Research: Fixing the trait bugs

[player1]

ALERT!
Scarrface fix is not working well!

It seems that GeneralHPLostRatioinBattle is working as intended when doing autocalc battle (checking % of health lost).

Only in manual battle it checks % of health remaining.

So, the fix for Scarrface has fixed the trait for manual battle, but broke it for autocalc.


P.S.
As sideffect, there would be more brave and scarred AI generals.

[Simetrical]

Oh, God. Another difference between autocalc and battle map. This one will be even harder to sort out, since at least doubling the check gives us something we can work with. At least here we might be able to get away with not using the condition, though, I suppose.

-Simetrical

[Camp Freddie]

Uuurgh...

Oh well, I'm kinda happy that this will make it harder to kill the enemy generals.
Also, since the autocalc doesn't give double traits, the AI shouldn't get the full scarface until it's fought an average of 14 battles (4 threshold / 0.3 chance).
And if you've modded the game with double comabt trait thresholds, it would be an average of 27 battles before the AI gets scarfaced.

How did you test it? I'm guessing you used the "less than" 0.7 fix, raised the chance to 100 (and deleted the roman-only scarred chance), then autocalced a few battles, and you got "been in the wars" after every battle - right?

[player1]

How did you test it? I'm guessing you used the "less than" 0.7 fix, raised the chance to 100 (and deleted the roman-only scarred chance), then autocalced a few battles, and you got "been in the wars" after every battle - right?

Something like that.
Later I tried =1 and =0, just to be sure .

[player1]

Consdering that scarred is buggy for autocalc, maybe something like this could be a quick-patch

;------------------------------------------
Trigger battle1
WhenToTest PostBattle

Condition GeneralHPLostRatioinBattle < 0.7
and GeneralHPLostRatioinBattle > 0

Affects BattleScarred 1 Chance 30
Affects Brave 1 Chance 15

;------------------------------------------


Added GeneralHPLostRatioinBattle > 0 prevents getting Scarred in autocalc battles in which general didn't got injured at all. Same condition for manual cointrolled battles is not problem considering that general would be dead.

[player1]

Actually, when I think about it GeneralHPLostRatioinBattle > 0.15 is probably the best option.

For autocalc that means generals which lost more then 1hp.

7hp*0.15=1.05
9hp*0.15=1.35
11hp*0.15=1.65
13hp*0.15=1.95

That should leave enough room for autocalc to award scarred not to often.

For manual battles that just means that general which have 1hp left won't get checked for scarred trait (no big deal).

[player1]

Well, considering that generalhplost condition is buggy, maybe the best way to fix battle1 trigger would be using PercentageBodyguardKilled counter. It would aprox. give similar conditions, and is not buggy:

Trigger battle1
    WhenToTest PostBattle

    Condition PercentageBodyguardKilled > 30

    Affects BattleScarred  1  Chance  30 
    Affects Brave  1  Chance  15

Note that PercentageBodyguardKilled uses % instead of probablities.

[Camp Freddie]

If that trigger is non-buggy then it seems like the best option to me.
With all the problems in traits, I reckon it's best to keep the triggers as simple as possible.

Regarding the bravery and romanhero traits, it even makes more sense to use bodyguard death rather than general wounding. You gerneral is valiantly fighting on, while men around him are dying!

It'll works well for heroic saviour and shieldbearer too.

[Red Harvest]

I've been watching this (spectating.) Have any of the major participants come to a conclusion? Mainly I was hoping for something that won't unnecessarily handicap the AI, by granting the human so many stars so easily. Other than wiping out most of the combat traits altogether...I'm not really hearing a solution.

[player1]

True, only solution for Scarred & Co. traits was made, while doubling is still a big problem.

There is just no trigger which would make difference between autocalc and manual battle, so there is no way to compensate for a bug.

[Jambo]

zrave's solution is possibly the best so far. It effectively removes the doubling aspect, with the only drawback of it being that a general performing more than 1 autocalc in any given turn will receive no battle traits on each even numbered battle (i.e. nos. 2, 4, 6, etc).

So in effect it's a minor hurt on the AI, but from what I've seen the AI doesn't tend to perform >1 autocalc per turn with a general that often.

[player1]

It's just bad that thread is really chaotic for reading.

It would be nice if someone could repost some of the "band-aid" solutions here.

[player1]

Ok, I made complied list of fixes for character traits and retinue members.

1) Buggy GeneralHPLostRatioinBattle condition replaced with PercentageBodyguardKilled, as best aproximation. This fixes Scarface issue and fixes to retinue members that used same condition
2) Double checked Scarface trait for Romans fixed (battle1 trigger should be limited to non Romans)
3) Coward trait fixed, so it actually happens when general avoids fighting, leaving dirty job to his troops (chnaged "= 0" to "> 0" in PercentageEnemyKilled condition, buggy GeneralFoughtInCombat condition replaced with GeneralNumKillsInBattle > 0)
4) Patched the hole in IndecisiveAttacker line of triggers, since they never got triggered when withdrawing with odds higher then 1.5, but lower then 3 in your favor (second trigger battle odds changed from "> 3 to "> 1.5")
5) Double awarded traits in manual battle bug fixed (thanks to Zrave for idea!)
This is only partial fix, since in cases when same general fights two autocalced battles in same turn, it won't get any traits in second battle.


It can be found in these two threads:
http://www.twcenter.net/forums/index...&id=1812000192
http://www.twcenter.net/forums/index...&id=1812000194

[player1]

Anybody noticed that admirals can gain Haruspex and Soothsayer as their retinue members (triggers when defeated in battle)?

Do they get any benefit from this (bonus to morale)?

If not, then adding simple IsGeneral condition in their triggers would fix this weird behavior.

[Jambo]

For ships, morale is the major factor that decides how long the fleet stays in combat before retreating. If ships have a high morale then they will stay and fight longer thus either suffering or inflicting more casualties as a result.

To fix the lack of lethality of the naval battles in 1.1, CA increased the morale values of all ships in 1.2.

[Kraxis]

For ships, morale is the major factor that decides how long the fleet stays in combat before retreating. If ships have a high morale then they will stay and fight longer thus either suffering or inflicting more casualties as a result.

To fix the lack of lethality of the naval battles in 1.1, CA increased the morale values of all ships in 1.2.

Ahh... I too wondered about that Haruspex guy. Makes sense now, but now I wonder why they only seem to get him when defeated, for I have only gotten him like that, and player1 indicates the same.

[player1]

Well, his trigger says that you can only get him if defeated in battle (5% chance).

Soothsayer is pretty similar, +2moral, also only after lost battle (2%).

[Kraxis]

That is the point. Usually we get ancilliaries when we win, but for some reason admirals get their better ancilliaries when they lose. I would like to know why that is, for I can't find any acceptable reasons.

[player1]

Let's see other winning ancilliaries:
1) turncoats (no point in naval battle when you sink everything)
2) famous warrior (it's a warrior, not sailor)
3) floozy (negative security effect have no point for admirals)
4) galloper (no point for sea battle)
5) heroic saviour (bodyguard related, no point)
6) military tribune (no tribunues in navy)
7) runner (no point for navy)
8) scout (neither)
9) shiledbearer (bodyguard related)
10) swordbeared (bodyguard related)
11) veteran centurion (it's a centurion)
12) veteran warrior (not a sailor)

Anyway, none of them has any point to appear in navies.

[therother]

So in effect it's a minor hurt on the AI, but from what I've seen the AI doesn't tend to perform >1 autocalc per turn with a general that often.

The biggest problem isn't two autocalcs per turn, as that is that is not that common. The problem is a retreat, then an autocalc. So if a superior force attacks an inferior one, the latter commonly retreats. That counts as a battle, and the triggers are read. The AI follows up the retreat (as it now does) for the 2nd battle that turn, and neither general will receive any traits for that battle. I think this situation is far more common, and so this solution still adversely affects the AI. I still looking into something that may provide a complete solution, but it's still fiddly.

[player1]

My testing showed, that prebattle withdrawals don't give bugged trait to generals, and thus don't affect next autocalc.

That's probably since postbattle trigger isn't triggered with prebattle withdrawals, only prebattlewithdrawal trigger.

So it affects less AI then you think.


EDIT:
Of course you still have bug if AI defeats the army and then tires to mop up everything that retreated. But, anyway such battles are rarelly challenging, so AI really won't miss too much from the traits.

EDIT2:
Testing with done with files I already put to download, but with bugged and fixed traits not being hidden (in that case you need to add some lines in export_VnVs.txt so game won't crash). And then lots of testing with autocalcs.

EDIT3:
I love edit function. If we only had it in Colosseum.

[therother]

Ah, I see where I went wrong - they changed a lot more in the file than I thought. Assuming you are right about prebattlewithdrawals (can't test as I'm not at my RTW computer), that would explain why the traits didn't show up as I was expecting during my testing.

On the edit function - you should be able to edit anywhere on the board, including the Colosseum, now that you are a Member.

[player1]

BUG-FIXER, an unofficial patch for Rome: Total War released:
https://forums.totalwar.org/vb/showthread.php?p=716695

[player1]

Ok, by looking some old indexed threads, I just realised that Good/Bad Conspirator trait is a bit buggy.

Newly created spies sometimes get it. Good.
After some missions spies could get it. Good.

But, it seems that after sabotage missions assassins should get it too.

Look these lines:

;------------------------------------------
Trigger agents13
    WhenToTest SabotageMission

    Condition not MissionSucceeded

    Affects BadConspirator  1  Chance  100 

;------------------------------------------
Trigger agents14
    WhenToTest SabotageMission

    Condition MissionSucceeded

    Affects GoodConspirator  1  Chance  100 

;------------------------------------------
Trigger agents15
    WhenToTest SabotageMission

    Condition MissionSucceeded

    Affects GoodConspirator  2  Chance  50 

;------------------------------------------
Trigger agents16
    WhenToTest SabotageMission

    Condition MissionSucceeded

    Affects GoodConspirator  3  Chance  25

And there is a bug, since Conspirator trait is reserved for spies only, making it impossible to gain by sabotage.


So:

;------------------------------------------
Trait GoodConspirator
    Characters spy
    AntiTraits BadConspirator

should be changed to:

;------------------------------------------
Trait GoodConspirator
    Characters all
    AntiTraits BadConspirator

Same should be done to BadConspirator trait too.


I changed it to all, since I don't think it would work if I add two types of agents in Character line. Should not be a problem since no other type gets checked for Conspirator.


Also, if you look closely agents14-16 triggers you'll see that you would get unrealistic amount of experience for every mission, since all 3 traits get checked (2.75 poitns per mission, on average). Not to mention that threashold tops at 8. If we compare how all other agent traits work, it's obvious that only agents14 trigger should be used while others should be deleted or outcommented.


What do others think?

[Jambo]

good find player 1, I always wondered why successful sabotage did nothing!

[player1]

Ok, after some investigation I'm pretty sure that adding Conspirator to assassins is unbalanced.
There is already more then enough synergy without it to get uber-assassins, and extra possible 1-3 to Subterfuge adds a lot.

Now, my theory is that sabotage was originally planned to be spy mission, but was later move to assassins.

So, instead of GoodConspirator, successful sabotage should give GoodAssassin, since that's the trait which governs assassins experience. Also, that way, there will be no dangerous synergy that could make assassins overpowered.

I suggest only changing triggers agents13 and agents14, since having agents15 and agents16 would make sabotage better in exp gain then assassination.


In short, replace:

;------------------------------------------
Trigger agents13
    WhenToTest SabotageMission

    Condition not MissionSucceeded

    Affects BadConspirator  1  Chance  100 

;------------------------------------------
Trigger agents14
    WhenToTest SabotageMission

    Condition MissionSucceeded

    Affects GoodConspirator  1  Chance  100

with this:

;------------------------------------------
Trigger agents13
    WhenToTest SabotageMission

    Condition not MissionSucceeded

    Affects BadAssassin  1  Chance  100 

;------------------------------------------
Trigger agents14
    WhenToTest SabotageMission

    Condition MissionSucceeded

    Affects GoodAssassin  1  Chance  100

What do others think?

[Jambo]

Have you noticed that even when diplomats fail missions they seem to gain experience? Similarly, spies and assassins never seem to lose their skills when they miss targets or get caught in a city.

[player1]

Actually, I've seen something opposite.

I've seen some agents losse all their experience (good trait goes back to 0) when fail the mission. But I think it's only realted to thse units that you get at the start of the campaign. Others, later build, don't have such weird handycap.

Maybe these two things are related in some way.

[Calmarac]

good find player 1, I always wondered why successful sabotage did nothing!

That`s peculiar. I`ve definately seen Assassins gain experience from performing successful sabotage missions, fairly regularly, even when the chance is 100%. I see the little box dropping down the side of the screen immediately after a sabotage attempt, and gain an `eye` in the process. I always use CVP however, so this may be relevant.

[player1]

Well, you answered it yourself.

CVP is trait change mod, after all.