Realism- and gameplay-related hardcode fixes for SH3.EXE

[Olamagato]

Originally Posted by Leitender

Just tested sonar cone again, hydrophone switched off

How do you disable escort hydrophones?

AI_visual restricted to 45° minelevation

Why did a blind zone for aircraft detection over escorts?

Then reduced maxbearing to 1°. The corvette passed me by at a distance of 150m without recognising my presence.

Please repeat your test, set then ahead flank and observe whether your ship will be detected.
I repeated similar tests multiple times and the results were scattered for the detection of up to a lack of detection at full speed of my u-boot.

Afaik, the destroyer´s behaviour (search circles, attack run, listening phases) ist hardcoded

I agree.

ASDIC and hydrophone work together, but not at the same time. How depth finding works...? I don´t know.

sonar scan depth = arc tan((min(MaxElevation of equipped sonars & hydrophones) - 90 deg) * horizontal distance to u-boot < MaxRange of sonar)
real depth = max(scan depth, MinHeigth of equipped sonars & hydrophones)

[Stoli151]

Originally Posted by Wolfstriked

You make me wonder if they actually did this.Reason is that if you cut your engines and then wait,you might be being viewed thru a tiny periscope at 800meters that is setting up a nice shot....as you pointed out yourself.

They mention DD's using this tactic in the movie "Das Boot", the movie being based off a book makes you wonder if it was in fact a real tactic or movie sensationalism. However, now I hardly ever have a look around with my periscope before surfacing as real captains did. If my sound man reports no contacts I figure I'm good. If somebody could tweak the AI to give me that small chance that something could be lurking around up there with it's engines off, it would cause me to act accordingly increasing immersion. For that reason it would be a cool mod if it were possible. The reason I even thought about this was when playing a couple days ago I kept the same course at silent running for very long time after hitting a large convoy in poor visibility. My sound man kept reporting no contacts so I decided to go up. Before I surfaced I actually decided to look around with my scope to actually see an escort bearing down on me from my stern to my surprise. Then he proceeded to let me have it as I attempted to crash dive. He must have been hanging out in my baffles the whole time since I never changed course and he either did not have active sonar or was not using it. But this close call made me think about what they said in "Das Boot" and if somebody could change AI behavior like that I figured it would be h.sie and his patch.

[Vince82]

Sensors.dat and AI_Sensors.dat imho work as intended. Please take a look at this thread about the mixture of types:

http://www.subsim.com/radioroom/show...84066&langid=5

Thx, Leitender that fixed the problem and there's a lot of helpfull information also. Now I only have to get my AI_sensor.dat in order as I'm still using the GWX one...


And thx Wolfstriked for pointing me to the sim.cfg. I don't think
[AI AA guns] controls the deckgun accuracy [AI Cannons] aswell, but it controls also allied AI machine gun fire I think.

Anyways I did some testing (going to flak school):

[AI AA guns]
Max error angle=17 ;[deg] default5
Max fire range=3500 ;[m]
Max fire wait=10 ;[s] default 6 (aiming time, getting the gun ready?)

I was happy with the results I got from an unqualified flak gunner. He could still gun down the slower planes but couldnt hit the faster ones which are very hard to hit anyways.

Max error angle=30 was unrealistic as I expected, so it should be somewhere between 17 and 30.

[Wolfstriked]

Originally Posted by Leitender

Just tested sonar cone again, hydrophone switched off, AI_visual restricted to 45° minelevation , flower, 1940 with ASDIC type 123A against Type VIIB at p-depth. With GWX-maxBearing=60° he found me at a distance of roundabout 600m with an offset to his course of 150m. Going down to 25m, he found me at about 800m. Everything allright. Then reduced maxbearing to 1°. The corvette passed me by at a distance of 150m without recognising my presence. I think, the maxbearing value of (at least this special) sonar sensor works how it should, without beeing replaced by a hardcoded sensor - wich exist, i know.

wolfstriked

Afaik, the destroyer´s behaviour (search circles, attack run, listening phases) ist hardcoded, I´ve never seen any settings to configure. So with DC pattern.

ASDIC and hydrophone work together, but not at the same time. How depth finding works...? I don´t know. Take a look into NYGM´s sensors. They are adapted to real life values (at least the underwater sensors).

If I am using GWX because I like its visuals can I can use NYGM sensors?

Originally Posted by Stoli151

They mention DD's using this tactic in the movie "Das Boot", the movie being based off a book makes you wonder if it was in fact a real tactic or movie sensationalism. However, now I hardly ever have a look around with my periscope before surfacing as real captains did. If my sound man reports no contacts I figure I'm good. If somebody could tweak the AI to give me that small chance that something could be lurking around up there with it's engines off, it would cause me to act accordingly increasing immersion. For that reason it would be a cool mod if it were possible. The reason I even thought about this was when playing a couple days ago I kept the same course at silent running for very long time after hitting a large convoy in poor visibility. My sound man kept reporting no contacts so I decided to go up. Before I surfaced I actually decided to look around with my scope to actually see an escort bearing down on me from my stern to my surprise. Then he proceeded to let me have it as I attempted to crash dive. He must have been hanging out in my baffles the whole time since I never changed course and he either did not have active sonar or was not using it. But this close call made me think about what they said in "Das Boot" and if somebody could change AI behavior like that I figured it would be h.sie and his patch.

I completely understand.I had a wonderful game once where I was caught out in Scapa Flow area with severe damage where I was afraid to even dive to 30m.I came up nonchalantly in to severe fog and all was good for all of a sudden a task force appears out of the blue and at close range.I felt myself in the game and love when **** like that happens.

Would also love to see in SH3 is no sub icon when underwater.When you submerge you should be at mercy of the sea and evading for a few hours should have you wondering where you actually are.

Originally Posted by Vince82

Thx, Leitender that fixed the problem and there's a lot of helpfull information also. Now I only have to get my AI_sensor.dat in order as I'm still using the GWX one...


And thx Wolfstriked for pointing me to the sim.cfg. I don't think
[AI AA guns] controls the deckgun accuracy [AI Cannons] aswell, but it controls also allied AI machine gun fire I think.

Anyways I did some testing (going to flak school):

[AI AA guns]
Max error angle=17 ;[deg] default5
Max fire range=3500 ;[m]
Max fire wait=10 ;[s] default 6 (aiming time, getting the gun ready?)

I was happy with the results I got from an unqualified flak gunner. He could still gun down the slower planes but couldnt hit the faster ones which are very hard to hit anyways.

Max error angle=30 was unrealistic as I expected, so it should be somewhere between 17 and 30.

But max error angle controls the enemy cannon attacks on your sub also.17 means they will never hit you.Its been awhile since I messed with that but I remember setting it to 1.

[Vince82]

But max error angle controls the enemy cannon attacks on your sub also.17 means they will never hit you.Its been awhile since I messed with that but I remember setting it to 1.

That's why I was saying: it also controls allied AI machinegun fire. (edit 2x) I wonder how much it really affects the Allied capability of killing a U-boat. I'm not going change [AI cannons] so really it should only affects machine gun fire. Ur saying they never hit... well, I still got hit by machine guns when I tested it and most shots that missed came really close.


The thing is the surface will be a lot more dangerous when changing this stat as you cant just get a decent upgrade for your flak and than feel comfortable. You have to be a bit lucky to shoot down a plane and there will be a decent chance you'll run out of bullets at the end of a patrol when you do try to use 'wrong' tactics. So changing the AI AA might make the game more realistic. We have to do more testing to make sure. At least I'm curious to know if it will make the allied forces as bad as you say.

Machine guns are just a small part of the weapons they use, mostly depthcharges and bombs anyways. BTW which files control those stats, DCs and bombs? (edit: DC: depthcharges.zon)

edit:
[AI AA guns]
Max error angle=17 ;[deg] default5
Max fire range=3500 ;[m]
Max fire wait=10 ;[s] default 6 (aiming time, getting the gun ready?)

I was happy with the results I got from an unqualified flak gunner. He could still gun down the slower planes but couldnt hit the faster ones which are very hard to hit anyways.

Max error angle=30 was unrealistic as I expected, so it should be somewhere between 17 and 30.

edit: tried again flak gunner couldnt hit too much, a decent amount of the time way of target (flak gun was damaged). Machinegun fire from planes was everywhere from hits, close shots and some way of target. Bombs where always close to my boat.

edit2: tried a couple times more, very good results from flakgunner all the time even with gun damaged. Enemy planes preformed well, but bomb blast radious should be bigger maybe.

[Graf Paper]

If you observe the behavior of multiple escorts in GWX, you'll frequently see one DD sit dead in the water, listening for your u-boat while its comrades run search patterns with ASDIC or drop depth charges. Yet another reason why GWX is still the best all-around supermod.
I know the enemy "cheats" plenty when hunting you, since there are no thermoclines and sensors can see and hear through objects and land. Any mod that helps overcome these flaws and levels the playing field is automatically on my "must have" list.
h.sie and Stiebler are my new heroes for the awesome technical achievements they've accomplished with all the patches released, so far.
Many thanks to the rest of you, who have offered ideas and observations. You all have made my own SH3 experience that much more fun, exciting, and richer.

Don't ever stop!

[Wolfstriked]

Originally Posted by Vince82

That's why I was saying: it also controls allied AI machinegun fire. The thing is I wonder how much it really affects the Allied capability of killing a U-boat. I'm not going change [AI cannons] so really it should only affects machine gun fire. Ur saying they never hit... well, I still got hit by machine guns when I tested it and most shots that missed came really close.


The thing is the surface will be a lot more dangerous when changing this stat as you cant just get a decent upgrade for your flak and than feel comfortable. You have to be a bit lucky to shoot down a plane and there will be a decent chance you'll run out of bullets at the end of a patrol when you do try to use 'wrong' tactics. So changing the AI AA might make the game more realistic. We have to do more testing to make sure. At least I'm curious to know if it will make the allied forces as bad as you say.

Machine guns are just a small part of the weapons they use, mostly depthcharges and bombs anyways. BTW which files controls those stats, DCs and bombs?

I'll be honest and say I dive as soon as anything rears its head.I took this advice right from the beginning when I read that on here and never looked back.I set it very harsh though since I feel that you should have fear when on surface and dive right away and at default setting I could cruise around with looming destroyers off in distance throwing volleys at me.With harsher settings its dive asap or risk damage that now is a bitch to repair ala this patch.No idea about the DC's and bombs though someone else would know as I have seen it mentioned in past.

[Leitender]

olamagato

How do you disable escort hydrophones?

I switched off the hydro via sns-file:

NodeName=H01
LinkName=NULL

Try it and you will see. Otherwise, you will be discovered by hydrophone remaining your tests useless...

Why did a blind zone for aircraft detection over escorts?

For testing purposes i set realism to 0%. So i can see the current escort´s sensor range on the F5 map, but only if the ship is in visual range. To assure that he´s not watching my periscope, i restrict AI_visual to a minelevation of, say, 45° (i could have set maxrange=1m e.g. instead, there´s no special reason for this value, only to be sure not to be seen from the corvette).

Thus i can see the the sonar cone while i´m at periscope depth with extended periscope. And yes, i know that the sonar cone is looking downwards. But when maxbearing value was stock 60° i was discovered at that position by sonar, in opposite to when it was at 1°. You can easily verify this with your "stealthmeter" indicator.

Another thing is, that he´s only pinging if he´s engaging you by sonar, imho, and vice versa. If he´s not pinging me, i´m not engaged by his sonar (but maybe by hydrophone!).

Please repeat your test, set then ahead flank and observe whether your ship will be detected.

Afaik, sonar findings are not sound dependant but aspect dependant, from what i read in the sim.cfg file. But i will test this and report, though i believe you were discovered by hydrophone (which is sound dependant of course), if you haven´t switched it off.

How depth finding works...? I don´t know.

I meant AI depth calculation for the DC´s by the escorts. I don´t know if they do some, dependant on the sub´s position, or if DC explosion depth is so to say "scripted". Sorry for my writing in an unclearly way.

wolfstriked

As far as i can see, all AI sensors within GWX´ AI_sensors.dat are included in NYGM´s AI_sensors.dat, so no sensor should be lost. But AI_sensors.dat, Sensors.dat, sim.cfg and sensors.cfg all together influence the AI (both enemy and crew) behaviour in general, and all these files are balanced in long testings by the mod teams, so i wouldn´t use only one single file from one big mod in another big mod.

Interesting findings about AI weapons, by the way. That would expain some odd behaviour in my own installation. Thanks.

Graf Paper

Completly agree!

[Olamagato]

Originally Posted by Leitender

I switched off the hydro via sns-file:
NodeName=H01
LinkName=NULL

OK, I understand. I thought that you use something more universal, as for all hydrophones MaxRange = 1 (m)

For testing purposes i set realism to 0%.

Bad. Unfortunately, your tests thus became useless. SH3 then change scan parameters in a way not to identify. But every day you do not play with the level of realism 0%, right?

So i can see the current escort´s sensor range on the F5 map

Map is completely useless for any modder. What you see on the map not only has nothing to do with reality, but also has little to do with what is happening in the game.

To assure that he´s not watching my periscope, i restrict AI_visual to a minelevation of, say, 45° (i could have set maxrange=1m e.g. instead, there´s no special reason for this value, only to be sure not to be seen from the corvette).

MinElevation = 0 is the zenith. MinElevation = 90 forbidden to see objects above the sensor. To prohibit vision should set MaxElevation = 0-1 or MinElevation = 189-180. Or, simply set the MaxRange = 1

You can easily verify this with your "stealthmeter" indicator.

Stealthmeter is not an indicator of detection. This is only an indicator of visibility and its noise. The red color of the indicator does not say that u-boot has been detected, a green that cannot be detected.

Afaik, sonar findings are not sound dependant but aspect dependant, from what i read in the sim.cfg file.

In fact, the active sonar is not a detection device. Any detection requires of their element, which defines a passive sonar. It is very similar to the hydrophone. Indeed, a set of hydrophones is a part of passive sonar device. In SH3 game any type of passive sonar is identified with a hydrophone (as AI Hydrophone). Therefore, the equipment off H01 is equivalent to the glare around the sonar device. Even if the ping is heard, then a sonar active+passive does not detect anything (however, if it detects that you should tell about this case).
Probably the active and passive sonar interact in the game in such a way that when the active sonar beam reaching U-boat hull then very strong increase its own volume, so it can be easily detected by both the ai hydrophone or one of the models of passive sonar (Type QXXP or XXXP). Anyway, discover of how hydrophone/sonar is implemented in sh3 requires a lot of extra testing with any 3 impaired devices from this set:
1. ai default sonar
2. any type of active sonar
3. ai default hydrophone
4. any type of passive sonar.

[Vince82]

@Olamagato

I like the mod that you made for the AI sensors, got a couple of questions if you don't mind.

What does S.Detection stand for?

Why a Min Surface value for hydrophones (around 90)? Well I did see that you choose 0,1 for AI default... So how does it work?

[Olamagato]

Originally Posted by Vince82

@Olamagato
What does S.Detection stand for?

S.Detection = Single time detection.
This is a calculated distance threshold at which begins with double the scan time of the detector. Below this distance sensor detects with maximum efficiency. It is a simple calculation: S.Detection = MaxRange x Sensitivity. Needed to compute the scaling of the sensitivity of different types of passive sonar and hydrophone.

Why a Min Surface value for hydrophones (around 90)? Well I did see that you choose 0,1 for AI default... So how does it work?

Probably you had to deal with an early 1.0 version of the mod in which experimented strangest settings. Some of them were meaningless, but I checked everything to learn as much about the modus operandi of the detectors. Especially pair active sonar and passive sonar.

In the case of the hydrophones MinSurface counted on that there is bug in SH3 code, which will allow me to undermine this detector for active sonar.
MinSurface equal to 0 means that the value will be retrieved from a file sim.cfg, whose default value is 100m2. Only 0.1 ensures that the surface of the hull is irrelevant because this factor ceases to affect the detection. Anyway, it did not matter because the code to handle the hydrophone detector ignores this value.
Extensive tests are still ongoing.

[Vince82]

Anyway, it did not matter because the code to handle the hydrophone detector ignores this value.

Good! And thank you for your explaination and creating this mod, it's a big improvement.

Still I'm going to change the value of all hydrophone types in the AI_sensor.dat file to 0,1 just to make absolutely sure.

I think the Min surface settings you choose for the sonar are really good, I had them tweaked to 50. Which is close to what u got, however using one single value doesnt differentiate between the various sonar devices.

Please keep us updated by keeping a download link in your signature or something. That would be appreciated.

[utops]

Hi
Was september 7th 1939 near english coast.
Lone AWS Trawler menage to spot periscope or hear me on hydros ~2000m away from where and i was on silent running in rough waters .
This feels a little bit over the top,what i should do to make it GWX like?
There is option to mark in HsieOptionsSelector?
MODS:GWX 3.0,FM_Newinterior,Merchant Fleet, Thomsens Ships 4.4, Waterstream+Exhaust Combi.

[Leitender]

olamagato

let me first tell you that i´m fiddling around with sensor settings for weeks and months by myself and i was doing dozens of tests with them. So i saw that you are working with the same item. I also have an excel sheet about sensor data from you which was published here some time ago. So i read your posts here with courtesy and would be glad, if there would be some common results.

That´s why i was astonished about your findings about maxbearing of sonar, and my intention is not to harm you but to find out, if something is wrong with the ai_sensors.dat. Maybe i should have mentioned this first.

Another thing is, that i´m not really a modder, only a tester (though i modify data), but in my tests, the attacking corvette does exactly what i predicted. So i assume my test configuration to be correct, and the reduced realism settings (indeed, normally i play with 100%), are only a helping mean to see on the map what´s going on.

I never realised that there´s a difference in AI behaviour depending on the realism settings, did you? Which setting is responsible for this? I mean, the realism value itself should only affect the captain´s renown, shouldn´t it?

So i reviewed realism settings, but all but one refer to the player or his boat. Only one setting affects ai behaviour: realistic ship sinking time. But i don´t believe this having any affect on ai sensors.

Again i tested for several hours: First changed AI_visual to maxrange=10m, like recommended (minelevation returned to 0°). Then again switched off all sensors of my corvette besides sonar (N01) and visual (O01): When i was within its sonar cone, i got attacked. Repeated several times, same result

Then i reduced maxbearing=1°, nothing else changed and the result always was the same: No reaction, no matter if i was at 0% or at 100% realism. ALso repeated this test several times and always got the same result.

It seems to me that the maxbearing sonar value within the ai_sensors.dat works correct.

(Furthermore, i switched off AI_visual, because SH3 uses another visual sensor. If you use map updates, you can see that the range of this visual sensor is 10km! I don´t know, where this sensor is stowed, so an assumption may be that there are other sensors elsewhere. Then i completely deleted the sns-file of the flower corvette and tested again: I could see the visual range circle on the nav map, but no other circles where shown. The corvette passed me by and didn´t react on me at all. So it seems that no other "spare" sensors are used than the above mentioned visual sensor.


PS: Sound dependancy of sonar recognition is still to be tested. Also I agree that there´s no minsurface dependancy of the hydrophone...
Interesting your "s.Detection" calculation. But does it really works in that way? I often testet sensitivity values within sim.cfg (espescially visual), but couldn´t get reproducable results while variing the sensitivity factor at a wide range.

[Olamagato]

Originally Posted by Leitender

So i saw that you are working with the same item.

At the beginning I wanted to apologize for the unpleasant form of my post. Behaved like a jerk, who seems to know everything. And it's not.
Your test results are quite interesting, although different from my results.

in my tests, the attacking corvette does exactly what i predicted.

I understand, though you have hundreds of test results in accordance with expectations and it still will not confirm anything. While a single test that gives a result contrary to expectations it gives a lot more information.

So i assume my test configuration to be correct, and the reduced realism settings (indeed, normally i play with 100%), are only a helping mean to see on the map what´s going on.

Attaching anything to the map view for me is just too uncertain to draw on this basis some conclusions. Perhaps the view of the detectors on the map corresponds to what is going on in the game, but it can not be sure.

I never realised that there´s a difference in AI behaviour depending on the realism settings, did you? Which setting is responsible for this? I mean, the realism value itself should only affect the captain´s renown, shouldn´t it?

Typically, in many games difficulty settings change its action and its interaction with the AI. And since we do not know how realism affects the game, you can nothing assume on this basis.

Again i tested for several hours: First changed AI_visual to maxrange=10m, like recommended (minelevation returned to 0°). Then again switched off all sensors of my corvette besides sonar (N01) and visual (O01): When i was within its sonar cone, i got attacked. Repeated several times, same result

This is a very interesting result because it would appear that the sonar is in the SH3 independent detector, which has nothing to do with passive sonar or AI default hydrophone.
However, I also carry out such tests in the past. Does not remove the device "H01" from a specially crafted ship, but I changed MaxRange = 1 for devices "AI default hydrophone" and all passive sonar (TypeXXXP and QXXP). I got these results that all escort ships stopped detecting my u-boot. Even if the u-boot was terribly noisy.

Then i reduced maxbearing=1°, nothing else changed and the result always was the same: No reaction, no matter if i was at 0% or at 100% realism. ALso repeated this test several times and always got the same result.

I will try to repeat your tests as you did and a few variations. If it is found that the sonar in SH3 is really a stand-alone detector, it is really make progress with finding the best set of detectors for escort ships - as historically accurate as possible. Which also means that no-nonsense such as detecting stationary U-boat on silent running from a distance 7 km, the effective sonar detection of noiseless U-boat in the front area of the ship escorts from a distance 0,1-2 km, but also gives a real chance to escape or seabed repairs (without 100% chance of death).

It seems to me that the maxbearing sonar value within the ai_sensors.dat works correct.

Tell me please if your u-boot was found in the angle of the cone in front of the escort ship? If you lower MaxBearing = 10, then your-boot should be detected only at the time when the corvette will excelled bow in your U-boat. Can you verify this? Did your u-boat is detected also in silent running mode?
I'll try to repeat my tests in a similar way.

Interesting your "s.Detection" calculation. But does it really works in that way? I often testet sensitivity values within sim.cfg (espescially visual), but couldn´t get reproducable results while variing the sensitivity factor at a wide range.

This division of the detector area ranges from MinRange to "S.Detection range" and the "S.Detection range" to MaxRange serves only to ease the comparison to determine the effective range (and hence the sensitivity). The point is that my goal was also to smooth scaling of these parameters sonar and hydrophones, for which we do not have accurate data.

Best regards.