Up on the FTP site!
Changes:
- Smart garrison/ungarrisoning of troops especially for IG for both buildings and transports (Arkhan and Corsix)
- Smarter deepstriking code
- Overall AI gameplay tweaks suggested by users since v2.1 in early December
- Control Panel now includes ability to turn Donations On or Off
- De-synching multiplayer problems should now all be resolved
IMPORTANT: Please test the Multiplayer aspect of this mod!
Skirmish AI 2.2 Beta 1 - Post Comments In Thread!
Started by thudo, Jan 19 2007 05:27 AM
59 replies to this topic
#1
thudo
-
- Division Leaders
-
- 12,164 posts
Wacko AI Guy!
- Location:Lemonville North, Canada
- Projects:DoW AI Scripting Project
- Division:DoW
- Job:Division Leader
Posted 19 January 2007 - 05:27 AM
Advanced Skirmish AI Team Lead for the coolest Warhammer40k PC RTS out there:
Dawn of War Advanced AI Headquarters
Latest DoW Advanced AI Download!
Dawn of War Advanced AI Headquarters
Latest DoW Advanced AI Download!
#2
ThetaOrion
-
- Members
-
- 676 posts
title available
Posted 19 January 2007 - 07:36 AM
After a month of play with 2.1, the only real complaint that I had against the AI that I could level at the AI was that the Chaos (and Space Marine) AI wasn't building secondary bases or building secondary strongholds outside of their home base on the large maps -- either not building them or not building them soon enough.
Anyway, Arkhan says that he has fixed this now, and it would be nice to hear from some betatester who plays the large maps or the long games that this has been fixed. It had me scratching my head for awhile wondering what was going on.
--
I really don't have any other specific complaints. This AI plays the large maps as well as any AI does, in my opinion. There were some small insignificant things that I ran into now and then, and a few mysteries that I think I solved for myself. It's like Thud is afraid to hear about reports of his AI on large maps or afraid to do a fix if it pertains to a large map, but there's really so few things that the AI is doing wrong on the large maps, that I don't see why Thud has to be paranoid about large map gameplay with the AI Skirmish Mod. It's doing fine, and the changes or problems are rather small and insignificant for the most part.
--
Oh well, I'll try again, since I has now been a couple of months since I last put in a real wish list . . .
First let me say that I think the current 2.1 AI plays the IG, the Tau, and the Orks better than I play them on the large maps!! IG, Tau, and Orks are lots of buildings or lots of research, and the AI never misses a beat, but I sometimes overlook something important while playing these that the AI never overlooks. Read this and weap, Thud, "THE IG AI AND ORKS AI AND TAU AI DOES INFINITELY BETTER THE LARGER THE MAP IS, AS THEY NEED THE LARGE SPACE TO BUILD THAT ONLY A LARGE MAP CAN GIVE THEM!!!" The Orks and IG never shine or reach their full potential unless played on a large map. On those small pathetic maps, all of their good stuff is usually landlocked behind buildings and stuff, and they never get used until the end of the game when their base is being overrun, assuming that they haven't been overrun and eliminated early on a small map. The Tau take awhile to Tier up, but are unbeatable in Tier 4, and again, only a large map really shows them off to their full potential in my not so humble opinion. Nobody builds secondary bases like the Tau, and only on the large maps do you see the AI building lots of huge secondary bases.
Second, I know that I play the Necrons on large maps better than the current 2.1 AI plays them, but only after a couple of weeks of practice and study. I time my march of death onto the enemy base to cooincide with completion of all the Necron research. The AI goes in too soon, and often dies or fails. But, if you go in fully capped and fully researched on your very first march of death, the Necrons are practically indestructible and unbeatable. Really nasty piece of work. My timing is better on the large maps with the Necrons than the AI's timing is. No big deal, as long as the Necrons have two slag deposits back home they are drawing upon. But, I get better results from my initial march of death because I wait a minute or few longer than the Necron AI usually does. AGAIN, NO BIG DEAL! Nothing to flame me about. Just put two slag deposits in each base, and if the Necron AI's first march of death fails, they have all they need to recover and do it again.
Third, I know I play the Chaos better than the 2.1 AI plays them, as I can win, and the Chaos AI often can't on the large maps, and I found that part of the problem was the fact that the Chaos AI wasn't building much if anything outside of their home base. They have no backup. Arkhan says he has fixed this.
Fourth, the Space Marines are a mystery. Sometimes they come alive, and sometimes they do nothing and just simply die. I can't figure it out on the large maps in 2.1. But, if someone dies early, it's always the Space Marines AI or the Eldar AI in 2.1. I know that the Eldar are weak, and Aralez put a day into beefing them up so that they are comparable to everyone else in the UC Mod Rebalance now. But, the Space Marines confuse me. They sometimes deep strike alot, and sometimes they don't do anything. I do get the feeling that the Space Marines AI isn't using heavy cover as well as some of the other factions can or do, but maybe it's just that the Orks or Tau or IG just always outnumber them, and in the end, it's numbers that win? I can't figure it out. Maybe it's the build program -- some really good SM build programs and some really bad ones? Maybe Aralez has weakened them too much in the UC Rebalance Mod in order to compensate for their inherent SM overpowered deep-striking abilities? I have my eyes on the Space Marines, but I have nothing concrete to suggest to improve them on a large map, except they should make sure that they get a servitor or two outside of their base early building a secondary stronghold outside of their base. The SM base seems to be their weakest point, always getting overrun easily. So, I always send my second SM servitor builder someplace else to start a base, and maybe that's why I always do better than the 2.1 AI when it comes to the Space Marines on a large map.
Fifth, I know that I and everyone else can play the Eldar better than the 2.1 AI Mod can play the Eldar. Even in the UC Mod where the Eldar got some massive perks, the 2.1 AI always loses as the Eldar, and I mean always. The Eldar are just no good in the hands of the AI, but in the hands of a human, I win every time. The AI just can't do all of the nice trickery that the human can do, period. I build secondary bases in the middle of nowhere relatively near to an enemy base on a large map, and launch my attacks from there. I build a Webway gate in each of my AI ally's bases so that I can get all of my army from one hotspot to the next instantly. The AI can't do any of this, or doesn't do any of this. The Eldar AI doesn't plant a Webway gate and a secondary base in the middle of nowhere and build a secondary base around it. And, the Eldar AI doesn't go around building webway gates inside of allied bases in heavy cover so that you can keep all of your AI allies alive simply by teleporting from one base to the next as they come under fire. At least the Eldar AI now seems to cloak or shroud their buildings and listening posts. But, The Eldar AI doesn't do all of the nice little extras that make the UC Mod Eldar overpowered in the hands of a human. Instead the UC Mod AI Mod Eldar are still underpowered severely in the hands of the AI. And, I don't think there is anything that can be done about any of this. Did you read that Thud? This is a wish list here, and I don't think there's anything to be done. But, whether you are dealing with a large map or a small map, the Eldar AI is where all the work needs to be done, and it probably starts with Relic or SCAR. Until you have Bonesingers teleporting into allied bases and building webway gates and secondary bases there, or going someplace in the middle of nowhere near an enemy base and building some webway gates and actually using those gates for surprise commando attacks out of nowhere and transporting Eldar troops through webway gates, the Eldar AI will still never be as competetive or as robust as a human play it. And, before Thud freaks out again, we all know that there's not much Thud or Arkhan or anyone can do about it. It's just the reality of the thing. They have done all they can do, unless SCAR can now be used to make the Eldar use the webway gates to get around quickly from one hotspot to the next. The Eldar were the hardest ones for Aralez to try to balance -- aka impossible. Do you balance them for the human or for the AI? He kind of achieved a balance. They are still underpowered and weak if played by the AI, and almost always the first to die. But, I think they are overpowered or just fine when I, a human, play them. I win handily with the UC Mod Rebalance and the 2.1 AI Mod Eldar! But, the Eldar are a challenge for the Mod makers, no doubt about it!!
Anyway, Arkhan says that he has fixed this now, and it would be nice to hear from some betatester who plays the large maps or the long games that this has been fixed. It had me scratching my head for awhile wondering what was going on.
--
I really don't have any other specific complaints. This AI plays the large maps as well as any AI does, in my opinion. There were some small insignificant things that I ran into now and then, and a few mysteries that I think I solved for myself. It's like Thud is afraid to hear about reports of his AI on large maps or afraid to do a fix if it pertains to a large map, but there's really so few things that the AI is doing wrong on the large maps, that I don't see why Thud has to be paranoid about large map gameplay with the AI Skirmish Mod. It's doing fine, and the changes or problems are rather small and insignificant for the most part.
--
Oh well, I'll try again, since I has now been a couple of months since I last put in a real wish list . . .
First let me say that I think the current 2.1 AI plays the IG, the Tau, and the Orks better than I play them on the large maps!! IG, Tau, and Orks are lots of buildings or lots of research, and the AI never misses a beat, but I sometimes overlook something important while playing these that the AI never overlooks. Read this and weap, Thud, "THE IG AI AND ORKS AI AND TAU AI DOES INFINITELY BETTER THE LARGER THE MAP IS, AS THEY NEED THE LARGE SPACE TO BUILD THAT ONLY A LARGE MAP CAN GIVE THEM!!!" The Orks and IG never shine or reach their full potential unless played on a large map. On those small pathetic maps, all of their good stuff is usually landlocked behind buildings and stuff, and they never get used until the end of the game when their base is being overrun, assuming that they haven't been overrun and eliminated early on a small map. The Tau take awhile to Tier up, but are unbeatable in Tier 4, and again, only a large map really shows them off to their full potential in my not so humble opinion. Nobody builds secondary bases like the Tau, and only on the large maps do you see the AI building lots of huge secondary bases.
Second, I know that I play the Necrons on large maps better than the current 2.1 AI plays them, but only after a couple of weeks of practice and study. I time my march of death onto the enemy base to cooincide with completion of all the Necron research. The AI goes in too soon, and often dies or fails. But, if you go in fully capped and fully researched on your very first march of death, the Necrons are practically indestructible and unbeatable. Really nasty piece of work. My timing is better on the large maps with the Necrons than the AI's timing is. No big deal, as long as the Necrons have two slag deposits back home they are drawing upon. But, I get better results from my initial march of death because I wait a minute or few longer than the Necron AI usually does. AGAIN, NO BIG DEAL! Nothing to flame me about. Just put two slag deposits in each base, and if the Necron AI's first march of death fails, they have all they need to recover and do it again.
Third, I know I play the Chaos better than the 2.1 AI plays them, as I can win, and the Chaos AI often can't on the large maps, and I found that part of the problem was the fact that the Chaos AI wasn't building much if anything outside of their home base. They have no backup. Arkhan says he has fixed this.
Fourth, the Space Marines are a mystery. Sometimes they come alive, and sometimes they do nothing and just simply die. I can't figure it out on the large maps in 2.1. But, if someone dies early, it's always the Space Marines AI or the Eldar AI in 2.1. I know that the Eldar are weak, and Aralez put a day into beefing them up so that they are comparable to everyone else in the UC Mod Rebalance now. But, the Space Marines confuse me. They sometimes deep strike alot, and sometimes they don't do anything. I do get the feeling that the Space Marines AI isn't using heavy cover as well as some of the other factions can or do, but maybe it's just that the Orks or Tau or IG just always outnumber them, and in the end, it's numbers that win? I can't figure it out. Maybe it's the build program -- some really good SM build programs and some really bad ones? Maybe Aralez has weakened them too much in the UC Rebalance Mod in order to compensate for their inherent SM overpowered deep-striking abilities? I have my eyes on the Space Marines, but I have nothing concrete to suggest to improve them on a large map, except they should make sure that they get a servitor or two outside of their base early building a secondary stronghold outside of their base. The SM base seems to be their weakest point, always getting overrun easily. So, I always send my second SM servitor builder someplace else to start a base, and maybe that's why I always do better than the 2.1 AI when it comes to the Space Marines on a large map.
Fifth, I know that I and everyone else can play the Eldar better than the 2.1 AI Mod can play the Eldar. Even in the UC Mod where the Eldar got some massive perks, the 2.1 AI always loses as the Eldar, and I mean always. The Eldar are just no good in the hands of the AI, but in the hands of a human, I win every time. The AI just can't do all of the nice trickery that the human can do, period. I build secondary bases in the middle of nowhere relatively near to an enemy base on a large map, and launch my attacks from there. I build a Webway gate in each of my AI ally's bases so that I can get all of my army from one hotspot to the next instantly. The AI can't do any of this, or doesn't do any of this. The Eldar AI doesn't plant a Webway gate and a secondary base in the middle of nowhere and build a secondary base around it. And, the Eldar AI doesn't go around building webway gates inside of allied bases in heavy cover so that you can keep all of your AI allies alive simply by teleporting from one base to the next as they come under fire. At least the Eldar AI now seems to cloak or shroud their buildings and listening posts. But, The Eldar AI doesn't do all of the nice little extras that make the UC Mod Eldar overpowered in the hands of a human. Instead the UC Mod AI Mod Eldar are still underpowered severely in the hands of the AI. And, I don't think there is anything that can be done about any of this. Did you read that Thud? This is a wish list here, and I don't think there's anything to be done. But, whether you are dealing with a large map or a small map, the Eldar AI is where all the work needs to be done, and it probably starts with Relic or SCAR. Until you have Bonesingers teleporting into allied bases and building webway gates and secondary bases there, or going someplace in the middle of nowhere near an enemy base and building some webway gates and actually using those gates for surprise commando attacks out of nowhere and transporting Eldar troops through webway gates, the Eldar AI will still never be as competetive or as robust as a human play it. And, before Thud freaks out again, we all know that there's not much Thud or Arkhan or anyone can do about it. It's just the reality of the thing. They have done all they can do, unless SCAR can now be used to make the Eldar use the webway gates to get around quickly from one hotspot to the next. The Eldar were the hardest ones for Aralez to try to balance -- aka impossible. Do you balance them for the human or for the AI? He kind of achieved a balance. They are still underpowered and weak if played by the AI, and almost always the first to die. But, I think they are overpowered or just fine when I, a human, play them. I win handily with the UC Mod Rebalance and the 2.1 AI Mod Eldar! But, the Eldar are a challenge for the Mod makers, no doubt about it!!
Edited by ThetaOrion, 19 January 2007 - 07:57 AM.
#3
LarkinVB
-
- Members
-
- 1,488 posts
title available
Posted 19 January 2007 - 09:04 AM
I could NOT launch v2.2.
The installer will deinstall v2.1 but seems to not include WXPMod_DCMod.dll which is missing afterwards.
So I get :
On a second thought it seems dangerous to deinstall WXPMod_DCMod.dll at all as it may be installed/used by other mods too.
Also the msvcr71.dll should be included by default when WXPMod_DCMod.dll is added.
Btw a nice tool to to check dependancies of exe and dll is here http://www.dependencywalker.com/
I used it to check that WXPMod_DCMod.dll is dependant on msvcr71.dll
The installer will deinstall v2.1 but seems to not include WXPMod_DCMod.dll which is missing afterwards.
So I get :
09:57:39.82 MOD -- (mod manager) Failed to load mod dll 'WXPMod_DCMod.dll' from MOD definition file 'DowXP.module' (Das angegebene Modul wurde nicht gefunden.)
09:57:39.82 MOD -- Initializing Mod dowai, 2.1
09:57:39.82 MOD -- failed to load Terrain Type definitions
09:57:39.82 WXP Mod Init -- Failed load for step: 'Terrain TYPE Manager'
09:57:39.82 Mod Interface Init -- Failed load for step: 'Mod Init'
09:57:39.82 GAME -- Failed to initialize MOD 'dowai'!
09:57:39.82 FE -- Failed to init MOD!
09:57:39.82 GAME -- Ending FE
09:57:39.82 GAME -- Shutdown quit
09:57:40.32 SPOOGE -- 362.00MB available texture memory
09:57:40.32 SPOOGE -- 362.00MB available texture memory
09:57:40.34
Application closed with error code 4
On a second thought it seems dangerous to deinstall WXPMod_DCMod.dll at all as it may be installed/used by other mods too.
Also the msvcr71.dll should be included by default when WXPMod_DCMod.dll is added.
Btw a nice tool to to check dependancies of exe and dll is here http://www.dependencywalker.com/
I used it to check that WXPMod_DCMod.dll is dependant on msvcr71.dll
Edited by LarkinVB, 19 January 2007 - 09:58 AM.
#4
LarkinVB
-
- Members
-
- 1,488 posts
title available
Posted 19 January 2007 - 09:31 AM
Even if I fix problem mentioned above I get
Unable to load the DLL for MOD 'dowai' -- please use the matching executable! and on the startscreen I get in top left
"Dawn of Skirmish DC AI Projekt 1.0" and no mod in the ingame manager.
Something fishy here....
Unable to load the DLL for MOD 'dowai' -- please use the matching executable! and on the startscreen I get in top left
"Dawn of Skirmish DC AI Projekt 1.0" and no mod in the ingame manager.
Something fishy here....
10:25:17.31 FE -- Loading front end.
10:25:17.31 FE -- Cursor hidden.
10:25:20.18 GAME -- Unable to load the DLL for MOD 'dowai' -- please use the matching executable!
10:25:20.18 MOD -- Initializing Mod dxp2, 1.0
10:25:20.18 MOD -- Warning: Failed to load archive 'W40k\W40kDataGOTY.sga'
10:25:20.18 MOD -- Warning: Failed to open folder 'DXP2\Locale\english\Data'.
10:25:20.18 MOD -- Warning: Failed to open folder 'DXP2\Data_Shared_Textures\Full'.
10:25:20.18 MOD -- Warning: Failed to open folder 'DXP2\Data_Sound\Full'.
10:25:20.18 MOD -- Warning: Failed to open folder 'DXP2\Data_Music'.
10:25:20.18 MOD -- Warning: Failed to open folder 'DXP2\Data_Whm\High'.
10:25:20.18 MOD -- Warning: Failed to open folder 'W40k\Locale\english\Data'.
10:25:20.18 MOD -- Warning: Failed to open folder 'W40k\Data_Shared_Textures\Full'.
10:25:20.18 MOD -- Warning: Failed to open folder 'W40k\Data_Sound\Full'.
10:25:20.18 MOD -- Warning: Failed to open folder 'W40k\Data_Whm\High'.
10:25:21.90 GAME -- Using player profile LarkinVB
10:25:55.25 GAME -- Ending FE
10:25:55.25 GAME -- Shutdown quit
10:25:55.25 MOD -- Shutting down Mod 'dxp2'...
10:25:55.65 SPOOGE -- 361.00MB available texture memory
10:25:55.65 SPOOGE -- 362.00MB available texture memory
10:25:55.70
Application closed without errors
#5
ArkhanTheBlack
-
- Members
-
- 814 posts
title available
Posted 19 January 2007 - 10:43 AM
This is a pure beta installation. Corsix' Mod manager is neither included nor supported here. The mod must be started with the default DC shortcut and not be changed with the mod manager if there's another mod which activates Corsix' mod manager.
The DXP2.module file must use Corsix new DC dll and not the default one and include dowai as required mod. The AI mod message at the top is supposed to be so at the moment. DowXP didn't make any problems with my installation so I guess the problem was the old AI deinstalling the gamemanager file and DOWXP.module didn't find it anymore.
Did you get the last dowai dll error when trying to start a skirmish game or before. Make sure if only loads the basic DC dll and not the new one. The new one should only be loaded in the dxp2.module file.
The DXP2.module file must use Corsix new DC dll and not the default one and include dowai as required mod. The AI mod message at the top is supposed to be so at the moment. DowXP didn't make any problems with my installation so I guess the problem was the old AI deinstalling the gamemanager file and DOWXP.module didn't find it anymore.
Did you get the last dowai dll error when trying to start a skirmish game or before. Make sure if only loads the basic DC dll and not the new one. The new one should only be loaded in the dxp2.module file.
#7
ArkhanTheBlack
-
- Members
-
- 814 posts
title available
Posted 19 January 2007 - 11:57 AM
Exactly! The reason for this unusual procedure is that 2.2 is intended to be released after Relic fixed their stupid mod manager. Therefore Corsix dll doesn't support his current mod manager work around (This was my own wish!). Since Relics patch seems to take years, we are forced to use a rather exotic mod setup at the moment. Just like I said, it's only a beta installation and not intended to be released that way. I'm not happy with the current config either, but thanks to Relics great patch support, we have to live with it at the moment.??? I'm confused. You want me to start the 2.2 with the default DC shortcut though there is a new 2.2 shortcut too ?
#8
thudo
-
- Division Leaders
-
- 12,164 posts
Wacko AI Guy!
- Location:Lemonville North, Canada
- Projects:DoW AI Scripting Project
- Division:DoW
- Job:Division Leader
Posted 19 January 2007 - 03:33 PM
Hmmm.. When Relic does release their new balance patch then we'll have to re-tweak v2.2 due to inevitable balance changes required. Nevertheless, as long as the testers can proceed with testing the current build especially for multiplayer than we would have accomplished our present goals. Well done team!
Advanced Skirmish AI Team Lead for the coolest Warhammer40k PC RTS out there:
Dawn of War Advanced AI Headquarters
Latest DoW Advanced AI Download!
Dawn of War Advanced AI Headquarters
Latest DoW Advanced AI Download!
#9
Zenoth
-
- Members
-
- 469 posts
title available
Posted 19 January 2007 - 05:25 PM
I have two things to say right away about 2.2.
First, in my case, the installation went perfectly well, and I received no error messages at all. It automatically un-installed 2.1, and installed 2.2 afterwards. I do know it's 2.2 running, since I saw the IG using transports and garrisoning very, very efficiently (more on that later).
Second, I can already say and confirm that 2.2 runs MUCH smoother than ANY previous builds I've seen. And I do not mean "faster" here, I really mean smoother as in less stuttering game-play, more constant frame-rates and less performance decrease over-time, even on large maps, official or not (I made three tests solely for performance analysis), in 4 Vs 4's.
Why the installer went wrong for you Larkin I have no idea why
First, in my case, the installation went perfectly well, and I received no error messages at all. It automatically un-installed 2.1, and installed 2.2 afterwards. I do know it's 2.2 running, since I saw the IG using transports and garrisoning very, very efficiently (more on that later).
Second, I can already say and confirm that 2.2 runs MUCH smoother than ANY previous builds I've seen. And I do not mean "faster" here, I really mean smoother as in less stuttering game-play, more constant frame-rates and less performance decrease over-time, even on large maps, official or not (I made three tests solely for performance analysis), in 4 Vs 4's.
Why the installer went wrong for you Larkin I have no idea why
#10
thudo
-
- Division Leaders
-
- 12,164 posts
Wacko AI Guy!
- Location:Lemonville North, Canada
- Projects:DoW AI Scripting Project
- Division:DoW
- Job:Division Leader
Posted 19 January 2007 - 05:45 PM
Well thats easily superb news indeed! Great way to start our weekend! 
Advanced Skirmish AI Team Lead for the coolest Warhammer40k PC RTS out there:
Dawn of War Advanced AI Headquarters
Latest DoW Advanced AI Download!
Dawn of War Advanced AI Headquarters
Latest DoW Advanced AI Download!
#12
LarkinVB
-
- Members
-
- 1,488 posts
title available
Posted 20 January 2007 - 01:43 PM
Orks have a fatal AI error
Sluggas are no scarabs
Sluggas are no scarabs
class 'SluggaBoyzTactic' (OrkInfantryTactic)
function SluggaBoyzTactic:__init( squad_ai ) super( squad_ai )
self:SetName("SluggaBoyz Tactic")
end
function AttackScarabTactic:AlwaysAttack() <<<<<<<<<<<<<< !!!!!!!!
return cpu_manager.cpu_player:IsResearchComplete("ork_recruitment")
end
#13
ArkhanTheBlack
-
- Members
-
- 814 posts
title available
Posted 20 January 2007 - 05:57 PM
Oops! The typical copy paste mistake...
Okay, as a quick fix for the other tester, open Dowai_mod\Data\ai\races\ork_race\tactics\sluggaboyztactic.ai and simply change the line "function AttackScarabTactic:AlwaysAttack()" to "function SluggaBoyzTactic:AlwaysAttack()". That should be enough to fix it.
Thanks for reporting Larkin.
Okay, as a quick fix for the other tester, open Dowai_mod\Data\ai\races\ork_race\tactics\sluggaboyztactic.ai and simply change the line "function AttackScarabTactic:AlwaysAttack()" to "function SluggaBoyzTactic:AlwaysAttack()". That should be enough to fix it.
Thanks for reporting Larkin.
#14
Corsix
-
- Hosted
-
- 290 posts
Code Monkey
- Location:Berkeley, UK
- Projects:DoW AI, DoW Mod Studio
- Blue Text :)
Posted 20 January 2007 - 11:02 PM
Regarding next patch / DLLs:
Next DC patch is currently undergoing closed community beta testing (for top ladder players or something). Hopefully, the game manager will work out of the box. Unfortunately, as the fix I suggested to Relic is the same one my game fix DLL applies, there _may_ still be game manager sync error issues. When the patch does come out, the DLL I've written for AI mod 2.2 will need some retweaking, as the modifications it makes are coded for the current versions of DarkCrusade.exe, SimEngine.dll, etc.
Next DC patch is currently undergoing closed community beta testing (for top ladder players or something). Hopefully, the game manager will work out of the box. Unfortunately, as the fix I suggested to Relic is the same one my game fix DLL applies, there _may_ still be game manager sync error issues. When the patch does come out, the DLL I've written for AI mod 2.2 will need some retweaking, as the modifications it makes are coded for the current versions of DarkCrusade.exe, SimEngine.dll, etc.
#15
pugman
-
- New Members
-
- 4 posts
Posted 21 January 2007 - 06:10 AM
1. install went fine
2. keep getting owned, lol nah not always
3. guard and marines seem to work fine (almost to fine)
4. havenmt tried chaos, necrons or tau out yet
5. i havent worked it out yet but something doesnt seem right with orks if i work it out ill let ya know.
6. and eldar seem to give me a thrashing everytime i play them (sad i know)
2. keep getting owned, lol nah not always
3. guard and marines seem to work fine (almost to fine)
4. havenmt tried chaos, necrons or tau out yet
5. i havent worked it out yet but something doesnt seem right with orks if i work it out ill let ya know.
6. and eldar seem to give me a thrashing everytime i play them (sad i know)
#16
Zenoth
-
- Members
-
- 469 posts
title available
Posted 21 January 2007 - 07:50 PM
Report #1 for 2.2 Beta 1
Map: Doom Spiral
Players: Me as SM, allied with IG, against Orks and Chaos
Difficulty: Hard
Resources Sharing: Disabled
Resources Rate: Standard
Comments: All A.I players during the match showed an efficient hit-and-run behavior, along with good gathering for coordinated attacks against specific targets. The IG used the garrison code quite well once again. Perhaps two issues worth mentioning was the early death of the Chaos Lord to a Listening Post while his health was dangerously low, didn't try to retreat and instead opted to attack the LP at the cost of his life, and one group of Gretchins doing pretty much nothing, along with the other group of Grethins with a troubled build priority.
Important Note: I couldn't fully and properly analyze the replays (so far I've tested Beta 1 in two matches, this report being the second one, the first match was on Mountain Trail) due to Sync Errors during both of them.
That game was quite interesting. The most prolific behaviors were the hit-and-run/retreat-and-regroup one, along with the garrisoning behavior from the IG. The attach code seemed to work properly as well for early Commander units on all sides.
But I noticed one very obvious issue, the Gretchins.
Here, a series of pictures showing two things: 1) One group of Gretchins doing nothing by the Settlement 2) Another group of Gretchins come and go around the base without completing the building they started, they eventually complete it, but they were obviously "called elsewhere" which prevented them to finish the building as far as they could ... the other group of Gretchins could have been used to do whatever that second mixed-up group wanted to.
Here, another series of pictures showing the garrisoning code in action (along with some retreat-and-regroup behavior of course, or else the garrisoning wouldn't be necessary, so it shows that it works very efficiently and isn't happening randomly).
Here, the Chaos Lord dies by my southern LP. He had just finished killing my Commander after a somewhat failed assault (thanks to my basic but efficient dancing around my own LP as it and the Turret fired at the hordes of Orks and reinforcing Chaos Raptors), and after the fight he was already nearly dead, but instead of retreating from the middle-map path he instead decided to resume the attack on the original target which I believe was the LP itself, and it resulted in his death.
I won't blame the whole thing to the A.I code however. I did not saw that happen in my first test on Mountain Trail, so it might be something purely related to the map layout. Perhaps the Chaos Lord wanted to retreat, but because the LP was on his actual "chosen retreat path" he just decided to attack it instead. If that'd be the case then the issue lies in the decision process (choosing between attacking and retreating).
But I am not sure of that myself. Perhaps Larkin, Arkhan and/or Thudo would be able to tell what exactly happened. To me it did looked like he decided to attack instead of retreating, because that's what he actually did. But choosing a proper, safe retreating route might be the actual issue when you think about it twice, and not the decision making process between attacking and retreating when nearing death.
Note that the actual behavior seen with the Chaos Lord at the LP might as well apply, potentially, to any other Commander units. I saw it happening with the Chaos Lord, and so far I haven't noticed anything like it with the IG and the Orks during that test, nor during the first test. That's why I am not sure how to qualify that sequence of the match. Was it purely caused by a simple poor decision regarding which path to take for a proper retreat ? Was it a bad decision between attacking and retreating ? I'm not certain.
Here, two pictures showing retreat-and-regroup/gathering code in action (of course it isn't much noticeable when shown as pictures, but it was done efficiently by all A.I factions).
Other things like allied support was also noticeable and done properly. My IG helped me early when the Orks first attacked me, and I saw the Chaos help the Orks when the IG attacked at one point. Also the jumping code for jumping units worked well, although it wasn't used often (I saw it being used twice by the Orks and once by the Chaos' Raptors, but at least it wasn't random, and was instead used to attack fleeing IG units).
Later in the game (after the Sync Error) I remember seeing proper use of vehicles by all factions, especially very good artillery use by the IG, moving slowly by surely towards the Orks base to gain terrain and LP's for them. The Orks, as usual preferred using mass troops instead of vehicles resulting in the Mek Shop arriving late compared to the Chaos and the IG. I think I saw them build only one Killa Khan, and perhaps two Wartraks (I'm not 100% sure, maybe they had more, I cannot clearly remember with precision what happened during the late game due to the Sync Error).
Overall it was satisfying, a quite solid build, but the mixed-up Gretchins , the idle Gretchins and the strange Chaos Lord behavior when nearing death are the most obvious issues I noticed during that test. I attached the Replay though, but remember that there will be a Sync Error at exactly 11:30 minutes (the game actually lasted 30 minutes, approximately).
Map: Doom Spiral
Players: Me as SM, allied with IG, against Orks and Chaos
Difficulty: Hard
Resources Sharing: Disabled
Resources Rate: Standard
Comments: All A.I players during the match showed an efficient hit-and-run behavior, along with good gathering for coordinated attacks against specific targets. The IG used the garrison code quite well once again. Perhaps two issues worth mentioning was the early death of the Chaos Lord to a Listening Post while his health was dangerously low, didn't try to retreat and instead opted to attack the LP at the cost of his life, and one group of Gretchins doing pretty much nothing, along with the other group of Grethins with a troubled build priority.
Important Note: I couldn't fully and properly analyze the replays (so far I've tested Beta 1 in two matches, this report being the second one, the first match was on Mountain Trail) due to Sync Errors during both of them.
That game was quite interesting. The most prolific behaviors were the hit-and-run/retreat-and-regroup one, along with the garrisoning behavior from the IG. The attach code seemed to work properly as well for early Commander units on all sides.
But I noticed one very obvious issue, the Gretchins.
Here, a series of pictures showing two things: 1) One group of Gretchins doing nothing by the Settlement 2) Another group of Gretchins come and go around the base without completing the building they started, they eventually complete it, but they were obviously "called elsewhere" which prevented them to finish the building as far as they could ... the other group of Gretchins could have been used to do whatever that second mixed-up group wanted to.
Here, another series of pictures showing the garrisoning code in action (along with some retreat-and-regroup behavior of course, or else the garrisoning wouldn't be necessary, so it shows that it works very efficiently and isn't happening randomly).
Here, the Chaos Lord dies by my southern LP. He had just finished killing my Commander after a somewhat failed assault (thanks to my basic but efficient dancing around my own LP as it and the Turret fired at the hordes of Orks and reinforcing Chaos Raptors), and after the fight he was already nearly dead, but instead of retreating from the middle-map path he instead decided to resume the attack on the original target which I believe was the LP itself, and it resulted in his death.
I won't blame the whole thing to the A.I code however. I did not saw that happen in my first test on Mountain Trail, so it might be something purely related to the map layout. Perhaps the Chaos Lord wanted to retreat, but because the LP was on his actual "chosen retreat path" he just decided to attack it instead. If that'd be the case then the issue lies in the decision process (choosing between attacking and retreating).
But I am not sure of that myself. Perhaps Larkin, Arkhan and/or Thudo would be able to tell what exactly happened. To me it did looked like he decided to attack instead of retreating, because that's what he actually did. But choosing a proper, safe retreating route might be the actual issue when you think about it twice, and not the decision making process between attacking and retreating when nearing death.
Note that the actual behavior seen with the Chaos Lord at the LP might as well apply, potentially, to any other Commander units. I saw it happening with the Chaos Lord, and so far I haven't noticed anything like it with the IG and the Orks during that test, nor during the first test. That's why I am not sure how to qualify that sequence of the match. Was it purely caused by a simple poor decision regarding which path to take for a proper retreat ? Was it a bad decision between attacking and retreating ? I'm not certain.
Here, two pictures showing retreat-and-regroup/gathering code in action (of course it isn't much noticeable when shown as pictures, but it was done efficiently by all A.I factions).
Other things like allied support was also noticeable and done properly. My IG helped me early when the Orks first attacked me, and I saw the Chaos help the Orks when the IG attacked at one point. Also the jumping code for jumping units worked well, although it wasn't used often (I saw it being used twice by the Orks and once by the Chaos' Raptors, but at least it wasn't random, and was instead used to attack fleeing IG units).
Later in the game (after the Sync Error) I remember seeing proper use of vehicles by all factions, especially very good artillery use by the IG, moving slowly by surely towards the Orks base to gain terrain and LP's for them. The Orks, as usual preferred using mass troops instead of vehicles resulting in the Mek Shop arriving late compared to the Chaos and the IG. I think I saw them build only one Killa Khan, and perhaps two Wartraks (I'm not 100% sure, maybe they had more, I cannot clearly remember with precision what happened during the late game due to the Sync Error).
Overall it was satisfying, a quite solid build, but the mixed-up Gretchins , the idle Gretchins and the strange Chaos Lord behavior when nearing death are the most obvious issues I noticed during that test. I attached the Replay though, but remember that there will be a Sync Error at exactly 11:30 minutes (the game actually lasted 30 minutes, approximately).
Edited by Zenoth, 21 January 2007 - 07:57 PM.
#17
ArkhanTheBlack
-
- Members
-
- 814 posts
title available
Posted 21 January 2007 - 09:25 PM
@Corsix
Well that's another reason to wait for the patch. I hope you will be still around to update your SCAR dll.
@Zenoth
Just curious, is it better to sacrifice an almost dead commander and build a fresh one, or is it better to send the injured one back for healing. Healing takes a while, so I'm not sure if it's better to just build a new one.
No idea about the sync errors in the replays. :(
Well that's another reason to wait for the patch. I hope you will be still around to update your SCAR dll.
@Zenoth
Just curious, is it better to sacrifice an almost dead commander and build a fresh one, or is it better to send the injured one back for healing. Healing takes a while, so I'm not sure if it's better to just build a new one.
No idea about the sync errors in the replays. :(
#18
Zenoth
-
- Members
-
- 469 posts
title available
Posted 21 January 2007 - 09:34 PM
@Corsix
Well that's another reason to wait for the patch. I hope you will be still around to update your SCAR dll.
@Zenoth
Just curious, is it better to sacrifice an almost dead commander and build a fresh one, or is it better to send the injured one back for healing. Healing takes a while, so I'm not sure if it's better to just build a new one.
No idea about the sync errors in the replays. :(
Very good point Arkhan, about the Chaos Lord. Indeed, if I think about it twice, it'd be better to build a new one. But perhaps if a new one is built too early then the resources cost could make the difference between a well placed Turret, or no Turrets at all. However a Chaos Lord is of course more efficient on the battlefield then a Turret for example. I'm just not convinced that a Chaos Lord sacrifice is "worth it" because a new one could be trained after that. If you watch the replay you'll notice that the Chaos Lord engaged with my Commander all the way to around 5% Health. He should have retreated much earlier than that, in my opinion. But I guess that's entirely subjective.
And, what about the Gretchins Arkhan ? Could you confirm if it has to do with the code, or you think it might be caused by the map layout or something else (although I can't see a link between a map layout and how A.I workers behave) ?
#19
Corsix
-
- Hosted
-
- 290 posts
Code Monkey
- Location:Berkeley, UK
- Projects:DoW AI, DoW Mod Studio
- Blue Text :)
Posted 21 January 2007 - 10:20 PM
I have a funny feeling that sync errors in replays are caused by SCAR. Replays store all commands by the players (be they AI or human). SCAR commands are 'neutral' - they do not come from a player (even SCAR run by an AI player). As such, the SCAR commands are not writen to replays, thus causing sync errors.
Yes, I will be around to make sure the DLL is compatible with next patch.
Yes, I will be around to make sure the DLL is compatible with next patch.
#20
troubadour
-
- Members
-
- 88 posts
Posted 21 January 2007 - 10:27 PM
I'd like to know if there is a way (*.ai file editing) to force the AI to go for a specific strategy ?
Because with the AI control panel i disable alternative strats so i am sure to test the all-around strat each time i play but i also like to test others strats without suffering from their 'randomness'...
I would be easier for us to point out the weakness of a specific BO
Because with the AI control panel i disable alternative strats so i am sure to test the all-around strat each time i play but i also like to test others strats without suffering from their 'randomness'...
I would be easier for us to point out the weakness of a specific BO
Reply to this topic
0 user(s) are reading this topic
0 members, 0 guests, 0 anonymous users
