April 10, 2021, 06:32:48 PM

Show Posts

This section allows you to view all posts made by this member. Note that you can only see posts made in areas you currently have access to.


Messages - Kernel64

Pages: 1 2 [3] 4 5 ... 14
31
AI Development / Re: How to order a global retreat?
« on: April 12, 2010, 01:23:01 PM »
This: AIWaveTargetGatherD(..) could be your problem. Use this target only if you wish the AI to leave units behind from the wave you're controlling.

A retreat with this as target will cause the AI to leave a portion of its force behind.

32
AI Development / Re: AIFindDropAttackTarget() always returning false?
« on: April 11, 2010, 11:20:26 AM »
I thought drop attacks only occur when the opponent has no base that doesn't require transports?

Or maybe it returns false if the AI has no transports?

33
AI Discussion / Re: HexBox AI Needs Your Strategy
« on: April 10, 2010, 03:48:17 PM »
Chainaw,Yeah, it is possible. The most apparent work-around is typing something. As simple as 1,2,3, or Enter, 12, Enter, will then set the AI to the 12th opening, which then leads to the 12th tier 1.5, and so forth.

This is an excellent idea, however, we do really need to at least have a well-cooked metagame before even knowing if a certain strat is worth practicing over.

But, yeah, it is possible, and so the more stuff we get listed down in this thread, the merrier AIs become as practice partners.

Vods, replays, etc.. these are invaluable as well.

34
AI Development / Re: [AI Editing Tool] MilkyWAI Editor (BETA)
« on: April 10, 2010, 03:33:38 PM »
if ( AwesomenessEval(MWAIE) > PureAwesome) {
     Kernel64("NoseBleed", AwesomenessEval(MWAIE) ) == true;
}

^^ funny shit. but shouldn't it only be 1 equal sign, since your setting it to true?

Yeah, MilkyWAI gave an error and said something like:

ERROR in Line 204: You're kidding me right?

35
This is awesome! Nice work man.

I'll wait for the release version.

36
AI Development / Re: Understanding the AI
« on: April 10, 2010, 05:27:03 AM »
Yeah, dude, I will check them out. I wish I had their link or something, my pain though is that most AIs are not compatible with my approach.

I do believe though that in time, there will be more things to get a grasp about, and specifically the XML part of unit control, much like the reapers found in Blizzard's code.

It's very exciting. I'll give you a heads up if ever I find the time and get a decent and extensive squad control function up.

It's only a matter of time that all these AIs will be taken apart and built into one AI, and it might just be your project. That's it. I'll post on this thread as soon as I get something up and running.

Cheers!

37
AI Discussion / Re: HexBox AI Needs Your Strategy
« on: April 10, 2010, 04:13:53 AM »
That's awesome man. I'm adding your thoughts into the list.

Indeed, race specific strats is the only way to go, and that's definitely what HexBoxAI should accomplish.

Again, this is very helpful and will eventually lead to a better AI not just for HexBox but for the many other AI projects as well.

Thanks, and please don't hesitate to post more strats.

38
AI Development / Re: Understanding the AI
« on: April 10, 2010, 04:04:30 AM »
Indeed. TBH, my gripe is when people don't share, at least promise to give credit, nor simply just offer to group up with someone.

Nonetheless, I trust in the process, in that maybe, just maybe, someone might just get a good solution figured out, a module or something that he/she would share now, or eventually when it's working.

AKA, you can always scan through HexBox AI, it contains the stuff you are looking for. They are prototypes, but you can add a few lines and it's all done.

Controlling armies is much like the MutaControlTest function. Determining units is there in IsThereEnemysWithinRange function.

It's all there, but I understand that it takes time to learn how things work. Maybe throw in a few thanks here and there, that should be enough for me. It makes me feel confident that somehow, I'm not operating with a knife poised at my kidneys, and that someday, we might actually be able to create a decent AI separately, or together even if there wasn't an official agreement.

After all, it's the same engine and there's only much about it we can do ATM.

Thanks.

39
AI Discussion / Re: HexBox AI Needs Your Strategy
« on: April 08, 2010, 10:24:08 AM »
Awesome. :)

40
AI Discussion / Re: HexBox AI Needs Your Strategy
« on: April 08, 2010, 09:56:22 AM »
obfuscated sounds cool. I apologize if it looks a little confusing.

I've been testing these strats, and yes, most strats do require that the AI should know how to manage their forces. In sc2 or even in Sc1, losing 30% of your forces at a crucial timing can cost you the game.

It is very possible though, just as you said, and it can be done given HexBox AI allows you to give the AI protocols on how to use its army, when, and where.

There are still functions needed in HexBox AI, this is one of the prime reasons why it is made public, so the guys who have more time, energy, and ideas can add more functions, protocols, etc to the AI and eventually make it easier for the non-programmer to add their own decision making, counters, tactics, army control, etc into the AI.

I am hoping that as soon as HexBox AI matures, a simple editor can be made for it so that it would take only clicks to add in new commands for the AI to use, new conditions to weigh things on, etc.

I'm working on these strats, and I'm waiting for someone to post a strat that could effectively counter a mass tier 1 protoss strat with 4 gate that evolves to 6 gate +1 expo. This protoss strat involves:

rush/harass with zeals,
+stalker support
(switch) stalker harass (no blink, at about 10 stalkers, 5 zeals)
then eventually a mass of 14 stalkers, 8 zeals.


41
AI Development / Re: Understanding the AI
« on: April 08, 2010, 09:38:23 AM »
The solution to this is obviously making a custom function that picks one unit for a scout and give it an order.

Here's personally where my problem comes up: Path finding, or simply choosing the proper/decent point to where we cast the "move" order.

Anyone got a solution to this? An algorithm or what not that chooses points around a certain unit? From point A to B, maintaining a distance of N away from an enemy unit in the path?

42
AI Discussion / HexBox AI Needs Your Strategy
« on: April 06, 2010, 11:59:48 PM »
Hey guys, I'm currently developing and coding HexBox AI. It can contain potentially hundreds of strats, including timed pushes.

It seems that if the AI had stuff that people do it will make for a good semi-practice partner. It's a non-cheating AI, and does well if given a good strat/BO.

I would gladly include the strats you want to be in it. So, please don't hesitate to post them in detail them below.


43
Dude, I'm starting to think you're one wicked tease. :D

Seriously, come on man! ;)

44
This is cool. All it needs is more variations.

If I may suggest, the protoss seemed to stick with zeals and immortals, with a few phoenixes, which gets owned by mutalisks. Having it respond more heavily with the proper unit counters would be nice.

Thanks.

45
AI Discussion / Re: Learning AI posible ?
« on: April 03, 2010, 03:17:58 PM »
Why not have a replay file parsed, then the BOs target locations, make enemy actions as triggers, etc, and given to the AI? Modularize AI executions, and bam! you got the AI to learn, and use what it has learned.

Pages: 1 2 [3] 4 5 ... 14