A few people have asked me how I go about making my bullet patterns, or how I did certain things. Some people have noted that they love my Cave style of shmups. Some of my games have even been called fun. This obviously makes me an authority on all matters having to do with making things good.
First, a foreword. I have broken every single guideline in this guide before, owing to inexperience and being dumb. But, having several bullet hell shmups under my belt, I figured I would share my acquired wisdom on this subject with the Development forum, and hopefully some aspiring new developers will see it and take what I have to say into account. Also, I will point out that I'm not exactly an encyclopedia of shmup knowledge, so the only examples I'll use will be from games that I've played.
I will further add that this is NOT A PROGRAMMING GUIDE. I'm not here to teach you bullet generation algorithms; this guide is all about design.
There is one cardinal rule with boss fights, and any shmup should follow it whether or not it is a bullet hell. That rule is:
FORCE THE PLAYER TO STAY ALERT.
A voice in your head might be telling you this means "make it harder lol." Find that voice and punch him in the face, because he's an idiot. Even easy bosses can keep you on your toes.
So how, then, do you do this? Basically you want to avoid this rule's natural opposite, which would be if the player ever feels like he's simply spinning his wheels, constantly doing the same thing over and over. Therefore, the easiest way to do this would be to simply have the boss switch up its attacks frequently. Cave games do this a lot; the first boss in DoDonPachi has several attacks that it switches between, using one attack for no more than five to ten seconds. As the fight gets closer and closer to ending, the boss gets more aggressive, occasionally tossing in a wave of faster bullets amidst its regular attacks. This boss keeps you on your toes, even if you do find it easy, because you're doing something different every few seconds; it doesn't simply feel like filler between stage 1 and 2.
The method of changing attacks frequently has another benefit: it gives boss fights a much better sense of pace, as though things are moving along more quickly, even if the boss's lifebar isn't moving any faster. This is an important consideration depending on how fast-paced your game's stages are; you would want the boss fights to be paced about evenly with the stages themselves.
Touhou games, and any games modeled after Touhou, use a very different method of keeping the player on their toes. Given their "one phase, one attack" way of doing things, they instead have to rely on making the attacks themselves more interesting. In general this involves close-quarter bullet dodging, wherein the player is always in danger regardless of how they set themselves up to deal with the attack. Because of the slow bullet speed in Touhou games, they're also better forgiven for employing random bullet trajectories that avoid making it so an attack can be completely memorized. Thus, it's fair to say that a player has to rely a bit more on reflex dodging than on memorization. RANDOMNESS WILL BE DISCUSSED MORE LATER.
Cave games will often combine the two approaches. The stage 4 boss in DoDonPachi DaiOuJou uses many attacks that are sustained for quite a while, but because of how those attacks form, you're still called upon to dodge skillfully, and even then, the attacks never stick around long enough to become monotonous.
SO HOW THE HELL DO I MAKE BULLET PATTERNS GOOD?
Make your boss's first bullet pattern. Then make its second bullet pattern. Is the second bullet pattern clearly and obviously different from the first? Good. Make the third bullet pattern. Is it clearly and obviously different from the first and second? etc etc etc
This will essentially get you to try out new things, to think outside the box as it were. If you put yourself in this mindset when making a boss's bullet patterns, you're going to make some cool ones whether or not you actually know what mental or creative process you went through to do it.
Note that when I say "clearly and obviously different," I do not mean "the color of the bullets is different" or "the bullet speed is a bit higher." I mean the two attacks need to FEEL different as well. The player should have to approach it differently, and they should have to make different decisions in the middle of it to figure out how to avoid it.
Note that the more attacks you give a boss while following this rule, the more likely it is that you'll come up with a truly unique one, of the likes nobody's really seen before.
SO HOW LONG IS A BOSS FIGHT ALLOWED TO BE ANYWAY??
Depends on where in the game this boss fight takes place. A stage 1 boss shouldn't exceed 40 to 50 seconds, if it even gets that long. After that, a minute and a half is probably a good place to call it. If you're talking about the final boss or true last boss, there is a lot of wiggle room. I find the most fun final bosses/TLBs don't exceed two and a half minutes though.
You could probably get away with making a boss fight as long as you can keep its attacks interesting, but in following this train of thought you might fall into a trap of making the boss itself outstay its welcome. You also need to consider the length of the boss fight in the context of the entire game. A lot of players seem to find that 25 minutes is a good length for a shmup, so be sure to consider that when figuring out how much of that time you want to be spent fighting bosses.
Regarding timeouts... I personally think a good length of time for a timeout is twice as long as it would take to kill the boss if you were just constantly shooting it, but obviously this depends on the game and what other mechanics there are. If there is absolutely no benefit to keeping a boss alive, you probably don't even need timeouts at all, unless you intend to have replays, in which case timeouts are a necessity to keep the replay filesize from being theoretically infinite. Furthermore, timeouts should definitely exist if there is a way to milk the boss somehow. On that note...
WHAT ELSE CAN I DO TO MAKE A BOSS AWESOME?
Make bosses worth a damn to play for score.
That voice in your head from earlier is probably telling you to add in milking opportunities. Punch him again, that voice is an unspeakable frothing dipshit. Punch him again. Keep punching him until he starts to resemble a being that isn't excrutiatingly retarded.
A lot of games make bosses worth playing for score by making their bullets turn into point items during a phase transition or the boss's explosion, thus encouraging the player to wait until the screen has as many bullets on it as possible before ending a phase. Some games add onto this by only having this happen if you're doing a specific thing (the Kakusei mechanic in Espgaluda, for example).
Some games go a more basic route; as you shoot the boss, you get points. Or even more simple, the boss gives you a bunch of points when it dies. This is okay, but it doesn't really add any emphasis on style or technique.
One of the easiest, but also most fun, ways to add scoring potential to a boss is to add destructable parts to it. Maybe the destructable parts turn bullets into point items, maybe the parts drop point items of their own, or maybe they're just worth a lot to destroy. In any case, this adds another dimension to scoring off of bosses.
To up the stakes a bit, you can have your performance in a stage somehow affect the score you can get from a boss. Ketsui does this by having your boss multiplier accumulate throughout the entire level, and during the boss fight you can finally cash it in. DDP DFK does this by letting your chain carry over from the stage itself. Mushihimesama Futari does this in all of its modes by way of its stage counter/multiplier.
I decried milking tactics earlier, but it's also worth considering that some players actually find milking fun sometimes. You can add milking tactics if you want to, provided the first cardinal rule is obeyed. If you need a reminder of the first cardinal rule of shmup boss design, scroll back up to the top.
HOW DO I KNOW IF MY BOSS IS TOO HARD/EASY/WHATEVER
Well, there's the obvious way...
Difficulty is one of those things that's extremely difficult to judge, but I have a handy little tip that'll help you to figure it out, and it's probably not one you've ever thought of or even heard about, but for bullet hell shmups in particular it'll go a long way to helping you figure out how hard your bosses are to an almost mathematical degree.
ADD GAMEPLAY TIMERS TO YOUR GAME.
A "gameplay timer" simply records how much total time you have spent in your game across all sessions of it. By this time you're probably wondering what the fuck this has to do with anything. Let me furnish you with a scenario.
You're well into developing your game, and quite frequently, you test each stage just to make sure the flow is consistent between stages, the game still feels all right, and all that stuff. At some point you're going to start thinking that your game is too easy, or "that part was kinda hard, but not as hard as I intended." Before you change ANYTHING AT ALL, look at the gameplay timer.
IF YOUR GAME IS CHALLENGING TO YOU AND YOU HAVE FORTY HOURS ON THE GAMEPLAY TIMER, IT IS NOT TOO EASY.
I will repeat that.
IF YOUR GAME IS CHALLENGING TO YOU AND YOU HAVE FORTY HOURS ON THE GAMEPLAY TIMER, IT IS NOT TOO EASY.
Consider that any given player who picks up your game will need to spend at least that long in practice just getting to your skill level. This time COULD be halved if you're dealing with, say, one of the Touhou Is Harder guys, but in general, players will need to spend quite a bit of time in practice mode figuring out things that you already know because you made the game and have spent hours upon hours testing it. You may not realize it, but a test run is also a practice run!
For bosses specifically, you can add either gameplay timers specific to them (showing how much time you have spent fighting that one boss) or just an "attempt counter" that shows how many times you've fought it in total. If you've put 100 attempts into every boss, then you should be able to head into every boss battle in order and notice that the first boss or two are piss-easy, and from there get progressively more challenging. THIS IS THE SITUATION YOU WANT.
WELL JEEZ, WHAT OTHER SUGGESTIONS DO YOU HAVE FOR ME BROSKY?
Here's a handy one, and it's not specific to bosses; you can apply it to basically anything that's intended to be played competetively.
THE RANDOM NUMBER GENERATOR IS AN ASSHOLE WHO HATES PLAYERS AND DEVELOPERS.
Since my first draft of this section, a few people have spoken up declaring that I was wrong and a kitten-molester for saying that randomness sucks. Actually they didn't say that, it was more like they respectfully disagreed with my position on the matter.
The RNG is a tool. It can be used to good effect. It's also extremely easy to screw up. This is because it is an asshole who will screw your game up if you let it. I will first give a few examples of how randomness can be used poorly, and then I'll proceed to give guidelines on how to use it to a more positive effect.
A reasonably well-known example of randomness going wrong is in Touhou 6: EoSD, the Sakuya and Remilia fights specifically. Remilia in particular has an attack that's intended to have pretty consistent density most of the time, but in the event that God feels like fucking with you, it can toss a wall at you. You cannot plan for this wall to be there. You cannot do anything about it once it's there. You just have to hope it doesn't appear. THIS IS A TERRIBLE THING IN ANY COMPETETIVE GAME BECAUSE IT LEAVES SCORING PARTIALLY UP TO COMPLETELY RANDOM, UNFEELING CHANCE.
I will illustrate another problem with the RNG with a scenario. Let's say (for whatever reason) you make a boss with four attacks, and it can choose between these four attacks randomly. Once it finishes with its attack, it once again chooses from the four randomly. Seems like an okay setup, right? Four attacks is a decently varied array of abilities, and the unpredictability will keep players on their toes.
Wanna know what will necessarily end up happening?
At least one player will report that the boss used its hardest attack over and over and over. Another player will find that the boss was too easy because it only used its two easiest attacks. You as the developer will test this boss and find that you can't get it to use its attacks equally no matter what the fuck you do.
This is the nature of randomness. You think you can get it to work how you want it to, but the truth is that it is a completely unfeeling entity that does not give a fuck about you.
Is there a time when random chance is acceptable? Yes, and that time is "when you're careful". If you're going to use randomness to decide bullet trajectories, try to find a way to keep the bullet density consistent, and above all, try to make sure that a bullet pattern has the same general "shape" each time you see it. An easy example of this is in a circle of bullets radiating out from a point; if each wave has an angle of ten degrees between each of its bullets, then every wave will have consistent density, and you can simply randomize its trajectory by ten degrees. This keeps the attack from being memorizeable (forcing players to dodge on reflex, which can be more fun), but it also ensures that the attack has the same difficulty every time you see it. Even this simple example has drawbacks though. If you use more than one of these radiating circle-waves of bullets in a given attack, randomness could make the waves overlap awkwardly sometimes.
General guidelines when using randomness:
* Keep attack densities consistent
* Do not have it play a part in the scoring mechanics at all
* Don't decide attack orders based on randomness
If you're a new developer, I'm pretty sure you're better off not using randomness at all, in favor of crafting your patterns manually. This will force you to better understand what makes bullets patterns do the things they do, and you will by proxy learn how randomness can fuck everything up.
But Giest, I hear you saying, won't static bullet patterns make the game predictable and boring?
That depends. How clever are you feeling?
In Trigonometry Wars 3 Redux, I did not use the random function for any bullet patterns at all. But if you play that game and pay attention, you may notice that some bullet patterns toss the bullets on different trajectories each time you play. This is because I varied the patterns a bit based on the player's position. The patterns are not aimed, they just change based on weird things like the sine of the player's x and y positions. Unless a player can perfectly replicate their movement at all times, the patterns will be different each time they play, but I am also in complete control of how the patterns can form. I know what the ranges and potential values of the bullet trajectories, so in the event that I do accidentally make a pattern stupid or randomly wall-y, I know exactly what number I can change to fix it.
The sine function can be a wonderful alternative to randomness, because it can generate the look of something being random without actually using randomness. To demonstrate what I'm talking about, take a look at this fucking pattern right here:
Each wave of this attack shoots nine bullets, evenly spaced by 40 degrees from each other. Every two frames, a new wave is spat out. The bullets move along a trajectory decided by this function:
Code: Select all
1800*sin(Theta)Now, look at that screenshot. A discerning eye could see how the sine function is playing a part in it, but during gameplay, the bullets are moving too fast for a player to take that fact in, and all they will know is that they need to read and dodge the attack. It's too much of a smattering of bullets to really memorize a pattern in it, even though it will do the same general thing every time you see it. Also, because the sine function will basically always spit out an irrational number, each 'wave' of the sine function(when it's used in this way) will spit out a different set of values. In other words, even after the pattern loops, the second round of it will spit the bullets on a different set of trajectories.
Also I varied the initial value of Theta based on the player's position when the attack starts, naturally.
A note: How the function above works (and even IF it works) is dependent on your development environment. I use Game Maker, which means I have the luxury of an environment where that kind of thing just works without any extra consideration from me.
I could go into further discussion about the pros and cons of randomness in shmups, but that's beyond the scope of this guide, which is just about bosses.
HOW DO I DEAL WITH SAFESPOTS?
First of all, safespots can be fucking rad sometimes. I would almost suggest intentionally putting two or three of these in your game. As long as they're hard to find and to trigger, they can actually work beautifully.
Most of the time though, you want to prevent them. An easy way to do this is to throw in an aimed component into your boss's attack. This will ensure that there can never be a safespot.
An even better consideration is to use real numbers (as opposed to ints) for bullet directions. If an pattern's directional coverage is that it can potentially throw bullets in 1500 directions, that's a hell of a lot less likely to have a safespot in it than a pattern that can only throw bullets in 360 directions.
WHAT DO YOU THINK OF SMALL BOSSES?
Definition: A "small boss" is horizontally small, and therefore player position matters considerably more as far as actually hitting the thing. The main thing worth considering with this kind of boss is how much health it has. A small boss with equal health to a large boss will likely take longer to kill due to being much harder to hit, but that's not a guaranteed thing. The thing to be mindful of when making one of these bosses is how small it is, and how much it moves around; a boss that moves around a lot while filling the screen with dense bullet patterns will be nigh impossible to do consistent damage to, so be sure to test it thoroughly and see how much health the boss should have for how long you want the fight to last.
WHAT ELSE DO YOU WANNA TELL US?
Hell, I dunno. Post questions, and I'll update this guide with clarifications.
)
