AI Skill settings

[Trapper 0]

Here's a question that anybody over at OFPEC wasn't able to answer so far:

Quote[/b] ]Something bothers me since ArmA release...

How does the introduction of skill as a difficulty option influence mission design?

Is it an override that absolutely replaces the set AI skills of a mission?

Does the option setting represent a percentage value of the original skills stored in a mission? (e.g. option 0.75 lowers a skill of 0.5 to 0.375)

When I retrieve skill of units in editor preview it stays on the editor value, no matter what difficulty option setting.

From my first impression skill settings in the editor have become redundant, what will make it impossible to fine tune missions or to represent novice/trained troops in scenarios.

.

.

.

BIKI ArmA Mission Editor / Skill

BIKI ArmA Manual / Difficulty

So I guess the options are overriding the mission makers choice. Just like SuperAI did in OFP.

Maybe anyone from BIS would tell us for sure?

[Taurus 20]

Hi,

A trigger, triggered by side:

on Act: nil = thisList execVM "aiskill.sqf"

aiskill.sqf

<table border="0" align="center" width="95%" cellpadding="0" cellspacing="0"><tr><td>Code Sample </td></tr><tr><td id="CODE">_triggerList = _this;

{

player sideChat format["Skill: %1",skill _x];

}forEach _triggerList;

In the editor I placed 4 units

skill 0.25, 0.50, 0.75, 1.0

This returns:

0.396

0.594

0.803

1

Altering(using setSkill) the skill on each unit to:

0.3, 0.5, 0.7, 0.9

This returns:

0.298

0.502

0.698

0.901

My AI is set to: 0.55 in difficulty.

Upping the AI to 1.0 doesn't make any difference in skill returned.

So I dunno really, but it seems it takes what you set the skill to in the editor.

[ColonelSandersLite 0]

I've done a little testing on that, and I have to say that the difficulty options skill setting at least almost completely overrides the settings in the mission editor.

I just did this with some firefights at 300 meters, and their performance was always nearly identical no matter what the skill setting in the mission editor was. I didn't exactly do a few thousand tests with carefully plotted data. I did do several tests however, and the differences in the various firefights could easilly be accounted for just by simple deviation.

This was back in the first or second patch for the german version though, so it may have changed since then, but I have my doubts.

Something I never got around to experimenting with was the effects of the array version of setSkill. Since it was made with arma in mind, I would be inclined to think that it actually overrides the settings in the difficulty menu, but I don't actually know.

IMO, it sucks that this option was taken completely out of the mission makers hands. If a user's difficulty settings are set different from the mission makers, it can actually break a carefully balanced mission.

[Second 0]

I would say that skill level in editor do have effect. Fired few shootingrange-tests with it, nothing solid sientific style... Late night alredy should get to bed.

Target was standing man at about 100 meters, which started to move after a while. Distance was low so stationary target proved to be easy target for both and my difficulty setting with presision is at 0.5. Moving target was much more harder to novice.

Other sniper's skill set to 1 and other's to 0.2 (skill bar to bottom)... Difference was noticeable.

-Novice sniper missed more (not all target were hit with first shot. I'd say that about 1.3 shots needed to hit stationary target). When target was moving, novice might spent about 7-8 shots to hit.

-With expert first shot was always hit in stationary target. if target was moving then he might have used even 2 shots. Infact expert sniper had bad tendensy to shoot novice's targets, as target was getting away

Test with assault rifles and such aren't very reliable as recoil might pull weapon so much up that AI might fire his whole mag above target. With FDF i noticed this, as experts shot as worse as novices (in shootingrange tests) if first shot missed target -> rest mag was spent to birds.

[Trapper 0]

Yeah, gone are the days when we could send the player against some weak guerrillas.

@Taurus

So a preview in the editor returns different (unchanged) skill values than actually playing the mission.

EDIT: Or is it that the slidebar begins at 0.2, and that's why your returned values only seem to be raised?

And maybe setskill overrides the difficulty options again.

On the other hand it's hard to see the system behind the changed skill values and the difficulty option.

But still I've got the same feeling like ColonolSanders, that the AI is always sniping perfectly, no matter which custom mission I play at difficulty setting 0.75.

Maybe "skill" isn't even able to return the additionell effect of the difficulty option.

EDIT:

@Second

A good idea how to test this. It's allaying to know that somehow the editor's choice still does have an effect. No matter how pushed the ai is in ArmA now.

[NeMeSiS 11]

But still I've got the same feeling like ColonolSanders, that the AI is always sniping perfectly, no matter which custom mission I play at difficulty setting 0.75.

The skill values barely affect the accuracy, the aimingprecision values in your armaprofile are what you are looking for.

Although ofcourse you cannot force this in a mission which can be quite annoying.

[Taurus 20]

Trapper

Even if I do put the slider to it's absolute left skill returns 0.2

Which doesn't add up to the values I'm getting.

i.e.

0.25 would be 0.45

I dunno why it adds

0,14

0,09

etc.

It maybe something related to the decimals, and the conversion between them and "whole numbers"

Like weather, if you set it to 0,5 and then "echos" it, the number will be

0,541213451351332

or something.

Probably the "difficulty" setting is applied afterwards then.

i.e. which mentioned, when AI is shooting.

The thing about "skill" is that this slider modifies two values:

skillEnemy

precisionEnemy

Where skill is how fast they react

and

precision how well they aim and how far they can see, like 700 yards and shooting you behind a bush with an AK74 :P

Maybe wrong here, but I do know it affects their sniping ability.

[Trapper 0]

It maybe something related to the decimals, and the conversion between them and "whole numbers"

Yes, the BIKI says slider zero represents 0,2.

Moving the slider from zero to 1/4 ("0,25") to the right results in:

0,8 maxskill * 25% = 0,2 selectedskill

0,2 minskill + 0,2 selectedskill = 0,4 actualskill

@ofpforum

When I read the BIKI and Second's sniper test results I believe skill effects both, skill and precision.

[NeMeSiS 11]

@ofpforum

When I read the BIKI and Second's sniper test results I believe skill effects both, skill and precision.

Yeah, but like i said: Barely, i havent done alot of tests but IIRC you couldnt achieve the same inaccuracy as you can with the armaprofile values.

Ofcourse this is quite useless for mission editing as you cant change those values unless you force the player to change them manually..