38 replies to this topic

[thudo]

Testers! Get at it!

- Some Necron changes (Necron Lord research was tweaked), 4th Power Generator now built @ gamestart
- Updated docs (for future AI modders)
- Orcs build free slugga squads now

[thudo]

Found something interesting.. I am currently majorly tweaking the IG to survive against our new Necron AI.. Its frick'n tough. Necrons just spit right over Guard no matter what I can do. However.. I'm trying..

If anyone can recommend a way for the IG to survive against our Necron AI in a 1vs1 HARD situation then I'm open. I simply cannot keep IG alive for very long after the first IG HQ_addon is built. I tried getting HellHounds built then Basilisks out the same way we have it now (Basiliks then Sentinels) but no change at all. Brutal!!!!

Now..

I found out why the AI sometimes doesn't build an LP on a local captured strat. Reason. Its that initial builder scripted in the first build sequence. Remove it from there and bingo.. no issues. Can this be addressed so we can put that builder back into the sequence? It doesn't matter to the speed of building up at gamestart but it makes a major difference if its NOT there since an extra LP is built which would otherwise not be.

Edit: found that the above doesn't work all the time - put the IG AI on the 2nd startpoint on 2p_BloodRiver and it will not build an LP on the river there until late. Not sure why not.

Also 3 other observations:

- Heroes will sometimes switch to ranged when attacking an enemy LP
- With Heroes mode on, a deployed IG HWT team does not show what XP level its at. Is this a Relic bug?
- Individual units will sometimes pause in the middle of combat and do nothing while being attacked. 9/10 times they are killed.

I know about the latter but Arkhan likely cannot resolve this one as we discussed. Also.. its somewhat semi-rare.

[LarkinVB]

Question :

If the necron AI has 100 power and will enter this code won't the phase shifter research be build before the solar pulse ? If yes is this what was intended ?

	elseif (self.m_iNecronLordResearchPattern == 3) then
	
		-- Target Pos: Infiltrated units revealed, blinds units causing them firing inaccuracies.
		self:DynamicResearch("necron_solar_pulse_research", 1, 0, 125, 0, 0, "necron_lord_squad", 1)
	
		-- Timed: Lord and phase-shifted units are immune to all forms of damage, but cannot attack.	
		self:DynamicResearch("necron_phase_shifter_research", 1, 0, 75, 0, 0, "necron_lord_squad", 1)

[Inquisitor]

Testers! Get at it!

- Some Necron changes (Necron Lord research was tweaked), 4th Power Generator now built @ gamestart
- Updated docs (for future AI modders)
- Orcs build free slugga squads now

Nice work!

Found something interesting.. I am currently majorly tweaking the IG to survive against our new Necron AI.. Its frick'n tough. Necrons just spit right over Guard no matter what I can do. However.. I'm trying..

If anyone can recommend a way for the IG to survive against our Necron AI in a 1vs1 HARD situation then I'm open. I simply cannot keep IG alive for very long after the first IG HQ_addon is built. I tried getting HellHounds built then Basilisks out the same way we have it now (Basiliks then Sentinels) but no change at all. Brutal!!!!

It is a really hard MU that might require some special tweaking of IG to make it stand a chance against the resilient Necrons. I have previously mentioned that GL are essential in the beginning, so if you could make the AI make GL high priority vs. Necrons they might survive long enough to get Hellhounds out. The more GL the better, so you might also consider going for the Weapon Specialization (in Tactica Control) research as early as possible. The CS is also essential in the early game and it should have one or more Priests to give it more CC punch. The Basilisk got a hard nerf in 1.2, so it is not as important as in 1.1. Later in the game the GL become less important and then plasma is probably preferred as they are more versatile and have no setup time. Ogryns also make short work of most Necron troops.

@Inquisitor - we cannot control how buildings are arranged at the startpoint. We can "bias them" towards or away from the enemy but if you see Necron AI getting units stuck then imagine OTHER factions with larger units? Blame Relic. To compensate, we use the IG's Mars Pattern building to be built on the front so the BaneBlade would normally have enough space to operate.

Alright, I suspected that we could not control the building placement in great detail, but i would really like to hear your view on my suggestion to "fixing" stuck units that I posted in the other thread:

• ...
• When a unit is given a DoMove(...) its location should be cached and the next time it is given a location it should be compaired with the previous location. If its location is the same as the last time then the unit is marked as stuck (a boolean). If the unit is then given a DoMove(...) and the locations match and it IsStuck it should be simply be deleted. This solution would also "fix" units stuck to bad pathing (which sometimes happens on a lot of maps).

Could it work? I know it might hurt the AI resources if it kept building units that would just get stuck, but if we find that it keeps doing that, we might make the stuck boolean an integer instead and only return true on IsStuck() if the integer value is has reached a given value. This would make the stuck code tweakable and should probably be preferred to my initial suggestion.

[ArkhanTheBlack]

I found out why the AI sometimes doesn't build an LP on a local captured strat. Reason. Its that initial builder scripted in the first build sequence. Remove it from there and bingo.. no issues. Can this be addressed so we can put that builder back into the sequence? It doesn't matter to the speed of building up at gamestart but it makes a major difference if its NOT there since an extra LP is built which would otherwise not be.

Edit: found that the above doesn't work all the time - put the IG AI on the 2nd startpoint on 2p_BloodRiver and it will not build an LP on the river there until late. Not sure why not.

I think it's more a problem of locked builders and resource problems.

With Heroes mode on, a deployed IG HWT team does not show what XP level its at. Is this a Relic bug?

HWT respawn respawn each time they change their entrenched state. This is exactly like dying for SCAR and they therefore loose their rank.

Individual units will sometimes pause in the middle of combat and do nothing while being attacked. 9/10 times they are killed.

Relic bug!

If the necron AI has 100 power and will enter this code won't the phase shifter research be build before the solar pulse ? If yes is this what was intended ?

The order is not really important to me. Both abilities can be usefull at the start although I usually prefer solar pulse.

The Basilisk got a hard nerf in 1.2, so it is not as important as in 1.1.

Okay, that's what I wanted to know. I always like the hellhound more than the basiliskbecause of the cool pyro effects , therefore I definitely support the change.

The CS is also essential in the early game and it should have one or more Priests to give it more CC punch.

In 1.1 everyone said the psychos are more important because of its detection skill (at least one).

Could it work? I know it might hurt the AI resources if it kept building units that would just get stuck, but if we find that it keeps doing that, we might make the stuck boolean an integer instead and only return true on IsStuck() if the integer value is has reached a given value. This would make the stuck code tweakable and should probably be preferred to my initial suggestion.

We can't delete units.

[LarkinVB]

I think it's more a problem of locked builders and resource problems.

Had the same thing happen. Builder got locked and didn't move and LP remained unbuild for a very long time.

[Inquisitor]

The Basilisk got a hard nerf in 1.2, so it is not as important as in 1.1.

Okay, that's what I wanted to know. I always like the hellhound more than the basiliskbecause of the cool pyro effects , therefore I definitely support the change.

Yeah, the cooldown was increased by 50% from 6 to 9 seconds, so they fire less frequently.

The CS is also essential in the early game and it should have one or more Priests to give it more CC punch.

In 1.1 everyone said the psychos are more important because of its detection skill (at least one).

Probably still true, but vs. Necrons detection is a rare problem.

We can't delete units.

Argh! Didn't think of that possible limitation. Thank you for your reply!

[ArkhanTheBlack]

Probably still true, but vs. Necrons detection is a rare problem.

Well, all suggestions should made as a general improvement and not just versus special races. We run in circles if we don't keep that in mind.

Builder got locked and didn't move and LP remained unbuild for a very long time.

That's the "builder gets trapped" problem. I thought Thudo was talking about LP's which are just built very late (or not at all).

[ThetaOrion]

If anyone can recommend a way for the IG to survive against our Necron AI in a 1vs1 HARD situation then I'm open. I simply cannot keep IG alive for very long after the first IG HQ_addon is built. I tried getting HellHounds built then Basilisks out the same way we have it now (Basiliks then Sentinels) but no change at all. Brutal!!!!

--

There has been a lot of complaint the past couple weeks about the nerfed nature of the IG over on RelicNews. Most people are converging on the belief that the IG are the weakest link or the weakest faction in 1.2. There always has to be a weakest faction. The question is, "How bad is it?"

Of course, I already knew that the plain vanilla Relic AI was still broken where the 1.2 IG AI was concerned. They didn't tier up or produce anything, and I kind of suggested on RelicNews that all of that would probably change once the IG had a real AI.

I suspected that the 1.2 IG might be fine in the hands of a human, but certainly weren't fine in the hands of Relic's AI.

But, it's indeed possible that the IG will still just be way too weak to stay in the game, even in the hands of a human, even with the 2.3 AI Mod in place.

That's the way it was in 1.1 as well and with AI mod 2.2, and Aralez put about a fourth of his effort into the IG alone while making his UC Rebalance Mod, just so that the IG could stand up to the Necrons.

I suppose from your input and the input of others, that Aralez is going to have to do the same once again in his Rebalance Mod. Aralez has already stated that he intends to do another updated UC Mod after the release of the 2.3 AI Mod, if the UC Rebalance Mod is really needed. The UC Rebalance Mod for the 2.3 AI mod might indeed really be needed for the IG, if for nothing else, from the input that keeps coming in

Indeed, a Rebalance Mod might be the only way to really fix the problem to most people's satisfaction. I'm also glad to know that Aralez is willing to give it another go, as I really liked the 2.2 AI mod UC Mod combination the last time around! And of course, Aralez's original UC mod for plain vanilla DC 1.1 was rather worthless until the AI Skirmish Mod for DC came along and became the core foundation for it all, and then it was all suddenly better than anything I had ever experienced before, simply by combining the two of them together and having a lot of fun!

--

For myself, I have never been able to see or understand how Relic could ever balance the game for Skirmish mode without first making a real AI for themselves, and it has been a year or so since version 1.51, where Relic actually seemed to have a real AI and a real balance of their own, or at least an AI that was good enough that they could actually produce a balance for Skirmish Mode in-house that actually worked and was good enough all around.

How can Relic truly balance the game for skirmish mode gameplay if they don't even have a working AI of their own?

Edited by ThetaOrion, 25 April 2007 - 11:33 AM.

[thudo]

I would honestly like you, as an IG played, to counter Necrons in Tier1-2 on that map, 2p_BloodRiver against our new Necron AI. No matter what I bet you can't. Guardsmen and their buildings are like sand-paper to Necron Guass ordnance. Against everyone else, IG is well balanced but once "the march of the warriors" starts your f**ked beyond all telling!

THERE IS NOTHING an IG player can do. N-O-T-H-I-N-G ! ! ! I tried everything with more Hellhounds or Chimeras and they are "utter poo" and cannot get enough out or tech fast enough before being totally overcome.

I'm all ears folks but its a lost cause with IG vs Necrons.

As for the builder doesn't build an LP on a captured strat - this is reproducable on the map and startpoint and faction I mentioned. If this could be addressed..

So with IG.. are we going with Hellhounds/Baslisks or Hellhounds/Sentinels? IG really needs good anti-vehicle by then as they have none so Sentinels would be best. Baslisks can come third.

[Inquisitor]

Probably still true, but vs. Necrons detection is a rare problem.

Well, all suggestions should made as a general improvement and not just versus special races. We run in circles if we don't keep that in mind.

I was not suggesting. I was just answering thudo's question on how to beat Necrons with IG.

There has been a lot of complaint the past couple weeks about the nerfed nature of the IG over on RelicNews. Most people are converging on the belief that the IG are the weakest link or the weakest faction in 1.2. There always has to be a weakest faction. The question is, "How bad is it?"

My impression of the mood on the strategy forum is that it is not as bad as initially thought. There have been thought up some anti-necron builds that work quite well. I outlined some of it above. There will always be people that think that they have found the Holy Grail of balance, but I bet if every one got it their way the game would be even worse than it is now. Just browse the DOWpro forums - there are just as many heated debates about what needs to be nerfed or buffed.

...
I suspected that the 1.2 IG might be fine in the hands of a human, but certainly weren't fine in the hands of Relic's AI.

But, it's indeed possible that the IG will still just be way too weak to stay in the game, even in the hands of a human, even with the 2.3 AI Mod in place.

It requires good micro and a very specialized anti-Necron build, so probably not in this version, since a specialized anti-Necron build is not part of it.

For myself, I have never been able to see or understand how Relic could ever balance the game for Skirmish mode without first making a real AI for themselves, and it has been a year or so since version 1.51, where Relic actually seemed to have a real AI and a real balance of their own, or at least an AI that was good enough that they could actually produce a balance for Skirmish Mode in-house that actually worked and was good enough all around.

How can Relic truly balance the game for skirmish mode gameplay if they don't even have a working AI of their own?

I think that a good AI needs to imitate humans to be worth playing against. For instance, an AI that calculates dps and always builds the right forces to kill your deployed units, would just be found abusing and games against it would just not be fun. There are some things you can never code, like why people prefer some things over the other. It might have something to do with the control interface or that it is just not something humans would think of doing. That's my theory anyway...

Another fact is Relic first started making real progress in balancing the game when they stopped calculating dps and instead watched how people played the game. This was back in the WA days and they have stuck with that approach and has balanced the seven races pretty well. I think the most problematic MU atm is Tau vs. Necrons, since Tau have big problems with SP, Wraiths and FO mainly because FW have a setup time and/or lack of decent early CC. Later in the game they don't really get any troops that makes a big difference like the IG does.

[thudo]

Oh so grenade launchers for Guardsmen will win against Necrons? Ya nuts? Guardsmen need to be standing still and then the Necrons just walk to them and because IG in the early tiers have such paper armour they get slaughtered. Anyway, IG just has a serious problem with Necrons and thats it although I'm sure Tau also slaughter em as well but thats a given.

[ArkhanTheBlack]

So with IG.. are we going with Hellhounds/Baslisks or Hellhounds/Sentinels? IG really needs good anti-vehicle by then as they have none so Sentinels would be best. Baslisks can come third.

Hellhounds and Sentinels were fine before the Basilisk hype came up, therefore I'd go back to that constellation.

I was not suggesting. I was just answering thudo's question on how to beat Necrons with IG.

Sorry, my mistake.

As for the builder doesn't build an LP on a captured strat - this is reproducable on the map and startpoint and faction I mentioned. If this could be addressed..

Do they never build the LP or just very late?

[thudo]

Bah! Just tried it and the IG AI actually built the LP on the strat they consistently didn't last nite. Arghh... The joys of AI scripting.

Oh and Inquisitor? I forced the IG AI to build GrenadeLaunchers on their troops real early. That helped... for like 20seconds longer!!!!!!!!! IG have zero counters to mass Warrior rushes headed by the Necron Lord that early on. None!!!!!!! I'm shocked how ineffective Tier1-2 IG units are. Its criminal. Warrior swarms just steamroll like an unstoppable locust armada! Arrgghhh!

[Inquisitor]

Oh so grenade launchers for Guardsmen will win against Necrons? Ya nuts?

Nuts? Probably, but I read a lot of strategy forums and often play Necrons against a pretty decent IG player that tend to win over me lately - after he has "found the formula". His formula is GL, CS and turrets until Hellhounds in large numbers.

Guardsmen need to be standing still and then the Necrons just walk to them and because IG in the early tiers have such paper armour they get slaughtered.

Standing still? All troops need to be standing still (except NW), since firing on the move is only at 15% accuracy in DC 1.2 (up from 10% in DC 1.1) - that goes for vehicles as well! And GLs have vastly superior range to NW until T3.
You have to stay out of NW range. There is no way you can win by thinking you can outshoot the resilient and hard-hitting NW. Luckily it is not particularly difficult, since they are slow and have extremely short range.

Anyway, IG just has a serious problem with Necrons and thats it although I'm sure Tau also slaughter em as well but thats a given.

Might be, but Tau have no chance in 1v1 against a decent Necron player. The match up (MU) is hopeless for the Tau. If the AI Tau is winning it is because the AI Necron is not played very well.

Let me know if you need references for all of this. I can provide links to discussions and pro replays if needed.

[ArkhanTheBlack]

Let's just fix the obvious issues! We don't have to reach a 50/50 win rate for all race combis. The AI likes tough and simple races. As soon as skill is required it looks bad for the AI...

[kreative]

Sorry for my lack of feedback lately, been on vacation.
New changes look good, getting back into the fray asap.

[LarkinVB]

SM + tau vs chaos + orks at mountain trail :

1. Great slugga spam Arkhan. I tested and it worked fine.

2. Concept of fast walkers (first vehicle to be build) is somewhat limiting (at least orks). They could get out trakks even faster which are very usefull as ranged shooters during early T2 combats. Orks are in need of ranged support much more than in need for another slow CC vehicle. Therefore maybe it is possible to just build enough banners + vehicle shop for a quick trak as allround strategy ? There is already a dedicated killa kan strategy.

3. Chaos did build second sac pit + 2 turrets + vehicle building near LP which was far far away in enemies territory and undefendable. I think there should be a maximum distance limit for outpost buildings. Same for orks. I think it is fine to build them in front but not just across the map.

4. Tau commander is rarely if ever using his jump to escape from hopeless CC situations. I think he should get a modified EmergencyRetreat code like the necron lord.

5. Chaos did build three raptors just after lord and one marine squad at allround strategy. Since the AI is not that good with jumpers (old attacking to CC, then retreating just before enemy problem) I think they should be limited to max 2.

6. Any reason why tau are the only race to build first generator before barracks ? It's very minor, I just wondered. Tau commander is the only hero who is appearing late in some build programs.

7. SM is researching sniper rifles fast but rarely (if ever) upgrading to rifles in T1. Either force some upgrades to rifles or delay the research to T2.

8. KaiserSoze from DoWPro changed the auspex code so it will be thrown towards the infiltrated squad, not just dropped. Perhaps it is usefull for this mod too.

	if (self.squad_ai:CanDoAbility(ChaosMarineInfantry.detector_id)) then
--~		 aitrace ("ChaosMarineInfantryTactic:Squad can do chaos detect")
		local oEnemy = Ability.Filters.CloseInfiltratedEnemy(self.squad_ai:GetPosition(), 35, 1)
		local range = self.squad_ai:GetAbilityRange( ChaosMarineInfantry.detector_id )
		if (oEnemy ~= nil) then
			aitrace ("  Squad using Chaos Detector")
			--print ("Squad using Chaos Detector at range: "..tostring (range))
			local vSquadPos = self.squad_ai:GetPosition()
			local vEnemyPos = oEnemy:GetPosition()
			local vTargetPos = Vector3f(vSquadPos)
			local iDistance = distance (vSquadPos, vEnemyPos)
			--If target is too far awy launche auspex in range
			if (iDistance>range) then
				--print ("Cloak is away: "..tostring (iDistance))
				local fThrowFactor = range / iDistance
				vTargetPos.x = vTargetPos.x + (vEnemyPos.x - vSquadPos.x) * fThrowFactor 
				vTargetPos.z = vTargetPos.z + (vEnemyPos.z - vSquadPos.z) * fThrowFactor
			else
				--print ("Cloak is close: "..tostring (iDistance))
				local vTargetPos = Vector3f(vEnemyPos)
			end
			self.squad_ai:DoSpecialAbilityPos(ChaosMarineInfantry.detector_id, vTargetPos)
			
		end
	end

Edited by LarkinVB, 25 April 2007 - 05:51 PM.

[thudo]

2 - We can force out Ork Trakks moreso than KillKans in a buildprogram but the Ork Unitstats are now perfect. We get a good battlefield experience now.

3 - Hmm.. this script is universal for all factions. The outpost you saw might have been a rare construction. Further, the reason that happens as testers liked seeing the AI "push the front" and build production buildings forward to send the troops closer to the enemy. Mind you, if those outposts are overthrown, the AI just builds em back.

4 - Interesting.. Arkhan?

5 - Easy enough to limit assault marines for both SM and CSM.

6 - Fixed! Thanks for that observation!

7 - Actually, I've seen the AI research this as far as Tier3. Its kind of hit-and-miss since it works in some games well and others not so much. You really need to play alot of games to see the average. Generally the Marine's Sniper research does work.

Also.. I gather we need code so that All Hero units should not attach to infiltrated units like the Farseer doesn't.

Oh and Inquisitor - Necrons don't have a chance against a good Tau player because Tau can piss over the enemy using its insanely unfair range attacks and then start spamming its Tier3 crap. See this too many times hence why Tau will likely always been too uber. Necrons are good with the "death march" but if they can't get to their distant enemies then all Undead have to die eventually.

[ArkhanTheBlack]

2. Concept of fast walkers (first vehicle to be build) is somewhat limiting (at least orks). They could get out trakks even faster which are very usefull as ranged shooters during early T2 combats. Orks are in need of ranged support much more than in need for another slow CC vehicle. Therefore maybe it is possible to just build enough banners + vehicle shop for a quick trak as allround strategy ? There is already a dedicated killa kan strategy.

We can add a trak before the kan. This should be a good compromise.

3. Chaos did build second sac pit + 2 turrets + vehicle building near LP which was far far away in enemies territory and undefendable. I think there should be a maximum distance limit for outpost buildings. Same for orks. I think it is fine to build them in front but not just across the map.

Hmm, I like their front base approach since they usually build at the choke point position where their army gathers. This makes it a lot harder for the enemy to conquer the LP back.

4. Tau commander is rarely if ever using his jump to escape from hopeless CC situations. I think he should get a modified EmergencyRetreat code like the necron lord.

Why not? Will do!

5. Chaos did build three raptors just after lord and one marine squad at allround strategy. Since the AI is not that good with jumpers (old attacking to CC, then retreating just before enemy problem) I think they should be limited to max 2.

I'll reduce their rating and limit them to 2. This should keep them in track.

6. Any reason why tau are the only race to build first generator before barracks ? It's very minor, I just wondered. Tau commander is the only hero who is appearing late in some build programs.

Yes, Tau are the only race which don't have their barracks in the starting build order. The reason for that is simple: They have two barracks, the common Tau one and the Kroot hut. The Kroot strategy starts with the Kroot barracks first and builds the common Tau barracks later, while the other strategies build the common Tau barracks first. That's also the reason why the Tau commander is built very late in the Kroot strategy.

7. SM is researching sniper rifles fast but rarely (if ever) upgrading to rifles in T1. Either force some upgrades to rifles or delay the research to T2.

Yes, we can delay this to tier 2.

8. KaiserSoze from DoWPro changed the auspex code so it will be thrown towards the infiltrated squad, not just dropped. Perhaps it is usefull for this mod too.

Why not. Will be included... Thanks!