38 replies to this topic

[Zenoth]

Ok, everyone, let's take a deep breath and let's all calm down with the Necrons.

1) Dawn of Skirmish 3.10 is meant to be first and foremost a compatibility build to allow fans of the mod to play it with the latest patch.

2) Dawn of Skirmish 3.10 is not meant in re-balancing the Necrons economy via new A.I tactics.

3) If any of you wish to have a better performing Necrons in Soulstorm ask Relic to make a balance-focused patch à-la Dark Crusade patch 1.2, re-balancing the game has never ever been the goal of the Dawn of Skirmish team.

4) If anything of importance shall be done for the Necrons in this modification it should not be done in this version, but a future one, when Soulstorm receives its own major balance-related patch, and if that doesn't happen, which wouldn't surprise me, then the A.I team won't play Relic and won't try to imitate DOWPro either.

And what is all about competitive circles and whatnots? Since when is the A.I meant to be able to compete against the greatest of the SS or DC players? I've always thought that RTS experts played with and against themselves on-line with the A.I completely excluded from the scene. If A.I has somehow become the reference for expert play styles then I must have missed something somewhere within the past decade or so.

The Necrons in Dawn of Skirmish as of now will destroy pretty much any players around except for the competitive players, and since competitive players usually don't like to play against or with A.I why should it matter what they think of the Dawn of Skirmish Necrons all of a sudden? I personally think that the Necrons are annihilators as it stands at the moment, I'm entirely on Thudo's side, I've been testing this modification since the Winter Assault days and ever since Dark Crusade introduced the Necrons I've always thought that they were the most dangerous of the factions that this mod can put its hands on.

Now, let's focus on maintaining bugs away from this build so it can get the heck out of the testing phase so that the fans can finally have a reason to play it off-line outside of the competitive values. If they want to play with a better A.I or against better Necrons they should stay away from off-line play and spend more time on-line, that's how I see it anyway.

Edited by Zenoth, 20 November 2008 - 03:10 AM.

[thudo]

Works for me, good olde Zen!

Thanks DoS 3.10 testers for the discussion.

[Inquisitor]

Ok, everyone, let's take a deep breath and let's all calm down with the Necrons.

I agree on most of your points, Zenoth. One thing we must no loose sight off is the original motivation for this mod: That Relic's AI suck. The second thing we also ought to keep in mind is what the mod is built upon: Humans playing the game. A couple of examples, where the MOD learned from humans:

What do Relics AI do when outnumbered?
- Send units to their meaningless death.
What do humans do?
- Retreat when outnumbered.
What do the AI mod do?
- Retreat when outnumbered.

What do Relic's AI do when ranged units are attacked in CC?
- Stand and fight.
What do humans do?
- Retreat to safe distance.
What do the AI mod do?
- Retreat to safe distance.

- Build orders.

So, now that we see that this mod wouldn't exist if humans hadn't shown the way, what do you propose in the future? That we play the dumb AI or that we try to mimic humans even more? Which choice would be in the spirit of the MOD? Do we really have any other choice than looking at how the best players play and try to mimic that? Should we just lean back and say that Necrons play way too hard already and shouldn't be improved?

I'm sorry that I only can suggest improvements for the Necrons. When I look at the other races they seem to do it right, but I guess that e.g. expert SB players could propose many tricks and and point out mistakes and such. I'm no real expert in other than Necrons and SM I suggest improvements as I see fit. Besides, I just point out what could be improved.

If you feel that Necrons are too strong it's just your problem. This mod is not about balancing or anything. It's about making the AI play better, no matter if you dislike the race or how it plays.

Hope this came out it was meant to!

[Zenoth]

Ok, everyone, let's take a deep breath and let's all calm down with the Necrons.

I agree on most of your points, Zenoth. One thing we must no loose sight off is the original motivation for this mod: That Relic's AI suck. The second thing we also ought to keep in mind is what the mod is built upon: Humans playing the game. A couple of examples, where the MOD learned from humans:

What do Relics AI do when outnumbered?
- Send units to their meaningless death.
What do humans do?
- Retreat when outnumbered.
What do the AI mod do?
- Retreat when outnumbered.

What do Relic's AI do when ranged units are attacked in CC?
- Stand and fight.
What do humans do?
- Retreat to safe distance.
What do the AI mod do?
- Retreat to safe distance.

- Build orders.

So, now that we see that this mod wouldn't exist if humans hadn't shown the way, what do you propose in the future? That we play the dumb AI or that we try to mimic humans even more? Which choice would be in the spirit of the MOD? Do we really have any other choice than looking at how the best players play and try to mimic that? Should we just lean back and say that Necrons play way too hard already and shouldn't be improved?

I'm sorry that I only can suggest improvements for the Necrons. When I look at the other races they seem to do it right, but I guess that e.g. expert SB players could propose many tricks and and point out mistakes and such. I'm no real expert in other than Necrons and SM I suggest improvements as I see fit. Besides, I just point out what could be improved.

If you feel that Necrons are too strong it's just your problem. This mod is not about balancing or anything. It's about making the AI play better, no matter if you dislike the race or how it plays.

Hope this came out it was meant to!

I never implied that humans didn't influenced the mod, and without wanting to miss any respect towards you Inquisitor (I'm honestly trying not to), I'm asking from what source exactly can we expect human work to take its reference from, other than from other humans? I mean of course humans (players) influenced this mod. The potential problem here is that it's the so called experts that seemingly always dictate how the A.I should be playing to the team, and with deep balance-related arguments. The thing is the A.I is built within the available balance given by Relic. It's not the team of this mod that suddenly change the balance of the game by making the A.I think "like an expert". That's a category of humans (experts), not humans as a whole.

And why saying that in the future I'd propose to go with the "dumb A.I" path? Don't you think that the A.I is good already? Well, at least, you seem not to think so about the Necrons, the thing is we as a team must ensure that we don't impose expert tactics upon all of the mod's fanbase, not all players are experts and some people just like to have a good time. Just seeing the A.I having good build orders and seeing the builder units repairing your structures and vehicles thanks to A.I changes is already enough for them. But the coding team already did miracles, and without forgetting the superb contribution from Corsix and the arrival of Scar in the whole thing allowing even more better A.I in the end than the team ever expected back during the Winter Assault days. As of now there's very little improvements to be done in terms of balance, I want to be clear here. I repeat it, only Relic can really change the balance and make the Necrons as Uber as they were in Dark Crusade, if that's what you really want.

My personal problem with all of this is the damn experts influence argument. I don't want to feel like I'm playing a tournament when I'm playing against the A.I. If I want a real challenge I'll go play on-line. We all know, Arkhan knows, Larkin knows, Thudo knows, I know, the other testers including you Inquisitor should know that humans will always have an advantage over the A.I, we think with emotions and we aren't predictable, the A.I can be even with randomized choices, it still has to chose within a set list of actions it can do. I'm still surprised that you brought the "learning from humans" stuff, of course it is Inquisitor! This mod isn't SkyNet, it's not self-learning nor educating itself. Yes the team based many of the decisions from the community and how people play. But this A.I has a control panel and sliders and difficulty settings. Which is where all of my arguments are eventually falling to, namely accommodation and game-play variety.

Everything I've said so far is very subjective, but as a tester my duty is to stay objective and be open-minded to changes. We are making suggestions, and the coders decide what's possible, what's worth looking at, they have the final words, and believe me I prefer it that way. Now, with all that said Inquisitor, this debacle is justified, yes, objectively I will not deny that looking into new possibilities would be good. Which is why in my previous posts I've clearly mentioned that I would certainly be willing to test such economy-related changes for the Necrons, or for any other races. But in the end, whatever change is made, I want to make it clear that my personal recommendation is that they are made optional changes, and not imposed ones, simply because it's stuff that affects economy, and that will inevitably cause the style of play to be quite different. Not "better", not "worse", but different. It's all about variety, that's fine then. It's what I'd like to see. If changes are done and accepted, let's make them optional. And the number of turrets debate as well, that's still stuff that will make the game-play pace and style noticeably change.

So, to reiterate, yes, let's try new things, I'm not against it, but if we accept it let's please use the existing control panel to the advantage of this mod and let's just add more variety to what the A.I can offer. And I'll say it again, there are things that can be imposed, but others ought to be considered as being "for accommodation" material. It just so happens that what you're suggesting Inquisitor falls within that category (defense-related and economy-related changes, it's not small changes enough to be imposed, it's all I say).

Edited by Zenoth, 21 November 2008 - 04:19 AM.

[Zenoth]

Now, for a proper report.

From Beta 3, which I missed on reporting:

- Improved Eldar FoF behavior when retreating

I can confirm that it has been improved indeed. They seem (so far) to turn it On faster when they run away from danger. I've seen I think two or three occurrences where a squad of Dark Reapers had it Off while firing, then they eventually received enemy fire, then they turned it On and ran away for a short distance, and then they came back firing but it remained On for some time, not very long though, maybe two or three seconds. I'll have to observe more carefully and come back with more details on this.

But, yes, it's definitely better this time around.

- DE soul abilities won't target units out of view range anymore (Rend soul, soul storm, corrosive cloud)

That works now, and it's a pleasure to finally not being forced to garrison my Tech Priests because they were getting mowed down by Rend Soul non-stop, and the same thing goes for any faction's hero units. Nice job.

- Some capturing improvements (No biggy...)

Cool... but, care to tell me what you've done? I'd like to see if it does make a noticeable difference, if not then, no biggy indeed

And from Beta 4 / 5:

- Necrons LP upgrades reduced to 50 % in T1. They upgrade the rest after T3 upgrade was started

By "reduced to 50%" do you mean the number of LP's that get to be upgraded? Or is it the time it takes to upgrade has been recuded to 50%? Or the amount of resources they're waiting to get? I'm asking these seemingly innocent questions simply because I haven't observed much difference in terms of T1 LP upgrades so far when compared to Beta 3, or maybe I just missed it, which would surprise me but it's certainly possible. I'll look at it more carefully for sure.

EDIT: Never mind about that, I just realized it's the number of LP's. Now that I see that I'll go observe the difference it brings when compared to the previous build. For that I think 1 Vs 1's will be best. If it works properly it should help their economy enough to make them more offensive sooner in T1 but especially in early T2 which is where they need their punch.

EDIT #2: Have you actually done that same change for other factions? Now that I think of it, I'm not sure of it's just a coincidence but I'm sure that they (all factions) all upgrade half of their LP's in T1. Unless it's always been like that and I've simply never payed attention to it until now. If that's the case, my apologies for missing on the obvious.

- Melee units don't approach units they can't see, except the infiltrators are attacking

So, similar to the change done for the DE Soul Essence powers? By that I mean that melee units were seeing enemies further than the the Fog of War would make it possible and engaged them? Or is it something else?

- Fixed Heroes Scar bug where after 30mins multiplayer games would de-sync

I'll have to test on that, but if another tester could do it I'd greatly appreciate because the only other player with whom I can play on-line doesn't answer my call, but I'll try again later today.

EDIT: Also, Arkhan, I'd like to ask, technically speaking, under which condition(s) does the Necron Lord goes Nightbringer or Deceiver?

EDIT #2: And what's the situation on the Haemonculus attachment?

EDIT: #3: Something else I'd like to point at on the Necrons is the Plasma Generators placement. I'm just curious, before elaborating on why I'm asking, to know if on a technical side of thing it would be possible to force the placement of the first Generators (number to be determined) when the game starts around LP's instead of the Monolith?

Edited by Zenoth, 21 November 2008 - 09:35 AM.

[thudo]

Also, Arkhan, I'd like to ask, technically speaking, under which condition(s) does the Necron Lord goes Nightbringer or Deceiver?

It cannot: Relic or IL doesn't have the hard-code included for DoPossess() when it comes to two options. Before DoPossess() accounted only for the NB. But since SS we have two options now. Relic has been informed of this back before SS came out.

And what's the situation on the Haemonculus attachment?

Unsure.. the Tactic code for the Haemonculus attachment looks sound.

Something else I'd like to point at on the Necrons is the Plasma Generators placement. I'm just curious, before elaborating on why I'm asking, to know if on a technical side of thing it would be possible to force the placement of the first Generators (number to be determined) when the game starts around LP's instead of the Monolith?

Nope. Thats hardcoded and always shall be since Day0. All factions do it. Further, why is this an issue? The gens are small enough to be protected.

[ArkhanTheBlack]

EDIT #2: And what's the situation on the Haemonculus attachment?

Hmm, they don't attach in ranged mode according to the code but I don't know anymore why it was set up this way. It might be a mistake, opinions?!?

Have you actually done that same change for other factions?

No, and I certainly won't since in contrast to Necrons other races get an increased ressource output from the LPs. However, the full upgrades are also delayed after the tier 2 upgrade.

[Inquisitor]

My personal problem with all of this is the damn experts influence argument. I don't want to feel like I'm playing a tournament when I'm playing against the A.I. If I want a real challenge I'll go play on-line.

What you want can't be done, can't you see that? We can't make it easy to beat while making it hard to beat. This mod is not about cheating or making people comfortable. It about making a good game great. When you play vs. any race you should fear their strong sides and try to hit them on the weak side. If you play vs. Relic's AI you will face an even stream of more or less cheating troops. If you play this AI you will play much more like you would against a well playing human because it learned to play from humans years back. Many humans play the game on increasing levels to prepare them for online play. I think this mod is highly respected amongst those guys. Why? Because it plays 'more' like what you'd expect online. I suggest you reduce the difficulty to suit your preferred challenge.

This release is about tuning the AI to play with the new release and since SS the Necron economy is on a knife's edge. They don't have room for wasting resources unless they want to oppose T4 with T2.

So, to reiterate, yes, let's try new things, I'm not against it, but if we accept it let's please use the existing control panel to the advantage of this mod and let's just add more variety to what the A.I can offer. And I'll say it again, there are things that can be imposed, but others ought to be considered as being "for accommodation" material. It just so happens that what you're suggesting Inquisitor falls within that category (defense-related and economy-related changes, it's not small changes enough to be imposed, it's all I say).

I agree that most of my suggestions are out of scope for this release. I just propose them anyway.

EDIT #2: And what's the situation on the Haemonculus attachment?

Hmm, they don't attach in ranged mode according to the code but I don't know anymore why it was set up this way. It might be a mistake, opinions?!?

I think it's a mistake. The Haemonculus is seen attached in expert games sometimes. Maybe you removed it because it would attach to cc-only (e.g. Mandrakes) troops and thereby prevent them from engaging?

[Maktaka]

Quick question: Where any refinements made to the Dark Eldar build orders for 3.10, specifically the Hall of Blood/Dark Foundry combo build order? Its unusually low early troop count with that one (nothing but unupgraded Mandrakes until nearly minute 5) came up in the Relicnews thread, and I'm wondering if that was examined for any possible tweaks in this release (unless that's also the fast-tech build order, in which case a low troop count is to be expected).

[ArkhanTheBlack]

Quick question: Where any refinements made to the Dark Eldar build orders for 3.10, specifically the Hall of Blood/Dark Foundry combo build order? Its unusually low early troop count with that one (nothing but unupgraded Mandrakes until nearly minute 5) came up in the Relicnews thread, and I'm wondering if that was examined for any possible tweaks in this release (unless that's also the fast-tech build order, in which case a low troop count is to be expected).

There are no changes so far in the DE strategies. I guess the Hellion strategy is the problem. The man drakes and the buildings take too much resources for a successfull Hellion strategy...

[Pseudonymn]

I had a few points to make, but Inquisitor has already made them all for me. Two things I'd like to highlight, however, are:

[snip] If you play vs. Relic's AI you will face an even stream of more or less cheating troops.

It's confirmed "more" rather than "less" cheating troops.

The SP campaign is rife with hair-pulling moments where the stock AI is given abundant cheats in order to win the day rather than presenting an actual tactical challenge with any strategic depth. Reference the AI's borked Honor Guard system, and the its further ability to start maps with two separate and distinct, fully autonomous T4 bases on both attack and defense, providing it with double the pop cap and economic output in the first 5 minutes of play. I call definite shenanigans!

I've said it elsewhere many times before; Relic should have given the DoS boys jobs over there rather than slapped together the pathetic AI that they did. It would have been a far more interesting game. They've known about the mod since DC at least, and what they allowed IL to do with the SP campaign is unforgivable.

[snip]
If you play this AI you will play much more like you would against a well playing human because it learned to play from humans years back. Many humans play the game on increasing levels to prepare them for online play. I think this mod is highly respected amongst those guys. Why? Because it plays 'more' like what you'd expect online. I suggest you reduce the difficulty to suit your preferred challenge.

This release is about tuning the AI to play with the new release and since SS the Necron economy is on a knife's edge. They don't

I'd be one of those people who learned (and still learns) about online play through experimentation with the AI mod. It's an invaluable tool.

My earlier comments to Thudo with respect to perceived Necron IMBA since DC were not meant to address balance tuning issues for this mod. I know that balance tuning is outside the scope of what they're trying to achieve. I only wished to dispel this myth that the Necron are inherently IMBA, as it has persisted into SS where it just isn't true any longer. Yes, the AI mod does some pretty nifty things with the Cronz but they are nevertheless seated squarely at the bottom of the pile and draw either a last choice competitively or a "ZOMG O NOES! I Randomed Cronz!" by anyone who actually wants to win. That is all.

Edited by Pseudonymn, 22 November 2008 - 04:39 AM.

[thudo]

Ok folks..

We need more beta feedback as we need to get this out to the public.

I do have one observation:

o Has anyone noticed that the AI will send one capping squad to an LP, they'll cap it, then just sit there for minutes and minutes? I've seen this for awhile actually and not just in this current build. Sometimes during the first 5mins, a squad will go out (any kind of Tier1 capping squad and not just scouts), cap a point, then hang there for what seems tens of minutes as if the squad is "locked".

Anyone else seen this? Seems to happen to a single squad.

[ArkhanTheBlack]

Anyone else seen this? Seems to happen to a single squad.

I think I've seen that before. A squad captures a point and stays in capture position although they're already finished capturing the point. I guess the capture flag returns a capture status value which is extremely close to the captured value but not identical. I guess it's a Relic bug of not using epsilon-threshold values in this case, since the capture status is a floating point value. It's rare though. Haven't seen that for a while now.

I remove the homunculus attach restriction and then make a release version. Looks pretty finished to me...

[RobertW7928]

I'm just curious about certain abilities is the AI able to utilize them or cannot be helped due to Relic hard coding or other reasons, such as the Necron Lord Destroyer Posses Vehicle or the Necron lords Lightning Field, I'm a little concerned about the Campaign and finding out that the Stock AI cheats, I do hope this isn't present for the AI mod and what about War gear Upgrades with Heroes Feature? do the adjusting statistics apply correctly? and personally the Haemonculus attachment doesn't add any statistical benefits to the squad only seems to be less likely to be killed so what's the point of attaching? plus hero's with the unit alone will achieve at Lv4 - Upgrade to lv2 squads being built.

Edit #2 - just like to add about the hero's feature and transformation process do these units retain there level? - Chaos Lord - Daemon prince, Necron Lord - Deceiver/Night Bringer - Necron lord? i recall you did work on this previously but has this been covered *Completely* with all units in v3.1?

Edit #3 - Doe's the Haemonculus use the torture lamp ability?

Keep up the good Work guys and hope this is the best functioning AI to date & hopefully released Shortly. (I'm Bursting With Anticipation )

Edited by RobertW7928, 27 November 2008 - 02:53 PM.

[Inquisitor]

Looks pretty finished to me...

Yeah, me too.

I'm just curious about certain abilities is the AI able to utilize them or cannot be helped due to Relic hard coding or other reasons, such as the Necron Lord Destroyer Posses Vehicle or the Necron lords Lightning Field, I'm a little concerned about the Campaign and finding out that the Stock AI cheats, I do hope this isn't present for the AI mod and what about War gear Upgrades with Heroes Feature? do the adjusting statistics apply correctly? and personally the Haemonculus attachment doesn't add any statistical benefits to the squad only seems to be less likely to be killed so what's the point of attaching? plus hero's with the unit alone will achieve at Lv4 - Upgrade to lv2 squads being built.

Edit #2 - just like to add about the hero's feature and transformation process do these units retain there level? - Chaos Lord - Daemon prince, Necron Lord - Deceiver/Night Bringer - Necron lord? i recall you did work on this previously but has this been covered *Completely* with all units in v3.1?

Edit #3 - Doe's the Haemonculus use the torture lamp ability?

Some of the abilities can't be used by the AI because Relic made it so. It's a pity, but that's just how it is. Relic left us with a very buggy modding interface and simple stuff simply can't be done without the game crashing.

I think you should expect more or less the same difficulty in the campaign as with 3.0 of this mod.

[thudo]

Certain extension abilities like Ork Squiggoth Rampage or Tau Drone Burrow cannot be used by the AI because Relic didn't have time to export those functions for the AI to use.

The AI should use the Haemonculus' Torture Lamp ability methinks as its a standard ability and its scripted in our AI.

[Pseudonymn]

Ok folks..

We need more beta feedback as we need to get this out to the public.

I do have one observation:

o Has anyone noticed that the AI will send one capping squad to an LP, they'll cap it, then just sit there for minutes and minutes? I've seen this for awhile actually and not just in this current build. Sometimes during the first 5mins, a squad will go out (any kind of Tier1 capping squad and not just scouts), cap a point, then hang there for what seems tens of minutes as if the squad is "locked".

Anyone else seen this? Seems to happen to a single squad.

I've noticed something similar, although I'm not sure it this is related or not, or if it's been discussed elsewhere, but here's a shot in the dark:

Sometimes when I am queuing my build and capping orders, my capping squads will arrive at their target SP/CP/Relic, kneel as though to erect a flag and then sit their indefinitely. If left like that, the whole while, the flag animation will not play and the point will not be capped no matter how long it is left. It seems to happen most often when I queue several cap orders in advance, and seems especially prevalent where the squad had to first decap an enemy flag - it decaps, but does not progress beyond having a hole in the ground. The immediate workaround has been to select the squad, move it away from the point a step or two and resume capping. I've also found that, strangely, issuing orders to cap the same point twice (essentially a double right-click) appears to limit or even eliminate outright the occurrence of this issue.

I've also seen AI units stuck in similar fashion but become active again once attacked, provoking them to stop what they are doing and either fight or fall back.

I'm just guessing, but if your AI build orders have squads following a series of queued commands, they may well be manifesting this same bug.

Edited by Pseudonymn, 28 November 2008 - 07:12 AM.

[RobertW7928]

after playing a few games with v3 of the mod it seems that Dark Eldar & Sisters of Battle seem to be lacking of sorts and don't seem to utilize ther army fully & there use of units/Abilities, thus seem weaker, unlike the Imperial Guard which seems to out do vs SoB & DE, I might be cutting to conclusions but this is how it proceeded Oasis of Sharr 2 games of 4 v 4 (Mass Race War - 4 DE vs 4 IG & 4 SoB vs 4 IG) with Hero's, Fortress, AI Donations, Hard Difficulty, Standard Resources, No Resource Sharing, and lost both matches, these could contribute to many conditions but I'd just like to mention this for any improvements perhaps give these 2 races the likeliness of furious execution of the Imperial Guard, after that i thought I'll try Tau vs IG and won so there you have it.

[Psychobabas]

after playing a few games with v3 of the mod it seems that Dark Eldar & Sisters of Battle seem to be lacking of sorts and don't seem to utilize ther army fully & there use of units/Abilities, thus seem weaker, unlike the Imperial Guard which seems to out do vs SoB & DE, I might be cutting to conclusions but this is how it proceeded Oasis of Sharr 2 games of 4 v 4 (Mass Race War - 4 DE vs 4 IG & 4 SoB vs 4 IG) with Hero's, Fortress, AI Donations, Hard Difficulty, Standard Resources, No Resource Sharing, and lost both matches, these could contribute to many conditions but I'd just like to mention this for any improvements perhaps give these 2 races the likeliness of furious execution of the Imperial Guard, after that i thought I'll try Tau vs IG and won so there you have it.

I'd like to reply to this!

Well IG are a whole different deal on team games. In fact they are considered the best race for 2v2s, 3v3s because players do no get harrassed as much as in 1v1 due to distances, tendency to camp a bit during early game etc. Dark Eldar especially rely on harrassment and are downright lethal in 1v1 early game. True, their late game is one the weakest but they usually deal the killing blow in early game and slowly grind their enemy in later games.

Please release!!!