next »

[Alec S.]

GunGen attempt 2:  Results

My second method of Procedurally Generating guns is proving much more effective than the first.

How it works:

As I said in the last GunGen post, how it works is it spends points on attributes in 3 categories (Shooting type, Curve type and Secondary type).  It then spends remaining points on modifying variables of those attributes.

The Current Attributes are as follows

Shooting Type (How the bullets come out of the gun):

 - Standard

 - Rapid Fire Burst

 - Spread Burst

Curve Type (How the bullets move):

 - Standard (None)

 - Wave

 - Bounce

 - Homing

Secondary Type (How the bullets generate more bullets)

 - Standard (None)

 - Trail

 - Split

 - burst


Each aspect has a cost.  The system will split up the total amount of points into a value for each category.  It will then determine which aspects it can buy with those points, and pick one randomly.  It will then split the remaining points among that aspect's variables. 

The one exception I had to add to this system is if it has a lot of points to spend in  a category, but still randomly picks Standard.  At first I tried to make it spend the points on general variables like bullet speed and fire rate, but this could easily break the system if too man points ended up being spent in Standard.  What I ended up doing was making it so that if it tried to put more than 3 points into Standard, it would automatically upgrade it to the next aspect.  So if the system tries to spend 5 points in a Standard curve, it'll end up with a Wave curve.  Not the most elegant solution, but it will have to do for now.

Anyway, how about those ol' criteria I posted back in the first GunGen post

How well did this fit the requirements:

 - There must be significant emergent variety in the guns:  There is.  There are plenty of combinations of aspects, which is further expanded upon by the variables which can have a big impact on the gun.
 - The variation in the guns must be interesting and meaningful to gameplay:   Looks to be the case.  Plenty of interesting patterns come out of the different combinations, and guns will handle significantly differently.
 - The system must scale based on level.  The last king you fight needs to have a more powerful gun than the first king you fight:  Different amounts of points can go into the system to make stronger or weaker guns.
 - The guns must be more or less balanced within each level. This still needs some work.  Some combinations can make seriously under-powered and downright silly guns, even (or especially) at higher levels.  But this can be fixed within the system through balancing and tweaking.

Finally, here are some examples of cool guns that came out of the system:






On the other hand, here's an example of an incredibly stupid gun created by the system:

[Pixelulsar]

I love the guns that it created, they have really cool effects.  I like the two in the first GIF, they are my favourite ones.  In the really stupid one, does it have a lot of recoil, or are you just moving backwords?  I think it would be cool if you didn't go flying back.

Looks very cool so far.

[Alec S.]

Thanks!  Yeah, part of that is the recoil which I still need to work on.  The other part was that the bullets were curving so much that they were moving slowly backwards.  I've fixed the problem, but I posted the gif anyway as I thought it was entertaining.

[Kit630]

Looks to be off to a great start. Loving the graphics and the gameplay idea/story! Screenshots look amazing, and the gun examples promising.

Though I will say this, the "stupid gun" example isn't being given a fair shake :3 I can already see a use for it, as an example. The bullets travel in the same direction the player is propelled, a bit of spread. It could easily be used to rocket past an enemy or boss as a dodge move, and the spread of the bullets would still get some hits in.

With procedural generation, you could potentially spend all your time optimizing it, and never weed out all the "stupid" combinations (This is not to tell you to discontinue trying to do so though! :D). I think that it's up to us as the playerbase to learn how to use them to our advantage. I guess my point is, in the least and most non-offensive, hopefully constructive way possible, to say to try not to let any personal bias or preference have an effect on what types of guns the system creates. Sometimes the silly guns are the most fun, and one reason I like them is to get a good laugh out of them or still become skilled with them even if they are ridiculous. But! I can't speak for everyone, so that is just advice. Take or leave it at your pleasure :D

Although, as a thought; You could add a fallback to reduce the possibility of getting one of these types of weapons by adding in some sort of character npc with the role of gunsmith, maybe someone who knows about the artifacts and wants to assist the player in the overthrow of the kings. Kindof cliche, but hell. You could let the player have a set number of "resets" per gun, nothing gamebreaking that allows them to find "The perfect gun" but something that makes it up if they get something truly broken.

Just some thoughts, can't wait for a demo of some kind <3

[Alec S.]

Yeah, I'm all for allowing rather silly guns into the mix, that's one of the joys of procedural generation.  The only two considerations are that the guns also double as boss battles (to get each gun, you have to defeat the king that uses the gun) plus the element of scarcity, I don't want the player to feel ripped off by one of the guns being massively under-powered.  The main problem with the "stupid gun" wasn't so much the knockback (although that was probably the most comical part of it), but the fact that the bullets moved slowly backwards.  The knockback had some issues of it's own, aside from sometimes being very high, as at the time the spread of the bullets added more to knockback than anything else.

But, anyway, the point is I want to balance having a system that creates many surprising weapons, but also craft in limitations in certain areas to make sure none are extra terrible or break the game in some way (some of the early guns would essentially create a swarm of bullets that flew around the screen and never left or moved far from their origin.  Spawn enough of those and the game seriously lags)

[baconman]

You do realize, that "stupid gun" has a lot of crazy potential, right? Like zipping past enemies by carefully aiming your player sprite, then those backwards-curving bullets following you and possibly hitting enemies from behind!

Doesn't mean it's not stupid for a gun, but don't dismiss such potential. You might wanna have a looksee at MegaMan games for inspiration, too.

[Alec S.]

I don't deny that a gun like that gun has potential.  But as I said, I can't really program a boss AI to figure out how to use it.  And, again, as I said, the same variable that had the potential to make bullets move slowly backwards can also make them circle around one location.  They could also move slowly backwards and not launch the player backwards, which would essentially turn the gun into a really slow gun that you just need to aim backwards.

[Alec S.]

[johnki]

This is looking really cool. Even more stoked for it after playing Venusian Vengeance.

I probably read right over where it explicitly stated it, but the world itself is static, right?

[Alec S.]

Yeah, the only things that are procedurally generated are the guns.  Although there might be some small changes in the world depending on what order you choose to fight the kings.

[johnki]

Cool.

Yeah, I actually prefer it that way. One of the downsides of procedurally generated worlds is that the level design just isn't the same.

But the gungen system is looking to be really versatile. It'll be neat to see the effects of that on play.

[Alec S.]

A kings gotta have a castle

[amidos2006]

That's really an incredible idea, I was thinking about that idea before and I think you implemented it in an amazing way Great Job Looking forward for it

[Alec S.]

Working on some dialogue for the game.