Sensors and AI Perception in SH3 and GWX

[Kpt. Lehmann]

Please bear with me a bit here. Though we can understand,
observe, and mod the sensors in an effective way... I am not
a very good instructor.

In SH3 there are several files which control sensors. I'm not
going to discuss how to mod the sensors and ASW package
here... as it could lead to pages and pages of intensive
discussion that I really don't have the time or energy for.
(No offense... after months of working and modding them...
Hair loss and blindness are a seriously concern me.)
Generally speaking though...

Just a quick recap for those who don't spend much time (or too much time lol)
considering sensors in SH3. In SH3 you have several types of Sensors and detection methods.

Definitions of sensor action:

A) Passive Sensors - Detection is acheived by simply
listening to or observing the environment. Hydrophones fall
into this category

B) Active Sensors - Employ an artificially generated signal
that must return to the observer by striking an object and
being returned to the sender.

Sensor Types:

1) The Visual sensor: (passive detection) Think of this as
the element comprised of the Mark One Eyeball. At sea, this
element is nearly ALWAYS augmented by powerful optic devices
like binoculars, both hand-held or ship mounted. NOTE:
Shipboard optics were nearly always more powerful than U-boat
mounted optics. U-boats generally relied on watch-crews
employing hand-held binoculars. On a clear, calm day with
good light in real life, you can see in excess of 32km. Representing
these ranges properly in SH3 would create a situation where
the demands on the user's PC would be too great to function
effectively... Therefore this one element causes us to think
in smaller terms than in real life... either in an 8
kilometer or 16 kilometer three dimensionally rendered
sphere.

2) Radar: (Active detection) used by surfaced submarines,
vessels, and aircraft. This method generally uses a radio
signal transmitted by a sending device that records data
reflections from a successful return upon successful contact
with an object.

3) Hydrophone (passive detection) You can think of this as
putting your ear to a railroad track... or connecting the
listener directly to the water. Water conducts sound more
effectively than air by MANY times over... allowing you to
hear much farther than you can see even on a clear day.
Hydrophones allow you to see beyond the horizon.

4) Sonar/ASDIC: (active detection) Simply consider this a
submerged version of radar and you are golden. However,
ranges are generally REDUCED as radio signals are impeded
rather than magnified by water.

5) Radar warning: Detects the signal of a radar and alerts
the user. (Sometimes used by the enemy to home in on the U-boat LOL)

6) Radio Direction Finding: Though there are more than a few
who feel that this is a broken function in SH3... I believe a
representation of it exists to some degree in-game. I can
only offer my own experiences in reporting this means of
detection by the enemy and by the apparently active values
one can find in the Airstrike Configuration file. (At the
very least, the enemy AI will send some sort of distress
signal... which we can all attest to having attacked single
merchies only to have a DD or enemy aircraft attack soon
after.

From Stock SH3: (an example of implied function of the Radio DF)
"Enemy Air Strike Probability Increase on Radio Messages
Sent=30 ;[>0] Increase over the default probability
on a radio message sent
Friendly Air Strike Probability Increase on Contact Report
Sent=50 ;[>0] Increase over the default probability on a
contact message sent
Enemy Air Strike Probability Increase on Player Detection=50
;[>0] Increase over the default probability on player
detection."

OTHER IMPORTANT SENSOR FACTORS AND NOTES!

The "Command sensor" in SH3:
In SH3 only ONE sensor is actively in use at any given
time... This is true whether it is your own U-boat and
crew... or an enemy vessel or aircraft. The available sensor
with the highest maximum range is typically the "command
sensor" and OVERRIDES other sensors that may be available at
the moment.

The periscope in SH3 created the night vision bug. The
player U-boat crew unreasonably spotted distant objects...
due to the fact that in effect, the periscope could see
through its protective shroud even though it was retracted.
Additionally, to understand this better... you can think of
the periscope as a watch officer AI crewman. The trigger for
this to occur in-game is the moment the red light goes on in
the F5 navmap screen. (The actual moment that "night"
in-game begins to dominate certain behaviors of the AI.)
The enemy also subject to the night vision bug, though not
quite as badly as the player U-boat crew is in stock SH3.
They see too well at night all the time and the only way to
completely correct them at night causes other problems such
as blindness during broad daylight.

(Something else I've noticed regarding players speaking about
the vampire night vision bug... Their radar is "on" and as a
result is the "command sensor" which reaches farther than the
maximum visual range of the player AI watch crew. Players
sometimes mistake this as a visual spotting. The AI watch
crew cannot see beyond 5000 meters at night.)

Light factor: Both your own AI crew and enemy AI crews are
heavily influenced by this value.

Crew efficiency values: Enemy aggression is
HEAVILY influenced by crew effieciency rating. Crew efficiency
values below "veteran" often appear quite useless... and even
large numbers of otherwise effective "elite" crews can be
subject to traffic jams allowing the player to motor away
quite readily if the modder isn't careful.

Progression of technology in the war and AI crew efficiency
values are represented in SH3. This was preserved by the GWX
dev team.

Limitations born of design and necessity in SH3:
The size of the three dimensionally rendered world is an
understandeable limitation... 8km or 16 km is the norm...
However, on a calm clear day with good light as mentioned above...
visibility in the real world can exceed 32 km. Mirroring such things
ACCURATELY in SH3 would keep most of us from playing due to
impact on the system. (Maybe in a few years aye?)

In SH3 there are only two PRIMARY types of player hydrophones
modelled. (GHG and KDB) In SH3 only one range value can be
set for the maximum range of each hydrophone. In real life, single
ships could be heard a little over 20 km and convoys could be
heard at over 200 KM!!!! Again, in SH3 we can only set one
maximum range for each hydrophone... In GWX these values were
set VERY conservatively. (Somewhere between 20-30 km... I
don't want to just come right out and say it here... because
I don't want to ruin things by giving players exact
information on a sensor where real-life variables cannot be
modded in a more interesting way.)

Water temperature, salinity, and many other variables
affecting sensors are not modelled in SH3. There is no real
apparent random factor modelled for that matter. In real
life, maximum performance values are condition dependant,
meaning that even a directly stated value is an averaged or
generalized value.

Many people do not realize that in real life a single depth
charge could ruin local hydrophonic acoustics for as much as
10 minutes... I've heard this described as a "sizzling" sound
by the massive amount of bubbles released. This has been
modelled in part by using one method or another several major
mods... The same is true of GWX.

In SH3 any ship moving faster than 15 kts is not using its
hydrophone.

Acquisition of contact vs maintaining contact: For both
player and enemy AI crewmen in GWX, it is much harder to make
the initial contact... than it is to maintain that contact.
You can practice this in real life. Simply spot an airplane
and watch it until it is almost invisible... then look away
for a moment... It will be harder for you to re-aquire
visual contact. :O) The same can be said of hydrophone,
sonar, and radar contacts. In one way or another all means
of detection are subject to interference or "noise."

Reminder- The AI itself in SH3 cannot be modded... only its
peripheral perceptions of the environment can be altered positively or negatively.

For GWX players. Further modding WILL cause negative
side-effects in-game. Modding the sensors in SH3 really is
like trying to tie your boots without enough string.

For the most part in GWX... all we needed to do to make the
sensors more effective... was to heighten their sensitivity
to the environment... and tolerating what felt like AGES testing the
effectiveness of each class of sensor... with each AI crew
efficiency level... one value at a time... This translated
into months of focused, boring, and tediously repetitive
work.

For me personally, it feels like I spent a lifetime waiting
on the loading bar!

No doubt some will complain... or continue to make their
subtle little digs at what we've done. I've seen the word
"ubered" tossed around quite readily by those that feel we've
made things too tough... and the implication is made that we
made changes to the sensors in an irrational fashion. I view
those as simply opinions born of players that are having to
get out of their comfort zone! Others may even claim that we
don't know what we are doing.

Whatever.

At the end of the day... in all cases regarding the sensors
and ASW package in GWX, arbitrary decisions had to be made
based on in-game performance to offer the player a stiff, but
not un-breakable challenge that increases in difficulty with
the progression of the war.

Every modder who has ever worked on the sensors in SH3 has
had to make similar decisions weighing real-life (often
foggy) currently available research data in comparision to
the means of simulation.

Seriously, all of us have our own ideas of what "realism" is
based on what we see and interpret.

Only the guys who fought the real battles have any true right
to talk about "how things really were."

Now before anyone starts complaining about the SH3 Devs and
the limitations of SH3 itself... I have to say, throughout
the course of developing GW, I've grown to have what I feel
to be a special understanding of the challenges involved in
bringing such a complex piece of work to the table.

To try and describe the process of building GW and GWX, it
felt like walking into a famous cathedral that needed
renovation from the ground up... masonry, painting,
artwork... the entire massive lot. Expectations abounded,
and nay-sayers surrounded. To stand back and look at the
work as a whole prematurely, was imposing and frightening.

You guys are a demanding lot LOL...
but in the end it was the ghosts of the past that drove us
along with unrelenting force. Their story must be
remembered.

We are lucky to have SH3... and luckier still to have SH3
with a dynamic campaign. We all know it has its quirks...
but it does the job better than anything else out there in my
opinion.

The problem with modding and/or being an artist... is the
simple act of choosing a place to stop... and calling a
creation that will never be perfect for everyone on every
occasion... "FINISHED!!!"

Add-in real-life trials and tribulations... and a deadline or
two... and you can find yourself in a personal hell of your
own making LOL.

Sorry for the ramble and speaking in such general terms. I
just don't have the diesel left in the tanks anymore for a
long winded tech-discussion on such matters. I hope you will
settle for this overview. Except for the stray detail here and there,
it is likely to be a detailed as I ever get in reference to what we've done.

If you find it un-satisfactory, feel free to use other
mod-packages apart from GWX. In the end, it still boils down to
individual taste based on in-game experience... not miles and
miles of text... or publicly and loudly stated opinions.

Sink them all!!!

[Ducimus]

For GWX players. Further modding WILL cause negative
side-effects in-game. Modding the sensors in SH3 really is
like trying to tie your boots without enough string.

This is why, in my own modding, ive done my upmost to avoid three files in particuar.
SIM.cfg, Sensors.dat, andAI_sensors.dat

In only one instance did i feel compelled to edit one of those files, and I did so limiting myself to only airborne radar's maximum range to achieve a desired effect. It was the best solution i could find without messing with other variables that would throw everything else out of whack.

The entire sensor package as a whole, is like a big pile of spagetti. I know this better then most, so i resist touching it. The top two lines dealing with the waves in the Sim.cfg alone cause HUGE ranging effects. Some good, some not so good, but it's probably the best that can be done with what there is to work with.

A piece of advice, and i say this not as someone you rubbed the wrong way, but as an avid gamer and human being. Your entire post strikes me coming from a person worn out from having obsessed over something for entirely too long and has become sick either physically, mentally or both. TAKE A BREAK. Go do something else. Do not even look at subsim or SH3 for at least 3 weeks or (preferbly) longer. Beleive me, it helps.

Excellent info there! I have copied it to my notes. Thank you.

I have always felt that the Silent Hunter games claiming to be simulation
was a cop out of sorts.

I consider a simulation to be :

An enviroment that replicates as closley as possible,
the events and the equipment.

GWX comes as close to that as I have seen to this point.

Can it be better?

Yep.

No way ANYONE should hit the top of the ACES list nor bring in the Tonnage
I see listed all the time.

You want a game? Play stock.
GWX to hard? Play another mod.
You like GWX except (add yer complaint here). figure out how to do it yourself!
All the info is here in the forums!
The GWX guys have WAY better things to do then give us a whimpy version.

Give me the same odds that the TRUE Submariners faced and I'll call it a simulation
then.

@priater

I agree wholeheartedly with your wise words old man.....but what about those people who feel GWX is too "easy" ?!! I know, I know...its blasphemie to say it but I still feel/think that NYGM is best mod out there from a realistic tonnage point of view and, even then, it too is not as hard as i'd like it.

A simple glance over U-Boat stats reveals that the tonnage we're all getting from all these mods is way too much historically speaking. That said, most U-Boat commanders never got as much "exercise" as we are....i mean we don't die when we make mistakes, so its only natural that our tonnage will increase over time as we "master" the elements required for making a good kill.

I have to ask though, in fact I really want to know....why did so many fish launched against ships in WWII fail to hit their target ? Torpedo problems not withstanding, I believe there must be other, perhapps more subtle, factors that influenced a firing solution which are not modeled in this game.

Wave motion effect on ship speed
Incorrect range estimates
TDCs mechanical component wear
Natural wave mechanics influence on torpedo trajectory

Any more ? There must be loads!

[monsoonrat]

Any game that one buys, when you start your career, you are given a choice of easy, normal, hard, or realistic. This is an option that game designers know that some people want a closer to real life experience, but some people want an arcade experience. With that said, I would not stress the fact that some people are finding GWX too hard, what they need to do is play some other big mod or wait until someone else comes out with a addition mod that has the very thing that they are looking for. The other option is for them to learn to mod themselves, which they will quickly know how much time it takes to tweak and test, but they must also realize know that you might not come to bail them out when they painted themselves into a corner.

Now some people will want GWX dev team to provide them with an easy, harder version, but they must remember that you do not get paid for doing this and the fact that you do this between your job and your life does not leave much time to please everyone.

I myself know that one can get 10 positive comments about one's creation, but that 1 negative comment will put one big hole in my feeling good about a job well done. With the negative comments that I have read from various whiners you must be feeling like swiss cheese by now.

I haven't officially thanked the GWX Dev team for all their hard work, until I was waiting to finish my first mission. But, since you sound so down... THANK YOU very much GWX Dev Team, I really appreciate all your hard work as I am truely enjoying being pinned down in the english channel in 1939 (I know, everyone keeps saying, stay away from the english channel!) traveling at 2 knots and hoping that one of those 3 bastards (english and french) that are trying to kill me does not drop a depth charge directly on me, as I only have 40 sometime meters of water to hide in.

One last thing, thanks for the explanation about the sensors, it helped me better understand all the different variables and limitations.

[Notewire]

@Uber

I have been struggleing with torpedo accuracy for a long time, and what I can't figure out is - how do so many people hit their targets? I have been playing this game since it came out, and I have come to the solution that I just suck at solutions . I play on 100% realism, manual solutions, and I don't use the WO range estimates, as I think the fact that he can call the range down to the meter is completely inaccurate. Playing this way, I have had a heck of a hard time doing the Fast 90 method, as plotting course takes range and bearing, and using the periscope, it seems my ranges vary by about 5-10%, even when stopped. At very short range, this isn't a problem, but beyond 1000m - it gets pretty damn hard to estimate range down to the 100m increment, at least for me. So this leaves me with a gut feel for AOB, because enemy course is so inaccurate given a 5-10% range error. Anyway, even if I get it all right, within 3 min and 15 sec, and I get the fish to fly quick and true - I don't even know if it hits!!! Personally, I bet that German torpedo makers got a raw deal - I bet half of the duds were actually misses - because - how would you know? Seriously, I am a military guy, I can't imagine the crew watching me as I manuever, hunt, dive, hunt, manuever, calculate, set up a shot, and fire - and I am watching through the periscope and can't see crap because it's an electric - and no explosion - with the crew watching me holding their breath. I think I would be tempted to say "Damn, a dud". So my only option has been to close to knife-fighting range to fire - which - with GWX sensors (to come back on topic) has been very hard (as was realistic). So to bring my rant back on topic - given the above information on sensor accuracy and effeciency - anybody have any better torpedo advice than the Fast 90 - because I personally don't see how they hit as much as they did?!?

Tschuss,
Notewire

[Ducimus]

Originally Posted by Notewire

I have been struggleing with torpedo accuracy for a long time, and what I can't figure out is - how do so many people hit their targets?

Electric torpedo's? Those give me issues occasionally. I'll use WE assist once in awhile, and even then ill miss with electric fish. It has something to do with the targets speed, and the game possibly rounding internal calculations, im not sure.

But, once in awhile, no matter what range, the fish always falls aft of the target. I could shoot a steam fish at the exact same target, exact same position, exact same formula and hit. But the instant an electric fish goes out the tube, its a miss.

The best "fix" ive been able to come up with, is to take whatever the targets speed is, and set the TDC to 1 more kt then its acutal speed.

@notewire

I hear you mate, I really do....and as for not using the WO, well if you use NYGM, teh WO doesn't even give a range estimate....which means you're on your own with the stadiscope. At first I was resistant to this but now, well, I think it helps make the game more difficult....which is cool.

I will say this, please persevere with the manual targeting....don't bother trying to get ship speed from a target at 10,000M...just plot the contact and then take another plot after 3mins 15 secs. That should give the targets general course, whilst you're doing that, get the boat ahead and a little nearer. Take a couple of these "vague" readings and draw a line linking them up...this will give you a good idea of course.

Once the taget is 7000m or nearer, then start taking 3min 15sec plots to get the spead. Ducimus is right, you'll never get it spot on but you'll get a "feel" for for the speed....somehwere between 7 and 9 kts for eample. You'll also be improving the target course whilst doing these....then set the speed for something in the middle of your results and motor on to a position where you'll be about 600 to 900m AOB 90 of the projected course. Forget speed from now on, just get as near to 500m as you can, fire a spread of 2 fish with 1deg spread, no "under keel" shot...so 5M depth say and i bet my aunties nipples you get a hit. (Which is why even NYGM is too easy sometimes).

Of course, if the target is zigzagging then you should aim to get say 600 to 700 m from mean couse and adjust as he gets nearer. Also, use all your plots to appro speed, not just distance covered between 3mins 5 secs.

I'm probably telling a granny how to suck eggs here...buy hey, I like the way you play so feel the need to encourage it :-)

Also, joing WaW.....at least when you miss your target in WaW its, well, realistic :-)

And checkout these links for some very interesting target intercept techniques: http://hometown.aol.co.uk/dominicobaggio/intercept.html
http://forums.ubi.com/eve/ubb.x?a=tp...3&m=5051056603
http://www.paulwasserman.net/SHIII/

Lookforward to seeing you in WaW!

Cheers, Hans.