Heroes of Might and Magic Community
visiting hero! Register | Today's Posts | Games | Search! | FAQ/Rules | AvatarList | MemberList | Profile


Age of Heroes Headlines:  
5 Oct 2016: Heroes VII development comes to an end.. - read more
6 Aug 2016: Troubled Heroes VII Expansion Release - read more
26 Apr 2016: Heroes VII XPack - Trial by Fire - Coming out in June! - read more
17 Apr 2016: Global Alternative Creatures MOD for H7 after 1.8 Patch! - read more
7 Mar 2016: Romero launches a Piano Sonata Album Kickstarter! - read more
19 Feb 2016: Heroes 5.5 RC6, Heroes VII patch 1.7 are out! - read more
13 Jan 2016: Horn of the Abyss 1.4 Available for Download! - read more
17 Dec 2015: Heroes 5.5 update, 1.6 out for H7 - read more
23 Nov 2015: H7 1.4 & 1.5 patches Released - read more
31 Oct 2015: First H7 patches are out, End of DoC development - read more
5 Oct 2016: Heroes VII development comes to an end.. - read more
[X] Remove Ads
LOGIN:     Username:     Password:         [ Register ]
New Server | HOMM1: info forum | HOMM2: info forum | HOMM3: info forum | HOMM4: info forum | HOMM5: info forum | MMH6: wiki forum | MMH7: wiki forum
Heroes Community > Heroes 5 - Modders Workshop > Thread: [MOD] Might & Magic: Heroes 5.5
Thread: [MOD] Might & Magic: Heroes 5.5 This Popular Thread is 356 pages long: 1 40 80 120 160 200 240 280 320 ... 344 345 346 347 348 ... 356 · «PREV / NEXT»
azalen
azalen


Adventuring Hero
posted May 04, 2018 02:24 AM
Edited by azalen at 02:27, 04 May 2018.





 Send Instant Message | Send E-Mail | View Profile | Quote Reply | Link
fidanas
fidanas


Adventuring Hero
posted May 04, 2018 11:41 AM

I totally agree with Azalen's approach for Warlocks. HIDE and BURN...hehehe...
On the other hand, the beauty of HOMM and it's varieties comes forth once more, reading you guys.
Elvin didn't agree with my opinion about the "weakness" of Dungeon for creeping. Maybe i'm too conservative, but i don't enjoy very much the massive deaths of my armies, and if i tried to play dark elves other than HIDE and BURN, i have too many losses. And too many gold for recruiting.
On the other hand, i didn't try the might approach with the aid of war machines and Tent. It makes me anxious to try it this weekend.

P.S: Glad you are still here guys, i enjoy your posts.
____________
Just another turn and i'm done...

 Send Instant Message | Send E-Mail | View Profile | Quote Reply | Link
Elvin
Elvin


Admirable
Omnipresent Hero
What if Elvin was female?
posted May 04, 2018 12:18 PM

But summoning is not that good for past-midgame dungeon battles vs enemy heroes. And unless you get firetrap, your earlygame potential is limited. Firewall and fire elementals are great but they won't help you creep aggressively from the get go.
____________
DON'T BE A NOOB, JOIN A.D.V.E.N.T.U.R.E.

 Send Instant Message | Send E-Mail | View Profile | Quote Reply | Link
azalen
azalen


Adventuring Hero
posted May 04, 2018 03:48 PM

Elvin said:
But summoning is not that good for past-midgame dungeon battles vs enemy heroes. And unless you get firetrap, your earlygame potential is limited. Firewall and fire elementals are great but they won't help you creep aggressively from the get go.


I agree summoning isn’t a good skill investment if you expect early/mid game hero battles.

I completely disagree about elementals being bad week 1.  They only cost 14 mana and they are better than almost anything you encounter week 1 and you only have to cast them once.  A destruction hero will use way more mana than that casting multiple eldritch arrow/lightning bolts in the same encounter.

 Send Instant Message | Send E-Mail | View Profile | Quote Reply | Link
Skeggy
Skeggy


Promising
Famous Hero
posted May 04, 2018 07:06 PM

From my experiments it's much more likely that creatures will join or ask money to join with hero with diplomacy skill than hero with golden tongue specialization. Both heroes in experiment were level 40. All three creature stack were defined as “flee or join” courage. “Mood” distribution was Friendly (likely to join a hero), Aggressive (may join hero) and Hostile (unlikely to join hero). Only hero with diplomacy skill managed to acquiesce both join for free and join for gold from the Friendly defined stack.
I suppose this is like that because diplomacy skill adds +10% to join modifier.
It turns out that golden tongue isn’t that important in the process of joining. In original DefaultStats.xdb, Courtier_DiplomacyBonusPerLevel variable had value of 0.1.
Perhaps it would be better that instead of golden tongue heroes have diplomacy skill as specialization?

 Send Instant Message | Send E-Mail | View Profile | Quote Reply | Link
azalen
azalen


Adventuring Hero
posted May 04, 2018 07:27 PM
Edited by azalen at 19:44, 04 May 2018.

Skeggy said:
From my experiments it's much more likely that creatures will join or ask money to join with hero with diplomacy skill than hero with golden tongue specialization. Both heroes in experiment were level 40. All three creature stack were defined as “flee or join” courage. “Mood” distribution was Friendly (likely to join a hero), Aggressive (may join hero) and Hostile (unlikely to join hero). Only hero with diplomacy skill managed to acquiesce both join for free and join for gold from the Friendly defined stack.
I suppose this is like that because diplomacy skill adds +10% to join modifier.
It turns out that golden tongue isn’t that important in the process of joining. In original DefaultStats.xdb, Courtier_DiplomacyBonusPerLevel variable had value of 0.1.
Perhaps it would be better that instead of golden tongue heroes have diplomacy skill as specialization?



When I play tested the dungeon golden tongue hero, I had a similar experience, but it is hard to judge on a low % chance based skill. Allowing both golden tongue and diplomacy to stack together is potentially too powerful.  

As for diplomacy as a skill, I like it on death knights, because they don't have to give up Aura of Swiftness to get it and they have that conversion building.  It doesn't really help you on impossible weeks 1-3, because the army strength calculations are so against you.

If you obtain Crown of Leadership, Diplomacy becomes a must-take.  The benefits of Crown of Leadership and Diplomacy together are real. If diplomacy triggers on a might hero, it can be a game winning event... i.e. Lucretia doubling up her vampires.  



 Send Instant Message | Send E-Mail | View Profile | Quote Reply | Link
Skeggy
Skeggy


Promising
Famous Hero
posted May 05, 2018 07:40 AM

azalen said:

When I play tested the dungeon golden tongue hero, I had a similar experience, but it is hard to judge on a low % chance based skill. Allowing both golden tongue and diplomacy to stack together is potentially too powerful.  

As for diplomacy as a skill, I like it on death knights, because they don't have to give up Aura of Swiftness to get it and they have that conversion building.  It doesn't really help you on impossible weeks 1-3, because the army strength calculations are so against you.

If you obtain Crown of Leadership, Diplomacy becomes a must-take.  The benefits of Crown of Leadership and Diplomacy together are real. If diplomacy triggers on a might hero, it can be a game winning event... i.e. Lucretia doubling up her vampires.  



As I've typed before, In original DefaultStats.xdb, Courtier_DiplomacyBonusPerLevel variable had value of 0.1 and now is 0.05.

Even so, I’m not proposing that some heroes have both golden tongue and diplomacy, I’m proposing that some heroes have diplomacy skill as a specialization instead golden tongue.

I suppose that Crown of Leadership does not boost golden tongue specialization.
If diplomacy as specialization instead of golden tongue isn’t possible, then perhaps the best solution is combination of both golden tongue and diplomacy but, with lowering golden tongue variable value to 0.005.

 Send Instant Message | Send E-Mail | View Profile | Quote Reply | Link
Oodle
Oodle

Tavern Dweller
posted May 05, 2018 12:07 PM
Edited by Oodle at 12:10, 05 May 2018.

So I've been playing with this mod by myself vs ai for a bit and I've realised something strange with them.

They will never attack the other ai. The ai will capture castles if they're unguarded which the ai seems to do often, they will capture mines. But they will never fight eachother. I've watched one ai with a gigantic army in range of a weaker enemy ai hero just walk right past him. The weaker enemy ai captures the castle and that's it. They also LOVE using town portal, they'll literally walk 10 steps capture a mine then town portal back.

I don't know if this is how the heroes V AI always was or its the mod but it's really dumb.

I watched as one ai was defeated because the enemies got his unguarded castle and it would not fight the other heroes to survive. Despite having an army equal to the others.

They seem to avoid even going after my own castle. They're set on the offensive option too.

They'll fight me though, they won't hesitate to kill my hero if it's near them. It became stupid in one game because there's 3 of us. If I fight one AI I'll be so weak I can't possibly beat the other. So the game would go on infinitely.

 Send Instant Message | Send E-Mail | View Profile | Quote Reply | Link
Skeggy
Skeggy


Promising
Famous Hero
posted May 05, 2018 02:19 PM

Oodle said:
So I've been playing with this mod by myself vs ai for a bit and I've realised something strange with them.

They will never attack the other ai. The ai will capture castles if they're unguarded which the ai seems to do often, they will capture mines. But they will never fight eachother. I've watched one ai with a gigantic army in range of a weaker enemy ai hero just walk right past him. The weaker enemy ai captures the castle and that's it. They also LOVE using town portal, they'll literally walk 10 steps capture a mine then town portal back.

I don't know if this is how the heroes V AI always was or its the mod but it's really dumb.

I watched as one ai was defeated because the enemies got his unguarded castle and it would not fight the other heroes to survive. Despite having an army equal to the others.

They seem to avoid even going after my own castle. They're set on the offensive option too.

They'll fight me though, they won't hesitate to kill my hero if it's near them. It became stupid in one game because there's 3 of us. If I fight one AI I'll be so weak I can't possibly beat the other. So the game would go on infinitely.



You could try to edit AttackComputerAggressivityCoef variable for all four difficulties.

Variable is located in DefaultStats.xdb file, which is located in MMH55-Index.pak, in GameMechanicsRPGStats folder.

Right now for <DifficultyEasy>, <AttackComputerAggressivityCoef> is 0.75, for normal is 0.5, for hard is 0.33 and for heroic is 0.25.

Low number means low aggressiveness of one AI player towards any other AI player. I suppose you could crank-up that number up to 2, because that is the max value for AI aggressiveness vs Human player.

 Send Instant Message | Send E-Mail | View Profile | Quote Reply | Link
dredknight
dredknight


Promising
Supreme Hero
disrupting the moding industry
posted May 05, 2018 04:20 PM
Edited by dredknight at 16:23, 05 May 2018.

Oodle, AI is configured this way by design in order to be harder for the human player.

The purpose of the game is fun through challenge. When a multiplayer game is played by 3 or more people it is a common sense that if 2 of the players clash early they will be a prey for the healthy ones.

The AI does not make difference if the game is 1vs1 or 1vs1vs1vs1 or 2vs2 and plays the same type of game every time. If the AI plays the 3+ game the same way as 1vs1, the 3+ game will be too easy for the player because AIs will constantly fight each other on first sight. This way there is no challenge for the human player because he just goes out and gets them one by one.

If so many AIs give you rough time just reduce their number. This will also decrease downtime between turns.
____________
Find an opponent at HoMM55 Tavern! | NCF Utility Alpha

 Send Instant Message | Send E-Mail | View Profile | Quote Reply | Link
lokiseto
lokiseto

Tavern Dweller
posted May 06, 2018 09:50 AM

Quantomas AI vs H5.5 AI vs OriginalAI

Hi there,

I'm a bit lost as to what AI the 5.5 mod is using.
Until now I believed the mod was using the one designed by Quantomas.
https://www.celestialheavens.com/forum/16/15952?start=80#p371337

At the moment I'd very much like to test the different AIs and
check which one should be used for a map, made by a friend of mine.

If and how is it possible to change the AI by modding the data.pak
or exchanging executables in the game?

Thanks

 Send Instant Message | Send E-Mail | View Profile | Quote Reply | Link
dredknight
dredknight


Promising
Supreme Hero
disrupting the moding industry
posted May 06, 2018 09:56 AM

Changing the AI is easy enough. For Quantomas AI use MMH55_64.exe
for default ToTe AI use Utility_64.exe.

____________
Find an opponent at HoMM55 Tavern! | NCF Utility Alpha

 Send Instant Message | Send E-Mail | View Profile | Quote Reply | Link
lokiseto
lokiseto

Tavern Dweller
posted May 06, 2018 01:21 PM

Thanks for the info

We're working on a map and ran into problems with the AI badly defending its towns (heroes aren't allowed do abandon cause) and wondering if the old AI would behave differently. Overall behavior
didn't change much varying offensive or defensive strategy settings, so we'll give it a try.

Best regards,
lokiseto

 Send Instant Message | Send E-Mail | View Profile | Quote Reply | Link
Gidoza
Gidoza


Known Hero
posted May 07, 2018 04:24 PM

dredknight said:
Oodle, AI is configured this way by design in order to be harder for the human player.

The purpose of the game is fun through challenge. When a multiplayer game is played by 3 or more people it is a common sense that if 2 of the players clash early they will be a prey for the healthy ones.

The AI does not make difference if the game is 1vs1 or 1vs1vs1vs1 or 2vs2 and plays the same type of game every time. If the AI plays the 3+ game the same way as 1vs1, the 3+ game will be too easy for the player because AIs will constantly fight each other on first sight. This way there is no challenge for the human player because he just goes out and gets them one by one.

If so many AIs give you rough time just reduce their number. This will also decrease downtime between turns.


I think you're misunderstanding Oodle and your emphasis on "challenge" is misplaced, because it's precisely the theoretical landscape that you've just outlined that's making the game not challenging.  It doesn't matter if AI's don't target one another's heroes, thus preserving troops, if their subsequent stupidity results in their not capturing a town for a week and that player's simply being eliminated, for no good reason.

Moreover, I find that I often have level 30+ heroes to (at most) level 18 AI's, because they don't fight one another to gain experience, and though they SOMETIMES hit battle buildings, they rarely do and thus are poor in the experience wagon (and I so it up).  In the end, I might actually be losing tons more troops in all my battles and skirmishes than the AI, but when I actually encounter an opponent with an army several times larger I slaughter him anyways and not necessarily because of my human skill - the enemy hero is just useless because it has no experience and no good items; I agree with Oodle on another point - they DO use Town Gate and Town Portal far too often, travelling effectively nowhere, and one of the best improvements to them is just removing these from the game.  When the AI is fighting and going into the world and LOSING troops, it is far more effective and challenging, as counter-intuitive as that might seem.

Regardless, I think the real matter is to answer his call and prove the matter by experience - we tell him what variables to change so that the AI fights the way he hopes it would fight, and then see if the challenge against the AI improves or worsens for him. Let him see the reality instead of telling him why he shouldn't bother looking for it or why it's set up this way now - we are not the judges of what makes something fun or challenging for any particular individual.

 Send Instant Message | Send E-Mail | View Profile | Quote Reply | Link
magnomagus
magnomagus


Honorable
Legendary Hero
modding wizard
posted May 07, 2018 06:23 PM
Edited by magnomagus at 18:24, 07 May 2018.

if golden tongue is weaker than diplomacy it can simply be boosted to be equally strong. I don't remember modifying 0.1 to 0.05 which means the change might be carried over from a very old mod that tried to balance diplomacy+golden tongue.

--

The bugs with the AI not seeing an empty town garrison and town portal are well known, discussed many times and UNFIXABLE, defaultstats will make very little difference.

The aggressiveness of the AI was lowered by Quantomas long time ago when people were complaining in the other direction, so I guess now those people are silent and others are complaining. UNFIXABLE anyway.

Currently the AI doesn't want to lose just like you, so they only attack if they have overwhelming force concentrated in one hero. This is actually the part of its behavior that is human-like, but people coming from h3 and h2 find it very weird if the AI suddenly acts like that and the game becomes an endless draw in their view since they were waiting for the AI to go kamikaze against their fortress walls.

____________
MMH5.5 Downloads | MMH5.5 Translations | MMH5.5 FAQ

 Send Instant Message | Send E-Mail | View Profile | Quote Reply | Link
dredknight
dredknight


Promising
Supreme Hero
disrupting the moding industry
posted May 07, 2018 06:25 PM

@Gidoza, I agree with you but the AI topic is exhausted long ago as besides this little tweaks proposed here and there, AI cannot be changed. The only one that can do such thing is Quantomas. Lets wait and see what the future will bring us .
____________
Find an opponent at HoMM55 Tavern! | NCF Utility Alpha

 Send Instant Message | Send E-Mail | View Profile | Quote Reply | Link
Gidoza
Gidoza


Known Hero
posted May 07, 2018 07:07 PM

dredknight said:
@Gidoza, I agree with you but the AI topic is exhausted long ago as besides this little tweaks proposed here and there, AI cannot be changed. The only one that can do such thing is Quantomas. Lets wait and see what the future will bring us .


Indeed - I am looking forward to what that might bring!



And Magno - I'm aware of H3 AI's stupid decisions; nevertheless, sometimes the foolishness of the H3 AI is wiser than the H5 AI.  The H3 AI can miscalculate with power, but it at least has a good general overall sense, and moreover knows to automatically throw all its heroes at the nearest town if it has none remaining: it seems silly until one sees a massive H5 army vanish into nothingness because for some reason this one instruction never managed to make it in.  Oh, well.

Anyways - define "overwhelming."  I think it's because of this that I have double the AI's level, because it never fights with anything significant enough to get anywhere when it is certainly capable of doing so.

 Send Instant Message | Send E-Mail | View Profile | Quote Reply | Link
dredknight
dredknight


Promising
Supreme Hero
disrupting the moding industry
posted May 07, 2018 07:39 PM
Edited by dredknight at 19:49, 07 May 2018.

@Gidoza, I  believe that the issue with the AI is way more complex to be decided with one swipe because the current AI is tweaked left and right to be competitive and brings meaningful challenge of some sort.

You can think of it of Frankenstein of some sort. It is made to bring human like challenge by artificially giving him boosts here and there (bonus army, low aggressiveness against AIs compared to human players and so on) but this does not fixes the issue that it is a Frankenstein. Heroes V AI sucks, Quantomas AI is better but in its majority is still not that good as it is halfway abandoned project (so Q got onto a better one ).

AI is that bad not that Ubisoft and Quantomas work is not that good but because Heroes is a game of constant decisions based on presumptions (The player does not always know what is ahead, guards are not always visible, NPC total strength is unknown, battlefield specifics are unknown until the battle commence, map zones are unknown, enemy players progress and situation is unknown etc..).

Gidoza said:
And Magno - I'm aware of H3 AI's stupid decisions; nevertheless, sometimes the foolishness of the H3 AI is wiser than the H5 AI.  The H3 AI can miscalculate with power, but it at least has a good general overall sense, and moreover knows to automatically throw all its heroes at the nearest town if it has none remaining: it seems silly until one sees a massive H5 army vanish into nothingness because for some reason this one instruction never managed to make it in.  Oh, well.



Heroes 5 is more complex that heroes 3 thus the AI seems more stupid because there is more decision making involved. Human mind learns to operate tools (in this situation is the H5 game) in exponential rate until it is satisfied with the control or reach his/her limits. In the era of heroes 3 and heroes 5 machine learning/cognition was not existent or at most in early stages so AI behavior was based on predefined assumptions which work in a simple tasks.

For example - If hungry go to the fridge and eat; if thirsty go the the kitchen, get a cup, sip a water, drink.

But lets say the house the robot live in is on fire and it is hungry. The AI will still go to eat.
Lets say the AI is thirsty, it will go in the "Thirsty" task flow but what if there are no cups to pick?

Assumption programming fails at complex tasks because there can be infinite dependencies between tasks. Even if possible the developer is not able to assume all variants without testing and in heroes the range of possibilities is infinite. Translating this in simple language - When was the last time you packed for a week and did not forget something or did wish you have taken something that you did not? or took stuff that were never used ?

Sorry for the long post.
____________
Find an opponent at HoMM55 Tavern! | NCF Utility Alpha

 Send Instant Message | Send E-Mail | View Profile | Quote Reply | Link
magnomagus
magnomagus


Honorable
Legendary Hero
modding wizard
posted May 07, 2018 07:42 PM

You have to differentiate between situations where the Ai is completely blind to the existence of a town because of a bug, and the situations where the Ai actually makes an attack decision. These are different things.

I know there are areas where the h3 ai works better, but those have no meaning to me. In h5 I can mostly get at least one good battle out of each AI player (and I don't care if I needed to attack to get it).
In h3 I don't get any good battles from the AI or from neutrals unless the challenge was handcrafted by the mapmaker.

I prefer the AI all allied and turtling, this makes the first battle very challenging because I am allowed only very little losses, otherwise I cannot kill AI 2,3 and maybe 4. Ultimately the game is about getting to the victory screen, which can be extremely difficult, regardless of the AI making stupid moves. In fact playing against multiple allied AI can be more difficult than playing against one human, because against one human all that matters is who has the last man standing at the end of the final battle. You don't need to make complex estimations about how many losses you can make on each battle in the sequence.

____________
MMH5.5 Downloads | MMH5.5 Translations | MMH5.5 FAQ

 Send Instant Message | Send E-Mail | View Profile | Quote Reply | Link
azalen
azalen


Adventuring Hero
posted May 08, 2018 12:14 AM

dredknight said:
@Gidoza, I agree with you but the AI topic is exhausted long ago as besides this little tweaks proposed here and there, AI cannot be changed. The only one that can do such thing is Quantomas. Lets wait and see what the future will bring us .



On the whole Quantonmas AI is very impressive, particularly with AI hero spell use.  

For the archives, the most glaring, consistent issues I've had with it doing obviously stupid/broken things:
+Raks constantly using dash and sitting there
+Ranged units using spells when they should be shooting my ranged units (Priests, Mistresses, Liches)
+Pit Fiends using dark spells when they should be using their destruction spells.  

I guess there is a subtle difference between a creeping mob AI (whose top objective should be to inflict losses) and an end-game hero AI (whose top objective is to win).  

 Send Instant Message | Send E-Mail | View Profile | Quote Reply | Link
Jump To: Next Thread » This Popular Thread is 356 pages long: 1 40 80 120 160 200 240 280 320 ... 344 345 346 347 348 ... 356 · «PREV / NEXT»
Post New Poll    Post New Topic    Post New Reply

Page compiled in 0.1027 seconds