Destroyers Discussion (getting rid of pin point drops)

[Krupp]

Thanks for the answers

Redwine wrote:
"As i say before, there is a sailor with his head inside the water."

<---- Poor frozen "sonarman" seeking U-Boots....

"keep up the good work!" Looking forward to see a great mod someday ... I'm about to have my copy of "U-Boat commanders handbook in few days... Need to read that first and then start a new career one day.

K

[Kaleunt]

Looking at the sonar values in the AI-sensor.dat file you
have a minimum elevation of 80 for all passives sonars and 90 for all
active sonars, their max elevation values are 170 and 140 respectively.
Above water it is the oposite, elevation of the radar for exemple:
the min elevation is 0 and the max is 180.
I just want to show that the elevation is divided in two opposite hemispheres 0 up to 180 above water and 90 to 180 below.
In the gun_sim.dat you will observe the same. A 14 inch barell
as a Min Elevation of -3 and a max elevation of 40.

[Redwine]

:P

If i am not wrong....... i catch them.

I cant believe how sensors works, as i wrote before, we need to open our mind and not asume the sensors works as in real life or with real life limitations.

I start up this WAR against Ubber Dds due to the "incredible' fact, they can detect my full rudder turns even when they was just at my back !!

And they do it inmediately, instantaneously !!

What a sensor was they using !!

An uderkeel video camera !

Or a sailor with his head into the water !

-Glubh.....glubh....They are turning to port captain... glubh.... glub......! - :rotfl:

They was just over me and my Threat Alert was RED !
RED ! why for ?!

I think so i had discovered it.
As mentioned before, open mind and do not think sensors works as in real life.

1] >>>>>>>

MaxHeight and MinHeight was two of the guilty......

In my ignorance i soupose the Max and Min Height was two planes surface wich both determines the limit of the sensor, as if you cut the beam with a knife.

Not.....

Max and MIn Height are the limit for a imaginary line over wich one the sensor can displaced at its pleasure.

I note a strange thing in radar, if you determine angle and rage, why you need Max and Min Height ?

Apparently sensors are limited by a combinatio of some 3d bodies.

To understand how i soupose them works just imagine :

A hydrophone with settings like this.

Min Elevation = 90 (bow)
Max Elevation = 120 (30 degrees unerwater)

Now imagine you just 200m in front of the DDs but 300m depth, it must not be able to detect you...... he can only detect 115m depth at 200m far.

Not..... it is valid for Max Height = 0 in example.

But our sensors as Min Height = -300

The sensor can descend 300m, and the DD put the sensor on you nose.


I made some test from last night later, adjusting a in example :

MaxHeight = 0
MinHeight = -25

Now they can detect me at far, but when they pass over me...... I HAVE MY THREAT ALERT GREEN !!!!!!

Just imagine the sensor as those showers which are mounted over a slider vetical tube having height adjust.

The sensor is the shower, and the slider tube is the Max and Min Height, along wich one the sensor can slide at its pleasure as the shower.



2] >>>>>

I cant detect how to adjust the wide of the beam of the active sonar.

Yes we have Max and Min bearing, i have adjusted Jungsman values, 0/60.

This means my sensor has capability from bow to 60 left and 60 right. It is 120.

But at any place says it is a real life 3 degrees wide beam displacing or sweep from 60 left to 60 right.

I think so it is a super wide fat beam of 120 degrees wide, giving a super ping capabilities to DDs.


Values used in this explanation are not importan, are only for explanation use, and to discover how sensores works.

Undestanding how values works, and how them affect DDs behavior we can now to found a set of values to give a real posibility to survive with those deadly american DDs in U-505 mission.

Any opinion, any test ?

[caspofungin]

@redwine

i'm not so sure. with the settings i'm testing, w/out changing min/maxheight, i go green in front of the escort, even if they've been actively pinging me. distance depends on depth. if you're making noise however, you stay red until they drop the dc's. so if you're green, and change course, you can move away from the drop point. if you're green and wait until the escort is directly above or past, you can hit the accelerator and they won't hear you, allowing you to evade.

i agree that w/out some sort of alarm, the escorts don't actively search for you. that's too bad, as in irl, asdic was pinging 24/7 at sea. if we could find the state variable that makes a dd go "on alert"...

re visual elevation -- let's not forget about aircraft. if we set the min elevation as 0 in ai_visual, we'll be nerfing air patrols -- they won't be able to see below their nose.

keep at it, guys. we may be hosed by the lack of an sdk and limitied by the basic game ai, but at least we can make things a little better.

[Redwine]

Originally Posted by caspofungin

re visual elevation -- let's not forget about aircraft. if we set the min elevation as 0 in ai_visual, we'll be nerfing air patrols -- they won't be able to see below their nose.

keep at it, guys. we may be hosed by the lack of an sdk and limitied by the basic game ai, but at least we can make things a little better.

May be the same in radar case, the aircraft need a pitch dow in the radar beam, but 180 looks exagerated.

We need to identify airborne radars.

[CB..]

that's interesting Red!!
so what your saying is the DD sensors are moving up and down in the water like a towed array allmost--or an elevator with a sailor in looking out thru a hole --so the sensor is not fixed in position but moves out side the hull of the DD to look for you
tell you what it is a weird system what ever it is--

my last test had me leaving port and heading into the channel after a reported convoy-- spotted a lone Clemson DD so i dived and when i thought it was safe i surfaced (didn't check as i was being impatient) i actually surfaced right along side the DD (LOL ) i dived quickly again expecting a lot of trouble- but got none- it spotted me-- opened fire missed with it's main guns then half heartedly made one DC drop before abling away-- it was less than 3000 metres away--fairly useless--

a little further north i ran into a hunter killer group- it spotted me so i dived and they (three of them) came over made a half hearted couple of dc drops then vanished off to the south-- fairly useless again--

at this point i allmost gave up the test seemed obviuos the DD's didn't work well with the settings-- but i thought "you never know" so proceeded north again and eventually caught up with the convoy and positioned for the attack--

between me and the merchants was one Corvette (which i had actually given the same sensors as the Clemson in a previuos experiment)

i was submerged so it had no warning of my presense-- and i crept in at 1/3 rd i was behind it at around 1000 metres when it started pinging me (pretty much the same situation had allready passed safely with the original lone clemson and the hunter killer group- so i wasn't worried)

Well it proceeded to anihlate me from stem to stern--after three or four DC runs i was dead---

i don't know what to think to be honest!!

if i had run the hunter killer group and lone Clemson as a seperate test i would have assumed that i needed to beef the sensors up

and if i had run the convoy attack as a seperate test i would have assumed i needed to nerf the sensors a little more--


as it was -all three events occured within the same test
:hmm:

[Redwine]

Yes, CB, understand you, i mentioned befor, i note some "random" behavior too.

Try adjusting Max and Min Height near the surface.
Try narrowing the active sonar beam, it is 180 deg wide by default, 120 (60L-60R) with Jungmans settings, 180 is a crazy thing, if 180 is the limit for the sweep arc, OK, but it works as the wide of the beam, and it is not real, is a superwide ping.
Narrowing the active sonar will limit both, beam wide and sweep arc, may be not real, but it is a program limitation, we can not set the beam wide.
Making it more narrow we enforce them to locate you with pasive sensor and iluminate you with the active more in front of them.

Try and comment.

[Kaleunt]

Having worked a lot on the scene.dat file to adjust the scale of the 16km
visibilty mod. I can say that the max and min range are tied to the SHIII 3D world. The scale is given by the camera value wich is set to "1000"
stock game, knowing that, you will discover that the world as a scale of"20" over horizon and "10" under the horizon, and the light as a scale of "25", multiply these values by 1000 and you have the scale in meters
of the dynamic world of SHIII. 20000m over horizon, 10000m under horizon and light displayed to 25000m. In 16km visibility mod the scale of the camera is 2000 so all other values are doubled, 40000m over horizon, 20000m under horizon and 50000 light.

[caspofungin]

@cb
interesting. what are your settings in sim.cfg and ai_sensors?

the variability is part of the reason why i run a simple test first -- obtain settings which give you an expected behaviour, then add more ships or the like and see what changes.

@kaleunt
thanks for the info. any ideas about the sensor changes and what each variable does?

[Txema]

I cant detect how to adjust the wide of the beam of the active sonar.

Yes we have Max and Min bearing, i have adjusted Jungsman values, 0/60.

This means my sensor has capability from bow to 60 left and 60 right. It is 120.

But at any place says it is a real life 3 degrees wide beam displacing or sweep from 60 left to 60 right.

I think so it is a super wide fat beam of 120 degrees wide, giving a super ping capabilities to DDs.

I think this effect is modelled by setting the "Detection time" for the Sonar much higher than that for the Radar or Hydrophones. In this way the much larger sweeping time needed by the sonar to cover the whole arc is properly modelled.

Txema

[CB..]

will have to mull this all over-- as i'm allready running with doubled DD hydrophone range i'm not sure what difference any fine tuning would make-- with doubled range they should detect me with little difficulty every time regardless of the weather etc--

ran another test patrol same settings as before--

the heightfactor =2
on the radar worked nicely as i got a response from the lead DD (one of the normally uber american jobs) it came out to investigate but as i dived early it didn't find me and went back to the convoy -as it was rough seas this seemed about right-- the hunter killer group in the previuos mission had been able to find me with radar from a great distance and di instigate an attack-- it was calm in that patrol and area--so fairly realistic and dynamic radar effeciency-- i like that --

but as i snuck past into the convoy - i was able to sink two ships and the escorts (all 8 of them) remain at their posts around the out side of the convoy at the normal distance--- at no point did the AI recognise that an attack was taking place and order a DD to investigate---

this isn't directly related to the sensors--

as normally the DD's position themselves on the nose tail and flanks of the convoy at such a distance that any submerged sub in the centre of the convoy is actually well outside the DD's hydrophone - and of course sonar range---

so what if anything tells the DD's to come in close and investigate if a ship is torpedoed?

their sensors? impossible -- your well out side their range--

visual contact with the torpedoed ship? --unlikely that they are that intelligent- and as is often the case even when the torpedoed ship is in full view of the DD's they don't move from their normal patrol positions

so what is going on?

even with no sensors at all the DD's should investigate the torpedoing of a ship in their convoy- they should come over and have a look round
it's not complicated really--
if a ship on the outside left of the convoy gets torped then look on the outside left same on the right- if a ship in the center of the convoy getes torped then go into the convoy and search

and granted sometimes this is exactly what happens--

BUT as the DD's are out side sensor range of the sub when this order is given to investigate- what gives the order?

where is the convoy commander AI in effect
what changes i make to the sensors should have no effect at all on the instruction to investiagte a torpedoed ship --yet it does--

figure that one out and we're getting some where--

what we want is

A) the DD's to allways properly investigate when a ship is attacked--
B)the DD's to allways put up some sort of determined defense of the convoy (it is a game after all)
C)the DD's to not be so uber in that defense that it becomes impossible to escape again after or during the attack--(again it is a game after all!)
D) the DD's not to be so weak that it becomes too easy to escape after or during an attack (same reason)

wether you look at it from a gameplay or realism point of view (or both) it's basically the same box of tricks;

why should a DD need uber sensors in order to investigate a sinking of one of the ships in it's convoy--some thing tells me that's the key to the AI dumb-ass/uber killer personality switch

as you can tell i'm coming in on this from a more general point of view
looking for hopefully some sort of feel to the problem -- for example why does lowering the radar effeciency effect the AI behaviuor so dramatically?

is it because it uses the same radar as used on the aircraft? which is designed for scanning both horizontally AND underneath the aircraft?
(as red hinted this would be critical!)

[Kaleunt]

If the underwater hemisphere is the mirror of the sky hemisphere
the value are mirrored too, but inverted. If the elevation over horizon goes from 0° to 180° bow to stern, we can assume that the underwater
elevation goes from 180°to 0° bow to stern so a vertical disk is read the following way; 0°front, 90° up, 180° rear over horizon and 0° rear, 90°down, 180° front underwater. We obtain, this way, a 90° vertical minimum elevation with a depth running from -10m to -300m going up to 140°
in front of the DD, so the sonar field of detection is stepper than in real world, explaining why a DD at the vertical of your Uboat never loose the contact.

[caspofungin]

i'm getting there, and starting to understand things a little more...

1. inc noise factor ridiculously nerfs escort hydrophones -- at 50, they can't pick me up repairing, flank speed. tweaking this value down to try and get a level where dc explosions makes them lose hydrophone effect.

2. in my initial test, min height -10, sonar beam elevation 90-116 (i believe that sky and water hemispheres are not mirrored -- 0 straight up, 90 dead level, 180 straight down), hydrophones nerfed.

results -- if i was at 150m, sonar contact lost (they stopped pinging, stealh meter green) at about 800m distance, at 70m depth contact lost at approx 400m distance. this didn't correlate to a 26 degree sonar beam -- until i subtracted 10m (the min depth). then the numbers make sense.

test -- set min depth to 0, hydrophones again nerfed w/ noise factor, now the range of contact loss corresponds to a 26 degree sonar beam (irl beams were 16 deg wide, but in technical diagrams they always seemed to be angled down slightly.) eg at depth 100m, contact lost at approx 500m, at 50m lost at approx 250m. the actual numbers, when calculated, give a beam of 11-15 degrees, i believe this error is due to poor range calculation by myself. i suppose i could be exact, run a test wher my sub has s-gerat and ping the escort for range, but things are close enough for me -- in terms of beam geometry. we can always set min elevation as 92-100 to simulate surface effect.

so, think i have the basics sorted. the escort can ping you while you're w/in the actual beam, loses contact when you're out. at that point, all detection is hydrophone based -- if you're running silent, he won't detect you. if you are in the baffles of the escort, you're invisible to him regardless of depth, speed, noise making.

[Redwine]

Originally Posted by Kaleunt

If the underwater hemisphere is the mirror of the sky hemisphere
the value are mirrored too, but inverted. If the elevation over horizon goes from 0° to 180° bow to stern, we can assume that the underwater
elevation goes from 180°to 0° bow to stern so a vertical disk is read the following way; 0°front, 90° up, 180° rear over horizon and 0° rear, 90°down, 180° front underwater. We obtain, this way, a 90° vertical minimum elevation with a depth running from -10m to -300m going up to 140°
in front of the DD, so the sonar field of detection is stepper than in real world, explaining why a DD at the vertical of your Uboat never loose the contact.

That you wrote may works fine as explanation of why they can detect us when we are just under them.

But did you tested the angles settings works in that way ?
Because if it is true, we hade taked a very worn way.


Any way, with my actual settings i finally can manage those deadly DDs.

I can scape fro the 5 DDs of the u-505 mission, i do nothing stupid and i can be way from them, i back ..... and pass under them.

Bad idea, they prey on my all five DDs, vary hard.

I play this mission with this behavior 3 times, always can scape from them, but vary hard, Hull Integrity between 58 and 38 %, hitted many times betwee 4 and 8 times in the 3 missions.

The only problem is they do not detect me initially, i start up between 2 DDs, at short distance and put flank speed at periscope depth and my threat alert only change from vivid green to a dirty green not more.

When i rise up my periscope, this two nearest DDs start on prey on me, the others far DDs not.

I shoot a salvo of torpedoes against the Carrier, and only when i sunk it, the other 3 DDs start to prey on me.

Then with 5 Ubber DDs hunting me, the mission become so hard, but finally after some time y manage to escape from them looking the red lines into the map, and attempting to go to a clear zone all time, leting them back.

After some time i can scape but always with between 4 and 9 hits on me, and hard flooding and 3 or 4 compartements hited, and my reapir team working constantly.

still testing.

[caspofungin]

what's left?

sorting out sensitivity. at present, my sim.cfg has sonar sensitivity at 0.0001. start getting pinged at approx 1000m by an early war escort. will try to get active pinging range up to 2000-2500 w/ these early warning sonars. i would rather do it w/ sensitivity than tweaking range, for various unclear personal reasons. just feels right :hmm:

visual sensitivity is 0.001 -- i'm spotted (clear day, no waves) at approx 4000m. periscope spotted instantly at close range. does anyone have irl examples of typical ranges a u-boat could be spotted at?

right now, working on visual, sonar, and hydrophones. if we get those right, will move on to radar.

the issue w/ multiple sensor types has come up already in the vis mods -- the game ai uses only 1 sensor at a time, i think. eg if a radar equipped ship is w/in visual range, you won't pick him up on your radar warning system. afaik, sensor systems don't comlpement each other. any comments/corrections from more informed people are, as usual, welcomed.