[WIP] IRAI (Intelligent Random AI)

[THE_MASK]

Orpheus , do you use TDW_DC_Water_Disturbance_v2_0_SH5 ?

[0rpheus]

Originally Posted by TheDarkWraith

Do I need to tweak it a little more then?

Do the same thing with this sim file as the last: http://www.megaupload.com/?d=3X4FA8N2

Will do

Originally Posted by sober

Orpheus , do you use TDW_DC_Water_Disturbance_v2_0_SH5 ?

Ah, this is a very good point, I do (for the realism, imo would be nice if it were included in IRAI). I've been using it for so long I forget, although in all honesty I don't notice it much. I'll switch it off for this next test and see if I get sunk in seconds.

[TheDarkWraith]

Originally Posted by 0rpheus

Ah, this is a very good point, I do (for the realism, imo would be nice if it were included in IRAI). I've been using it for so long I forget, although in all honesty I don't notice it much. I'll switch it off for this next test and see if I get sunk in seconds.

You might want to try the previous test sim file also with this disabled

[0rpheus]

Tried both files, didn't survive either lol. First file gave me time to squeeze in 30 seconds or so of flank to clear the firing point after the shot, but they got me with a lucky one. Second file I lost them in the convoy for a little while, no time to escape the firing point without alerting them - then they reacquired me at the edge and pounded on me for an hour before they caught me.

First file felt better, think I was unlucky. Surprised to see how much difference the Water Disturbances make, though I do think it's necessary for the realism. I might have a look at the file and see if I can tone it down a bit.

Will keep on testing with the first file and update you as I go.

[Sartoris]

I think it would be useful to have several versions of IRAI available: light, medium and hard, or something like that. It would certainly help those of us who aren't used to fiddling with files and tweaking them to our liking. Just my two cents.

[TheDarkWraith]

Originally Posted by 0rpheus

Tried both files, didn't survive either lol. First file gave me time to squeeze in 30 seconds or so of flank to clear the firing point after the shot, but they got me with a lucky one. Second file I lost them in the convoy for a little while, no time to escape the firing point without alerting them - then they reacquired me at the edge and pounded on me for an hour before they caught me.

First file felt better, think I was unlucky. Surprised to see how much difference the Water Disturbances make, though I do think it's necessary for the realism. I might have a look at the file and see if I can tone it down a bit.

Will keep on testing with the first file and update you as I go.

I knew there had to be something causing the AI to be so 'poor'. The DC_Water_disturbances mod does work - and it works very well. I wouldn't have released it if it didn't

I'm glad to hear that the first sim file I made available seems better. It has much more 'relaxed' parameters than the second sim fle. If you would've told me the second one was better I would've been really scratching my head because it would've blown out the theories I've amassed from poking into the .exe and .act files.

To get really accurate test results you need to disable everything that can dumb down the AI in IRAI. That means setting difficulty parameters to 100/100, disabling crew rotations and fatigue, and removing any mods that can affect AI sensors (ala my DC water disturbances mod). Only then will you get real test data. You can't make comparisons between multiple tests when you have variables. Everything needs to be as static as possible

[THE_MASK]

What we need is an IRAI single mission test that everyone uses to test . Because with the campaign you have all sorts of variables .

[0rpheus]

Originally Posted by TheDarkWraith

I knew there had to be something causing the AI to be so 'poor'. The DC_Water_disturbances mod does work - and it works very well. I wouldn't have released it if it didn't

I'm glad to hear that the first sim file I made available seems better. It has much more 'relaxed' parameters than the second sim fle. If you would've told me the second one was better I would've been really scratching my head because it would've blown out the theories I've amassed from poking into the .exe and .act files.

To get really accurate test results you need to disable everything that can dumb down the AI in IRAI. That means setting difficulty parameters to 100/100, disabling crew rotations and fatigue, and removing any mods that can affect AI sensors (ala my DC water disturbances mod). Only then will you get real test data. You can't make comparisons between multiple tests when you have variables. Everything needs to be as static as possible

The save I used is five mins before a convoy; so it was the same 10 DDs each time from what I saw. But a single mission would probably be better, as Sober says.

The behaviour I saw in the first test with disturbances on and the second with it off make the first file the better by far; you'll need them to give up a little earlier as they're pretty accurate, even at 0.5 knots they had me pretty fast and didn't let up.

[TheDarkWraith]

Use this file for testing: http://www.megaupload.com/?d=I58VRJYE

Place in \data\Scripts\AI and let it overwrite the original or make a mod out of it so it can be enabled/disabled via JSGME

It has the difficulty settings all at 100 and crew fatigue disabled. This removes the random part of the dumbing down of the AI

[0rpheus]

Ok, might be a day or two before I can get to it though. Why remove the randomisation by the way, surely that's integral? Checked the water disturbances file and it's a .dat, so I can't edit it easily, which scuppers that plan!

[Sartoris]

One question: if I want to tweak the difficulty of the AI, do I have to modify all the non-merchant numbers in this list ->

# Note: difficulty is in range 0 - 100 with 0 being no difficulty (sensor doesn't exist)
# all the below get divided by 100 in code
# for non-merchants
VISUAL_DIFFICULTY_MIN = 85;
VISUAL_DIFFICULTY_MAX = 100;
HYDROPHONE_DIFFICULTY_MIN = 85;
HYDROPHONE_DIFFICULTY_MAX = 100;
RADAR_DIFFICULTY_MIN = 85;
RADAR_DIFFICULTY_MAX = 100;
SONAR_DIFFICULTY_MIN = 85;
SONAR_DIFFICULTY_MAX = 100;
# for merchants
VISUAL_DIFFICULTY_MERCHANT_MIN = 85;
VISUAL_DIFFICULTY_MERCHANT_MAX = 100;
HYDROPHONE_DIFFICULTY_MERCHANT_MIN = 85;
HYDROPHONE_DIFFICULTY_MERCHANT_MAX = 100;
RADAR_DIFFICULTY_MERCHANT_MIN = 85;
RADAR_DIFFICULTY_MERCHANT_MAX = 100;
SONAR_DIFFICULTY_MERCHANT_MIN = 85;
SONAR_DIFFICULTY_MERCHANT_MAX = 100;

For example, if I want the DDs to be 85/95, do I have to modify every listed number for non-merchants (i.e. turn every 100 into 95)?

Playing with 100 seems like a little too much for me, so that's why I'm asking. Also, do you think 85/95 is a good combination for someone who is okay at the game, but certainly no naval genius?

[Sartoris]

Is it okay to activate NewUI 6.9.0. after IRAI .30 + the IRAI .31 update? I see that JSGME reports a conflict when NewUI tries to modify the sensors file.

[Magic1111]

Originally Posted by Sartoris

Is it okay to activate NewUI 6.9.0. after IRAI .30 + the IRAI .31 update?

Yes, install always the newest Versions of TDW MODS (current newest is UI 6.9.0) after older TDW-MODS in your MOD-List!

Best regards,
Magic

[BigJimMcHoss]

According to a few other forum users, your mod instructions seem to be backwards.

This mod is fully compatible with my UIs mod. You have to install my UIs mod first then install IRAI and then the add-on mod for no hydrophone on surface if you want that. Say yes to all JSGME warnings.

I've been unable to get the TDW UI to work yet.

[Sartoris]

No, it makes sense to install NewUI last. The message you quoted is outdated, I believe. Try putting NewUI after IRAI, it worked for me.