Why shmups are such a niche genre

[shmupsorrow77]

randomly generated levels would probably be nice.

It's been done.

It's also proof for why this is not a valid proposal for 'evolving' the genre. As much as I love Kenta Cho's games (and I do), they're still just small time. Simple ideas put into practice by some guy on his free time. The more complex something is, the easier it is to break it. Procedurally generated content is the fastest way to break any game mechanic known to man.

Is there any non indie game that uses it that is actually good? Mediocre implementation doesn't count. If they did Diablo 1 and 2 would not have been as successful in their random dungeon generation. Quality of execution in generation is what matters.

[shmupsorrow77]

Game developer
Ted Brown
24 Mar 2010 at 12:04 pm PST

Gentlemen, please. Not all games are art, or even have a chance to aspire to be. This is a fantastic foundation for what the majority of studio developers are trying to make: well-crafted products designed to serve an audience. It's not a guarantee for success, but at the beginning of a multi-year endeavor, with tens of millions of dollars on the line, these things must be taken into account.

"Making games is very little about you and very much about millions of other people."

If you do not believe this, then you are making art. And nobody is going to hand you Gears or God of War money to turn your artistic goals into a game. Of course, I hope I'm wrong, but there it is.

As for the parting shots on "new and original" being left out, well, if you re-read the first part of the article, he sets them aside as phenomenons that can't be planned for.

Source --
http://www.gamasutra.com/view/feature/4 ... y_the_.php

So is Ted brown right, is gaming very much about millions of other people?

This means if the shmup community wants to expand it would do well to stop being so insular and hostile. Since the only ideas the shmup community is going to come up with by and large is those that are insular and niche.

[Rob]

Yes because we all know diablo 1 and 2 were such failures at random level generation.

This just shows you how meaningless level design is when the goals are to click mindlessly, level up and collect.

[Zeether]

somedody ban this clown

[shmupsorrow77]

Yes because we all know diablo 1 and 2 were such failures at random level generation.

This just shows you how meaningless level design is when the goals are to click mindlessly, level up and collect.

Your comment does not in anyway disqualify the idea that random level generation well designed and implemented would be a bad thing.

[BIL]

This means if the shmup community wants to expand it would do well to stop being so insular and hostile. Since the only ideas the shmup community is going to come up with is those that are insular and niche and by and large that is what has been keeping the genre niche.

You still don't get it. We know full well that the kind of games we enjoy aren't compatible with mainstream popularity. The most shooters can hope for is to gain a less tenuous foothold than the one they currently occupy in the West.

The kind of randomly-generated Descent knockoff you're rabbiting on about is exactly what nobody here wants the genre to become, even if it were to mean mainstream "success."

edit: and if you think this is a "hostile" reaction, wow. The most we've done is told you your ideas suck. You started this exchange by declaring much the same about the games we enjoy. A forum like SRK would've torn you a new e-asshole by now.

[JoshF]

Jean-Claude Van Damme movies need more romance and comedy to appeal to a broader audience. Why are they so insular with that martial arts action nonsense? Moreover, mass appeal is always synonymous with par excellence, ergo Two And A Half Men is the greatest television program ever. You really hit that one out of the park!

Completely unrelated, but I'm looking for a burnt-out twenty-something gamer who feels he has to justify to society why he still likes playing with toys, inevitably leading him down the GAMES ARE ART path. Just in case you find him tell him to check out this forum as he'll be in better company there.

http://www.selectbutton.net/

[szycag]

These games are about scoring and learning boss patterns, having attoned reflexes to compensate for very calculated but still difficult, fast patterns. Of course randomly generated levels work for dungeon crawlers, they're not action games. A totally randomized shmup experience would mean scores aren't comparable, not to mention patterns that were either small variations on predictable patterns or just noisy chaotic (and ultimately boring) patterns. We want fast and quick arcade experiences, not freeform space battles. It doesn't matter how small the market for these games has gotten, people are always going to want them, even if it's just playing already released games, because they offer something the kinds of games you like do not offer, BY DESIGN. See you talk a lot about game design while ignoring why shmups work and have worked for so long, even as the majority of people now would rather play interactive movies. Or in your case, overwrought messy boring simulations. Just hold your views and don't challenge ours, we're not the ones with small minds, it's just that your daydreamy ideas of what these games are all about are fundamentally flawed because you don't seem to have an understanding of why we play them.

[Rob]

Exactly. The random stage design suggestion is especially funny, since it would mean reducing the genre to a test of hand-eye coordination. I could enjoy a few games like that, but I don't think further simplification is what (he thinks) he's after.

[NR777]

This means if the shmup community wants to expand it would do well to stop being so insular and hostile. Since the only ideas the shmup community is going to come up with by and large is those that are insular and niche.

To be perfectly honest, no one here gives a flying fuck if Joe Casual gets into the genre. What you've said here is the equivalent of going into a horror movie forum and saying "well, in order to appeal to more people, you really need to cut down on the gore and the tits and that fear stuff." In either case, no real fan of the genre you're excoriating is going to respect your opinion because it is, quite simply, utter horseshit. Your statements boil down to "the only way to improve the genre and make it suitable for mass consumption is to alter it to the the point where it ceases to be." Again, if you don't like it, that's fine. As is being demonstrated daily by places such as this, there are plenty of people around who do enjoy the genre for what it is. This is an enthusiast's forum and you are clearly not an enthusiast. I ask again: why the fuck are you here?

[Taylor]

Is there any non indie game that uses it that is actually good? Mediocre implementation doesn't count. If they did Diablo 1 and 2 would not have been as successful in their random dungeon generation. Quality of execution in generation is what matters.

Diablo is orientated around stats, and player goals are not second to second. This means the little pacing it requires is easy to implement in a random algorithm, and difficulty can easily be spread out on per area basis. The areas themselves have strict, static guidelines so you'll always find the same enemies, bespoke objects and area links - as randomising these would screw up actual progression. Its generation is far from flawless anyway; it's very common to find bad exit positioning and horrible compositions of named enemies. Oh, and last time I played that game everyone used map hacks to remove the randomness. Though of course that's all irrelevant, because hack and slash dungeon crawlers themselves are irrelevant.

But supposed you did create an algorithm for creating random levels for a shmup that worked perfectly, the existing shmup players you clearly know and love would not like it. Because consistency and iteration are pillars of the genre.

[Zeether]

Shmup levels are repetitive, i.e. there is no surprise or unknown factor about what's coming next, randomly generated levels would probably be nice.

[Rob]

Let's play basketball on a court with no markings or hoops. I thought of this after playing dodgeball.

[NR777]

Let's play basketball on a court with no markings or hoops. I thought of this after playing dodgeball.

/thread

[BryanM]

Exactly. The random stage design suggestion is especially funny, since it would mean reducing the genre to a test of hand-eye coordination.

The level generator can use a seed. So it's the same levels over and over and over. That can be learned and mastered.

The commercial companies have no incentive to create a multi-course game this way, for lots of reasons. It doesn't mean the concept utterly sucks.

I know I know "my genre is a special snowflake." ... but it really isn't. Those stages in Super Mario Brothers are just screens of layouts knitted together, Chrono Trigger is a series of very loosely connected, at best, scenarios linked together, Shoot Them Aparts are a series of aesthetically pleasing encounters. And the internets is a series of tubes.

I would like to demonstrate this with an actual product someday, but the future is long, dark and uncertain, and always bleak.

[lgb]

these games are not "dead", someone still plays them; all genres are like this. they evolve and they change, and then you have more genres, more ways to waste time and have fun. there is no change needed for "shmups", they are fine as is.

any genre is going to have similar features between the games within them. the genre does need to "evolve" so radically, or it would just become a new genre. we already have such new genres, many of them, some of which are confused with what we at SHMUPS! would consider a "shmup", and some which are even being considered "shmups". thus, this genre has already evolved, and it's actually doing quite well. then you have the people who would like to play what we would consider "shmups".

so what is the problem, here?

[Davey]

randomly generated levels would probably be nice.

It's been done.

It's also proof for why this is not a valid proposal for 'evolving' the genre. As much as I love Kenta Cho's games (and I do), they're still just small time. Simple ideas put into practice by some guy on his free time. The more complex something is, the easier it is to break it. Procedurally generated content is the fastest way to break any game mechanic known to man.

Is there any non indie game that uses it that is actually good? Mediocre implementation doesn't count.

While not perfect, Parsec47 is awesome.

Game developer
Ted Brown
24 Mar 2010 at 12:04 pm PST

Gentlemen, please. Not all games are art, or even have a chance to aspire to be. This is a fantastic foundation for what the majority of studio developers are trying to make: well-crafted products designed to serve an audience. It's not a guarantee for success, but at the beginning of a multi-year endeavor, with tens of millions of dollars on the line, these things must be taken into account.

"Making games is very little about you and very much about millions of other people."

If you do not believe this, then you are making art. And nobody is going to hand you Gears or God of War money to turn your artistic goals into a game. Of course, I hope I'm wrong, but there it is.

As for the parting shots on "new and original" being left out, well, if you re-read the first part of the article, he sets them aside as phenomenons that can't be planned for.

Source --
http://www.gamasutra.com/view/feature/4 ... y_the_.php

So is Ted brown right, is gaming very much about millions of other people?

This means if the shmup community wants to expand it would do well to stop being so insular and hostile. Since the only ideas the shmup community is going to come up with by and large is those that are insular and niche.

I think you're misinterpreting his message. Note that he says "Of course, I hope I'm wrong, but there it is.". So yes, he is saying that most games are tailored towards the tastes of millions of other people, because that's how you make money. But more importantly, he's saying that games without mass appeal won't get funding, which is unfortunate because unique games have a poor chance of ever seeing the light of day. You've been proposing the opposite: make shmups less unique so that more people like them.

As for the randomization issue, I don't agree with the rest of the forum. If well implemented, I think it could be a great thing.

However,
- AFAIK, only indie devs have really dabbled in it. Kenta Cho games would be a good example, as well as nomltest. Geometry Wars is random as well, and the difficulty is fairly consistent between runs IMHO. Yeah, technically GW is a commercial game, but it shares more in common with indie games than commercial ones.

- Speaking of nomltest, that had to have been one of the most bitched about games in STGT history, which is saying a lot. But I think it brought to light that even within this very niche genre, different people want different things. People who like to break shmups down systematically would never like procedurally generated stages because the game is different every time. OTOH, randomization might be well received by people who mostly enjoy playing using their intuition and general skills.

- I don't think randomly generated levels are inherently unbalanced, or at least not enough to make them an inherently flawed idea. Parsec47 isn't terribly consistent, but it was made by a single amateur in his spare time (albeit a very talented amateur). At least it has a fairly simple system to try to smooth out the inconsistencies: it gives you frequent extends, which helps counteract unfairly hard parts, but it caps the extra life stock at 4, so you can't overcapitalize on easy parts. It's certainly not perfect, but it works pretty well for an indie game. I'm sure a team of professional developers working full-time could come up with something that's pretty evenly difficult between runs.

Also, during STGT, my nomltest scores were pretty consistent between runs and rose gradually over time. If it were unfair you'd expect more spikes and valleys, and that the general trend line would stay flat because it's based mostly on inherent skill. But I'm not sure if that was everybody else's experience.

- Even random shmups have some structure to them. nomltest can only throw a few different types of patterns at you; the variation comes in what combination of enemies you get and how they overlap. Parsec47 throws the same set of randomly generated enemies at you over and over for a few waves, so basically the flow of the game is to see a new pattern, learn it quickly, and apply that knowledge for maybe 15 seconds until the next wave starts. Geometry Wars has a general flow to it as well; the enemies are different every time but there is a general template that the game follows.

- Professionally made puzzle games are generally randomized yet fair, but they often have checks in place to keep things consistent. For example, in Tetris the Grand Master, the randomizer keeps track of history and skews its selection so that you don't get 10 Z-pieces in a row. The levels in Mr. Driller are different every time to you play, but the difficulty doesn't really change between runs. However, I'm not sure if the levels are generated by an algorithm or if there are just a lot of pre-built stages that get picked randomly. Granted, balancing a few pieces or blocks is probably way easier than ensuring consistent difficulty among complex, overlapping bullet patterns, but I'm not so sure it's technically impossible.

- Some types of bullet patterns wouldn't be workable. For example, Psikyo-style patterns with crazy fast bullets would have to go, since they're not really possible to navigate the first time you see them.

[Taylor]

Good level design isn't a science.

[Pirate1019]

Is there any non indie game that uses it that is actually good? Mediocre implementation doesn't count.

I said small time. Remove the reality-warping goggles.

big post

I still don't see how you can make a shooter with the complexity of ESPgaluda or DOJ, and still incorporate random elements while insuring balanced scores. You used Nomltest as an example, but unless I'm mistaken nomltest lacks any score mechanics. Parsec47 has item collection , but it's still fairly simple and the game is very much about the dodging and surviving. The mechanics need to be simple to accommodate the variation. Not a problem, but I think Kenta Cho fills the quota for procedurally generated content.

[lgb]

I still don't see how you can make a shooter with the complexity of ESPgaluda or DOJ, and still incorporate random elements while insuring balanced scores. You used Nomltest as an example, but unless I'm mistaken nomltest lacks any score mechanics.

bullet eating

spend a weekend with it

[Pirate1019]

bullet eating

spend a weekend with it

My bad, it's been 4 years. Point rescinded.

[TrevHead (TVR)]

Nomltest is more of a scorers shmup then anything else. too bad i suck that much that i cant unlock the extra modes.

Plus i think Nomltest and all the other random indies make good pratice shmups. I used to play on them all the time when i first started playing shooters. And by how badly ive been playing lately, i think i could do with going back to them

[Davey]

Good level design isn't a science.

Levels don't have to be masterfully fine-tuned if they're not going to be played over and over and dissected for maximum exploitation*. You don't even really notice all the nuances in Cave's stages until you've played them several times and really try to milk them for as much as you can (full chain, or whatever).

Like I mentioned earlier, even randomized indie games follow some sort of template for pacing. For example, Parsec47 repeats as follows:
- One wave of small enemies (easy)
- A couple waves of small and mid-sized enemies (medium)
- Midboss (medium/hard)
- Short break, followed by more waves of small/medium enemies
- Boss (hard)

Not particularly innovative, but for throw-away levels it works well enough.

I still don't see how you can make a shooter with the complexity of ESPgaluda or DOJ, and still incorporate random elements while insuring balanced scores. You used Nomltest as an example, but unless I'm mistaken nomltest lacks any score mechanics. Parsec47 has item collection , but it's still fairly simple and the game is very much about the dodging and surviving. The mechanics need to be simple to accommodate the variation. Not a problem, but I think Kenta Cho fills the quota for procedurally generated content.

There would certainly have to be some design constraints, since nobody's going to beat something like a Cave TLB on the first shot. And like I mentioned earlier, extremely fast patterns wouldn't be fair. Those would be trade-offs for getting something solvable in real-time. Think of fast puzzle games like Tetris or Magical Drop. Individual moves in those games usually aren't very tricky, but being able to repeatedly make simple decisions very quickly makes the game more complex. You just have to learn general tactics instead of learning specific levels or patterns.

I don't think random shmups could replace traditional shmups, but I think they could complement them as a subgenre (sticking with the puzzle analogy, kind of like comparing Tetris or Magical Drop or Klax to something like Lolo or Braid or Solomon's Key). But commercial developers know the business better than I do, so they've probably already figured out that there's either no market for it or that a good implementation is too hard.



* Yeah I know this is what a lot of people like about modern shmups, and if you fall into that category, randomized attacks wouldn't be your thing. That's cool.

[TrevHead (TVR)]

Hmmm im wondering what comercial games shmups out there make use of them. I doubt therell be any modern shooters, but i bet the really old arcade shooters before gradius used random elements. Like the order that the enemies shooter their bullets in space invaders, Galaxian etc. Also i wonder if any of the remakes like space invaders exteme use any random elements.

[t0yrobo]

I think random generation could work to an extent. Not so much in bullet hell games, but in games with fewer/faster bullets I think it could work. Still I think a certain amount a set elements would be important so that it wouldn't be totally twitch reactions.

[Exarion]

You could also take the warning forever approach, and have the levels change based upon player actions. Player hiding on the bottom of the screen? send in more snipers. heavy point blanking? make the enemies try to point blank. Using straight shot and ignoring spread shot powerups? send in wide waves, etc. This would also make scoring fair.

[AraraSPAMWitch]

You could also take the warning forever approach, and have the levels change based upon player actions. Player hiding on the bottom of the screen? send in more snipers. heavy point blanking? make the enemies try to point blank. Using straight shot and ignoring spread shot powerups? send in wide waves, etc. This would also make scoring fair.

Warning Forever is a fun game and that kind of approach has intriguing possibilities, but the scoring system will remain wonky. I can see people trying to score in this sort of game by manipulating the game using the "randomizing" factors to produce certain enemy patterns that give better scoring opportunities. i.e., if it is decided that enemy X gives the best scoring opportunities and enemy X is spawned by action Y, players will do action Y to make X appear.

This doesn't sound like fun to me. I'd rather have the solid, carefully designed levels that test players' mastery of mechanics in a game like Raiden or Espgaluda. A good game will have a carefully-planned flow that could not be generated randomly.

[TrevHead (TVR)]

I would imagine that any scoring shmup that had a element of chance as to force someone to replay over and over in the hope that he gets lucky and gets loads of enemies that give the best score in the game. Not my cup of tea really. that kind of randomness is done best in Nomltest, as the amount of random ship spawning is too high as to make hardly any difference to scores from run to run.

[dai jou bu]

Senko no ronde used from gamestop was $2.99. Nobody liked it, including me.

I'd like to know why, just out of curiosity.

[Pirate1019]

Senko no ronde used from gamestop was $2.99. Nobody liked it, including me.

I'd like to know why, just out of curiosity.

The controls felt like riding a unicycle underwater.