pit, etc.

 

Dude has made, simply put, one of the best topics I've read here. I can't help but to add my two cents about level design to what is already an excellent editorial on the subject. I'll not keep it short, either, as per Tweaker's request (heh), so be warned, because this is a long-ass post.

Level design is not chaos. Anything but. Even from the tentative steps into the genre that were games such as Pitfall!, things were seriously considered by the designer/author -- in Pitfall!, lower routes were generally harder but skipped about five screens (Compared to the upper route which advanced only a screen at a time), however both routes were required to collect all items and therefore complete the game. Although the screens are actually generated algorithmically, their function was by design; to allow freedom in the player's actions to an extent reaching beyond merely reacting to the obstacles onscreen (As was the case in, say, Donkey Kong). Ergo, a game widely seen to be the first real platformer (Traversing multiple screens comprising a whole, instead of tackling one self-contained area at a time) was actually non-linear. It could be argued that many games have take a step backwards since.

Like anything which is part-engineered and part-designed, there are two main 'strands' to maintain in balance when designing a good level; form and function. Form is an art unto itself, and I won't discuss that at length here (Possibly some other time), but is, at it's most basic level, applying artistic skills to ensure that a level fits into a cohesive whole that is The Game (Oh, by the way you just lost it =P) -- if everything looks and 'feels' like it is part of the same universe, it will be a more enjoyable experience, because immersion becomes easier; jarring awkardly into another art style will remind a player that they're in a game, and breaking the fourth wall in this way can be quite distracting, sometimes to the point of a player losing interest once they'd actually put some stock, mentally, into what the game appeared to be offering.

Function is where people trip up big time and so I'll be going into some serious depth about this.

Firstly, let us define the parameters of the vessel which contains a level, The Game;

Looking at the first definition (The one in question here), we can see that game have rules; these are clearly-defined structures which dictate what the participant of the game (The player) can and can not do. Levels are extensions of these rules, and represent puzzles within the constructs of the game; before a new level can be played, the current one must be solved -- which in Sonic's case usually means starting at the left of the level and traversing to the right. Any obstacles which make this less-than-straightforward (Such as a boss to defeat) are in fact macro-puzzles, this is to say a puzzle-in-a-puzzle.

"When is Jay going to get to the point?" - shortly, I had to convey to you the idea that a platform game level is a puzzle and contains puzzles, this is an idea I shall be referencing for the sake of clarity.

Puzzles in games have a few forms, all of which can be seen in various Sonic zones. All of which follow the narrative structure of conflict/resolution.

• Lock/key
• Pattern Recognition
• Application of the game rules/physics
• Application of real world rules/physics
• Skill test
• Memory test

I shall explain these and cite examples.

Lock/key puzzles - At it's simplest, there is a locked door and you cannot pass. Sure, this was seen in levels such as Pyramid Cave in Sonic Adventure 2, but where is it in, say, the original Sonic? The 'key' in this instance is a figurative device, and merely means an additional parameter which must be present in order to progress. A locked door in Labyrinth Zone requires a switch to be pressed - the key. A lowered weight in Marble Zone also requires a switch to be pressed; however, it may be part of a nested lock/key puzzle because in turn, this switch (In order for the solution to be permanent) requires a block to be pushed upon it. As a general rule, nested puzzles such as this one should either be avoided or used extremely sparingly; three layers of abstraction are the most you can attain before a player will get confused, but remember the level in itself is a puzzle, therefore within it, you should go for only two layers at most, as is the case with the Marble Zone act 1 example here. Anything more than this would be too much.

Of course, there is a non-literal application of this. Say, there is a rock face in Angel Island Zone act 2 which impedes progress. There is a parameter that must be met in order to continue. Try spin-dashing; in which case, the parameter is met. The player himself was in fact part of the solution, the ability used was the key. As another example, a loop cannot be traversed in Green Hill Zone act 1. This represents the locked door, and the key? Speed. Simply backtrack until you have enough room to accelerate to full speed. The key/lock mechanism is the core structure defining a level obstacle, with every chunk and object that gets placed, make sure there is a clear and definite solution, even if it's not immediately obvious.

Some of these puzzles can be circumvented, not solved, by taking an alternate route. For more on alternate routes, re-read Dude's original post

Pattern Recognition - This is as old as gaming itself. Simply put, things have patterns, learn to recognise them to play the game. what about the other end of the spectrum? Well, let us look at Marble Zone. If you see a lava pit and a pushable block, then this actually forewarns the player that the lava pit may be an extensive affair, possibly unable to navigate via normal leaping over it. The first example of this (Marble Zone 2) has a series of platforms above the pit, so the new player, completely unaware of this mechanism at this point (Because, you should never take for granted what the player does or does not know about a level) should have a get-out clause. However, the next time the block-surfing section is seen, it is less forgiving; and yet can still be navigated out of upon failure. Each time the block-riding is seen, the risk is ramped up slightly, until in Marble Zone 3 when a block absolutely must be ridden to pass a sequence; which the player now knows because s/he has seen it a few times, the pattern is in their head.

As a golden rule, if a player hasn't learned by the third time they see something, they deserve the punishment; so don't feel the need to repeat a simple sequence 9,002 times before your killer segment following the same pattern.

Application of game rules/physics - this crosses over with lock/key problems, in fact the latter two examples I cited are examples of this. In Sonic's case, the game physics can be quite complex (Refer to Tweaker's post for more on this), and there may be no way in a hack of, say, Sonic 2 to explain away a mandatory move or sequence of movements (Like Chip would in Sonic Unleashed, for example), so anything which uses a specific and non-obvious application of game physics should not be the only solution to any particular macro-puzzle. For example, the loop in GHZ1 makes sense to almost anyone, especially with the graphical confirmation of acquired speed that is Sonic's circular leg animation. However, some curves in Chemical Plant Zone 2 must be rolled towards from the preceding decline in order to access the higher routes. Notice how there is always a lower route in these cases? This is because it will be a new player's instinct to simply hold right when running, they don't initially know about rolling to gather more momentum. Think very carefully before making a puzzle with this solution; is the solution apparent? Is it obvious? Is it actually obscured via other means?

Having said this, some of the more excellent secret areas and/or hidden 1ups have utilised this principle. For secrets, go wild; a player always feels great when they access an area which seemed impossible to get to, and find hidden stuff -- they feel like they've outsmarted the designer who placed two 1ups somewhere thinking "They'll never ever find these!" (As seen in Carnival Night Zone 2).

Application of real world rules/physics - this is the thing that can really make a game seem great. Things which work as you'd expect them to in real life. Examples are the ring-attracting electric shield, running over the surface tension of Hydrocity's water, etc. Unfortunately, if you only design layouts there's little scope for this sort of application. For the ASM hackers out there (And other coding types), these can be some of the most enjoyable puzzles. However, use in moderation like evrything else. And don't go crazy. Sonic would've been no fun at all if it required a Physics major to beat.

Skill test - Simply put, a test of the reactions, timing and perception of the player. Ways to do this are obstacles on fast sections (although not top-speed sections. As Dude pointed out, the player most likely isn't gifted with ESP), smaller platforms (again, use sparingly and don't go crazy. The ability to discern how much is 'too much' will be one of the most important factors here) and enemies placed where they're highly likely to have a successful attack. Again, no ESP required; if it's hidden in a wall or floor, at least give some indication that they're there, like the Burrobots in Labyrinth Zone.

Memory test - You can guess this one, I'm sure; the only way the player can pass a section is if they've memorised it. By no means employ this for main level routes -- it can be used well for bonus areas or access to alternate routes, for example the right angle jump in a lightspeed section could lead to an upper route, which may be more rewarding, but definitely do not make this mandatory.

Right. That was exhausting to type. I've defined the sorts of puzzles (Or methods of conflict/resolution) that can be found in levels, let's see how this all fits with what Dude has been saying.

Tempo and flow are the key proponents of a level design, indeed the design of a whole game. When designing a level, the bigger picture should be visualised in the mind's eye as a general rhythm and pace. Is this going to be a level which just escalates constantly in a crescendo of difficulty, or is it going to have a relaxed middle? Is the level going to be a speedfest or an all-out badnik assault. Think about the level theme you're working with, and if it can work. But mainly, think about the rhythm -- I'm of the opinion that one of the main things wrong with Sonic Rush was the whole stacatto pacing of the game; speedy rail sections broken up by enclosed and forced enemy encounters? This was not satisying. Also, when the game finally presented some platforming, it was directly after a speed section; the transition was jarring and disjointed. Be a little more gentle with your pacing; the player is not some form of robot who can switch electronically from one mindset to another.

'Flow' is slightly harder to define. As Dude said:

And quite right he is. Establishing flow can be difficult, but (And I made a post to this nature fucking ages ago) possible with a lot of playtesting. Seriously, if you try one route of your level, and are constantly 'sticking' on walls, or you're trying to establish a rhythm and are incessantly interrupted by little things, that's bad flow. I'm not saying this has to be an Iizukafest of boostpads and straights, but if you have to slow down to a snail's pace in the middle of a speed section just to avoid some spikes and then somehow attempt to go fast again; it's frustrating. If you can, play some classic pinball tables. Seriously. The way ramps are positioned at angles from the flippers, or in relation to each other are all by design, this is to say it's not accidental. It is all planned. A good pintable rewards the player for hitting one ramp directly after another, and this is how score combos are set up. The fact that pinball-themed stages are a regular feature says something indeed -- the developers knew of the similarities between pinball and a hedgehog who rolls up into a ball -- Sonic behaves very much like a marble when in a ball, so ramps and bumpers make perfect sense.

This can be applied to level design in more general ways, as well. I frequently refer to Emerald Hill Zone speedruns when talking about level design. EHZ is a level which flows well at the best of times, but here is this amazing bounce-combo which cuts the level short. And guess what? This was not by accident. Indeed, an issue of the UK magazine Retro Gamer (Which I don't have to hand, otherwise I'd cite the issue number) had Sega staff refer to this as the "Yasuhara method of design" in an interview. Little non-obvious, 'hidden' tricks like this can be most rewarding when found and can add a real amount of depth, and replayability for those who love to time-attack. In fact I have a personal anecdote; I hid stuff like this in my old layouts for Sonic One (Which was gone, but is back on) -- material that is now in S-Factor. It was largely not recognised until recently, when Aquaslash told me he was addicted to time-attacking Spectra Valley 2, a layout I did. Lo and behold, he'd found a combo-bounce using the Spin-dash (Which was not ported to Sonic One, since I dislike the move), which turned out to be one I'd stuck in there anyway, but worked previously with a jump at a particular angle. This was a great payoff for me, since it established that I'd worked some great flow into that part of the level -- it was 'found' by Aquaslash and then gave him a compulsion to repeatedly play the level. Don't underestimate how important this is.

Another thing I'd like to see developed more is landmarking, this is to say individual segments of level that separate an area from anywhere else. One big flaw of later handheld 2D Sonic games is that all areas pretty much look alike; also a pet peeve I have with Metropolis Zone and some of the larger acts of Sonic 3 and Knuckles. I don't mean that you should have 9,002 different level chunks for use, just vary the way in which you place rings and items; make some memorable formations every now and again so the player knows where they are. It's a good idea to do this in bland, nondescript sections so they at least have something about them to remember.

"Wait, I'm a superduper über designer, I have no bland sections" -- no. Sorry, either you have them for pacing (Breathers between intense sections of action, speed or difficulty) or your whole level is a paceless affair, and as a result, bland all over -- what have Dude and I said about pacing?

In all, it's a question of great perception, and not overdoing things. I'm worn out from this post, so although I have more to say, I'm capping it here. By all means discuss, ask questions.

 

I've got to remark that there are a lot of segments in the classics that do just that and quite deliberately. Sometimes maintaining a good flow is meant to be part of the player's challenge, especially in games that feature speed running as a featured challenge.

 

Like all rules of design, it's only a guideline, but here's one that can never be ignored, my personal Golden Rule:

Only break a rule if you mastered it's application.

So every rule posted can be made exception to; only if you know how the rule works can you also make the exception work.