[WIP] Realistic Hydrophone

[tscharlii]

Originally Posted by gap

Yes indeed, great mod an video.

Originally Posted by sober

Great video I like the mod .

Thank you!

Originally Posted by gap

- I have noticed that the mod package includes sound files apparently identical to the stock (German) ones. What is their purpose?

Those are .wav counterparts to the .ogg files of the game. My mod plays the speech through the directsound api, which understands .wav by default but not ogg. Obviously there has to be a ogg decoder included in sh5, but I have not looked for it yet. Some files (those with real names, eg. YesCaptain.wav) are cut from other sound file, so they can be used in a different context.

Originally Posted by gap

- With TDW's hydro follow target, we could set if the hydrophone operator must report bearings at fixed and user customizable intervals of time, or everytime target's bearing changes. Any way to restore this feature?

Yes, but probably after other things work. See below

Originally Posted by gap

- following the reasoning of my previous post, can you make bearing accuracy to vary depending on hydrophone model used, range to target, operator experience, target bearing (for the GHG only), own engine noise, sea state, etc.

This is what I want to focus on. I'll skip aspect/speed estimations for now. They can be added later on when realistic hydrophone information gathering works properly.

Originally Posted by gap

- If a contact is detected during normal sweep, the operator will report immediately whether it is closing/moving away. Can you introduce a margin of error to these early reports, marking their provisional nature with phrases like "probably moving away" in notepad logs. This data would then be calculated with better precision by your scripts and reported again in the log. Even better if the accuracy of early and "precise" reports is affected in part by operator's experience, noise factors, etc.

My idea here is to suppress the initial stock contact report altogether, and have my own initial contact report, as soon as the sonarman is able to hear the contact after considering all the relevant factors. His report will rather be something like "I can hear something on starport abeam". Starting from there he will gather more precise information. I'm currently in the process of determining whether and how it is possible to suppress stock speech and shiplog reports.

Originally Posted by gap

- can you make the "report estimated range" command only available after having followed the same contact for a while? At the very beginning our radioman should only be able to say wether it is a close, medium or long range contact, without throwing any number. Later on you could let him to report some numerical ranges, randomized according to experience, real range, noise factors, etc. If possible, you should also avoid the "dumb officer syndrome" where randomization leads to absolutely unconsistent estimations (i.e. if a target is esitimated as closing, the sonarman shouldn't report two consecutive range estimations of 1,500 and then 1,600 m).

Maybe these ideas should go into TDWs directions, since he is the author of "report estimated range". Also, I probably shouldn't try to fix his code again in the near future. In fact I had a similar function in my mod with voice reports and all, but removed it before releasing the first version of this mod, because I was not sure, how realistic it is to have the soundman tell distances +-100 or even +-1000 meters at all. I imagine this requires a huge deal of experience and this experience seems to be hard to build up. Different engines, different speeds, weather conditions, and you never get first hand visual feedback, whether your estimation was correct. Maybe in military training sessions before going on patrol for real.

Originally Posted by gap

- can you get rid of the "report exact range" command? I think is is simulating the usage of an active sonars, but type VII were hardly (if ever) equipped with them. Moreover it is a cheat, because the enemy won't be able to hear our virtual pings...

I'm not a huge fan of telling people how they have to play a game. What I do is, I never press that button, albeit it is there. It's a choice, like setting "no map contacts" in the options dialog or installing "real navigation".

[gap]

Originally Posted by tscharlii

Those are .wav counterparts to the .ogg files of the game. My mod plays the speech through the directsound api, which understands .wav by default but not ogg. Obviously there has to be a ogg decoder included in sh5, but I have not looked for it yet. Some files (those with real names, eg. YesCaptain.wav) are cut from other sound file, so they can be used in a different context.

I see now. Maybe you should also include English sound files then, for the ones who don't like German speech

Originally Posted by tscharlii

Yes, but probably after other things work. See below

Originally Posted by tscharlii

This is what I want to focus on. I'll skip aspect/speed estimations for now. They can be added later on when realistic hydrophone information gathering works properly.

Okay, a GHG hydrophone and an unexperienced operator should result in a bearing margin of error not lesser of ±1.5 deg on average; probably much higher at long range, near bow/stern, and when surrounding noise level is high.

Originally Posted by tscharlii

My idea here is to suppress the initial stock contact report altogether, and have my own initial contact report, as soon as the sonarman is able to hear the contact after considering all the relevant factors. His report will rather be something like "I can hear something on starport abeam". Starting from there he will gather more precise information. I'm currently in the process of determining whether and how it is possible to suppress stock speech and shiplog reports.

Cool, I like this feature. I wish you good luck with it

Originally Posted by tscharlii

Maybe these ideas should go into TDWs directions, since he is the author of "report estimated range". Also, I probably shouldn't try to fix his code again in the near future.

As far as you don't use his code there shouldn't be problems at all. I am probably overcautious on these matters though. If there was an easy way to get in touch with TDW, he would probably encourage your work. Unfortunately this is not possible at present. In any case, if you get inspirations from his code for introducing some improvements into it, and as far as you don't get money from it, you do it openly and you give him credit, and you support his mod, I don't think TDW will complain. But try outsmarting him or claiming his code as yours own, and soon or later he will know it

Originally Posted by tscharlii

In fact I had a similar function in my mod with voice reports and all, but removed it before releasing the first version of this mod, because I was not sure, how realistic it is to have the soundman tell distances +-100 or even +-1000 meters at all. I imagine this requires a huge deal of experience and this experience seems to be hard to build up. Different engines, different speeds, weather conditions, and you never get first hand visual feedback, whether your estimation was correct. Maybe in military training sessions before going on patrol for real.

Estimating rough ranges trough hydrophone was surely possible. Today I have read a British report on the interrogation of the survivors of a captured/destroyed u-boat, where the sonarman had (correctly) reported a range to the attackers of only 500 m, and the commander had discarded it based on his periscope visual (he was convinced that the pursuing ship was much bigger, thence his wrong range esitimation). Undortunately I didn't save the link, but I will try looking again for it tomorrow

Originally Posted by tscharlii

I'm not a huge fan of telling people how they have to play a game. What I do is, I never press that button, albeit it is there. It's a choice, like setting "no map contacts" in the options dialog or installing "real navigation".

People are not forced to use a mod either. I would gladly disable that button, if I had an option to do it. Why? you may ask. Well I suppose for UI cleannes, and for avoiding dispensable temptations in the most harsh situations

[tscharlii]

Originally Posted by gap

I see now. Maybe you should also include english sound files then, for the ones who don't like German speech

Who on earth would not like german speech? Up until now I didn't even realize there is english audio for SH5. Every sh5 video I saw on youtube featured localized menus/texts, but german ingame voices. I own a german copy of sh5 and I do not have access to the english speech files as they are not included in my copy.

[gap]

Originally Posted by tscharlii

Who on earth would not like german speech? Up until now I didn't even realize there is english audio for SH5. Every sh5 video I saw on youtube featured localized menus/texts, but german ingame voices. I own a german copy of sh5 and I do not have access to the english speech files as they are not included in my copy.

Though not understanding much of them, I like German dialogs myself. Nonetheless I am sure someone doesn't, and seeing that you are "not a huge fan of telling people how they have to play a game" I thought I would have given you this suggestion.

By the way, I have English speeches on my HD (my italian-localized version of the game uses them by default); if you want, I can send them over to you. If you care. If you don't, neither do I

[gap]

Originally Posted by gap

Estimating rough ranges trough hydrophone was surely possible. Today I have read a British report on the interrogation of the survivors of a captured/destroyed u-boat, where the sonarman had (correctly) reported a range to the attackers of only 500 m, and the commander had discarded it based on his periscope visual (he was convinced that the pursuing ship was much bigger, thence his wrong range esitimation). Undortunately I didn't save the link, but I will try looking again for it tomorrow

Found it!

http://www.uboatarchive.net/U-111INT.htm

Originally Posted by VI. SINKING OF "U 111"

In "U 111" the sounds of "Lady Shirley's" propellers were heard, and Kleinschmidt at first estimated that the British ship was about 5,000 metres (about 5,467 yards) distant.
The Junior Officer and the man at the listening gear were stated to have expressed surprise and to have queried Kleinschmidt's estimate, as they were of the opinion that their enemy was much closer.
A little later Kleinschmidt estimated that the British ship was 3,000 metres (about 3,280 yards) distant. The man at the hydrophones at once questioned this statement as he thought that the distance was about 500 metres (547 yards) only; but Kleinschmidt somewhat irritably repeated his estimate of 3,000 metres. It was emphatically stated that Kleinschmidt looked through the periscope just before making this statement; prisoners assumed that either something was wrong with the periscope or Kleinschmidt clung to his conviction that his adversary was a much larger ship at a much greater distance than was actually the case.
A few moments later the explosions of depth charges were heard.

You can get a lot of interesting clues from British Interrogation Reports. They are not always fully accurate, but on many topics they are the best (or the one) accessible source

[tscharlii]

Alright, some theory crafting regarding hydrophone range and bearing accuracy based on different sources. Thank you, gap, for pointing me to them.

[1] http://www.cdvandt.org/GHG1996.pdf
[2] http://de.wikipedia.org/wiki/Gruppenhorchger%C3%A4t
[3] http://www.sengpielaudio.com/calculator-spl.htm
[4] http://www.sengpielaudio.com/calculator-levelchange.htm
[5] http://www.forum-marinearchiv.de/smf...?topic=4469.15

The following calculations are based on very little data and lots of assumptions and last but not least, I'm not sure whether I have applied the maths involved correctly.

I'll assume a GHG (group listening apparatus) hydrophone and a merchant at 12-13 knots as target contact, since [1] provides some data on this scenario, and this is a common encounter in sh5.

First: max range under optimal conditions, that is:

• no noise
• contact on a relative bearing of 90 degrees starport or 270 degrees port

[1] says: 20 to 25 nautical miles, that is 37km to 46.3km for single contacts
[2] says: 20km for single contacts

I'll go with something in the middle, which, as far as i know, happens to be the max range the game starts to make contact pickups by hydrophone possible:
At R = 30km under optimal conditions, we can barely hear the contact.

At 90 (or 270) degrees the sound hits the hydrophones perpendicularily, i.e. the full sound pressure stimulates the hydrohpone membranes. If we deviate from that bearing, the sound waves come in aslant, and only a portion of the sound pressure is effective on the hydrophone. That portion is proportional to the cosine of the deviation from 90/270 degrees:

maxrange = R * cos(alpha), with alpha = deviation from 90/270 degrees

On to bearing accuracy: The GHG features 4 settings for the high pass filter[1]:

• off: no filtering, complete received sound usable, bearing accuracy 8 degrees (I understand this as +-4 degrees)
• 1kHz: frequencies below 1kHz are filtered out, uses frequencies from 1kHz - 7kHz, yields a bearing accuracy of 4 degrees
• 3Khz: frequencies from 3-7kHz used, yields a bearing accuracy of 1.5 degrees
• 6kHz: frequencies from 6-7kHz used, yields a bearing accuracy of < 1 degree

The maxrange depending on bearing, that we calulated above, applies to the "off" setting. As soon as we start filtering frequencies to achieve better bearing accuracy, the sound gets softer and we need to be closer to the contact, so we can still hear it. But how much closer?
[1] says: the merchant at 12-13 knots produces a maximum sonic spectrum at 100Hz, and only 10% at 4Khz, which we can hear using the 3Khz setting. Fortunately, filtering out 90% of a sound does not mean, that the sound pressure usable at the hydrophone drops by 90%. [3] has some explanations and calculators about this topic. I assume, the sounds at different frequencies produced by the engines are equally loud. If we add up 10 equally loud sound sources with different fequencies, the sound level increases by 10dB. Conversely, the sound level of those 10% at 4kHz is 10dB less than the sound level of the whole, unfiltered sound. This level change is called "Delta L". The formula is not quite as simple as it looks like from this particular example (see [3], Adding levels of equal loud sound sources).

What does a 10dB decrease of the sound level mean regarding the usable sound pressure at the hydrophones? According to [4], the sound pressure factor y = 10^(Delta L / 20): For -10dB, this is around 0.316.

Sound pressure is reciprocally proportinal to distance.
So, at a bearing of 90 degrees, under optimal conditions, we can barely hear our contact at 30km and determine its bearing +-4 degrees. If we wanted to use the 3KHz setting and determine its bearing +-0.75 degrees, we need to be closer than 30km * 0.316 = 9.48km.

Time for some more assumptions regarding the other hydrophone high pass filter settings, that are not covered by [1]:
Given that 10% are above 4kHz, let's assume 33% (i.e. one third) of the sound emissions of a merchant at 12-13 knots are above 1Khz, and 3% (i.e. 1/33rd) are above 6kHz.

This leads to the following sound pressure factors:

cutoff-frequency        None        1kHz        3kHz        6kHz
Portion of the sound    100%        33%         10%         3%
Decrease of sound level 0dB         -4.77dB     -10dB       -15.18dB
Sound pressure factor   1           0.577       0.316       0.174
bearing accuracy        +-4         +-2         +-0.75      +-0.5

Let's combine the two results (reduced maxrange depending on the bearing as well as on the selected high pass filter) and visualize them.
We have four functions, one for each high pass filter setting. The x-axis is the bearing deviation from 90/270 degrees. The y-axis shows the maximum range in km at which we can hear the contact using a specific high pass filter setting.

Blue(off)  : f(alpha) = 30km * 1     * cos(alpha)
Red(1kHz)  : f(alpha) = 30km * 0.577 * cos(alpha)
Green(3kHz): f(alpha) = 30km * 0.316 * cos(alpha)
Gray(6kHz) : f(alpha) = 30km * 0.174 * cos(alpha)

These results correspond to [5], which (I think) cites from E.Rössler, Die Sonaranlagen der deutschen U-Boote. In a GHG-Test performed in February 1943, the type IX sub U-194 was not able to detect a contact in a distance of 4km on a bearing between 348 and 10 degrees. This test report also has some data on the GHG range while the sub is moving at different speeds.

Feedback is very welcome. Maybe I should have used sound intensity instead of sound pressure? Maybe assumptions are just plain wrong? Otherwise I'm planning to base the mod on this model.

[plj]

Originally Posted by tscharlii

starport

That's starboard We're not playing X3 here :p

I cant comment on the rest, tho it sounds very cool!

[tscharlii]

Originally Posted by plj

That's starboard We're not playing X3 here :p

Yeah, thanks, i'd have continued typing (and speaking) it the wrong way. English is not my mothertongue. X3 is a nice game though, and I also play Starcraft II, which has starports as well. That probably puts me on the dark side of the force.

Originally Posted by gap

Individual hydrophone parameters (SensorData controller).

I guess you are talking about /data/Objects/Sensors/UBoot_Sensors.sim
What's the best way to have a look at the contents of this file? When I try to open it using the GoblinEditorApp.exe it just gives me a lot of "Lost Controllers" entries. That can't be right.
Currently I'm looking at it using a hex editor.

[tscharlii]

Ahh, n/m, I think I figured it out. Load UBoot_Sensors.GR2, Merge UBoot_Sensors.sim, and then:
Project Tree: SensorData @ GHG9 -> Existing Behaviours -> Render -> SensorData

[gap]

Originally Posted by tscharlii

Yeah, thanks, i'd have continued typing (and speaking) it the wrong way. English is not my mothertongue. X3 is a nice game though, and I also play Starcraft II, which has starports as well. That probably puts me on the dark side of the force.

"It takes strength to resist the dark side. Only the weak embrace it!"

Obi-Wan Kenobi

Originally Posted by tscharlii

I guess you are talking about /data/Objects/Sensors/UBoot_Sensors.sim
What's the best way to have a look at the contents of this file? When I try to open it using the GoblinEditorApp.exe it just gives me a lot of "Lost Controllers" entries. That can't be right.
Currently I'm looking at it using a hex editor.

Open UBoot_Sensors.GR2 first, and merge then UBoot_Sensors.sim

[gap]

tscharlii,
your graph, the calculations it is based on, and the few arbitrary assumptions you had to make, look perfectly plausible.

I have just a little amendment regarding SH5's "physical world horizon": IIRC, it is in a radius of ca. 40km from the player, not 30 as you are assuming. But this little mistake doesn't invalidate the rest of your reasoning. I wish to point you now to some of the parameters affecting hydrophone's efficiency in stock game. This is to make sure that you will take them into the due account; should you be already aware about the information below, please ignore it.

Configurable global parameters (data\Cfg\Sensors.cfg). They affect all the player's hydrophones (AI units' sensor parameters are found in data\Cfg\Sim.cfg, and are much more simplified):

Hydrophone range factor: how much target's range affects its detection by hydrophone

Hydrophone fog factor: how much fog affects hydrophone detection

Hydrophone light factor: how much darkness affects hydrophone detection

Hydrophone waves factor: how much rough seas affect hydrophone detection

Hydrophone speed factor: how much own speed affects hydrophone detection

Hydrophone enemy speed: how much enemie's speed affects its detection by hydrophone

Hydrophone aspect: how much a low profile offered by the enemy affects its detection by hydrophone

Hydrophone noise factor: how much surrounding noises affect hydrophone detection

Hydrophone sensor height factor: how much a low sensor height affects hydrophone detection

Hydrophone already tracking modifier: this is a bonus, making the hydrophone operator not to loose a contact which was already detected

Hydrophone decay time: this is the duration of the bonus, in seconds from the last detection

Hydrophone uses crew efficiency: (boolean) whether hydrophone detection will be affected by crew efficency or not

Numerical factor must be ≥ 0 and can be bigger than 1 (i.e. they are not percent values), even though most stock settings are comprised between 0 and 1. It is my understanding that the higher the settings, the worst is hydrophone's detection under the conditions described for each parameter. Indeed, some of the parameters (fog, light, aspect) are non-relevant for hydrophone detection and they are set to 0 in stock game: they are probably there for symmetry with other sensors. The meaning of other parameters which are also set to 0 in vanilla game is a bit obscure (namely: enemy speed, sensor height), and needs to be investigated, but again I suspect that they are designed to work for other sensors.

Individual hydrophone parameters (SensorData controller). They affect individually each player hydrophone or any other sensor which can be fitted aboard playable u-boats (another controller, obj_Sensor, is used for AI sensor, and as before its parameters are much more simplified):

SensorType: valid types are: 0 - visual, 1 - Radar, 2 - Sonar, 3 - Hydrophone, 4 - RadarWarning, 5 - RadioDF

PreciseRange: this is the range in meters within which sensor detection won't be affected by target's range

MaxRange: sensor's maximum detection range in meters. To my understanding, between PreciseRange and MaxRange target's detection chance is interpolated linearly.

MinHeight: min target depth/altitude in meters, for it to be detected

MaxHeight: max target depth/altitude in meters, for it to be detected

MinSensorHeight: min sensor working depth/altitude in meters

MaxSensorHeight: max sensor working depth/altitude in meters

Surface: Standard detectable surface at MaxRange (in square meters)

RPMDetLevel: Detection level as percentage of Maximum RPM (between 0 and 1). This is a bit tricky as it make a warship (with higher max RPM output) lesser detectable than a merchant travelling more or less at the same speed.

SweepPeriod: this is how long (in seconds) will take a 360 deg (bearing) sweep of the sensor.

SweepArc: I think this is the width (in degrees) that the sensor will be able to scan at each given moment; I ignore if it cam also affect sensors' bearing accuracy, but I doubt it.

ProbInsideArc: detection probability inside each SweepArc (between 0-1)

Revolving: (boolean) this is whether the sensor 3d model will revolve around its y axis during a sweep

SkipSweep: (boolean) whether the operator will skip invalid sector during sensor's sweep (see below)

Bearing (min, max): this let us to set valid bearing intervals (in degrees) within which sensor detection is possible. More than one interval can be set.

Elevation (min, max):this let us to set valid elevation intervals (in degrees) within which sensor detection is possible. More than one interval can be set.

That's all. As you can see, our current chances to customize hydrophones and other sensors are pretty much limited. In my opinion, the main limits are:

- inability to customize global parameters for each sensor.
- inability to customize sensors' accuracy (we should explore the SweepArc parameter though)
- inability to set different detection ranges and/or detection probabilities as a function of sensor's and targets' altitude/depth and of target's bearing.

TheBeast and XRundel have deviced a brilliant method to partially obviate some of the above limits: for their Repaired Equipment Mod, they have used more than one SensorData controller for each hydrophone, each set with different sensor depths and max detection ranges. This workaround is not exempt from flaws though: each controller must be placed in a precise order for the hydrophone not getting mute for the player below a certain depth; I and Volodya discovered it a while ago. Yet, after our fix the human player is not affected by detection range limits at varying depths, and he can always listen as far as the maximum set MaxRange. Moreover, detection ranges for the AI operator don't vary following a curve, but they "jump" from a level to the next one. A smooth depth/range curve could be simulated by using many conveniently set SensorData controllers, but I am afraid tha doing it could determine lagging or even worst problem.

This is really all for now. Keep up your brillian work

Originally Posted by plj

That's starboard We're not playing X3 here :p

I cant comment on the rest, tho it sounds very cool!