Jump to content


Photo

Skirmish AI 3.10 Beta 3 - Post Comments In Thread!


35 replies to this topic

#21 Inquisitor

Inquisitor
  • Members
  • 121 posts
  • Location:Ordo Malleus

Posted 14 November 2008 - 04:11 PM

Have you tried pitting cron AI vs any of the other races recently? It's not a pretty sight. Also, the word on the street are that Eldar, DE and Tau are currently the most OP races. After those probably Chaos, Orks and then the rest... It's not a coincidence that the 1.1 patch (which wasn't a balance patch) included a buff for the crons eco, you know.

If thats the case why break whats already fixed if in fact Crons got a boost?

Could you rephrase, I don't understand what you mean?

2nd HQ upgrades has no other purpose (nothing, zilsch, zero) than making the structure able to be Awakened. It costs 450 pow just to go Tier 2 and should only ever be built if power is abundant. Certainly not when vehicle cap is 3/20. Any Monolith can build all 'researched' troops - even if it is T1.

This will have to be investigated because I remember adding additional addons to additional Monoliths provides an advantage (ie. unlocking higher tiered units in that Monolith). Seems stupid if there were no benefit.

If anything, the Cron LP's issue should be moot since the Cron's LPs are there for distruption but its worth testing your suggestions by "delaying" the LP and addons on additional HQs a little.

Sorry! :p I just checked the wiki, and it seems I wasn't quite right, upgrading the second/third HQ has these benefits:
1. The NL gets more expensive (+50 pow for each upgrade), but gains more HP.
2. FO gets more HP for each upgrade.
3. Preparing the structure for becoming Engaged, i.e. after the second upgrade (450 pow for the first, 900! pow for the second) you can use 1000 pow to awaken the structure. (total is more than 9 destroyers!) ;)

Most of these benefits are nothing compared to getting a full vehicle cap, which you can almost imagine.
The Emperor knows, the Emperor is watching.
Want to zoom out further? Try this camera mod.

#22 thudo

thudo

    Wacko AI Guy!

  • Division Leaders
  • 12,164 posts
  • Location:Lemonville North, Canada
  • Projects:DoW AI Scripting Project
  • Division:DoW
  • Job:Division Leader

Posted 14 November 2008 - 04:28 PM

Hmmm.. Well I guess thats to balance the Crons because in DC they were much too uber.

I still don't see Cron AI weak at all -- they more times than not dominate the game still. With their eco boost then sh*t.... However, yer suggestions and observations should be taken into consideration for sure.
Advanced Skirmish AI Team Lead for the coolest Warhammer40k PC RTS out there:

Dawn of War Advanced AI Headquarters

Latest DoW Advanced AI Download!

#23 Inquisitor

Inquisitor
  • Members
  • 121 posts
  • Location:Ordo Malleus

Posted 14 November 2008 - 05:19 PM

Hmmm.. Well I guess thats to balance the Crons because in DC they were much too uber.

SS has been out for over 8 months and you're talking about DC still?

I still don't see Cron AI weak at all -- they more times than not dominate the game still.

I never said they played bad, I just wanted to provide feedback for improvement. I could mention a lot of stuff that's good, and where it plays nice, but I already did so on several earlier occasions.

With their eco boost then sh*t....

The OP races I referred to are in the 1.2 patch, so next to nothing was changed by the eco boost. I just mentioned it to help you understand that they're not DC uber anymore. Why is that so, you ask anyway? I think that it is obvious since Relic thought it was important to buff the eco in a non-balance patch.

However, yer suggestions and observations should be taken into consideration for sure.

Great! If it is possible, I think that some kind of reviewing of the likeliness for the second HQ upgrade would help a lot with little risk. :p
The Emperor knows, the Emperor is watching.
Want to zoom out further? Try this camera mod.

#24 ArkhanTheBlack

ArkhanTheBlack

    title available

  • Members
  • 814 posts

Posted 14 November 2008 - 05:36 PM

what about the ork waagh! banner population with massive battle? if i recollect correctly they were still at max 100 while other races had a 20 - 30 max pop change, shouldn't the ork pop be 150 with massive battle?.

That's pretty much a SCAR bug. I've no idea how to fix it. The pop modifier just doesn't work with the orc pop system.


Shouldn't there be a time restriction as the AI can spam the abilities otherwise ?

Not sure, nobody complained about that so far...


I see little NW dancing. Is this intentional? They're the best dancers in the game.

I heard exactly the opposite before. They are too slow to dance and good melee fighters. Therefore it was decided that they shouldn't dance but accept the melee fight instead.


Necrons still upgrades all LP's even though they're next to useless (when there's much else to spend power on). Suggest putting these wayyy down in the priority.

At the moment it's 100 % at the start of T2. What would you suggest?


Necrons upgrades second HQ (useless upgrade) even though Vehicle Cap is at 3.

Well, there's a Necron bug involved here. If the other monoliths are not upgraded, it's not (or not always) possible to awake the monolith.


Builds turrets like insane even though Vehicle Cap is at 3 with two idle Monos. Suggest reducing the AI's likeliness to build turrets.

There are lots of different opinions about turrets in general, and the effiency is highly map and opponent dependant. You can deactivate the turrets in the control panel if you like.


@Zenoth
Thanks, since I don't play games which aren't at least released for half a year, I still have some time to think about it. So far the only ego RPG I really liked is Vampire Bloodlines. Oblivion was probably the most beautifull RPG game I ever played, but unfortunately it was also one of the most boring RPGs I ever played. I have a bit of a love/hate relationship with that game...

#25 Inquisitor

Inquisitor
  • Members
  • 121 posts
  • Location:Ordo Malleus

Posted 14 November 2008 - 05:57 PM

I see little NW dancing. Is this intentional? They're the best dancers in the game.

I heard exactly the opposite before. They are too slow to dance and good melee fighters. Therefore it was decided that they shouldn't dance but accept the melee fight instead.

Alright. They just dance very well with humans because they have high accuracy on the move. Maybe it's better to let them fight with the AI, because they will dance very far. Humans dancing NW will just go in little circles.

Necrons still upgrades all LP's even though they're next to useless (when there's much else to spend power on). Suggest putting these wayyy down in the priority.

At the moment it's 100 % at the start of T2. What would you suggest?

Uhm, I don't know. Depends. Where can I see what they're up against? In which file?

Necrons upgrades second HQ (useless upgrade) even though Vehicle Cap is at 3.

Well, there's a Necron bug involved here. If the other monoliths are not upgraded, it's not (or not always) possible to awake the monolith.

Ah the bugs... ;)

Builds turrets like insane even though Vehicle Cap is at 3 with two idle Monos. Suggest reducing the AI's likeliness to build turrets.

There are lots of different opinions about turrets in general, and the effiency is highly map and opponent dependant. You can deactivate the turrets in the control panel if you like.

I just don't think that 2+ turrets is a wise choice when vehicle cap is very low. I think this goes for other races to some extent as well (though they need spend mostly req on turrets). I wouldn't disable them entirely personally. I just try to help you out on the priorities.

Thanks for your comments! :p
The Emperor knows, the Emperor is watching.
Want to zoom out further? Try this camera mod.

#26 Zenoth

Zenoth

    title available

  • Members
  • 469 posts

Posted 15 November 2008 - 01:43 AM

I will come back with a more detailed report later, but I just wanted to point at something concerning the economy of Necrons and pretty much all other factions. All of the suggestions so far about little new strategies to save more Power/Requisition early in Tier 1 and 2 are without the difficulty setting and the resources rate in mind.

I don't know if any of you played against 4 Necrons on Hard/Harder with High resources rate, but I can say that to win any such battle, on any maps, you need to be more than used to play this game. Just taking the turrets suggestion with the Necrons for example, trying to decrease the numbers built from 4 (per Monolith) to say... 1 or 2 to save Power (that's the ultimate goal isn't it) would do absolutely no difference even at Tier 2 at High resources rate. But it would certainly help some at Standard resources rate.

Now the thing is that economy-saving strategies isn't always good for pure off-line skirmishes, in my opinion, it's different on-line with humans at play, but off-line it's another story. It also depends on how you like your skirmishes. I myself am the type of player who likes seeing the A.I defend itself, because I like to attack a lot. If the Necrons for pure example are "forced" to change their tactics to go more offensively (having more resources by limiting defensive strategies will inevitably result in more offensive actions) and decrease their base defenses and delay the construction of a second Monolith (or a third one) it will certainly show up on the victory score board after their defeat.

I don't know if any of you see it the same way I do, but there's so many variables to consider when touching the economy, it's the reasoning behind it, and how much it will impact on how people like to play the game in majority. I don't read the Relic forums often, where to fans mostly comment on this modification, but here I ask: were there any "complaints" concerning the Necrons in particular with DoS 3.0? Other than the fact that they just kick serious arse?

I want to be clear here, I am not against having a discussion on all of this and even trying "new things" but I, myself testify that I like the Necrons like they are right now, they already kick some serious arse as it is, and to be honest I just don't see how "less" of a threat they are in Soulstorm than they were in Dark Crusade (prior to Patch 1.2 at least). I've seen enough on-line matches replays and seen enough A.I in action to know that the Necrons don't need much changes with their economy-related tactics, if none at all. But that's just me of course, which is why I'd be curious anyhow to see how it'd end up by changing some of the things that Inquisitor suggest.

As for the Necron Warriors dancing, in my head it's a simple thing, they shouldn't dance at all. We have enough to see Necrons infantry in general being affected by morale loss, which in my eyes makes absolutely no sense whatsoever, but yeah, I know, outside of the lore, considering only game-play and balance, it apparently "had" to be done, but even then, I wholeheartedly disagree, but anyway that's another subject.

I just think that, overall, the Necrons Warriors are just too durable even just in Tier 1 to take any advantage from dancing, and like Arkhan says, they are too slow for that to be effective (that's the end result, dancing and being slow is almost asking to die faster, it's better to stay put and fire what you can, but yes, if you're fast enough then dancing and retreating does help a lot). I guess we could try to see how it turns out if changes are made to the NW's dancing, I'm ready to try that out yes, of course, it's up to the coders to decide, and up to us to suggest.

And there's one thing I do agree completely with Inquisitor, about the Necron's turrets placement. It can be a problem at times. It's not always the case mind you, but it does happen to see for example three or four, or more (if two or three Monoliths are present) turrets around one or two Monoliths which in the end blocks the way to vehicles and Tomb Spiders from moving at all. As I said, at least from my observations, it's not always happening, and it does depend on the maps, but I have seen it happen on many different maps already. But you know, to be honest, it happens with all other factions too, and it's not just with turrets.

I have a suggestion for that in mind, but I don't know if it can be done at all.

Would it be possible to simply kill any inactive units if they don't move after a certain period of time? Because we're talking about A.I here, so if any units controlled by the A.I don't move for 10 minutes then it means either two things: 1) there's a bug in the A.I code making that unit inactive somehow, or 2) it's stuck and cannot move despite the A.I knowing what has to be done with that unit.

There's a potential problem with that however that I can see in advance. If for example killing stuck units would be possible, and let's presume it's being done, then a problematic consequence to that is simply that another unit, let's say a vehicle, will just get built at the exact same production building, and again it will get stuck because in the end the problem is the placement of structures around that building and not the unit itself.

At least each factions do build production buildings near LP's/Relics outside of their base in Tier 3 and when they have enough resources for that, but then again those structures help them get more units at once on the field in less time (which is certainly a good thing, let's keep it that way please), but it doesn't "force" the A.I to build from those new structures, they can still build from the ones they built in their base which by Tier 3 and in long-lasting games will end up being surrounded not only by turrets or just other buildings of all types but also often being surrounded by allies structures as well.

It's a simple problem visually to observe, but on paper it has to be complex. I'm putting myself in Arkhan's shoes and seeing that issue must result in eye rolls. I don't know how exactly it could be fixed. The very best way I believe is to restrict what can be built around a specific type of structure once it is built. You know, the area around a building before it is placed on the ground, it determines how far away from it can other buildings be built (or how close to it, depending on how you see it). Well, that "restriction" area is very small on turrets, and it's the main reason why we see them so close to a Monolith in the Necron's case, ending up in possibly blocking the way eventually. That's a tough one really, perhaps Arkhan and his magic will come up with something to "fix" that.

But coming back to my suggestion with inactive units killing... maybe if the source (the responsible building) of that unit can be identified then it would be possible to condemn that structure from building any ground vehicles anymore, so that way no future units built from there will get stuck and there will be no need to commit suicide on 'x' numbers of stuck units constantly, but of course the condemnation of the concerned structure(s) would only be possible when there's a third or even fourth one. But then again... wouldn't it just increase the problem if we just increase the numbers of production buildings in the end. I really don't see what can be done about that to be honest. And it's a problem seen in so many other RTS'es too (3D ones at least, since I've never seen that happen in Age of Empires 2 of course, fortunately 2D sprites cannot be stuck).

Edited by Zenoth, 15 November 2008 - 02:16 AM.


#27 Inquisitor

Inquisitor
  • Members
  • 121 posts
  • Location:Ordo Malleus

Posted 15 November 2008 - 02:15 PM

... and to be honest I just don't see how "less" of a threat they are in Soulstorm than they were in Dark Crusade (prior to Patch 1.2 at least). I've seen enough on-line matches replays and seen enough A.I in action to know that the Necrons don't need much changes with their economy-related tactics, if none at all. But that's just me of course, which is why I'd be curious anyhow to see how it'd end up by changing some of the things that Inquisitor suggest.

It's a fact that Necrons rate in the lower half of the nine races, when you ask players that play other humans. Browse the relic forums or anywhere else where they discuss SS strategy to verify this. It's primarily because it's very hard for them to get an eco up (compared to DC) because gens got a major increase in cost. Also, FO got a lot of kicking and good players just don't use them vs. other good players.

But the AI has a different audience and I'm aware of that. I just pointed out a few things that could be improved, and I honestly think that remarks about that Necrons already are hard enough to beat belongs somewhere else. Our mission is to improve what almost can't be improved: The brilliant Skirmish AI mod. Our purpose is to point out what can be improved, even though it will make it even harder for some people to play against a specific race.

As for the Necron Warriors dancing, in my head it's a simple thing, they shouldn't dance at all. We have enough to see Necrons infantry in general being affected by morale loss, which in my eyes makes absolutely no sense whatsoever, but yeah, I know, outside of the lore, considering only game-play and balance, it apparently "had" to be done, but even then, I wholeheartedly disagree, but anyway that's another subject.

Did you just say that they shouldn't dance because it's unfluffy? Fact is that dancing is a core gameplay mechanic in the DoW-series and no human player will last long against other human players if they don't dance. Besides, in the fluff the NW have much better shooting than CC capabilities. It is certain that if the AI would dance them like humans dance them they would last longer vs. CC strong troops.

I already withdrew my original request and I think it would be best to leave it as it is (i.e. don't dance NW). The AI dance logic is simply not suited for NW.

I just think that, overall, the Necrons Warriors are just too durable even just in Tier 1 to take any advantage from dancing, and like Arkhan says, they are too slow for that to be effective (that's the end result, dancing and being slow is almost asking to die faster, it's better to stay put and fire what you can, but yes, if you're fast enough then dancing and retreating does help a lot). I guess we could try to see how it turns out if changes are made to the NW's dancing, I'm ready to try that out yes, of course, it's up to the coders to decide, and up to us to suggest.

Humans just dance them in little circles to reduce the chance of CC-troops hitting them while still being able to fire ideally 80% of the guns (which have 80% accuracy otm).

And there's one thing I do agree completely with Inquisitor, about the Necron's turrets placement.
...
I have a suggestion for that in mind, but I don't know if it can be done at all.

Would it be possible to simply kill any inactive units if they don't move after a certain period of time? Because we're talking about A.I here, so if any units controlled by the A.I don't move for 10 minutes then it means either two things: 1) there's a bug in the A.I code making that unit inactive somehow, or 2) it's stuck and cannot move despite the A.I knowing what has to be done with that unit.

It would be nice to look at something like you suggest, but I think it's a big and risky feature that would be out of the scope of this release. The best would be if we could somehow ensure a free path from any spawn point, but I know that with the bugs it will be next to impossible.

...

@Arkhan: I tried digging up those percentages myself, but I couldn't find anything. I really need to see what the other probabilities are to offer a qualified suggestion. I would like to reduce the likeliness significantly, so that upgrading won't take precedence over building vehicles or other upgrades.

@all: I know this is OT, but I saw that Relic won't make DoW2 moddable. Johnny Ebbert stated that it is not feasible with the Essence engine. Anybody also saw this/have any comments? I think it's sad, but understandable. I'm not sure if it is wise not to make it a priority, though.
The Emperor knows, the Emperor is watching.
Want to zoom out further? Try this camera mod.

#28 ArkhanTheBlack

ArkhanTheBlack

    title available

  • Members
  • 814 posts

Posted 16 November 2008 - 02:25 AM

@Zenoth
The AI can't kill its own units and we don't have that much control about the building placement. Actually, the most efficient way would be indeed to reduce or at least delay the turrets. I'll have a look at it. They need at least one very early to detect invisible units, but further ones can probably be delayed.


@Arkhan: I tried digging up those percentages myself, but I couldn't find anything. I really need to see what the other probabilities are to offer a qualified suggestion. I would like to reduce the likeliness significantly, so that upgrading won't take precedence over building vehicles or other upgrades.


Open the file \Dowai_mod\Data\ai\races\necron_race\strategies\necronstrategyinfo.ai. Here you find the build programs for the Necrons. At the end of the first block of each build program you will find the line

{ 1, 0, 75, 250, 100, "TurretAddon", "addon_necron_list_post_1" },

The fifth number is the percentage number. I'm thinking about changing that to

{ 1, 0, 75, 250, 60, "TurretAddon", "addon_necron_list_post_1" },

which would reduce the upgrades to 60% of all LPs. Maybe 40% would be even better. The upgrades can still be made after they've started the T3 upgrades or even later.


@all: I know this is OT, but I saw that Relic won't make DoW2 moddable. Johnny Ebbert stated that it is not feasible with the Essence engine. Anybody also saw this/have any comments? I think it's sad, but understandable. I'm not sure if it is wise not to make it a priority, though.

Depends on what you want to mod I guess. COH AI can be modded, but new models are probably more difficult. I guess it will be similar with DOW2.

#29 troubadour

troubadour
  • Members
  • 88 posts

Posted 16 November 2008 - 09:27 AM

atm i play with turrets unchecked in the AI control panel because at Hard & Harder level, AI focus too much on building turrets instead of building squad and req is 'wasted' in a way. As early detector i think that every race can get one in T1.
Also 'multiple production buildings' got the same problem, i have unchecked it for the same reason.
Please note that i dont say it shall be removed, actually with the AI control panel everything is fine, i just post because i thought it has sthing to do with my playstyle but i see i am not the only one concerned.
I think it would be a bit better if you could change the turrets check box to a turrets slider in order to control better how AI build turrets, cause i know zero turrets is not the best option for a challenging game
I dont say the AI shall not build turrets at all, but at the moment it builds too many in T1 and it cripples its economy
I discuss 1 vs 1 game vs AI, for instance in multiplayer game with ressource sharing or QS game things are completly different and AI shall build early turrets, Well i guess there is no an easy solution, just my 2 euros cents :p

Edited by troubadour, 16 November 2008 - 09:29 AM.


#30 Inquisitor

Inquisitor
  • Members
  • 121 posts
  • Location:Ordo Malleus

Posted 16 November 2008 - 02:05 PM

I'm thinking about changing that to

{ 1, 0, 75, 250, 60, "TurretAddon", "addon_necron_list_post_1" },

which would reduce the upgrades to 60% of all LPs. Maybe 40% would be even better. The upgrades can still be made after they've started the T3 upgrades or even later.

I think that 40% sounds like a good suggestion for all the strategies. It would make it less likely to upgrade LPs than upgrading a turret (50%), which sounds good to me. The power saved on these upgrades will hopefully make it more likely to build useful researches and/or some troops instead. It will not make it totally unlikely that it upgrades them, which is also more like a human would play.

I looked in the code and I wonder why there are two almost similar lines in each of the different strategies:
...
{ 1, 0, 75, 0, 1,	"TurretAddon",  "addon_necron_list_post_1" },
...
{ 1, 0, 75, 250, 100,   "TurretAddon",  "addon_necron_list_post_1" },
...
I then looked in the SM strategy and found something similar, but not quite:
...
{ 1, 100, 75, 750, 60,	 "TurretAddon", "space_marine_list_post_addon_1" },
...
{ 1, 100, 75, 250, 100,	 "TurretAddon", "space_marine_list_post_addon_1" },
...
If it isn't too much trouble, I'd like to know why there are two lines and why they differ in the way they do? I would help me understand more and thereby provide more qualified feedback. Thanks. :p

...

I dont say the AI shall not build turrets at all, but at the moment it builds too many in T1 and it cripples its economy
I discuss 1 vs 1 game vs AI, for instance in multiplayer game with ressource sharing or QS game things are completly different and AI shall build early turrets, Well i guess there is no an easy solution, just my 2 euros cents ;)

I agree with this, but turrets sometimes seem to be a help to the AI, because they provide an immobile defense that will allow the sometimes 'confused' AI to muster some defense. Don't get me wrong: I think there should be less, but I think it would be a mistake to play without them because it make the AI more fickle on the defensive. I think the likeliness should be reduced somewhat to help the AI tech faster and get a bigger army. Ideally it should very rarely build 2+ turrets if it can build a troop/vehicle, but what do you say? Do you agree that it is sometimes helping the AI more than hurting?
The Emperor knows, the Emperor is watching.
Want to zoom out further? Try this camera mod.

#31 Zenoth

Zenoth

    title available

  • Members
  • 469 posts

Posted 17 November 2008 - 03:38 AM

I agree with this, but turrets sometimes seem to be a help to the AI, because they provide an immobile defense that will allow the sometimes 'confused' AI to muster some defense. Don't get me wrong: I think there should be less, but I think it would be a mistake to play without them because it make the AI more fickle on the defensive. I think the likeliness should be reduced somewhat to help the AI tech faster and get a bigger army. Ideally it should very rarely build 2+ turrets if it can build a troop/vehicle, but what do you say? Do you agree that it is sometimes helping the AI more than hurting?


I would simply say let's try it, and we'll see. Let's force the A.I to limit the number of turrets they can build in Tier 1, perhaps even Tier 2 (or maybe limit the turrets upgrade in Tier 2 for those that were built in Tier 1 and for new planned ones too), to maybe just one or a maximum of two. I'm curious about that and I would like to see how much of "help" it could be for the A.I especially with Standard Resources rate.

But there's one thing I would highly recommend about that. If we come to agree that in the end it does help the A.I perhaps tech faster and get a better army on the field faster, then I would ask Arkhan/Thudo to please consider making the turret building limitation an option in the A.I control panel, and not to force it upon everyone. I'm all for changes, but I'm also a fan of accommodation and modularity.

I know I myself like to see the A.I getting lots of defenses early, but that's more a personal taste than anything related to "good strategy". If the change is imposed and not made as an option then I know I will personally not like it even if the majority says it's better that way. I'd say that just an option in the control panel for that if at all possible, letting everyone chose what they prefer for the A.I defenses will be, in my opinion, the best thing to do for the final build.

#32 Inquisitor

Inquisitor
  • Members
  • 121 posts
  • Location:Ordo Malleus

Posted 17 November 2008 - 10:49 PM

I'd say that just an option in the control panel for that if at all possible, letting everyone chose what they prefer for the A.I defenses will be, in my opinion, the best thing to do for the final build.

I think it would be better to aim for the strongest AI out-of-the-box. In my humble opinion, the aim should be to create the greatest challenge without changing any options. If the AI plays better with many turrets it should be building them. If the AI plays better with less turrets it should refrain from building them. Regardless of the player's playing preferences.

How can we determine what is the most cost-effective for the AI? I use current human play as a benchmark (mostly derived from viewing experts replays and discussions), but the AI doesn't play like a human, so maybe it's a fundamental mistake? Maybe, if it would be possible to let one AI race be a guinea pig? Say, if SM were only allowed to build very few and we tested it vs. the another AI race (many games, same map/ starting loc/ opponent) and see how it fared compared to the current? If I could get a special build/instructions how to change the max turrets/tier, I could run say, 20 games of each. That would give us some statistical data and some pointers to what makes the AI the strongest.
The Emperor knows, the Emperor is watching.
Want to zoom out further? Try this camera mod.

#33 Zenoth

Zenoth

    title available

  • Members
  • 469 posts

Posted 17 November 2008 - 11:49 PM

I'd say that just an option in the control panel for that if at all possible, letting everyone chose what they prefer for the A.I defenses will be, in my opinion, the best thing to do for the final build.

I think it would be better to aim for the strongest AI out-of-the-box. In my humble opinion, the aim should be to create the greatest challenge without changing any options. If the AI plays better with many turrets it should be building them. If the AI plays better with less turrets it should refrain from building them. Regardless of the player's playing preferences.

How can we determine what is the most cost-effective for the AI? I use current human play as a benchmark (mostly derived from viewing experts replays and discussions), but the AI doesn't play like a human, so maybe it's a fundamental mistake? Maybe, if it would be possible to let one AI race be a guinea pig? Say, if SM were only allowed to build very few and we tested it vs. the another AI race (many games, same map/ starting loc/ opponent) and see how it fared compared to the current? If I could get a special build/instructions how to change the max turrets/tier, I could run say, 20 games of each. That would give us some statistical data and some pointers to what makes the AI the strongest.


There are things that can be imposed, and other things that shouldn't be, but that's how I see it. Now of course the coders have the final say, and I'm not here to oblige them to do it, I'm simply recommending them to make that an option because I feel it'd be an undue imposition. I will certainly live with it if it's not tweakable in the control panel, I won't cry about it, but I certainly won't like it, I'm just being honest here. I just believe that I have the right to play games like I want them if at all possible, if the coders have the time, will and knowledge to do it.

The "strongest A.I" principle and possibility stays the same. The A.I will be the strongest it can be, it will be the most efficient by the expert's views it can be, if the player sets the slider in the control panel to make it so. So no one will be deprived of the possibility to play against the best this mod can offer, and no one will be deprived of an alternative play style if they so wish, like I do. I myself will play with the restricted turrets, but not always, it's just the point. It'll simply add variety to the mod in the end, and that's certainly not a bad thing, at least not in my book.

As I said, as a tester I'm all for testing it if Arkhan/Thudo want to try it too, why not? There's nothing to lose by testing it, I'm just following the ideas and I respect all other tester's views and the fans' views as well. But in the final build I just don't want that type of decision in game-play change to be imposed, but instead suggested to the player. But anyway, I've made my point by now, I'm just repeating myself.

#34 ArkhanTheBlack

ArkhanTheBlack

    title available

  • Members
  • 814 posts

Posted 18 November 2008 - 01:59 AM

If it isn't too much trouble, I'd like to know why there are two lines and why they differ in the way they do? I would help me understand more and thereby provide more qualified feedback. Thanks

The build controller computes on build item after another and checks if it already exists. The first item in the list that doesn't exist will be built. Therefore if the build controller get a percentage value of 1% LP upgrades, it will be upgrade one LP because every percentage value will at least rounded up to 1 instance. If the AI techs further it will reach the next percentage value of x% and make sure that enough post are upgraded to fullfill the condition. I'd advice you to read the AI docs if you're interested in how the build programs work.


The turret count is already very low. There's one turret after HQ upgrade 1 was started and 2 after HQ upgrade 2 was started. The turret spam of 4 turrets starts at the end of tier 3 when all resource upgrades are done. That's either very late or the AI is doing especially well. At that point the AI uses already very large army sizes where sacrificing some turrets wouldn't make any difference. The only way that could get a problem if the AI is forced into a turret replacing loop if you manage to destroy an AI turret everytime it tries to replace the last one.
I'm more worried about the Necron placement problems.

Edited by ArkhanTheBlack, 18 November 2008 - 02:01 AM.


#35 troubadour

troubadour
  • Members
  • 88 posts

Posted 18 November 2008 - 08:14 AM

@Arkhan
* Many thxs for your explanations but then I think there is a problem with the turrets count, please watch the replay, once HQ1 upgrade was done,AI starts building turrets like mad, it built 5 turrets and never reach HQ upgrade 2 and none of these turrets were destroyed

* Moreover going straight to 3 servitors at T1 is a bit too much, 2 servs shall be enough in T1 (when not a QS game) when going to T2-T3 then a 3rd serv is fine

* AI manages skull probes in a strange way, AI lost a lot of squad to Tau (me) whereas probes remain near AI LP, attaching them to squad should help
More advice from other players needed about this problem

Settings were : AI harder, everything else set to STD (res rate)

Attached Files



#36 ArkhanTheBlack

ArkhanTheBlack

    title available

  • Members
  • 814 posts

Posted 18 November 2008 - 04:39 PM

* Many thxs for your explanations but then I think there is a problem with the turrets count, please watch the replay, once HQ1 upgrade was done,AI starts building turrets like mad, it built 5 turrets and never reach HQ upgrade 2 and none of these turrets were destroyed

Relic building spam bug! Happens form time to time with different buildings...


* Moreover going straight to 3 servitors at T1 is a bit too much, 2 servs shall be enough in T1 (when not a QS game) when going to T2-T3 then a 3rd serv is fine

They should only build 3 servitors if they make a fast tech.


* AI manages skull probes in a strange way, AI lost a lot of squad to Tau (me) whereas probes remain near AI LP, attaching them to squad should help
More advice from other players needed about this problem

They should attach to ranged squads, but they have a min unit strength or they won't attach.



Reply to this topic



  


0 user(s) are reading this topic

0 members, 0 guests, 0 anonymous users