Spring RTS Engine

Hello

I would like to test an opponent modelling approach that i implemented and would like to know if there are AI's that are known for aggressive and defensive behaviour and wether i can meddle around with the build tables such that the aggressive AI for example only builds tanks ect.

NTAI is the only AI whose building behavior you can directly influence, for all the others (KAI/AAI/TSI) you would have to poke around in their source.

Great to see another Spring AI project BTW!

this question has probably been asked before, but i cant find an answer.

I have a self-compiled version of Spring 0.74b3 and downloaded the precompiled one. The AI ive written , which is based on the AAI, only works with my self-compiled version of Spring, but not with the pre-compiled one. There i get the "Incorrect Global AI dll version"-error. The same happens if i try to run any other AI in with the self-compiled spring. Im pretty new to Spring, so its most likely something which is obvious to you with me being unaware of it. Any ideas?

Are you compiling .74b3 code, or the svn code?

the .74b3 code which is linked in the download section

If you're using the C++ API (aka everybody here) then the AI must be compiled using the same compiler as spring. 0.74b3 uses mingw32 (the same as Devc++/netbeansc++/codeblocks), I'd assume your compiling with VC++ 7/8?

i am using VC 7, so does this mean i need to change something about the compiler?

You cannot run a VS-compiled AI in a mingw-compiled Spring or vice versa, both have to be built by the same compiler (unless you employ the ABIC wrapper layer, which brings its own set of problems with it). Use either mingw or VS, but don't mix them.

so the game and all AI's out there are compiled with mingw? is there any way that i can change the compiler in VS to mingw?

Yes but doing so is a nightmare.

I develop in VS2005 and when I wanna release I fire up codeblocks and use that to build releases.

AF wrote:Yes but doing so is a nightmare.

I develop in VS2005 and when I wanna release I fire up codeblocks and use that to build releases.

thanks a lot. now i can finally do some test runs :)

why would you want a defensive option? noobs would just set it to defensive on a chokepoint map and then have games that last forever. maybe package it with a "real easy" setting that is infact super hard cheating mode

1v0ry_k1ng wrote:why would you want a defensive option? noobs would just set it to defensive on a chokepoint map and then have games that last forever. maybe package it with a "real easy" setting that is infact super hard cheating mode

because i want them to resemble human playstyles, which can also include defensive strategies such as nuking, bunkering or teching.

why would you want to recreate porcing? just make an NTAI profile where every factory makes cons, and every con makes defences

Why is porcing a bad thing to encourage? Some people think it is fun. If they think it is fun, good for them. Some people think (or used to, when it was a current game) Sim City is fun. Sim city is a single player RTS without any attacking or defense. So basically all you do is porc. And it can be fun.

Defense is not evil, people. Don't bad mouth a perfectly good part of the game here. Games are for fun. Anything which ads to the fun of some players without significantly killing the fun of other players is therefore good. Creating a mod or an AI that encourages porcing doesn't destroy your fun. Just don't play it. It does enable other people's fun. So it must be a good thing.

You can use my Ntai configfile for BA to if you want. It's build for fast attacking with massproduction of lowcostbots.



Any hints for improving it are welcome anyway.

Don't play too long against it. It will use to much of your cpu after a while of all the things it need to think over.

Porkin is an interesting tactic, and used correctly, can win...

Do not frown on a porker because the mod you play "isn't balanced enough" to have something that breaks porking.

Sheesh, you people and your "This unit is ONLY good against this unit or tactic" are weak...

Arty and radar = Broken Pork
Spaem weak things to soak up hits while constructors eat their fucken DT = Broken Pork

Think outside the damn box! Remember, it's called Real. Time. Strategy.

Make up something new...

Onto the AI guy, have fun making an AI, and make sure it does whatever you want. I don't know much about coding or configuring AI's, so, I can't give much advice. You do have all of my support, however.

tbh the only counter to a proper porcer is to spend all the resources they spend on defences on your economy, then your economy will grow much faster, and a point will be reached where your economy is enough to produce enough heavy units backed with artillery to just steamroller over. its boring for the porcer because they spent 25 minutes with no enemy units coming within range, then 1 minute being stomped. its boring for the attacker for 25 minutes having to build up.
In a teamgame, if you porc, the guy opposite will just help someone 2v1 on of your allies and your team will die.
porcing is one of those stages before you begin to really understand how to play RTS

porcing on metal maps is a viable tactic, boring but viable.

BUT

metal maps take all the strategy out of a real time strategy game so basically what you're left with is a long base building game...

The reason i started making a config for NTai is cos I got in a bet with a mate who reckoned no AI could beat him on a metal map (he's a big porcer). The bottom line is even on a metal map if you porc you're going to lose, or at the very most draw due to your opponents boredom which is a point that a decent config on NTai will prove very nicely...

DJ wrote:The reason i started making a config for NTai is cos I got in a bet with a mate who reckoned no AI could beat him on a metal map (he's a big porcer). The bottom line is even on a metal map if you porc you're going to lose, or at the very most draw due to your opponents boredom which is a point that a decent config on NTai will prove very nicely...

If you *know* the other person is going to build a defensive perimeter and wait to be attacked, it's probably easy to build an AI that techs really fast, then does something catastrophic like say carpet bombs with a hundred nukes or so.

The difficulty is the "knowing" bit. AIs are generally very poor at reacting to what the enemy is doing; they tend to play blindly.