Nightly Surface Attacks

[rik007]

One aspect easily forgotten are nightly surface attacks early in the war. So I hope that the requirement is met that you should be able to conduct a decent surface attack. Prien, Kretschmer etc actually surfaced inside a convoy picking out juicy targets. I have a BBC-documentary in which a u-boot crew member tells that inside the convoy and submerged they could see crew members of merchant ships on deck smoking thier cigarettes. I know a way to put out those cirgarettes...

:arrgh!:

[Webster]

the game never has been good at "seeing" visually, its always seeing with sensors even when your not supposed to be detected.

things like rain, fog, poor light conditions, should allow you to be masked but they dont matter to the game, its like their gunners have night vision and the ships have gps coordinates for you from miles away the second your scope goes up.

this is why you cant get near a convoy at night on the surface without being blown out of the water from miles away. :damn:

[Snestorm]

Getting close to a target in convoy isn't that bad.
Exitting the scene from 1.000 meters is nearly impossible.

The only ones that seemed to have their vision affected by wheather was me and my lookouts.

In reality, during heavy fog, a destroyer, without radar, would have a hard time just staying in contact with the convoy. The merchants could use a Stern Blue light to maintain position, but making any turns is out of the question. (The light is only visible for a short distance from DIRECTLY ASTERN.)

[karamazovnew]

Normally each uboat attack should raise the alarm state in the region. If a ship sees a close vessel being blown out of the water, you can bet that the entire crew will put on their glasses. However if you go near Patagonia, you might even dock with them while they're all asleep. Each ship should have dedicated lookout positions that cover a specific arc. The more danger there is, the more men on watch, some even with binoculars. And the distance at which they can see should be dependant on fog, wave height, time of day, if you have your back at the sun, and the sub's speed.

The devs have already explained that the map will show the detection ranges, but that should only be the median Chance of detection range for the current wather time of day and general type of ship. You should still be spotted outside of it if you're unlucky, or be able to put along side the ship if all the men are below decks and there's a drunk captain at the helm.

[Lord Justice]

the same goes for costal def as well !! what the heck usally hit by 1st time shell miles out and in pitch dark, do the gun battery controllers have sight beyond sight?

[karamazovnew]

the same goes for costal def as well !! what the heck usally hit by 1st time shell miles out and in pitch dark, do the gun battery controllers have sight beyond sight?

McScott: Laddie... ya smell something fishy?
Angus: Aye, I certainly do.
McScott: What's your nose telling ya?
Angus: Me nose itches of Fritz tin can.
McScott: You don't say. Me finger itches too on the trigger.
Angus: Already? Let the man come closer to give the bastard hope.
McScott: Ups...

A minute later, 20 km away, the unsuspecting Uboat of 4Para begins to flood.

[Lord Justice]

aye, then ma breeks are soakin and we aw get droont

[PL_Andrev]

REAL VISUAL CONTACT is possible and no too hard to implementation.
Sensor of visibility shouldn't be depend from real range (with corrections of fog, wave) but depend from single pixels color.
You have three variables for sensitivity:
- contact circle (how many pixel is tested in local range)
- difference between test colour and average colour of background
- time reaction

AI knows that this contact is enemy but this contact is not detected yet.
A is average colour of contact in tested range
B is average background colour
If difference between A and B is higher than critical value AI waits "reaction time" and contact is detected (is sighted).

Of course difference between colour of contact and background and pixels number depends from level of enemy...

This method is very close to "real human eye"...
Look on the screenshot:
Circle on ship is "tested colour pixels range". If difference between all colours is higher than 5 (at this example |77-68|+|86-77|+|85-76|=27) the enemy ship is detected and can be attacked.

http://img301.imageshack.us/img301/9743/sh42.png

[rik007]

REAL VISUAL CONTACT is possible and no too hard to implementation.
Sensor of visibility shouldn't be depend from real range (with corrections of fog, wave) but depend from single pixels color.
You have three variables for sensitivity:
- contact circle (how many pixel is tested in local range)
- difference between test colour and average colour of background
- time reaction

AI knows that this contact is enemy but this contact is not detected yet.
A is average colour of contact in tested range
B is average background colour
If difference between A and B is higher than critical value AI waits "reaction time" and contact is detected (is sighted).

Of course difference between colour of contact and background and pixels number depends from level of enemy...

This method is very close to "real human eye"...
Look on the screenshot:
Circle on ship is "tested colour pixels range". If difference between all colours is higher than 5 (at this example |77-68|+|86-77|+|85-76|=27) the enemy ship is detected and can be attacked.

http://img301.imageshack.us/img301/9743/sh42.png

Of course this is one aspect you also need to take distance into account and the waves, crew experience etc. I think still your idea is better than implementing using random generators.

[rik007]

Getting close to a target in convoy isn't that bad.
Exitting the scene from 1.000 meters is nearly impossible.

The only ones that seemed to have their vision affected by wheather was me and my lookouts.

In reality, during heavy fog, a destroyer, without radar, would have a hard time just staying in contact with the convoy. The merchants could use a Stern Blue light to maintain position, but making any turns is out of the question. (The light is only visible for a short distance from DIRECTLY ASTERN.)

Even at night the profile of an U-boat was raised only a few meters above the surface and very, very hard to detect. These guys then exited an attack pretty easily. In SH-3 the lights turn on and then you are detected again and again. In reality without radar that didnot happen that easily even when the u-boat surfaced within the convoy which was the way they did it.

[oscar19681]

One aspect easily forgotten are nightly surface attacks early in the war. So I hope that the requirement is met that you should be able to conduct a decent surface attack. Prien, Kretschmer etc actually surfaced inside a convoy picking out juicy targets. I have a BBC-documentary in which a u-boot crew member tells that inside the convoy and submerged they could see crew members of merchant ships on deck smoking thier cigarettes. I know a way to put out those cirgarettes...

:arrgh!:

I know a way to light a whole shipload of cigarettes . Esspecially on a tanker!!!

[rik007]

I know a way to light a whole shipload of cigarettes . Esspecially on a tanker!!!

Smoking is bad for health, like being on a tanker.

[Snestorm]

Even at night the profile of an U-boat was raised only a few meters above the surface and very, very hard to detect. These guys then exited an attack pretty easily. In SH-3 the lights turn on and then you are detected again and again. In reality without radar that didnot happen that easily even when the u-boat surfaced within the convoy which was the way they did it.

Think they'll fix it?

[JScones]

Would be nice, but how will they fit that in to their development schedule? What, with all the 3D modelling and all. :roll:

[rik007]

Would be nice, but how will they fit that in to their development schedule? What, with all the 3D modelling and all. :roll:

Bottom line is that detection during a nightly attack should be more realistic than it is now. It should be more refined. I do not think of radical implementations to get this in place. I do not hope that we can only attack submerged. I little variation will help. Just a small change please...

[Snestorm]

So true. Especialy if one realizes that a periscope is pretty much useless on a dark night.

[PL_Andrev]

I'll be happy if UBI put not only value of sensitivity but allow to use scripts to dynamic check of visual conditions (waves, fog, time). With script we can do everything, with solid value - not.

[Méo]

Bottom line is that detection during a nightly attack should be more realistic than it is now. It should be more refined.

Another good point, especially when the type VII was designed for it (got no proof but seems obvious to me).

We need a better surface night attack experience all around! :timeout:

[IanC]

The old DOS subsims had different detection ranges depending on day/night, weather, and moon phases (I think AOD had all of these).

[rik007]

Another good point, especially when the type VII was designed for it (got no proof but seems obvious to me).

We need a better surface night attack experience all around! :timeout:

And it doesnot mean much implementation to get that in place.

[rik007]

The old DOS subsims had different detection ranges depending on day/night, weather, and moon phases (I think AOD had all of these).

Strangely enough sometime we still compare AOD with the SH-series. What a great game that was.

[PL_Andrev]

The old DOS subsims had different detection ranges depending on day/night, weather, and moon phases (I think AOD had all of these).

Yes, but f.e. many missions are located near northpole.
Situation: no moon, midnight ... escort ships don't see you... but this is "white night"... bug of game?
Not - this is problem of "thousands values"...

[Zonke]

REAL VISUAL CONTACT is possible and no too hard to implementation.
Sensor of visibility shouldn't be depend from real range (with corrections of fog, wave) but depend from single pixels color.
You have three variables for sensitivity:
- contact circle (how many pixel is tested in local range)
- difference between test colour and average colour of background
- time reaction

AI knows that this contact is enemy but this contact is not detected yet.
A is average colour of contact in tested range
B is average background colour
If difference between A and B is higher than critical value AI waits "reaction time" and contact is detected (is sighted).

Of course difference between colour of contact and background and pixels number depends from level of enemy...

This method is very close to "real human eye"...
Look on the screenshot:
Circle on ship is "tested colour pixels range". If difference between all colours is higher than 5 (at this example |77-68|+|86-77|+|85-76|=27) the enemy ship is detected and can be attacked.



In order to do that, the game would have to render views for every ship that could potentially see another ship (if it were to be implemented player->AI , AI->AI , AI->player)...

[ichso]

REAL VISUAL CONTACT is possible and no too hard to implementation.
Sensor of visibility shouldn't be depend from real range (with corrections of fog, wave) but depend from single pixels color.
You have three variables for sensitivity:
- contact circle (how many pixel is tested in local range)
- difference between test colour and average colour of background
- time reaction

AI knows that this contact is enemy but this contact is not detected yet.
A is average colour of contact in tested range
B is average background colour
If difference between A and B is higher than critical value AI waits "reaction time" and contact is detected (is sighted).

Of course difference between colour of contact and background and pixels number depends from level of enemy...

This method is very close to "real human eye"...
Look on the screenshot:
Circle on ship is "tested colour pixels range". If difference between all colours is higher than 5 (at this example |77-68|+|86-77|+|85-76|=27) the enemy ship is detected and can be attacked.


What would you do if there is another ship behind the first one ?
The program would realize that the ship further away would be in your spotting range, the pixels within it's theoretical silhouette would be also different from the ones of the empty horizon but the ship further away is still not supposed to be seen.

Also, if the sky has different colors all at once, which colour do you exactly use to incorporate this test. E.g. if the sky is taking on a scale of yellow to red colours during sunrise. Clouds might also offer a different colour palette. So simulating the contrast of ship <-> sky is not that easy. Ships can have multiple colours in their painting as well.
I'm not even talking about camo schemes here.

And a last but very important point:
ships are often spotted not by their own silhouettes but by there smoke trail on the horizon. And a smoke could is a thing that constantly changes form, thickness and contrast to the sky. They would be a pain to be simulated in terms of real vision especially.

In an already complex simulation you better stick with the more abstract schemes of making visual contact like they did in the games before and try to improve on this approach.

[rik007]

What would you do if there is another ship behind the first one ?
The program would realize that the ship further away would be in your spotting range, the pixels within it's theoretical silhouette would be also different from the ones of the empty horizon but the ship further away is still not supposed to be seen.

Also, if the sky has different colors all at once, which colour do you exactly use to incorporate this test. E.g. if the sky is taking on a scale of yellow to red colours during sunrise. Clouds might also offer a different colour palette. So simulating the contrast of ship <-> sky is not that easy. Ships can have multiple colours in their painting as well.
I'm not even talking about camo schemes here.

And a last but very important point:
ships are often spotted not by their own silhouettes but by there smoke trail on the horizon. And a smoke could is a thing that constantly changes form, thickness and contrast to the sky. They would be a pain to be simulated in terms of real vision especially.

In an already complex simulation you better stick with the more abstract schemes of making visual contact like they did in the games before and try to improve on this approach.

Then at least agree upon an implementation in which you can sneak into a convoy, surface, make you attack and get out of it with a great change to escape unhurt. In SH-III you can certainly approach a convoy but after launching an attack, the lights flip on and you are detected. A sensor.cfg parameter for the before and after an attack would give us the ability to mod the detection chances before/after the torpedos hit.

Remember that surface attacks will provide emersion and variation to the game. It will be a shame if it isn't there. I miss comparable with the missing of Wolfpacks.

[PL_Andrev]

On my example I wanted to show how to realize the ship detection in case of bad weather conditions (fog, rain, dusk, storm) similar to human eyes.

Indeed, to check all ships AI must follow up on all enemy ships in the area (AI vs. player, AI vs. AI), but remember that GPU does not need to render the whole scene (as you see in monitor), but the only small area with size f.e. 100 pixels - and not "all the time" but once every 10 seconds for each ship...

[ichso]

O.k. how would you want to use this approach for ai ships <-> ai ships anyways. In the 3D model of the game world the program has the coordinates of the objects/ships. But to know whether one ship can see the other you would need to computer the vector of the first ship in direction of the other and then compute what the background would be in this direction and compare that to how the second ship would be rendered on screen.

Either you explained your approach wrong, I understood it wrong or this is overly complicated, no matter if done every 1/100th second or done every 10 seconds.

Using the player's POV you at least already have the rendered scene to use but even there, there would be too many factors to include to get satisfying results, I think.

[Zonke]

On my example I wanted to show how to realize the ship detection in case of bad weather conditions (fog, rain, dusk, storm) similar to human eyes.

Indeed, to check all ships AI must follow up on all enemy ships in the area (AI vs. player, AI vs. AI), but remember that GPU does not need to render the whole scene (as you see in monitor), but the only small area with size f.e. 100 pixels - and not "all the time" but once every 10 seconds for each ship...

I see, seams more plausible now I suppose with the 10sec interval and small render. It would ensure the AI has the same conditions as the human player..

Whatever the method, I hope they improve it. I like the ideas of ships being on different states of alert at different times (based on where they are, crew size/skill, time in the war and ofcourse whether they've been attacked recently)..

[timmyab]

I've read in 'Das Boot' and 'Iron Coffins' where the U-boat captain has used the submarine's exhaust fumes to create a kind of smoke screen to hide from destroyers at night.It would be cool if you could do this.Also it would be nice if you could wait for the moon to go behind a cloud before attacking. I'd like to see the whole business of whether you're detected or not to be far more randomised than it was in SH3, something I believe computers find difficult to do, with the advantage being decidedly with the sub in the early years at least.

[JScones]

I've read in 'Das Boot' and 'Iron Coffins' where the U-boat captain has used the submarine's exhaust fumes to create a kind of smoke screen to hide from destroyers at night.It would be cool if you could do this.Also it would be nice if you could wait for the moon to go behind a cloud before attacking. I'd like to see the whole business of whether you're detected or not to be far more randomised than it was in SH3, something I believe computers find difficult to do, with the advantage being decidedly with the sub in the early years at least.
Do diesel engines actually produce that much smoke? I know they throw a bit of soot on startup, but I thought otherwise, when maintained, they ran quite smoke free? :hmmm:

[TarJak]

You can make a well maintained and smooth running diesel throw out thick black smoke by fiddling with the fuel plate. No harm to the engine, but less fuel efficient.

In normal operation though it should be thin blue smoke that is hard to see unless you are accelerating hard.

[timmyab]

In the books they went full ahead and stern on to produce this screen.I just thought that as it was mentioned in both books that in might have been a common device.

[Malmer]

During the Snorkel development exhaust modifications were dropped just because the diesel fumes were barely visible...

[JU_88]

Do diesel engines actually produce that much smoke? I know they throw a bit of soot on startup, but I thought otherwise, when maintained, they ran quite smoke free? :hmmm:

This ^^
....and i woulndn't take anything from Das Boot or Iron coffins as the Gospel truth.
Both are 'based' on facts - but designed primarily for entertainment.

[looney]

even so it is easier to make a engine produce smoke than to make an engine produce no smoke while still running.

If the Kaleu wanted to produce a smoke screen just add a bit of oil in the exhaust. Easy peasy

[Malmer]

Neither would I vote for an option to blind enemy lookouts with some magic flare... It might seem a good idea, but chances are this will backfire.

The same for a smoke screen; you might be invisible but it can also attract unwanted attention or worse--block your own visibility.

[TarJak]

I'd say smokescreening whilst it was most likely not common practise would only be used in dire emergency when trying to get up enough speed to submerge and dive faster.

The best practise would be not to get into the situation where you have to use it.

Should it be in the game? TBH I wouldn't care if it was or not.

[Steeltrap]

Of the large mods - and I played both NYGM and GWX - I found NYGM to have a better simulation of night surface attacks.

Before radar started appearing with any regularity, it was entirely possible in NYGM to get close on the surface to the expected track of the target then flee after firing. I did so from as close as 800yds.

It was possible in GWX, but I found I was more likely to be detected by ships that really shouldn't have been able to do so. It was still better than 'stock' SHIII, mind you, only I feel NYGM was better still in this respect.

I remember Dick O'Kane in Clear the Bridge pointing out the effectiveness of the camo scheme on a fleet boat when he met with another boat at sea and could only just make it out at 500yds when he knew where to look. He was making successful surface attacks up to - and indeed the night of - the time Tang was taken out by her own torp (Sep '44? can't remember exact date).

One thing I found unrealistic was the mass searchlight frenzy that happened when a target was struck. Convoys kept strict darkness discipline, even when attacked. Pointing searchlights hither and thither did little other than advertise the precise location of ships for many miles at night (including beyond the horizon), and allow any potential attackers to track targets thanks to the fact that the lights gave away far more than they revealed. My understanding was that only escorts used searchlights with any regularity, and even then only in certain circumstances.

This whole issue of surface visibility/detection is one of THE crucial factors they need to get 'right' (in fact I've always found stock SH games to be unrealistic - sometimes to extremes - in this regard).

Cheers all

[rik007]

Of the large mods - and I played both NYGM and GWX - I found NYGM to have a better simulation of night surface attacks.

Before radar started appearing with any regularity, it was entirely possible in NYGM to get close on the surface to the expected track of the target then flee after firing. I did so from as close as 800yds.

It was possible in GWX, but I found I was more likely to be detected by ships that really shouldn't have been able to do so. It was still better than 'stock' SHIII, mind you, only I feel NYGM was better still in this respect.

I remember Dick O'Kane in Clear the Bridge pointing out the effectiveness of the camo scheme on a fleet boat when he met with another boat at sea and could only just make it out at 500yds when he knew where to look. He was making successful surface attacks up to - and indeed the night of - the time Tang was taken out by her own torp (Sep '44? can't remember exact date).

One thing I found unrealistic was the mass searchlight frenzy that happened when a target was struck. Convoys kept strict darkness discipline, even when attacked. Pointing searchlights hither and thither did little other than advertise the precise location of ships for many miles at night (including beyond the horizon), and allow any potential attackers to track targets thanks to the fact that the lights gave away far more than they revealed. My understanding was that only escorts used searchlights with any regularity, and even then only in certain circumstances.

This whole issue of surface visibility/detection is one of THE crucial factors they need to get 'right' (in fact I've always found stock SH games to be unrealistic - sometimes to extremes - in this regard).

Cheers all

Wow Steeltrap! You couldn't say it better. The searchlight switching on is ver unrealistic. I never read it was done that way. I do not even think merchants had searchlights and certainly didnot use them for the reason you gave. Nightly surface attacks are as much important as Wolfpacks and very easily overlooked.

[JScones]

One thing I found unrealistic was the mass searchlight frenzy that happened when a target was struck. Convoys kept strict darkness discipline, even when attacked. Pointing searchlights hither and thither did little other than advertise the precise location of ships for many miles at night (including beyond the horizon), and allow any potential attackers to track targets thanks to the fact that the lights gave away far more than they revealed. My understanding was that only escorts used searchlights with any regularity, and even then only in certain circumstances.

This whole issue of surface visibility/detection is one of THE crucial factors they need to get 'right' (in fact I've always found stock SH games to be unrealistic - sometimes to extremes - in this regard).
Yeah, I start to feel a bit like Susan Boyle with all the spotlights suddenly aimed at me. If I'm not misteaken, merchants would be trying to get away from the threat, not make themselves a greater target.

But aren't searchlights objects that can be added or removed by modders? :hmmm:

[Snestorm]

Ja, no more DDs with searchlights.

Dog Zebra = Darken ship. Set Material Condition Zebra.
No light showing.
Everything watertight gets (and stays) closed.

[rik007]

Yeah, I start to feel a bit like Susan Boyle with all the spotlights suddenly aimed at me. If I'm not misteaken, merchants would be trying to get away from the threat, not make themselves a greater target.

But aren't searchlights objects that can be added or removed by modders? :hmmm:

We never did mod that? Didn't we? Remove all spotlights from the ships and then leaving the possibility to be detected up to radar would be the spec:

http://www.subsim.com/radioroom/showthread.php?t=158891