Who knows about making AI ships fire torpedoes? your services are needed! - Page 3

polyfiller · 10-21-13, 02:17 PM

Jhapprich - thanks - just downloaded your files ... if I read the files correctly the approach is similar to what Darkfish originally released - a hidden gun turret with hidden turret / barrels (to help with tracking target ?) and the fire effect in the the .sim file for torplauncher93 in guns_radars.sim is - 0x38e9700135eb5250 which calls the node $Torp_Spawner_PG in torpspawngunfx.dat - which spawns 0x8f346dcc43bfbcb2 as a single particle - which I presume is a torpedo ( your zip file doesn't include torpedoes_us.dat .. but in my that corresponds too a US MK14 torpedo) ?

So the difference I can see betwen your files and Darkfish is that you call the ID of $Torp_Spawner_PG as the fire effect whereas darkfish calls the ID of it's parent node - $Torp_Spawner (0x6d21707950f0ac3c). Does this affect how the torpeo aims / fires ? I also note that you try to limit the range of the launcher by specifying 1000m as the range in the guns_radars.sim file. I did extensive testing of this and found the range specification has no effect whatsoever - I have a test mission with AI subs hunting a convoy at ranges varying from 1000M out to 6000m. The subs at 6000m fire just as soon as they can "see" a target and certainly take no notice of the range value in the .sim file. The range of the wepon (applies to all weapons according to my testing) is the shell speed - and you have this set to 2000 m/s. I have multiple launchers for submarines set to different shell speeds to try and force the AI to "lead" a moving target adequatley. I have a range of launchers with shell speeds from 10 m/s 12 m/s, 14 m/s and 16 m/s. I found this gave some sort of spread to the launch of torpedoes (4 forward launching launchers on a sub, one for each speed) ... however the AI's ability to use this to lead the target provded problematic and the values for the parent invisible turret had a bearing on how this works - for example I also set a .sim file entry for the hidden turret - and that's where I specify the 10 m/s shell speed.... this affects how the main launcher mount rotates / tracks the target, I then set a .sim entry fo the actual torp launcher barrel mounted on the turret with a shell speed of 1000000000000 (lots). This seemed to actually make the AI track a moving target at anything other than short (1000m or less) range.

I can understand that setting the torp speed to something unrealistically high would improve accuracy - but are your launchers actually leading the targets to try and aim a slow projectile, or is the fast torpedo speed compensating for almost no target leading by the AI ?

I'll load up and run your test missions.

The good news is we are all using the same sorts of approach and looks like we are all trying to compensate for the AI's inability to launch a torpedo / lead a target at realistic ranges.

Will post more later.

Thanks again for the upload / sharing :-)

polyfiller · 10-21-13, 03:09 PM

OK, completed the Fubuki test mission - and also modified it a bit to really see if the launcher can shoot by leading the target - NOTE : the speed for the Mk14 in my torpedoes_us.dat is stock. I removed all but one of the Fubuki's and set it to run slowly - so that it couldn't use it's speed to achieve an easy shooting angle. I had it running alongside the convoy, "broadside" style shooting. I also eeked out the range a little - and noticed that when I went out to about 3500m the Fubuki didn't engage - either guns or torps ... must be AI-sensor setting.... was getting excited for a second thinking that the Fubuki was taking note of the range of the topr launcher - but no, it also failed to fire it's guns, and when it clsoed to around 1700m it engaged..... and failed to register a single torpedo hit.

No while a destoyer can thump around at high speed and eventually get close enough ... in a mission .... in campaign it will try to engage at longer ranges (will with my sensors.dat anyhow) ... and wil potentiallye xhaust all it's torp ammo (unless we set a high ammo store for the launcher as a workaround).. What I found with my destoyer test missions was that when they engage at, say 6000m, the try launching torps all over the place at rediculous angles and if I had 3 to 4 destroyers running in a fleet together, they would invariably end up torpedoing each other via their rapdi speed and changes of angles in closing the distance to target, combined with poor torp aiming.

I can confirm that right now, I reckon my torp launcher config leads the targets better than yours. I'll see how quickly I can upload my full TSWSM development version so you can take a look - would take me too long to extract just the torp stuff - Ihave about 10 Ai subs, loads of AI launchers etc and probably simpler if I upload the entire mod.

merc4ulfate · 10-21-13, 08:24 PM

Quote:

Originally Posted by jhapprich

jap. here is the download.

link: http://www.gamefront.com/files/23782...dolauncher.zip

Does the AITorpedolauncher folder go into the single mission folder of the game to run this test?

Admiral Halsey · 10-21-13, 09:35 PM

Quote:

Originally Posted by merc4ulfate

Does the AITorpedolauncher folder go into the single mission folder of the game to run this test?

Use JGSME to run it. That's what I did and it worked fine.

merc4ulfate · 10-21-13, 10:04 PM

Thank you Admiral

Admiral Halsey · 10-21-13, 11:16 PM

Quote:

Originally Posted by merc4ulfate

Thank you Admiral

No problem.

10-22-13, 01:58 AM

@polyfiller: assigning the particle generator but not the parent node happened unintentionally and does not effekt anything. torpedo is launched as objekt particle, the parent effect is the torpedospawnfx.this effect replaces the muzzle flash, while darkfish used a cartridge effect. the gun still fires an invisible dummy shell with no damage at very high speed to have a stretched firing angle. the gun range of darkfish`s launcher cannot effect firing range, right, because the torpedo launching of his is independent from the gun. this is what i changed. my launcher actually fires a torpedo within the limitations you give it. you can adjust whether the torpedo will go staright below the surface, or you can have it dive within a specified angle(gun elevation) the range limitation to 1000m works, i have my fubukis fire their guns earlier than torpedos (however this does not always work, so i guess starting fire is depending on the weapon_cannon with the shortest range. i choose 1000m, since this is a reasonable distance to hit a moving target. further away, the launchers won`t fire, so range limitation works.the fubukis wont close in, right, because they are destroyers and the game is designed to have dd`s attack subs-this is when they will show their full offensive potential. and remember this is actually still a gun. changing the ai´s visual sensor range might help, havent tried this yet. but a good approach is to use very fast torpedos, this increased the sinkining rapidly. there is no effect in the placing of the weapon further to the bow or stern. hope ill have your files soon. best regards
ps what do you mean with "leading the target"?

merc4ulfate · 10-22-13, 03:34 PM

I just ran this and I am pleased. The Japanese destroyers on the outskirts of the convoy do indeed close in to attack the convoy. Now between the escorts that are there and the merchant deck guns they get completely obliterated but the torpedoes is what I was looking at most.

My feelings on the torpedoes is that it adds a delightful aspect to the game. I find the number of fish fired by the destroyers is realistic but I do notice that some fish run closer to the surface and are more apt to strike their intended targets while some run very deep and never hit anything at all. The fish that run deep tend to have about a 30-45 degree nose up orientation while running laterally in a straight line and appear to be at least 10 meters, if not more, below the keel of the vessels. Those to deep never explode.

From what I saw I would say about 50% are running too deep but over all I sure do love the look of this.

=======================

Test two:

First two ships fired only 4 fish before being destroyed. Second two ships fired none before being destroyed. 1 out of 4 fish ran deep.

Test three:

5 fish total from ship one all others 0. 2 out of 3 fish ran deep but I noticed they started out shallower and got deeper as they went. The depth was no consistent and varied from fish to fish.

polyfiller · 10-22-13, 07:03 PM

That is interesting - torps runing deep ... my setup the torps never run too deep.

Jhapprich - read your feedback - I think you might be confusing Darkwraith with Darjfish (or are they the same person) .... Darkfish's launcher always used a partile efect to spawn the torp ... so essentially same as what you have. Great minds thik alike I guess :-)

I've sent a message to our overlord to get permission to upload the TSWSM develoment mod ... then you'll get to see / test my config. for the torps.

10-23-13, 02:17 AM

really looking forward to this.yes, i meant darkfish. i now have an ai alfa, actually listed as unittype 200,that runs submerged, on its own power, with a submerged torpedo launcher. it attacks enemy targets, but it coul be a little more aggressive. any idea? very interested in your implementation of homing torpedos. would mean you would not use a particle generator, hm? ps treid the airtorpedo, which will not run , but stay on its spawnpoint like a mine, BUT it floats in the water straight, without nose up/down according to its defined depth.strange...

merc4ulfate · 10-23-13, 07:58 AM

This photo shows the initial aim and firing of the lanchers. A very nice animation for the lanchers to turn and fire.

This image shows an intial offset to the depth of the two fish just fired.

This image of the same two fish shows that the offset has left one near the surface but the other is now much deeper and would pass under a vessel.

Both the fish missed all targets but what I wanted to note here was that I followed only these two to the end of their fuel run. By the time they both ran out of fuel the separation in depth was so great that they can barly be seen in the same frame. The upper red circled one is still near the surface while the lower blue circled one is at least 100 meters below the surface. While they both ran straight one was affected but something which cause it to run deep. I wonder if the roll of the firing vessle from the first fish fireing, guns firing or sea state affected the trajectory of the second?

These two were fired at the same time and towards the same trajectory but the fish on the left took a port side turn perpendicular to the fish on the right.

All three of these fish ran deep with two going much deeper after firing. Notice the nose up attitude of the fish. They would run straight but their nose would remain turned up anywhere from 20 to 40 degrees.

My last test had the four desroyers firing over 50 fish with an estimated 50% running deep at varying levels. The "run to deep" nose up attitude and depth is not consistant which makes me wonder how much of this has to do with the ships roll on firing. The nose up attitude may be the fish trying to acheive its programed depth but unless programed to asend they will not. The downward trajectory may be causing it to remain on it's downward angle.

10-23-13, 08:41 AM

ok, took a look at your screenshots. first of all, are you using any mod that change your stock torpedos? 2nd, the different depth between the torpedos is caused by the launchers ability to raise/lower the barrel +/- 1 degree. the nose up/down caused by the programmed torpedo depth in the.sim of the torpedo, so is the different angle of some torpedos relatively to it`s curse. the reason is, that when firing the launcher, that is actually a cannon like the other gunturrets, the barrel aims towards the estimated target and shoots an invisble dummy shell. the torpedo is launched instead of the muzzle flash, this way i corrected the random targeting error of darkfish´s original spawner. so you have to think that way: the torpedo is a child of the parent node, the muzzle flash, and will always follow the vector of the parent, so it will run in the original firing direction +/-1 deg to the water level +tolerance. diving ones while be nose-up`s , as the torpedo will try to reach the running depth of his .sim; if there is a guidance enabled, it will also have a different horizontal orientation. this can best be seen using homing torpedos, they will turn towards the loudest soundsource but won`t reach it because curse and depth is dictatet by the muzzle flash the time the launcher fires. so what we can do is setting the elevation and tolerance both to "0", but i do not know if the launcher will properly fire(eg heavy whether, ship´s not steady-not time for targeting maybe?)

merc4ulfate · 10-23-13, 08:55 AM

This is my mod list

Generic Mod Enabler - v2.6.0.157

OpsMonsun_V705
OMv705_to_V720
OMv720_Patch5
OMEGU_v300
OMEGU_v300_Patch7
OM_Harder_Escort_L2
UBM Missions
Popular German Music From The 30´s
OM_Pac_Env
AITorpedolauncher

10-23-13, 09:19 AM

ok, you should try stock first. ps setting the elevation min/max to 0 or +/-0.01 and the elevation tolerance alike will cause the launcher not to fire at all. i just changed the date in the destroyers .sim/obj_turret and in the launchers sim in guns_radars as followed: elevation min0,max 1, fire 0/1 elevation min 0, max 1; weapon_cannon/ elevation min0,max 1, elevation tolerance 0,01. the torpedos are running virtually at the same depth now, almost no nose-up. ill give elevation tolerance 0 a try. curiusly, only the middle launcher of the fubuki would go lower at me.

hmm...

ps. elevation tolerance 0 doesn`t work though, 0.01 works fine for me. ill add an magnetic detonator to the assigned torpedo instead
pss. tried the same with 0.001, only random noseups with the middle launcher. best solution for now

merc4ulfate · 10-23-13, 10:13 AM

I quit playing the stock game some years ago. It just isn't a challenge.

I think the magnatic detonator would help a lot with any running a bit deeper than normal.

10-22-13, 01:58 AM	#37
jhapprich Stowaway Posts: n/a Downloads: Uploads:	@polyfiller: assigning the particle generator but not the parent node happened unintentionally and does not effekt anything. torpedo is launched as objekt particle, the parent effect is the torpedospawnfx.this effect replaces the muzzle flash, while darkfish used a cartridge effect. the gun still fires an invisible dummy shell with no damage at very high speed to have a stretched firing angle. the gun range of darkfish`s launcher cannot effect firing range, right, because the torpedo launching of his is independent from the gun. this is what i changed. my launcher actually fires a torpedo within the limitations you give it. you can adjust whether the torpedo will go staright below the surface, or you can have it dive within a specified angle(gun elevation) the range limitation to 1000m works, i have my fubukis fire their guns earlier than torpedos (however this does not always work, so i guess starting fire is depending on the weapon_cannon with the shortest range. i choose 1000m, since this is a reasonable distance to hit a moving target. further away, the launchers won`t fire, so range limitation works.the fubukis wont close in, right, because they are destroyers and the game is designed to have dd`s attack subs-this is when they will show their full offensive potential. and remember this is actually still a gun. changing the ai´s visual sensor range might help, havent tried this yet. but a good approach is to use very fast torpedos, this increased the sinkining rapidly. there is no effect in the placing of the weapon further to the bow or stern. hope ill have your files soon. best regards ps what do you mean with "leading the target"? Last edited by jhapprich; 10-22-13 at 02:32 AM.

10-22-13, 03:34 PM	#38
merc4ulfate DILLIGAF Join Date: Feb 2007 Location: florida Posts: 2,058 Downloads: 210 Uploads: 0	I just ran this and I am pleased. The Japanese destroyers on the outskirts of the convoy do indeed close in to attack the convoy. Now between the escorts that are there and the merchant deck guns they get completely obliterated but the torpedoes is what I was looking at most. My feelings on the torpedoes is that it adds a delightful aspect to the game. I find the number of fish fired by the destroyers is realistic but I do notice that some fish run closer to the surface and are more apt to strike their intended targets while some run very deep and never hit anything at all. The fish that run deep tend to have about a 30-45 degree nose up orientation while running laterally in a straight line and appear to be at least 10 meters, if not more, below the keel of the vessels. Those to deep never explode. From what I saw I would say about 50% are running too deep but over all I sure do love the look of this. ======================= Test two: First two ships fired only 4 fish before being destroyed. Second two ships fired none before being destroyed. 1 out of 4 fish ran deep. Test three: 5 fish total from ship one all others 0. 2 out of 3 fish ran deep but I noticed they started out shallower and got deeper as they went. The depth was no consistent and varied from fish to fish. Last edited by merc4ulfate; 10-22-13 at 05:11 PM.

10-23-13, 07:58 AM	#41
merc4ulfate DILLIGAF Join Date: Feb 2007 Location: florida Posts: 2,058 Downloads: 210 Uploads: 0	This photo shows the initial aim and firing of the lanchers. A very nice animation for the lanchers to turn and fire. This image shows an intial offset to the depth of the two fish just fired. This image of the same two fish shows that the offset has left one near the surface but the other is now much deeper and would pass under a vessel. Both the fish missed all targets but what I wanted to note here was that I followed only these two to the end of their fuel run. By the time they both ran out of fuel the separation in depth was so great that they can barly be seen in the same frame. The upper red circled one is still near the surface while the lower blue circled one is at least 100 meters below the surface. While they both ran straight one was affected but something which cause it to run deep. I wonder if the roll of the firing vessle from the first fish fireing, guns firing or sea state affected the trajectory of the second? These two were fired at the same time and towards the same trajectory but the fish on the left took a port side turn perpendicular to the fish on the right. All three of these fish ran deep with two going much deeper after firing. Notice the nose up attitude of the fish. They would run straight but their nose would remain turned up anywhere from 20 to 40 degrees. My last test had the four desroyers firing over 50 fish with an estimated 50% running deep at varying levels. The "run to deep" nose up attitude and depth is not consistant which makes me wonder how much of this has to do with the ships roll on firing. The nose up attitude may be the fish trying to acheive its programed depth but unless programed to asend they will not. The downward trajectory may be causing it to remain on it's downward angle. Last edited by merc4ulfate; 10-23-13 at 08:12 AM.

10-21-13, 02:17 PM	#31
polyfiller Weps Join Date: Sep 2005 Posts: 356 Downloads: 40 Uploads: 0	Jhapprich - thanks - just downloaded your files ... if I read the files correctly the approach is similar to what Darkfish originally released - a hidden gun turret with hidden turret / barrels (to help with tracking target ?) and the fire effect in the the .sim file for torplauncher93 in guns_radars.sim is - 0x38e9700135eb5250 which calls the node $Torp_Spawner_PG in torpspawngunfx.dat - which spawns 0x8f346dcc43bfbcb2 as a single particle - which I presume is a torpedo ( your zip file doesn't include torpedoes_us.dat .. but in my that corresponds too a US MK14 torpedo) ? So the difference I can see betwen your files and Darkfish is that you call the ID of $Torp_Spawner_PG as the fire effect whereas darkfish calls the ID of it's parent node - $Torp_Spawner (0x6d21707950f0ac3c). Does this affect how the torpeo aims / fires ? I also note that you try to limit the range of the launcher by specifying 1000m as the range in the guns_radars.sim file. I did extensive testing of this and found the range specification has no effect whatsoever - I have a test mission with AI subs hunting a convoy at ranges varying from 1000M out to 6000m. The subs at 6000m fire just as soon as they can "see" a target and certainly take no notice of the range value in the .sim file. The range of the wepon (applies to all weapons according to my testing) is the shell speed - and you have this set to 2000 m/s. I have multiple launchers for submarines set to different shell speeds to try and force the AI to "lead" a moving target adequatley. I have a range of launchers with shell speeds from 10 m/s 12 m/s, 14 m/s and 16 m/s. I found this gave some sort of spread to the launch of torpedoes (4 forward launching launchers on a sub, one for each speed) ... however the AI's ability to use this to lead the target provded problematic and the values for the parent invisible turret had a bearing on how this works - for example I also set a .sim file entry for the hidden turret - and that's where I specify the 10 m/s shell speed.... this affects how the main launcher mount rotates / tracks the target, I then set a .sim entry fo the actual torp launcher barrel mounted on the turret with a shell speed of 1000000000000 (lots). This seemed to actually make the AI track a moving target at anything other than short (1000m or less) range. I can understand that setting the torp speed to something unrealistically high would improve accuracy - but are your launchers actually leading the targets to try and aim a slow projectile, or is the fast torpedo speed compensating for almost no target leading by the AI ? I'll load up and run your test missions. The good news is we are all using the same sorts of approach and looks like we are all trying to compensate for the AI's inability to launch a torpedo / lead a target at realistic ranges. Will post more later. Thanks again for the upload / sharing :-)

10-21-13, 03:09 PM	#32
polyfiller Weps Join Date: Sep 2005 Posts: 356 Downloads: 40 Uploads: 0	OK, completed the Fubuki test mission - and also modified it a bit to really see if the launcher can shoot by leading the target - NOTE : the speed for the Mk14 in my torpedoes_us.dat is stock. I removed all but one of the Fubuki's and set it to run slowly - so that it couldn't use it's speed to achieve an easy shooting angle. I had it running alongside the convoy, "broadside" style shooting. I also eeked out the range a little - and noticed that when I went out to about 3500m the Fubuki didn't engage - either guns or torps ... must be AI-sensor setting.... was getting excited for a second thinking that the Fubuki was taking note of the range of the topr launcher - but no, it also failed to fire it's guns, and when it clsoed to around 1700m it engaged..... and failed to register a single torpedo hit. No while a destoyer can thump around at high speed and eventually get close enough ... in a mission .... in campaign it will try to engage at longer ranges (will with my sensors.dat anyhow) ... and wil potentiallye xhaust all it's torp ammo (unless we set a high ammo store for the launcher as a workaround).. What I found with my destoyer test missions was that when they engage at, say 6000m, the try launching torps all over the place at rediculous angles and if I had 3 to 4 destroyers running in a fleet together, they would invariably end up torpedoing each other via their rapdi speed and changes of angles in closing the distance to target, combined with poor torp aiming. I can confirm that right now, I reckon my torp launcher config leads the targets better than yours. I'll see how quickly I can upload my full TSWSM development version so you can take a look - would take me too long to extract just the torp stuff - Ihave about 10 Ai subs, loads of AI launchers etc and probably simpler if I upload the entire mod.

10-21-13, 10:04 PM	#35
merc4ulfate DILLIGAF Join Date: Feb 2007 Location: florida Posts: 2,058 Downloads: 210 Uploads: 0	Thank you Admiral

10-22-13, 07:03 PM	#39
polyfiller Weps Join Date: Sep 2005 Posts: 356 Downloads: 40 Uploads: 0	That is interesting - torps runing deep ... my setup the torps never run too deep. Jhapprich - read your feedback - I think you might be confusing Darkwraith with Darjfish (or are they the same person) .... Darkfish's launcher always used a partile efect to spawn the torp ... so essentially same as what you have. Great minds thik alike I guess :-) I've sent a message to our overlord to get permission to upload the TSWSM develoment mod ... then you'll get to see / test my config. for the torps.

10-23-13, 02:17 AM	#40
jhapprich Stowaway Posts: n/a Downloads: Uploads:	really looking forward to this.yes, i meant darkfish. i now have an ai alfa, actually listed as unittype 200,that runs submerged, on its own power, with a submerged torpedo launcher. it attacks enemy targets, but it coul be a little more aggressive. any idea? very interested in your implementation of homing torpedos. would mean you would not use a particle generator, hm? ps treid the airtorpedo, which will not run , but stay on its spawnpoint like a mine, BUT it floats in the water straight, without nose up/down according to its defined depth.strange...

10-23-13, 08:41 AM	#42
jhapprich Stowaway Posts: n/a Downloads: Uploads:	ok, took a look at your screenshots. first of all, are you using any mod that change your stock torpedos? 2nd, the different depth between the torpedos is caused by the launchers ability to raise/lower the barrel +/- 1 degree. the nose up/down caused by the programmed torpedo depth in the.sim of the torpedo, so is the different angle of some torpedos relatively to it`s curse. the reason is, that when firing the launcher, that is actually a cannon like the other gunturrets, the barrel aims towards the estimated target and shoots an invisble dummy shell. the torpedo is launched instead of the muzzle flash, this way i corrected the random targeting error of darkfish´s original spawner. so you have to think that way: the torpedo is a child of the parent node, the muzzle flash, and will always follow the vector of the parent, so it will run in the original firing direction +/-1 deg to the water level +tolerance. diving ones while be nose-up`s , as the torpedo will try to reach the running depth of his .sim; if there is a guidance enabled, it will also have a different horizontal orientation. this can best be seen using homing torpedos, they will turn towards the loudest soundsource but won`t reach it because curse and depth is dictatet by the muzzle flash the time the launcher fires. so what we can do is setting the elevation and tolerance both to "0", but i do not know if the launcher will properly fire(eg heavy whether, ship´s not steady-not time for targeting maybe?)

10-23-13, 08:55 AM	#43
merc4ulfate DILLIGAF Join Date: Feb 2007 Location: florida Posts: 2,058 Downloads: 210 Uploads: 0	This is my mod list Generic Mod Enabler - v2.6.0.157 OpsMonsun_V705 OMv705_to_V720 OMv720_Patch5 OMEGU_v300 OMEGU_v300_Patch7 OM_Harder_Escort_L2 UBM Missions Popular German Music From The 30´s OM_Pac_Env AITorpedolauncher

10-23-13, 09:19 AM	#44
jhapprich Stowaway Posts: n/a Downloads: Uploads:	ok, you should try stock first. ps setting the elevation min/max to 0 or +/-0.01 and the elevation tolerance alike will cause the launcher not to fire at all. i just changed the date in the destroyers .sim/obj_turret and in the launchers sim in guns_radars as followed: elevation min0,max 1, fire 0/1 elevation min 0, max 1; weapon_cannon/ elevation min0,max 1, elevation tolerance 0,01. the torpedos are running virtually at the same depth now, almost no nose-up. ill give elevation tolerance 0 a try. curiusly, only the middle launcher of the fubuki would go lower at me. hmm... ps. elevation tolerance 0 doesn`t work though, 0.01 works fine for me. ill add an magnetic detonator to the assigned torpedo instead pss. tried the same with 0.001, only random noseups with the middle launcher. best solution for now

10-23-13, 10:13 AM	#45
merc4ulfate DILLIGAF Join Date: Feb 2007 Location: florida Posts: 2,058 Downloads: 210 Uploads: 0	I quit playing the stock game some years ago. It just isn't a challenge. I think the magnatic detonator would help a lot with any running a bit deeper than normal.