Really Big Sky And The Issue Of Procedural Generation

[mjclark]

Anybody tried Really Big Sky yet?
Don't think this has been mentioned here before (the search function on this forum really wasn't up to it ...again) but it's a featured shmup at the upcoming Eurogamer Expo and looks interesting.
It's on the list of 12 indie games being showcased at the Expo along with Waves, an arena shooter which also looks rather good but hasn't generated a discussion here yet.

I've also been playing Leave Home and the Scoregasm demo and starting to see that if it can ditch it's embarassing euroshmup past, the UK and european indie dev scene has a lot on offer

[bcass]

To be fair to the UK, which produced great shmups like Uridium and Uridium 2, I think most of those naff "Euroshmups" were produced on mainland Europe, not in the UK. Put it this way, when people use the term "Euroshmup" I certainly don't think of the innovative shmups that have been made in the UK (which includes some of Jeff Minter's games like T2K, T3K, Space Giraffe and Gridrunner Revolution).

The gameplay in Really Big Sky looks nice but I think the whole "neon bloom" visual style has been done to death now and using that style just makes it look horribly generic. Let's not even mention the "Narration from a cheeky cockney AI" either...

[mjclark]

Yeah- you're right Minter's right up there although his games don't get as much love as they deserve on this forum eh?
I guess one of the deal breakers about a lot of these new indie shmups is the procedural generation. To begin with I thought that it really spoilt the gameplay, especially with a game like Self Destruct where fixed levels would have made it into a scoring masterpiece (I actually contacted the developer about a version with fixed levels but, understandably he'd moved on to new projects). Being raised on shmups with memorisation and stage strategies, procedurally generated levels seemed to me to be useless nonsense but I'm just starting to "get" it now and wonder if this is the way forwards.
Obviously it means that hi scores between different players aren't strictly comparible and there's less room for off game discussion since each playthrough will be somewhat unique but I'm seeing that procedural generation actually adds a dimension to gameplay for those who want it.
It's like the difference between chess and poker or backgammon, all real tests of skill but in the last two the player must play against not only their opponent but also seemingly random circumstance. Some feel that this element spoils the game otheres feel it makes it all the more exciting!
Discuss

[Herr Schatten]

I think most of those naff "Euroshmups" were produced on mainland Europe, not in the UK.

In absolute numbers, that's not much of a surprise, as mainland Europe is a lot bigger than the UK, hence more games were produced there in general. In relative terms, though, I strongly doubt that the hit/miss ratio was actually any better in the UK. Both areas definitely had their fair share of stinkers.

Now, on with the topic, I don't want to derail it.

[bcass]

I think most of those naff "Euroshmups" were produced on mainland Europe, not in the UK.

In absolute numbers, that's not much of a surprise, as mainland Europe is a lot bigger than the UK, hence more games were produced there in general. In relative terms, though, I strongly doubt that the hit/miss ratio was actually any better in the UK. Both areas definitely had their fair share of stinkers.

Totally disagree. No Euroshmup is anywhere near as good as the stuff that came out of the UK.

[Udderdude]

I guess it's the next step in making arena shmups for people who don't like to memorize anything at all and just want to dick around with their twin sticks while flashy colored lights appear everywhere.

[mjclark]

I guess it's the next step in making arena shmups for people who don't like to memorize anything at all and just want to dick around with their twin sticks while flashy colored lights appear everywhere.

That's what I thought initially but it actually presents a qualitatively different experience-it's less cerebral and more emotive. You have to stay on your toes cos you never know what's coming next however many times you play through.
As discussed in other threads, one of the appeals of CAVE games is the visual impact- cataracts of stars, rapidly filling guages etc. and this has a similar appeal once it clicks.

[Sumez]

people who don't like to memorize anything at all and just want to stick around with their twin dicks

[Treasurance]

somehow I like the tagline

[Bananamatic]

people who don't like to memorize anything at all and just want to stick around with their twin dicks

deathsmiles multiplayer

[Moniker]

Twin dicks don't get enough respect. Although I must admit the non-random ones tend to be better... certain modes in geo wars, ballistic, etc. Score rush, of course. I think a lot of twin stick games aren't as random as they seem: enemy generation is based on player position and quantity of enemies remaining. Best example here is maybe radiangames Fireball. You can work with it in similar fashion to how you work with boss bullet patterns in regular shmups.

In any case, I think procedural generation *can* work for shmups, it just hasn't yet. Locomalito's Viriax, for example, is procedural and is definitely a sort of mutated shmup. I think it succeeds quite well.

On the subject of RBS, I'd love to try it but my pc's not up to spec.

[Drum]

This space reserved for screaming internet war.

[Randorama]

Being raised on shmups with memorisation and stage strategies, procedurally generated levels seemed to me to be useless nonsense but I'm just starting to "get" it now and wonder if this is the way forwards.

I am afraid that you say why they're not "the way", just below:

Obviously it means that hi scores between different players aren't strictly comparible and there's less room for off game discussion since each playthrough will be somewhat unique but I'm seeing that procedural generation actually adds a dimension to gameplay for those who want it.

Basically, these games remove the "arcade feeling" that everyone tries the same sequence of stages. I actually wouldn't bother playing a game for score, if I can't be sure that if I play well and I get a stinker sequence of stages, I don't get a good score. However:

It's like the difference between chess and poker or backgammon, all real tests of skill but in the last two the player must play against not only their opponent but also seemingly random circumstance. Some feel that this element spoils the game otheres feel it makes it all the more exciting!

Well, in Backgammon one really needs to learn how to handle all possible "random" combinations. At a given point in a game, a pro player knows all the moves that should be performed from the possible result of rolling a dice. So, instead of learning "pieces are in X configuration, I make move Y", a pro learns "pieces are in a X configuration. If I roll 2, I make move Z; If I roll 3, I make move W; If I roll 4..." and so on. Basically, random factors make the game more complex but still a closed system. One can figure out all the possible outcomes, and have a ready-made decision all the time. It just takes more time.

So, procedural games that operate on a closed system of stages could be interesting score-wise, I think and ONE WAY forwards. Randomized stuff is not worth bothering about, except for killing time and being pretentious ("zomg this game is generative! whooo!").

[Drum]

What I'm doing with my game:
Enemies spawn from a pool of points. Pool is determined by rank and is directly tied to score - ie. the game will play different every time, but the total score you can ever get in a game is determined by ability. Amount of enemies/bullets there can be on-screen at a time is determined by rank also.

[mjclark]

Basically, random factors make the game more complex but still a closed system. One can figure out all the possible outcomes, and have a ready-made decision all the time...
So, procedural games that operate on a closed system of stages could be interesting score-wise, I think and ONE WAY forwards..

I absolutely agree, and Really Big Sky does this excellently- there's a fixed set of events to encounter but their order and difficulty is not predictable. You've got to play the hand your dealt. This isn't that much different from those many shmups in which the first stages are presented in random order, except here it's a bit more complex and rank plays a big part

What I'm doing with my game:
Enemies spawn from a pool of points. Pool is determined by rank and is directly tied to score - ie. the game will play different every time, but the total score you can ever get in a game is determined by ability. Amount of enemies/bullets there can be on-screen at a time is determined by rank also.

Have you played Leave Home? It's based on dynamic rank with procedural generation and works really well. It's a 5-minute score attack, but I like the idea of a full game with a similar mechanic although I imagine it could become very difficult very quickly eh?

In fact the incorporation of procedural generation in shmups could be seen as a natural progression from the much debated dynamic rank system we're all familiar with, notably in Battle Garegga. The events the player encounters are largely determined by how he has played up to that point in the game, adding a whole new dimension to the gameplay in which memorisation becomes less important than resource management.

[Estebang]

Omega Five is the only competently designed, worthwhile twinstick shmup I'm aware of.

Although Space Giraffe and Bangai-O HD technically use two sticks, I wouldn't quite put them in the same category.

In fact the incorporation of procedural generation in shmups could be seen as a natural progression from the much debated dynamic rank system we're all familiar with, notably in Battle Garegga

Garegga rank isn't random. It's completely a "reap what you sow" deal. The basic elements of the game don't change, just the quantity and speed of bullets/enemies. I think Zanac's rank is closer to what you're describing, but still far from it.

[mjclark]

Omega Five is the only competently designed, worthwhile twinstick shmup I'm aware of.

In fact the incorporation of procedural generation in shmups could be seen as a natural progression from the much debated dynamic rank system we're all familiar with, notably in Battle Garegga

Garegga rank isn't random. It's completely a "reap what you sow" deal. The basic elements of the game don't change, just the quantity and speed of bullets/enemies. I think Zanac's rank is closer to what you're describing, but still far from it.

Procedural generation as we're talking about it here in RBS and Leave Home isn't random either, the difficulty of the events you face is a direct function of how you've previously played. In these games you reap what you sow also, hence the comparison with Garegga.
We are surely all agreed that arbritrarily random stages in a game make a nonsense of gameplay, but that's not procedural generation and that's not what we're talking about here.

And I could never get on with Omega Five. Although far simpler in terms of graphics and mechanics, Score Rush gets my vote for twinstick shmup par excellence, but that of course is an entirely different discussion

[Obiwanshinobi]

Apocalypse (starring Bruce Willis) remains my favourite twin sticks shooter. The likes of Alien Breed or Alien Swarm don't seem to offer quite as much variety, explosions and awesome bosses (although they do have multiplayer, which Apocalypse lacks). At its time Apocalypse rivalled most flashy Japanese action games (Silent Bomber, S&P and whatnot).

[Drum]

Omega Five is the only competently designed, worthwhile twinstick shmup I'm aware of.

Although Space Giraffe and Bangai-O HD technically use two sticks, I wouldn't quite put them in the same category.

In fact the incorporation of procedural generation in shmups could be seen as a natural progression from the much debated dynamic rank system we're all familiar with, notably in Battle Garegga

Garegga rank isn't random. It's completely a "reap what you sow" deal. The basic elements of the game don't change, just the quantity and speed of bullets/enemies. I think Zanac's rank is closer to what you're describing, but still far from it.

Procedural generation follows a ... procedure. At it's basic form, that's the exact opposite of random. A lot of procedural systems introduce random elements to make the generation more interesting. Because games should be interesting.

[zaphod]

I think the main complaint is when procedural generation breaks your chain or forces you to drop a medal because something spawned wayy too far from your position.

I think it's fine, as long as it doesn't screw with the scoring system.

[Drum]

I think the main complaint is when procedural generation breaks your chain or forces you to drop a medal because something spawned wayy too far from your position.

I think it's fine, as long as it doesn't screw with the scoring system.

Those scoring systems were designed with rigidly pattern-based games in mind tho. That said, in adapting them for games with AI director systems you could actually improve them. Like you could freeze or lock-in your chain value at some risk/expense (maybe when you stop shooting?) if targets dry up. Putting the burden on the player makes it more fun. I think there are some shooters that allow you to keep you chain going if you use a bomb, but I can't recall names. Something like that.

[zaphod]

Well, with the exception of bonus enemies earned by destroying other enemies quickly or bumping up the rank, i think many people consider it unfair to not always have the same amount of points available in the level.


people dont want their scores beaten because someone lucked out with an easy formation.

[Drum]

Well, with the exception of bonus enemies earned by destroying other enemies quickly or bumping up the rank, i think many people consider it unfair to not always have the same amount of points available in the level.


people dont want their scores beaten because someone lucked out with an easy formation.

Yeah, but one has nothing to do with the other. Post I made up the page:

What I'm doing with my game:
Enemies spawn from a pool of points. Pool is determined by rank and is directly tied to score - ie. the game will play different every time, but the total score you can ever get in a game is determined by ability. Amount of enemies/bullets there can be on-screen at a time is determined by rank also.

[zaphod]

And that would be how to do it right, for the most part.

It's hard to design a scoring system that really handles that sort of gameplay, but not impossible.

[Moniker]

Heads up: the latest indieroyale.com bundle has RBS in it (as well as Cthulu Saves the World & Breath of Death VII). Last I checked the bundle cost ~$4, so if you're curious, now's the time to check it out.

[Observer]

It's an unplayable mess for me. Hoping to get support somewhere but I need to vent this, I've never faced so many problems all at once:

The game has NO SUPPORT for 4:3 resolutions. It could've at least included a 1024*768. So it's widescreen or screw you. And damn it screwed me as I'm still using a 4:3 display for TATE'ing purposes.

I have a USB 360 pad and the game refused to even acknowledge it. It assigns it to player 3 and god knows what else, thus leaving me stuck with keyboard+mouse no matter what. Can't even find a way to reconfigure controls. Pressing Esc just takes me out of the game, not even a pause menu? (Maybe there is one mapped to the P key?)

The game had incredible lag spikes and fps variations. I can run Waves no problem at max or Beat Hazard at the most brutal intensity and difficulty settings without even blinking. Yet this can barely hit 19fps with low? 9 fps with 3 enemies on screen? Something must not be working right between the game and the video card.

Oh, and the inertia.

But I would really love to know what's causing this absolutely messed up situation. The OS is a clean install done 3 weeks ago, the comp barely has stuff installed. I'm running under Windows 7 Professional 64-bit version. Could that have to do with this? The video card is a GeForce 220GT with 1GB, not really good. The CPU is an AMD Athlon X2 3600, not precisely state of the art either. But it does its job with most indie games.

[Aisha]

IThe game had incredible lag spikes and fps variations. I can run Waves no problem at max or Beat Hazard at the most brutal intensity and difficulty settings without even blinking. Yet this can barely hit 19fps with low? 9 fps with 3 enemies on screen? Something must not be working right between the game and the video card.

Oh, and the inertia.

But I would really love to know what's causing this absolutely messed up situation. The OS is a clean install done 3 weeks ago, the comp barely has stuff installed. I'm running under Windows 7 Professional 64-bit version. Could that have to do with this? The video card is a GeForce 220GT with 1GB, not really good. The CPU is an AMD Athlon X2 3600, not precisely state of the art either. But it does its job with most indie games.

Did you update DirectX 9.0c? If I understand correctly, the version of DirectX 9.0c that ships with Vista and 7 is a bare bones version that is not current. I remember that when installing Mountain of Faith back when it was released, I lazily opted not to update DirectX 9.0c which resulted in noticeable framerate variations.

[Moniker]

The inertia can be turned off in the options menu. As far as performance goes, you can try setting the resolution to 640X480 in the app's properties, and then select the smallest windowed option in the game. It's just barely playable for me.. I might have to do some more tweaking. But I've got a lousy pc for gaming, so YMMV.

[mjclark]

Weird- my game plays perfect on Athlon X2 with 4Gb RAM, GeForce 7900GT and Windows 7 x64 Ultimate.
Also 360 pad works flawlessly. Could be that DirectX or I remember some games were fucked if you had onboard sound instead of a seperate soundcard...