project-image

AI War II

Created by Chris Park/Arcen Games

Finally a sequel to the award-winning, genre breaking, asymmetric strategy cult classic. The most sentient AI in gaming.

Latest Updates from Our Project:

The AI From Within - Part 1: From 30,000 Feet
over 7 years ago – Fri, Dec 02, 2016 at 07:31:16 PM

Keith here. My favorite thing about AI War is watching players interact with "The AI". It's not a real person, but it often feels and acts like one. So much of what's fun about the game is the story our minds weave to make sense of the stuff on the screen. The more imagination you have, the more fun you have. Add to this the process of building your understanding of what the AI is doing and the knowledge of why it is doing it, and you have an immersive simulation in a sensible and fantastic universe. A universe where your opponent AI might just want to run sandwich-production factories in peace, and everything else it does is defending that noble aim against your unprovoked aggression.

So what is "The AI"?

At its heart: some C# functions that run over and over on every AI-controlled unit in the game. And some functions that look at the broader picture to determine how to allocate unit-spawning resources. And some functions to tell special-task AI units where to go. And, before the game even starts, some functions to place the initial defenses during galaxy generation.
We place some units, we define some rules, and the shenanigans emerge.
One of the things I most look forward to about AI War 2 is seeing what advanced modders will do with the AI. They'll be able to modify or replace those C# functions, and try to produce more effective results. Or just very different results, like an AI with tons of local shipyards that can crush the player under weight of numbers but which are in turn vulnerable to attack. Or an AI that cares for nothing except extinguishing all fighter ships in the universe, and anything that stands in the way of that goal. Obviously the burden's on the modder to make it actually work, but I've met some incredibly talented programmers in the player community.
Over the course of the kickstarter I’ll be going into more detail about the design of the AI. First, to entertain you with how it grinds hapless heroes into sausage. Second, to give you potential modders what you’ll need to know to make your own Humanity Extermination Engine. What could go wrong?
Let's begin.

The AI is not a single commander with a grand battleplan.

The AI is a collection of mostly-independent agents all working towards the same goal (the player’s demise), but with very different ideas on how to achieve that goal. Individual ships (mostly) make their own decisions, and the appearance of coordination emerges from that. In essence, the player is fighting an entire armada of evil Captain Kirks.

So we'll begin with an overview of the individual pieces of AI logic. We're going to start with the smallest, most local stuff because that's what shapes the feel of the game, and then progress to the higher level logics which exercise broader control. Even then, these broader algorithms are very much layered on top of the unit-level core.

(in case you missed this earlier update, here's a glossary of terms about the AI's tools: https://www.kickstarter.com/projects/arcengames/ai-war-ii-0/posts/1745189 )

Targeting

a) The algorithm controlling each and every weapon (not just AI-controlled) asks:

  • Am I loaded?  
  • Are there any enemy ships nearby?  
  • Are any of them eligible targets (in range, etc)?  
  • Which one do I most want to shoot at? 

Movement

b) The algorithm controlling each AI-controlled ship asks:  

  • If I'm in combat, what target should I pursue? Or should I retreat?  
  • If I'm not in combat, am I a Guard ship that should stay here? Or am I a Threat ship?  
  • If I'm about to attack a human system, should I wait for more help? Where should I wait?  
  • If I've been waiting for a really long time, should I ditch this and join a Threat Fleet?  
  • If I'm on a long trip to some other system, should I cancel that because of changing conditions?

Special Forces

c) The algorithm controlling each Special Forces Fleet asks:  

  • How much did my budget increase since the last time I checked?  
  • What should I build with that, or should I save it?  
  • Which system should my units to defend? Or should I call them back to our local ninja fortress?

Threat Fleets

d) The algorithm controlling each Threat Fleet asks:  

  • Does one of our important systems need emergency support?  
  • If I'm attacking a system, should I keep at it?  
  • Are there any human systems adjacent to AI territory that I think I can take down? Which one should I send my units to attack?  
  • Should I cancel the movement of any of my units (because it's going to an old target)?

Strategic Reserve

e) The algorithm controlling the Strategic Reserve asks:  
  • How much did my budget increase since the last time I checked?  
  • Are any of my deployed units no longer needed where they are, and should I bring them back into the reserve?  
  • Are there any systems that I need to defend?  
  • If so, which ship types should I build with my budget?

Waves

f) The algorithm controlling Waves asks:  

  • How much did my budget increase since the last time I checked?  
  • How small a Wave am I willing to send, and do I have enough for that?  
  • If so, when should I schedule the ships to arrive?  
  • And what system should I send it to?  
  • And what ship types should I build?

Cross Planet Attacks

g) The algorithm controlling Cross Planet Attacks (CPAs) asks:  

  • Is it time for another party yet?  
  • If so, what's the total budget for this attack?  
  • How much should I take from the Strategic Reserve?  
  • How many Guard ships should I release from our systems?  
  • Was I able to find enough available ships? If not, take more from the Strategic Reserve.  
  • If even that's not enough, release ships from Special Forces Fleets.  
  • How can I best organize the Guard ships being released?  
  • Where do I deploy the ships from the Strategic Reserve?

Exogalactic Strikeforces

h) The algorithm controlling any ongoing sources of Exogalactic Strikeforces (Exos) asks:  

  • How much did my Exo budget increase since the last time I checked?  
  • How small an Exo am I willing to send, and do I have enough for that?  
  • What target(s) do I want to destroy with this Exo?  
  • For each target, how much budget goes towards this one target?  
  • Which ship types should I build with each target's chunk of the budget?  
  • What systems can I spawn from? And which system should I launch each part of the attack from? 

Reinforcements

i) The algorithm controlling Reinforcements asks:  

  • Have I waited long enough since the last Reinforcement?  
  • Which of our systems are alerted? That is, which have adjacent non-friendly systems?
  • What's our budget per alerted system?  
  • If there's a ton of alerted systems, there's a war going on. 
  • Should I divert some of our resources to other algorithms (like the ones controlling CPAs and Waves) that are more likely to end that war?  
  • Which of our systems can I reinforce, and which are the highest priority?  
  • For each system I'm reinforcing, how many reinforceable positions (guard posts, and some guardians) are there?  
  • What ship types should I build with this system's budget?  
  • Which post/guardian do I assign each of these units to?

Emergence

Again, in contrast to most contemporary strategy game AIs, each of these AI systems are all (mostly) unaware of each other. They have tons of cross-interactions, but only a few are actually in the code. The rest implicitly emerge, and lead to all sorts of crazy stories we never anticipated.

Up Next

I hope you enjoyed that view into the AI, in the next part of this series I'll explain just how an AI Ship Captain decides what to do: make a sandwich at the guard post or perform a stealthed deep strike upon a defenseless factory on one of the player's "safe" systems?

Until then, take care (and keep an eye out for nightmare-deathmachine robots),  

Keith

War of the Models!
over 7 years ago – Thu, Dec 01, 2016 at 09:08:17 AM

Hey Guys, Blue here!  

Thought I'd pop in and talk a little about what I'm up to on my side. 

To be frank, Models! Our current progress is focused on creating a work flow for models that are clean but easy on the performance. 

This means I have a few different things I have to keep in mind. Polygon Count and Batching. Polygon Count is how many polygons a given model has, and batching is a procedure in which objects use the same textures and materials for different effects.

First round was something I was pretty happy with, but unfortunately it missed the mark in many different ways. It's poly-count was too high and it's texture is not one capable of batching.

The idea is that these models aren't individual. In most cases, there will be large groups of these, and after a certain extent they can start bogging down the game.

 Here we go. This meets a certain poly-count goal as well as uses a single texture set that can be used for several other models. Apply an emission map aaaaand boom!

You get this in game!

Here's another that follows the same process!

So yes, there's just a small preview of whole modeling process. Hopefully with this workflow, we can pump out lots of models that can fill your game with your very own armada.

-Blue

The Spire: Beam Weapons For Fun And Profit
over 7 years ago – Thu, Dec 01, 2016 at 08:55:42 AM

Keith here. Thanks so very much for funding our base goal and our first stretch goal (new soundtrack by Pablo Vega!). You folks are awesome.

So, what's the next Stretch Goal?

The Spire. (Cue epic soundtrack)

What are the Spire?

One of the key alien races in the AI War universe. They're really awesome.

Why are they awesome?

As in many things, it comes down to size. Human capital ships are pretty big compared to their other ships. But on a galactic scale, Humans are a brand new star-faring species, inexperienced in construction techniques, galactic knowledge, and (probably the worst problem) computer intelligence design. The largest human empires only covered a tiny fraction of the Milky Way galaxy and this is reflected in ship and fleet designs. So even the greatest and largest human capital ships are but a fraction of the size of the ancient Zenith golems. Thus when it comes to fights against the AI's more advanced Guardians (not to mention its Dire Guardians) humans have to rely on numbers rather than individually powerful craft.

The Spire, on the other hand, are an advanced elder race. They can build big ships, and we mean capital BIG. They have the technology, the resources, and the territory to build and maintain these immense platforms. Combined with this size advantage is their more advanced technology in weapons and shielding. Against the larger Spire capital ships even Dire Guardians simply melt.

Now, if you were engaged in a last-ditch guerrilla war against an implacable inhuman foe, wouldn't you want big ships on your side? Sure you would.

There is one catch, however. You know how your survival depends on not attracting too much AI attention? You know how the galaxy crashes down on your head unless you're subtle?

Yeah, the Spire are not subtle.

What could go wrong?

So how do the Spire work (non-spoiler edition)?

You start a new game with a Spire Colony Ship. You do some things. The AI gets mad. And then you get really cool toys (and the AI gets really mad).

So how do the Spire work (spoiler edition)?

Our current plan is that the Spire will work in much the same way as they did in the Fallen Spire campaign in AI War Classic, with a few key differences. It is a little complicated, so I’ll explain this general plan first and then focus on those key differences at the end.

1) When starting a new game you will be able to add a Spire Colony Ship to your initial fleet.

2) This colony ship is a fearsome combatant in its own right, but that is not its purpose. Its purpose is to transform into a Spire Citadel. Choose the time and place carefully, as the AI will sit up and take notice. Nonetheless, careful planning will let you ramp up production and power quite quickly.

3) The Citadel can be expanded with shard reactors, habitation centers, and shipyards. Each Citadel structure mounts incredibly powerful modular and customizable defenses. Devastating beam cannons, massive shield generators, and so on. You will need those defenses, as the AI will respond to your Human-Spire alliance with vicious Exogalactic Strikeforces.

  • A habitation center synthesizes large quantities of usable metal for you, and allows the citadel to support more structures.
  • A shard reactor provides substantial local power as well as the unique radiation needed by the shipyards.
  • A shipyard constructs and supports the real solution to your problems: Spire Capital Ships.

One Citadel can only support so many shipyards, which can only support so many of these ships (and only the smaller classes). And constructing additional colony ships is (initially) beyond your capacity.

4) Thankfully, the means to expand are out there. One of your Ark's special Hacking abilities allows it to scan for the remnants of ancient Spire outposts. Scanning will point you in the direction you’ll need to go to find these outposts. Once you've narrowed it down to a specific system, scanning there will reveal and bring the core of the ancient Spire outpost, the artifact, aboard your Ark.

Unfortunately, scanning literally bombards the local planetary systems with strong communication signals. As you might expect, that will attract the AI's attention to your Ark.. But that's nothing compared to what happens once the artifact is aboard. Suffice it to say that you'll want to get back to a Citadel pronto to weather the storm.

With each additional artifact, you can build an another Spire Colony Ship and found an additional Citadel.

5) As you construct more Citadels and shipyards, bigger and more powerful Spire ships and technology become available.

Once you reach a certain critical mass... well, you'll see :)

So what's different from AIW Classic's Fallen Spire?

AIW Classic's Fallen Spire (FS) had two major design objectives:  

  • Give the player the option of shifting from guerrilla war to open war.  
  • Provide something like a scripted campaign.

FS succeeded wonderfully in its first design objective. Many players play AIW Classic more often with FS than without. Some play it exclusively that way.

But AIW Classic's codebase was not supportive of scripted behavior, so it had to be hardcoded. It was extremely time-consuming to implement. It also took a long time to stamp out the bugs.

So FS only moderately succeeded in the second design objective.

As a result, and given our budget, the main difference in the sequel's implementation of the Spire is that it's only trying to meet the first objective. As is the Arcen-way, we are also going to be implementing and making improvements based on years of user feedback (but you already know that).

We do want to do scripted stuff again someday, but it means creating a solid scenario-building framework (basically a major extension of what's planned for tutorials) to provide the tools to do the job right. That will take a lot of time and money. It's a potential future stretch goal for this project, but it would be for delivery after the currently projected timeline.

Tell me more about these Improvements?

1) We are going to reduce the time it takes to have your first Citadel operational, so you can start getting to the real power that much faster. But you still have the option of sticking with a minor Spire ally (the initial Colony Ship, like FS's outpost and frigates) and expand the Spire side if and when you want to.

2) The process of finding additional artifacts (shards in the FS sense) used to be a "scan on your homeworld, and then we'll tell you exactly which system the thing is in" process. Now it will be much more a "you're getting warmer..." process where you take your Ark and a big fleet and go cruising for a bruising.

3) Because your Ark is actually involved in the recovery, the hunt and chase are both more interesting (and terrifying). Further, this change allows the force that chases you when you recover an artifact to be a finite one-time assault rather than an ongoing spawn that literally forces you to retreat to your homeworld. Now you get to choose where you make your stand, or just keep running.

4) I know graphics aren't the most important thing to most AI War players, but c'mon, huge battles with giant shielded capital ships firing off all those beam weapons... it's gonna look way better now :)

Will the secret alternate victory condition still be there?

Oh yes. Oh yes indeedy.

Subtlety? Hah.

Live long and conquer, 

Keith

(Note: To those in the US: Happy Thanksgiving! We're going to be fairly scarce around here until Monday, so the next video will be next week, but there will probably be another AI-focused update on Friday)

Music goal: Check! On to the Spire stretch goal!
over 7 years ago – Thu, Dec 01, 2016 at 08:01:59 AM

Chris here.  Apologies for three updates in one day, but we hadn't expected to be hitting this particular goal until tomorrow.  In the end, it's a nice problem to have!

A bunch of backers from the first campaign responded to our reminder on that side to check out the re-launch, and so we've had our best day since we hit the first funding goal.

We passed the $65k mark, which means our first stretch goal is in the bag, so here's where we are now:

We're really excited about the music goal that's already been hit, and we're also really excited about bringing back the Spire as a playable race.  That was one of the most popular features of the first campaign, and so having the chance to approach that again is really exciting.

Please do note that the way this is being handled is very different in this campaign compared to the first one.  Keith will have a more detailed post either Friday or Monday on this subject, but basically the idea is to be a little bit less divergent from the main game mode.

I'll break down the line of thinking, here, even though it's not going into all the details that Keith will later:

AI War Classic: Fallen Spire

I guess I'll avoid spoilers here, but basically this was a "scripted campaign within the dynamic campaign" that you could activate.  If you followed that for long enough, then you'd get the ability to have giant Spire ships of your own -- leading you into a Godzilla vs Mothra giant monster fight with the AI.  Very different from the usual guerrilla tactics.

Pros:

  • The scripted campaign itself was fun and added some structure that a certain segment of players appreciated.
  • The payoff at the end was super fun.

Cons:

  • The scripted campaign is the same every time, kinda-sorta.  It's layered into the procedural campaign and so the details differ a lot each time, but scripted by nature means that it's more hand-crafted than usual.
  • You don't get that big payoff until the very end, and it's not a guarantee you'll live to that point.

First Kickstarter Plan (Mine)

Cinth had the idea of playable races, and then it was off to the races pretty fast with that.  My goal was basically to take the Fallen Spire endgame and let you start there and play a whole game in that fashion (cutting out the scripted campaign in the middle).  

You'd have all-Spire stuff, play very differently from normal, and it would be giant monsters all the way down.

This was exciting, and we were planning on the Zenith being playable in that fashion, too.  However, it does lead into some problems where it quickly becomes two different games.

In Classic, there was a major feature called Defender Mode that I was implementing at the same time that Keith was implementing Fallen Spire.  Defender Mode was a completely separate game mode, and it was something people had asked for for ages... and then we couldn't hardly get anyone to do serious playtesting on it, let alone play it long term.

Fallen Spire, meanwhile, being more integrated into the main game, went on to glory and riches and is something a lot of people loved.

As one aside: I wasn't ignoring the past here, but basically in The Last Federation we had had success introducing Betrayal Mode and Invasion Mode in its first expansion, and the overlap in those was high enough with the main game mode that people were able to happily bounce between the modes.  So I figured the Spire and Zenith races as separate factions would fall into that territory, basically.

Whether that is true or not is debatable -- I'm honestly not sure at this point, but we'd set aside adequate budget to explore it thoroughly, and had some fallback positions we could try if the more audacious design didn't work out.

This Kickstarter's Plan (Keith's)

Giant Flashing Red Disclaimer!

I'm not the lead designer on this feature, and it's possible I'm about to say something that is accidentally outright false. I'm attempting to show you the thought process we used in order to get here, and then Keith will come through Friday or Monday and actually explain what this is in better detail.

Nothing I write below this point in this specific section should be taken as any sort of promise, since I don't want to accidentally commit Keith to something he didn't intend. I know what he has planned is very cool, and we've discussed it on numerous occasions, but I have not yet read the very latest iterations of his design on it.  

Normally I try to be as accurate as possible in my posts, but in this case I want to give you a bit of an idea of what is in the ballpark direction coming up, and then let Keith present his work when he's finished writing up all the bits for you in the manner that he's confident in.

Okay, On With It ;)

We knew that the ability to play as the Spire was a marquee feature that people were excited about, but there were a variety of dubious things about my proposed implementation.  Given that we're having to do this on a lower budget with a smaller staff, that also made for less margin for error.

The general approach Keith decided on was to make it so that you keep the basics of your human faction in place, but also get the Spire stuff added into it.  That gets at the core goal of "be like the endgame of Fallen Spire, but more from the start of the game," but without being so divergent in style that it's Defender-Mode levels of different.

The worry being that people wouldn't play the latter nearly as much... and that even if they did, there would have to be double work on most balancing and content additions in the future.  So it could make further expansion to the game a lot more sluggish since we'd have to give the Spire enough to keep parity with the humans.

The benefits of Keith's approach are many, anyhow, and they basically make it so that we can hone directly in on the core cool factor of the Spire as a playable race without the problems noted above.  He's spoken in the comments in a few places about other specific things with this, but as I said -- more on this will be coming Friday or Monday.

Art Style

As to the art style for the Spire, I can speak to that with some authority.  Blue and I have been working on various improvements for the art style in general, which you may have noticed in the videos we've posted thus far.  But there's even more goodies than we've been able to show just yet.

When it comes to the Spire, I created the original style for them in AI War Classic using ZBrush, and ultimately I'm really unhappy with how that has aged.  Cath helped to reinvigorate them in a major way in Stars Beyond Reach, but we had some enormous troubles with them just being too white and glowy and thus hard to differentiate.

During the first campaign, I had experimented around with a red and glowy style, and was able to get a lot more range out of it.  I'm still not really loving how that came out, though.  So in terms of art style, I think we have some more work to do for these guys, and we likely won't know exactly what they look like until sometime early next year.  

But if you've been enjoying the evolution of the graphics in this game thus far from the start of the first kickstarter to now, I think you'll only continue to be pleased. :)  And in terms of the just absolutely titanic scale... yeah, that's going to stay, haha.

Okay -- that's enough.  Goodnight!

Chris