[wip]Visual sensor revamp for TMaru

[Ducimus]

I know its kinda silly to be tinkering with anything with a patch looming large, but i coudlnt restrain myself any longer, as the AI visual detection schema is driving me nucking futs.


D/L link
http://www.ducimus.net/sh4/release/TEST_TMaru_AI_Visuals_1.03.7z

-- Format wordrap

ver 1.03

//AirS_visual->MinSurface from -15 to -4.35


----------------------
ver 1.02

//AirS_visual from -38 to -15.

---------------------------

//ver 1.01

//AirS_Visual
- MinHeight from -18, back to 0
- MinSurface from 0 to -38


-----------------






// AI visuals Test ver 1.0 for Trigger Maru


WHAT IT'S SUPPOSED TO DO:
a.) Adds additional visual nodes to assign to varying unit types, so visual detection is more dynamic in game.

b.) To make night surface attacks a little more feasible against merchants.

c.) To allow airplanes the ability to detect shallow submarines

d.) Extend visual range of capital warships.



HOW IT WORKS:
a.) adds a dat file that contains additional nodes for the game to use.

b.) Additional Visual nodes are as follows:
- 1.) M_visual. Max range of 6500 meters, assigned to merchants, and other misc non military units.

- 2.) F_Visual. Max range of 12,500 meters. Assigned to capital warships such as cruisers, carriers, etc.

- 3.) AirS_Visual. Max range of 21,000 meters. Assigned to aircraft. This node also has a min height of -10 meters. My Initial testing indicates that a
player can be detected while at periscope depth within a 6,000 meter radius of the unit equiped with this node. Below periscope depth, player is unable to be detected visually.

-4.) A_Visual. Unchanged, max range of 8,500 meters. Only warships use this node now.



KNOWN ISSUES:
- Max visual distances assigned to the node, do not reflect acutal sensor range in game. For example, 8500 meters equates out to 8,000 meters or so.


MISC ERRATA:
- Sensor information on the nav map has been renabled to better facilitate testing.

- sub marker has been renabled to better facilitate testing.

- If anyone provides feedback, do so in METRIC, not imperial.

- This version is a rough framework, and is not fine tuned. WIth what testing ive done, ive centered it around a 9KM visual sensor on the AI, with a 12KM max visual distance on the scene.dat. (in otherwords, close to stock settings)


CREDITS:
- skwasjer , for his amazing tool.
- NYGM, the dat file in this test modlet is based on NYGM's.
- Tater, This work is inspired by Taters work on the same thing.





Again, this is a WIP. It is not fine tuned. To install, just overlay it on top of TM.

[tater]

Did you make sure to arm the aircraft with bombs?

I have aircraft attack in test missions is why I ask. Some of the loadouts are bogus, though.

tater

[Ducimus]

Did you make sure to arm the aircraft with bombs?

I have aircraft attack in test missions is why I ask. Some of the loadouts are bogus, though.

tater

In the editor i did, but otherwise no. He didnt even strafe me. I ended up testing it by putting the AirS_visual sensor on a escort, park him (not docked), then travel underwater at him and see when he decides to move his ass and chase me down. Was around 5000, 6000 meters. Outside of that he didnt see me. I then cruised under him at 100 feet, and he never saw me. Went back up to periscope depth after i passed him, and i think he picked me up at 67 feet, at around 3,000, meters or so.

edit: using a min height of -12 meters, and i could get him to detect me at periscope depth from a range of 9,000 meters

[tater]

One very cool addition with air visuals messed with would be to start using plotted air searches. Meaning planes on the map following waypoints.

Then have the airgroups set to whatever, but have the airstrike changce near zero---but have the chance if the player is detected quite high.

Very few random aicraft encounters, but if a "scripted" patrol finds you...

tater

[tater]

BTW, there are 2 very related things in the 1.4 changelog posted.

The 20km view distance, and perhaps more importantly the ability for ships to detect torpedos.

The latter could be a fundamantal change in the way visuals are handled, OTOH, it could also just be that they created a new "unit type" that can be spotted (a lot of the AI seems tied to Unit Type=).

It would be cool to know what the technique for wake detection is, perhaps it could be used for shallow subs...

tater

[tater]

- 3.) AirS_Visual. Max range of 21,000 meters. Assigned to aircraft. This node also has a min height of -10 meters. My Initial testing indicates that a
player can be detected while at periscope depth within a 6,000 meter radius of the unit equiped with this node. Below periscope depth, player is unable to be detected visually.


One, does the 6000m radius include altitude?

Two, I think that 6000m might be too far for this. The problem is that planes will be found in all types of water, not just shallows.

An idea:

Make a few aircraft visual nodes. One for single engine planes to differentiate between them and crewed maritime patrol aircraft. The single seat might detect to PD, but only at very very short range. Perhaps less deep than 12 m, too. The idea would be setting the range such that if it was dead calm they could at some short range, but in even typical seas they could not. Another would be for crewed planes, and could detect at PD, but also VERY short range only. Very short, usually impossible die to sea state.

Another node will be the one that detects submerged subs most effectively. For argument, take it as the 6000m version you have now. We then BP-clone the desired maritime patrol planes and give them this visual, and probably shorter range---they'll also get depth charges as bomb load :) . Now there are 2 versions of of the Betty, H6K, and H8K. One with sensors that can pick up a PD boat at VERY short range, and the other at longer range. We then make special airgroups with the PD-detecting visual equipped planes, and place the bases such that they only cover SHALLOW WATERS.

We can then simulate the specific waters mattering. Anywhere near Java, etc, will be a nasty, dangerous place, for example. The mid pacific? Far less dangerous at PD.

Note that the PD-detecting airgroups can have a few regular planes in them, and the regular airgroups can have a few of the PD-detecting versions, too. Keeps everyone guessing.

tater

[Ducimus]

The 20km view distance, and perhaps more importantly the ability for ships to detect torpedos. The latter could be a fundamantal change in the way visuals are handled,


I think your right, that the 20KM distance will probaly change the way visuals are fundamentally handled. When testing with the 9KM , i think i ran into some of the problems you were encountering. It seemed like the variables in the sim.cfg didnt do as much as the range on the sensor itself. For example, a shorter max range, seemed to let me get in closer at night. I almost want to say some parmeters of the AI visuals seem, or feel hardcoded.

Torpedo detection, i think thats a hardcoded or code level change. Im not sure, but thats what my hunch is. I don't think it will effect the moddable files at all. If you remember version 1.1's sim.cfg, they were the exact same as stock SH3. And in SH3, the AI detected steam fish when they got in close. This is what makes me think the files we work with are irrelvant.


One, does the 6000m radius include altitude?


Nope, it sure doesn't. Thats one reason why i started another patrol last night to test it out, and posted the files incase someone wanted to try it out. I have no idea what it might do at height, im not even sure if the game takes height into account at all.


Two, I think that 6000m might be too far for this. The problem is that planes will be

Well the idea isn't to pester the player in shallow water, only to make him visible to aircraft if hes at periscope depth, which by all accounts is accurate. What im having trouble understanding is how the minheight is acutally working. What i mean is, periscope depth for a Tambor, gar, or gato is 18 meters. So why is -10 working? You'd think you'd need to set it at -18 or so. This has me wondering how the game measures depth.

From the bottom of the subs keel? It could be that at perisope depth, the AI is only seeing the conning tower. The surface area on that is probably what "trips" it at a given range. The effects of range and suface factor is another reason why im not terribly worried about height. Not to mention that planes travel so fast, that they might fly right by it before the detection time has expired. I may infact have to decrease the minsurface even more, but thats just tinfoil hat.


RE: Multiple aircraft nodes.
:hmm: not a bad idea you have there.

[tater]

Yeah, that makes sense (depth).

BP-cloning makes it not only easy to add new stuff, but the load required on the system isn't much.

I really think many of the air issues might be solved by many alternate types of planes. Right now I have CV airgroups with just a very few planes, and they are BP-clones with the range dropped to ~40km. That keeps the CAP actually flying CAP, and not super long-ranged maritime patrol.

So adding a few different AI_Sensor hard-coded skill levels is EASY. Put 1 of each plane in an airgroup, and 3 different AI sensors, and you have 3 different "skill levels" of AI capability under your control (plus crew skill levels). The same can be done with ships. The only problem there is adding new ships, even BP-clones adds to the rec manual (exception is if you make your own type up, but no escorts can be a made up type, they won't attack with DCs).

Still, I like the idea of making a new type, like UnitType=16, and BP cloning a bunch of the larger than DD ships to it. You'd instantly have 2 versions of each ship, and you could make some the stock unit type, others the new unit types with different everything.

[Hitman]

So why is -10 working? You'd think you'd need to set it at -18 or so. This has me wondering how the game measures depth.

From the bottom of the subs keel? It could be that at perisope depth, the AI is only seeing the conning tower. The surface area on that is probably what "trips" it at a given range. The effects of range and suface factor is another reason why im not terribly worried about height. Not to mention that planes travel so fast, that they might fly right by it before the detection time has expired. I may infact have to decrease the minsurface even more, but thats just tinfoil hat.


I believe the game computes the amount of surface (Using the surface factor as correction) of the 3D model that is INSIDE the detection area of a sensor. Probably setting 18 metres depth would make the sim compute the whole sub hull, conning tower and periscope shears, while 10 metres make it only compute what falls in that area, and less would only make it compute the periscope that rises above, effectively making it as hard to spot as if nothing else was visible (Like viewed from a destroyer f.e.):hmm:

[tater]

I bet that is correct, hitman.

[Ducimus]

Air visual needs more work. They're flying right by me as close as 2,000 - 3,000 meters and not seeing me.

detection time in sim.cfg is 0.5, or half a minute. I have to be exposed to them for 30 seconds in order to be detected. Maybe they're flying too fast and are not in the locale for that long, which is most likely the cause.

So i need to extend the time im exposed, without upping the detection time in the cfg because that effects ALL nodes, not just this one.


:hmm:

So my options are:
increase Sensitivty in node?
Lower the minheight to increase radius of exposure?
Use minsurface variable in node properties, assuming the game uses that value over whats in the sim.cfg?

edit:

Only logical thing to do at this point is increase the surface area exposed via lowering the min height to -18 and see what happends.

[tater]

You can also just increase the range, right?

The plane needs to be in range for a longer time period. Set the max range ridiculously high.

A BB should have a horizon range at ~35,000m. Make a planes max detection range 99,999 and see what happens :)

tater

[Ducimus]

Something isnt right with the AirS_visual node. Maybe its the enviormental variables, but for some odd reason, its behaving like patch 1.2. They are simply not seeing me. At the time it was 15 kt winds with a light fog, but as close as they came, they should have seen me regardless.


I know this node works,because i put it on a warship, and it detected me fine. But place it on a plane, and all of a sudden its not working. I'll have to try experimenting with the surface factor and sensitivty in the node itself.

[Ducimus]

More testing. I set the min height back to 0, and he detects me again. Not sure why. Ive also noticed that regardless of max visual distance or if i use the default AI visual node or not, i still get detected (Ie, he turns towards me) at about the same distance.

Im beginning to think this is a function of surface factor, sensitivity, or wave factor (saved game is in 15 kt wind, light fog).

I'll have to try with a lower wave factor, and lower surface factor and see if he spots me sooner. Failing that, trying a sensiivty of say 0.03, or 0.05.

edit: Not to self, try using a surface factor of 30 or 35 for starters, and go up or down from there. Also, lower the living hell out of the wave factor, to say 0.5 or 0.75

[Hitman]

I know this node works,because i put it on a warship, and it detected me fine. But place it on a plane, and all of a sudden its not working.

Since the obvious difference is the heigth of the observer, why don't you try to script a plane that flies very low (At the same heigth as the BB visual node is located) and see what happens? The other reason for this could be the speed :hmm:

[vanjast]

I know this node works,because i put it on a warship, and it detected me fine. But place it on a plane, and all of a sudden its not working.
While looking at the mapping side of things I've noticed that one (in camera view) cannot go higher than about 1000ft (300m) in altitude. It seems they might be using a quake equivalent 3D engine.

This might be the cause of the anomolies that your experiencing and maybe a plane flying at ~800ft might solve the problem.

[Ducimus]

Success, i got a plane to attack me at periscope depth in career mode. Now i just have to adjust the variables so their sane. MinHeight is not the variable to use, its MinSurface. Adjustments in the sim.cfg don't seem to do a whole lot.

[tater]

I would be really nice to get a definitive dev answer on what legacy SH3 files (the bulk of the contents of the Cfg folder) are actually used, and under what conditions.

tater

[Ducimus]

I would be really nice to get a definitive dev answer on what legacy SH3 files (the bulk of the contents of the Cfg folder) are actually used, and under what conditions.

tater


Up until patch 1.3, the sim.cfg could be adjusted with predictable results. Not so now. We know AI aircraft was having trouble spotting things in patch 1.2 (although it worked in patch 1.1). I can't shake the feeling something fundamentally changed in patch 1.3. Adjustments to the sim.cfg now, don't seem to effect things all that much, if at all. Although i hesitate to say, "if at all". It feels more like the effect of changes seem so minute at best. We know that "lost contact time" has changed how it works, so theres no reason to doubt other variables have changed as well.


edit: I should add that, no matter what i did to the sim.cfg, nor how large i increased the range of aircrafts visual node, they always detected me at about the same distance.

[lurker_hlb3]

I worked on this same concept about 6 weeks ago as a possible add-on to RSRDC. I to uses the NYGM AI_Sensor.dat file as a baseline. Using the “BombDummy” option in my aircraft .eqp file to force them to attack. Created a control test mission file using the G4M Betty as the test aircraft flying at 50m altitude. Created four aircraft that had a CPA (Close Point of Approach) of 20/15/10/5 km. When the test mission was run, no matter what values (test values where 30000, 25000, 20000, and 15000) I set in the AI_Sensor_vis.dat, the aircraft always reacted at 8kms. This 8km reaction ranges is the same when “stock” AI_Sensor.dat is used. I retested using “AI visuals Test ver 1.0 for Trigger Maru” after resetting the MinHeight to 0 and MinSurface to 2 and got the same results. Please noted that the reasons these test results where not posted before, is that in my view the desired results were not achieved. Please note that after 1.4 comes out and the visual range is bumped up to 20km the test results are going to be different.

[tater]

Yeah, I did rather a lot of testing with a BB based AI visual set to 35,000m.

My detections were always at ~8800 yards, +- ~1000 yards depending on other settings. Note that I not only tested vs a sub target (player sub), but also with a BBs vs each other.

Nothing seemed to change the visual limits.

I have a terrible feeling it's like the useless DC controls (det depth) in the sim. They all get overridden by the AI.

[Ducimus]

Please note that after 1.4 comes out and the visual range is bumped up to 20km the test results are going to be different.

Yeah i realize that. I hope for that honestly. this 8KM thing is ... BS.

Anyway, ive made someting resembleing progress.

After some trial and error testing ive come to a number for Minsurface, and right now, its looking like -38. At this range, ive been detected by aircraft while submerged at periscope depth, to a maximum range of round 4600 meters to maybe even has far as 5,000, but 4800 is probably the real value. Very hard to gauge reactions exactly in this kind of testing.

So heres a new version if anyones intrested:

http://www.ducimus.net/sh4/release/TEST_TMaru_AI_Visuals_1.01.7z


//ver 1.01

//AirS_Visual
- MinHeight from -18, back to 0
- MinSurface from 0 to -38


My testing shows a plane can now detect you at periscope depth at a max range of 4600 to 5,000 meters. 38 meters, is about 124 feet, so below that should safe crusing depth, if aircraft is on your position.

[Ducimus]

I just have to say, i now love that air visual node. Gives aircraft a whole new meaning and dimension. After a couple hours of play, i think i have broken my habit of cruising at periscope depth all the time while submerged. :rotfl:

[Ducimus]

I think i need to fine tune the Minsurface a bit more. Crash diving at initial detection to 160 feet, and am still getting attacked by aircraft that fly near me. Thats a bit overkill. So im trying a minsurface of -30 instead of the -38 and see how it performs then.

[Ducimus]

Mental note to self: Next time you save a career game with an aircraft contact to "play with", make sure its during calm weather and not rough weather :88)


Apparently the weather was effecting more then i gave it credit for. I'm Readjusting that minheight again during a 0 wind condition. I have a feeling the end result after im finished is detection radius while submerged won't be very large (probalby small enough to be inconsequential), and they wont be able to spot you at all in rough weather. Probably more realistic in that sense.

[tater]

Yeah, calm and clear should be easy, rough seas harder. You can control it a little by BP-cloning maritime patrol planes, and having the PD detecting ones more common in areas of clear, shallow water.

Ie:
open sea flying boat base:
6 H6Ks with an ai visual that is poor at PD detection
3 H6Ks with your new one

Reverse that (6 good at PD, 3 not as good) in clear water areas for another airgroup, "shallow water flying boat base."

tater

[Ducimus]

It just occured to me that the actual geometry of the visual node, is a probably three dimentional circle, and i will never get this perfect.


http://www.ducimus.net/sh4/min_height_problem.jpg

[tater]

This was about scope spotting, but it's a useful reality check:

Posted by Donut:
Having spent time served in USCG.air search,& rescue. Seeing a scope @2,000 feet is not possible,if there is no wake,due to same. My contention is perhaps from (stable platform) surface craft @ 1/2 visual range,air craft:less than 1,000ft.only if sub @ flank/full.(IMHO)If there is any wind ocean is white with foam,(No wake) . .With MK52 gun fire control radar, periscope will not return a pip signal to indicate same. In 1961.

For a submerged sub, I'd think that the plane would indeed be able to see it in the right kind of water, but the ANGLE would be very important. At a shallow viewing angle (total internal reflection). There is a point at which you see nothing underwater, regardless of clarity at all. The in the water version is the effect in a swimming pool where you can see the sky, but away a little bit the water/air interface looks like a mirror. It's how optical fiber works, too.

The easy visual way is to make sure the range is short. In reality, the altitude is critical.

Assuming your node detects surfaced subs at great range, and only submerged ones at very short ranges, it's likely pretty good. Otherwise, a what-if:

Make a visual node for planes that is very long ranged, but only detects surfaced units.

Make a sonar or hydrophone for planes that has the angle set at maybe 50 degrees, and a very shallow depth (using max range).

Since the sea state has an effect, that will sort for calm seas (clear). The angle will be set so that the plane will see the submerged sub based on the viewing angle, not the range.

tater

[Ducimus]

I think whats going to have to happen is im going to have to be conservative.

Set the node too shallow ( say 10 meters), and the air unit seemingly see's nothing at all. Maybe not enough surface factor exposed, or maybe the detection time requirements not being met, or the plane not getting close enough. Any of which can be because the radius at which it can see below the waterline is short at that minsurface.


If you set the node deeper, you can enlarge the radius, give it more time to detect, and increase likelyhood of an encounter. Problem is, that the closer the unit gets to the player, the deeper it can see.


I think its all going to boil down to a trade off. Getting it to detect the player at the "proper" distance, with the side effect of it being able to see deeper then wed like when on top of the players location. So in that light the sighting radius should be short.

I have a feeling the answer to this, lays somewhere in between -10 and -15 meters.

This version is -15 meters currently:
http://www.ducimus.net/sh4/release/TEST_TMaru_AI_Visuals_1.02.7z

[tater]

Because of the critical angle (48.6 degrees for fresh water), when you consider the sub as the source (the light reflected from it that you see), the sub will be invisible.

http://library.thinkquest.org/C006694F/Optical%20Fibres/TIR.gif

So adjusting it so that the plane needs to be close is not really a big problem, it's realistic.

tater

[Ducimus]

Heres an interesting observation. I reduced the MinHeight to - 12. When i got a SD radar i dove to 160 feet, and i was right on his track. He didnt see me until he was practically on top of me, where i could visually see him change course and start to swoop in on my location, but he didnt dive at me. It looked like the plane was so fixated on trying to maintain a visual contact, that he didnt have enough space to make an attack run. Instead he circled around a couple times, and went away.

If i was at periscope depth, i wonder if the same thing would happen? -12 could be the magic number.

[digitizedsoul]

Just wanted to add that i'm fascinated by this discussion, and if either one of you guys (tater, duci) needs a donation let me know :)

I'll add it to my normal subsim tithes :)

[tater]

If you want to make a donation, click the link up to the right ("Make a Donation") and support SubSim.

:up:

Thanks to Neal for letting us chat.

tater

[Peto]

I've been out of the loop for quite a while but am glad to see you guys still tweaking files and doing your voodoo!

Just thought I'd kick in this comment: In George Grider's book Warfish, he tells a story about a time they were attacked by a plane while they cruised at 200 feet. A rather rude shock (as he relates it).

Keep up the incredible work guys! And...

Thank You!

Peto

[MONOLITH]

Just wanted to add that i'm fascinated by this discussion

I've been out of the loop for quite a while but am glad to see you guys still tweaking files and doing your voodoo!

Keep up the incredible work guys! And...
Thank You!


Ditto. :rock:

[Ducimus]

While im still trying to get a reasonable min surface value for air visuals, i think this constitutes a major change in gameplay on two fronts.

First, Planes are once again deadly. From various readings i gather the pacific was rather clear, or rather, that a shallow submarine easy to detect, and that subs typically cruised at 100 ft or so, if they weren't doing a periscope patrol. Once a plane was spotted, they'd go down to 100+ feet regardless. Assumably because a subs silloutte wasnt visible at that depth. This mod is an attempt to recreate that. So, again, planes have teeth now. Where as before, theyre just a nuisance.


Second, plane detection is a start of a chain reaction that will make the game a whole lot more intresting. First, one must understand that once a unit has made contact with the player, it sends out an "SSS" of sorts. Nearby units will come around to see.

How long they do this is dictated by the lost contact time. In SH3, and SH4, up until patch 1.2, a lost contact time of say 30 mins, meant that once a player is detected, all nearby units will come to investigate, for as long as the lost contact time is. If at the end of 30 minutes no contact was made, they'd go home.

In patch 1.3, im told that how lost contact time work,s was fundamentally changed.

In short, for a non convoy escort unit, lost contact time is decided by the following formula:

2 * (skill * 4 + 2) * lost_contact_time * -/+30%. ( Skill being 0 - 1 where 1 is elite. )

So for a typical crew rating 3 unit, (forgive me if my math is ,off, its late and im tired)

2* (0.75 * 4 + 2) * 15 * +/- 0.30
2 * (3 +2) * 15 +/- 0.30
2* 5* 15 = 150 * +/- 0.30 = 45
150 + or - 45 = 105 to 195 minutes.

So a typical crew rating 3 escort, will hang around from 105, to 195 minutes. However, the kicker is, this timer is started when the escort arrives on location. So then.... a plane detects you, which sends out the red flag, which means more tin cans will start sniffing around the area it reported you at.


Ill give an example. Tonight, near Palau islands, i had an SD radar contact of two planes coming. I dived, but they detected me anyway. (still fine tuning that part). So they bomb the area, and i go away submernged at 160 feet. About a couple hours later, i get a sonar contact of a 2 ship convoy. I then go to periscope depth, (hoping theres no planes about), take a peek, dunk the scope, start plotting this convoy, and realize im in a good position for a submerged attack, which i press. Along the approach doing periscope observations, i came accross a frighting discovery. Two escorts that i didnt hear were in my baffles at about bearing 180. My first thought, was "hello, where did you come from?" I can only guess that they were sniffing around the vacinity the planes had reported me earlier. All told, i had 4 tin cans in the area. One was the convoy escort, two were sniffing around in the general direction of where i had crash dived to avoild the planes, and a 3rd was poking around a bit further south of the two that crept up behind me. With the exception of the convoy escort and the 2 merchants, all of them were looking for me.


Planes..... im starting to love them.

[Hitman]

Good work Ducimus

It certainly is very realistic (Readed several reports of that in Wahoo and Clear the bridge) that being spotted by a plane has later consequences in that DDs are sent after you to investigate, or reinforces ocnvoys that go in the vicinity, specially in Empire waters, yellow sea, and similar places.

My only concern is about the effects of this all at night....even with the in-game-built reduced vision at night, will planes spot you submerged at night if they fly right over you? :hmm: That would be a problem :damn:

Keep the good work :up:

[Prientje]

TMaru_AI_ Visuals_ 1.2

September 1942 rough sea

sub was from the beginning at 150 feet speed 2 kts and from plane detectet

plane throwing bombs

i go to 200 feet

plane throwing bombs too

is that MAD ?

[Ducimus]

RE: Planes at night
There are so few of them, its not even a factor that ive seen. You could adjust this however you like, but i honestly dont see it as a problem. Their vision is too limited at night, and the spawns are too few.

@Prientje

Thank you for the input. Its valuable to me in being able to fine tune this modlet so it behaves reasonbly well. Im current experimenting with another version and will post it later today (hopefully). If your curious about the mechanics of this stuff, and why you might be detected so deep, read further up the thread.


@ all
If anyone else is expermenting with this, the primary focus right now is on aircraft in a career game. ALl im doing is playing a career game, saving it when i have a plane contact in clear/calm weather, and then seeing how this plane behaves in terms of visual detection. If the results are unsatisficatiory, then i just quit the game, disable this test mod, adjust the minsuface using S3D , then reload the game and see how it effects the planes detection radius and depth.

Adjustments are being made to the following node:
AirS_visual->MinSurface

it is located in the following file:
/data/library/AI_Visual_sensors.dat/


In version 1.02 of this modlet ive used a minSurface of -15
I then tested a MinSurface of -12 (not published)
I am currently experimenting with a minSurface of -10.

I beleive im getting close. The more i observe, the more i understand it, but two heads are better then one. So if anyone wants to give this a go, please do.

Tommrrow i go pick up the moving truck, and will be taking down and packing up my computer, so ill be offline for awhile, very soon.

[tater]

It's interesting, in messing with minsurface, I used values that were small, but >0.

Will negative values always detect submerged units?

I'm wondering if -5 might allow my BB sensor with 35,000m range to react to something past ~9500.

[Ducimus]

>>Will negative values always detect submerged units?

That is my blind, uninstantiated assumption.

Another intresting food for thought, is a very small min surface value... maybe the AI can see periscopes better?


Whats kicking me in the jimmy, is how this number is scaling. I mean, -10, -12, -15 meters, (periscope depth is 18 meters!) and yet im being seen at 160 feet of water? I just have to scratch my head at that. However, i suspect that each time i raise the minsurface value closer to the surface, im also reducing the maxium radius the plane can see submerged objects. Its just when he gets RIGHT NEAR your position that he can see deeper i think.

[tater]

Well, min surface is an AREA. Negative numbers have no meaning in the real world.

When you were messing with min HEIGHT, the -15 was indeed METERS. With surface, all bets are off.

All the negative is doing is altering a calculation in an unexpected way, so I'd not expect it to scale well.

I'd tend to think it should be set up so that a plane would be able to detect you at any lateral offset only in a dead calm. Add in in some chop, and the depth and lateral offset from the sub should really drop.

tater

[Ducimus]

Thats true. I just wasnt getting any decent results with min height where i was with minsurface.

edit: acutally thanks for reminding me of that, ive been so fixated on this, im starting to not see the forest for the trees.

[tater]

Wonder how the two interact... Say a less negative minsurface, combined with a minheight that is negative.

[Ducimus]

Well, i just reminded myself why i switched to MinSurface. By setting a minHeight of -100, planes can no longer see me

[Ducimus]

Ok after more testing.

First, MinHeight, doesn't help. If i leave it at 0, planes detect me. If i change it to -X, they don't.


MinSurface, at -5 they don't see me, except if i come up to radar depth. At least this one test run thats what i observed.

At a -10 value, they see underwater. How far down i don't know. It could be depth is a total non factor in detection at this point.

I now think the value lays between -5, and -10.



I know some people are thinking, yeah, and you thoguht it was -38 before, and then this, and then that. Modding something like this, is alot of trial and error styled testing, as well as process of elimination. It takes time, and you'll stumble ALOT along the way before you narrow down what your looking for. Thats just the nature of the beast.

[Ducimus]

Somewhere, in the vacinty of..... -5.

However, i have a bad suspicion arising. Normally, with surface factor the farther you get from an object, the smaller it becomes. Using a negative number, it could be that its doing the exact opposite. The farther you get away from the unit, the bigger your profile.

Now, on the surface, plane detection works just as normal. Its below the surface that seems to be actign weird. And well, why shoudlnt it? im basicaly forcing it to do something it wasnt meant to do.

In my minds eye, im picturing a cone. With the tip of the cone near the surface, and the wide base of the cone deep below the surface. It almost feels like the deep you go, the quicker he see's you. at least, thats the thought that is developing. Im not sure if this is whats happening yet. It requires more testing. I have a feeling that if he sees you under water or not, is going to be a boolean thing.

If that ends up being the case, then we'll need two air visual nodes, and place this special node on specific aircraft.

[tater]

OR, make a good, long range AI visual for planes that can even detect a scope up close, and add a hydrophone to some planes. that is very short ranged. Or maybe a sonar (assuming you can remove the ping from one).

[Ducimus]

To borrow some of Lurker_Hib3 and tater's input, ive formulated a plan, assuming that underwater detection is as boolean as im beginning to think it is.

Two air nodes. One a normal one, the other with this negative min surface value.

Heres what i think. We put the negative node on a short range plane like a zero, and we shorten its maximum visual range (so hes not TOO uber) . We beef up this zero's guns, and equip him with the dummybombs. The end result is while not all planes are going to be calling in the calvary, the zero's, which should be nearer to shore anyway, will. Zero's probably shoudlnt be carrying much in the way of real bombs anyway so the fake bombs for zero's would be ideal. So under that idea cruising in close to shore, could be a whole lot more dangerous then before.

OR there's always the inverse approach, equip this on the long range patrol planes, not as dangerous, but still a greater chance to force a tin can encounter.?

[tater]

Zeros wouldn't be able to call anyone, they either had crappy radios, or none at all.

BP-clone the maritime patrol planes, and just put a small number with the special node in each air group.

tater

[tater]

Do we know what the altitude for airstrike generated planes is?

Is it always the same?

tater

[Ducimus]

Zeros wouldn't be able to call anyone, they either had crappy radios, or none at all.

BP-clone the maritime patrol planes, and just put a small number with the special node in each air group.

tater

Proablly the best idea.


>>Do we know what the altitude for airstrike generated planes is?

No idea.

[tater]

What does min and max elevation do?

[tater]

SONAR.

A sonar in the dat:

Type=3
MinRange=0
MaxRange=1200
MinHeight=-300
MaxHeight=-10
MinBearing=0
MaxBearing=90
MinElevation=90
MaxElevation=106
MinSurface=0
Sensitivity=0.05

Note that there is a SonarSound node attached, delete it, no sound.

So we make a new sonar:

Type=3
MinRange=0
MaxRange=1200
MinHeight=-25
MaxHeight=-1
MinBearing=0
MaxBearing=90
MinElevation=90 (assuming 90 means along the surface of the water)
MaxElevation=180
MinSurface=0
Sensitivity=0.05 (?)

Have a visual, AND a sonar.

[Ducimus]

What does min and max elevation do?

Angle of the beam. Think of a flashlight pointing downwards at an angle. The greater the max elevation, the greater the bottom side of the flashlights beam angle, the deeper it delves, and the shorter the blindspot at a given distance from the sending AI unit.

edit: Don't aircraft have to have the node to support it? IE, if theres no sonar node, in its data files, i don't think it will accept it, will it?

[tater]

What are the height values
on a sonar to have it work over the biggest possible area?

(I added an N node to an H6K to test)

[tater]

Doesn't seem to work. Wonder if it requires the node to be in the water...

I can always make the N node hang 500m below the plane or something.

I need to use one with a ping, then i'll know if it is working.

[Ducimus]

I think i may have what im looking for soon. I beleive ive narrowed the number down to -4.x. At -5 i get detected at periscope depth, 160 feet, 200 feet, and wasnt detected at 350 feet.

At -4, i dont remember being detected at all.

So, -4's too little, -5's too much, its gotta be a decimal in between!

[tater]

If the - value fouls up detection of a surfaced sub some how, you can always add a surface search radar with settings like AI_Visual should have.

Now we need to get them to track you for 3 days :)

tater

[Ducimus]

-4.5 too big GI!

-4.25 to small!

[Ducimus]

I do beleive, i smell the sweet smell of success!

Conditions:
saved career game, time, after noon, no fog, light wind of 4 kits.



With a MinSurface of -4.35, i was detected at periscope depth, without ever raising the scope. Reloading the game, and repeating the experiment, at a depth of 160 feet, i was not detected, at least not that i could tell. The only thing i can say is that detection is titering on the edge. Given harsher weather conditions it might not happen, and i cannot garuntee that you wont be detect at 160 feet, but in the game i just tried this in, i wasnt, albiet i think it was BARELY.

[Ducimus]

Ok this is the last test version i will probably upload anytime soon. Time here is short, and well, my home office is looking rather bare. Infact, you can hear sound echo off the walls now.

http://www.ducimus.net/sh4/release/TEST_TMaru_AI_Visuals_1.03.7z

[tater]

Nice.

Did a few quick tests.

I used an H6K armed with DCs.

It dropped on me at 80ft. Note these were scripted planes.

I was attacked by a few Bettys, but none bombed me. Looks like they didn't have DCs, but they sprayed the water with MG fire.

Looks like we'll want a few new bombs---even make a 30kg and 60kg for the zeke. What I was really thinking was tweaking the air-dropeed DC. The stock one has a sink rate of 5m/s, and a det depth of 5m +-5m.

In RL, they had 2 different DC det depths. Make 2 identical bombs, one to each depth, and put 1 of each on the planes.

Q. When making an attack on a submarine, were pilots allowed to drop depth charges after the submarine disappeared below the surface?
A. Yes.


Q. How many seconds after the submarine disappeared from sight were they allowed to attack?
A. Within 30 seconds.


Q. What size of depth bombs did your aircraft use?
A. 250 kg.


Q. Did you use any other airborne weapon besides depth bombs?
A. We had nothing besides bombs. The bomb was a standard 250 kg bomb with a modified flat nose attachment and a special tail, a nose and tail fuse was used.


Q. What depth setting did you use?
A. We used 25 and 45 meters. If the submarine was discovered near a convoy at periscope depth and immediately submerged, we dropped the 25 meter; if time passed and gave the submarine time to go to a lower depth, we would drop the 45 meter bomb.


Q. Did you drop more than one depth bomb at a time?
A. Generally we dropped only one, but sometimes two.


Q. When you dropped two, what spacing was used?
A. None, the bombs were dropped simultaneously. Only experienced pilots were allowed to drop all bombs at once.


Q. At what distance from the submarine was it necessary to explode the depth charge in order to sink the submarine?
A. With the 250 kg bomb, it had to hit within 13 meters of the submarine in order to sink it. The smaller planes equipped with 60 kg bombs had to make a direct hit on the submarine to sink it.


So 25m and 45m were the 2 real depth settings. I'd set the depth to that, no or little error. The burst charge on a standard 250kg makes it about as strong as the type 2 with a 162kg warhead, maybe more. In my current DC mod that's ~50% more powerful than the stock DC tyou are used to. I can make 2 versions, one for stock, 1 in line with my DC mod.

I'll get on this.

My gut says we should BP clone the maritime patrol planes, then give their basic load DCs in addition to the other loads, then sprinkle a few per appropriate air base.

tater

[Ducimus]

>>My gut says we should BP clone the maritime patrol planes

I think i have a backup of the dat file before i removed the 2nd air node. So this should be no problem.

[tater]

BTW, the wind picked up the next day, and I stayed along the coast under air cover. They didn't spot me with some chop. I think that makes sense.

tater

[tater]

New bombs added.

I have 2x250kg DCs, one to 25m, one to 45m (with some slop). Plane dropped DCs seem to blow at set depth, not aimed depth.

I also added a 30kg bomb, though I may change it to a 60.

Messed with plane loadouts, and corrected the eqps for all the wrong nodes (ain't S3D grand?).

Seems like airstrike generated planes pull a random loadout. Including the unarmed "basic." Some were dropping bombs, others dropped DCs on me.

tater

[Ducimus]

Thanks for the support tater!


I know its only been a few days, but I feel like ive been away for a week. Anyway, does anyone have any feedback for this WIP?

[tater]

I was "away" in halloween land for kids, then toddler sickness (AGAIN), and haven't done much in anticipation of 1.4.

tater

[Peto]

I've been using this test_viual mod for 3 patrols now. My findings are:

1. Periscope depth is no longer safe in decent weather conditions (which is good).

2. Once they find you at periscope depth, they keep coming back (no surprise there). But what did surprise me was after going to 120 feet (weather conditions: flat calm/ no clouds or fog) they found me again!!!! I had to go to 150 feet. Area was the East China Sea. This only happened once so far.

3. This mod rocks :rock: .

If tater combines his new arial dc'd listed above, the ocean will become a much more dangerous place.

Question: Is it possible to add a plane as an escort for a lone merchant?

Great Stuff (Again) Ducimus!!! :up: Hope your move is going well!

Peto

[Ducimus]

Thanks for input! Your post makes me feel better about having incorporated it into the posted version of TM. :88)

As for Taters modifications, i hope to include those sometime after patch 1.4 if he is still interested.

>>Is it possible to add a plane as an escort for a lone merchant?

Simple answer, i don't know. I am doubtful on that, but it is worth expermenting with to find out though :up:


edit: BTW, if anyone wants, heres a standalone version. Same thing as the previous versions, only with two additonal nodes that are not currently used:
http://www.ducimus.net/sh4/release/TEST_TMaru_AI_Visuals_1.04.7z

Those nodes are:
AirB_visual
E_visual

[tater]

I'll have a revamped bomb/aircraft mod out for you in a bit.

Added:
default air dropped DC depth setting of 25m
new air dropped DC depth setting of 45m
default air dropped DC upped in strength.

250kg bomb upped to nearly match explosive power of air dropped DC (casing is heavier, not as much explosive inside). Bomb has AP, however. I think I should up it even more, a direct hit by a ~500lb bomb should probably sink a sub. If they are virtually always crippled (mission killed) I'll be happy.

500kg bomb massively increased in strength. A single direct hit should ALWAYS sink your boat. Subs don't survive 1000lb bombs.

The 100kg bomb should also do more hurt. Right now unchanged, I'm thinking of upping the upper limit for damage keeping the lower limit near where it is. That will make them somewhat variable in killing power.

Adding a 30kg and a 60kg bomb. Thinking about adding an 800kg bomb for the Betty and Kate. They were converted BB shells, and will have a large AP value, perhaps a smaller radius. Note that thinking about that makes me want to check the large shells to make sure a single shell from a 14"+ gun will sink your sub 100% of the time, lol. But that's another mod I plan on doing. Bottom line is that if you are not terrified to be on the surface near CAs or large (meaning within maybe 20,000 yards), something is broken, cause you should be. If you come out of the fog to see a BB at close range, any order other than "crash dive" should be insanity.

tater

[Ducimus]

Im loving this.

As it is now, aircraft aren't really a threat at all. You start the war with SD radar, which has some really good range on it. You get a contact, you dive, and thats it. Cruise around at periscope depth, watching them zoom by through the periscope, and generally feeling nothing but frustration because you can't surface.

In short, they just annoy the piss out of you.

How now, they'll acutally be dangerous, and something to fear. SD radar or no SD radar, you have reason to pull the plug and get away from the surface.

[Gunner]

Im loving this.

nothing but frustration because you can't surface.
In short, they just annoy the piss out of you.
How now, they'll acutally be dangerous, and something to fear. SD radar or no SD radar, you have reason to pull the plug and get away from the surface.


Actually, it's you and Tater who are Dangerous and something to Fear, lol :D

This sounds really cool!, cannot wait to try this out :up: :arrgh!:

[tater]

BTW, only some of the planes will be armed with DCs, others didn't seem to carry them. Those that are so armed, will have pairs with one set to 25m, the other to 45m. The H8K is particularly nasty, I gave it 2x25m DCs, 2x45m DCs, and 4x250kg bombs in one load.

Since the planes pull randomly from the loadouts as far as I can tell, I might make more variant loads, too.

The Zeros have no bombs, though I can arm them with the dummy bombs from RSRD.

I have seen some references to smaller air-dropped DCs, but I have no specific data at this point. When I get some, I might make some more.

tater

[tater]

Wierd idea.

Make a gun, or a "rack" that drops the air dropped DCs based on a K gun with a range of 0. The 3d model for them is a bomb instead of a DC. The rack is invisible. This would be for a plane like a Betty with a bomb bay.

The rack only has 3 DCs or so. We set the reload time on it to 600 seconds or something so the plane orbits around, and only drops a DC every 10 minutes :)

Actually, I might do this with some escorts, too... borrow the roll rack 3d model, and have it really be a K gun with 0 range, and no shooting visual effect. Set the reload time LONG so that at least one "roll rack" will keep dropping for a while. 18 DCs only dropping 1 every 10 minutes...

tater

[Angus89]

If only you could get allied planes to drop drums of deisel, or the odd unarmed torpedo. Refit with wings. :lol:

[Peto]

Im loving this.

As it is now, aircraft aren't really a threat at all. You start the war with SD radar, which has some really good range on it. You get a contact, you dive, and thats it. Cruise around at periscope depth, watching them zoom by through the periscope, and generally feeling nothing but frustration because you can't surface.

SD Radar IMO is far too powerfull in the game. If it worked that well, why were subs getting caught on the surface? The answer to that follows...

OK. Idea. SD Radar was handy but it also had some serious short-comings. The search pattern emitted by SD radar was shaped like a 4 leaf clover. this meant that a plane coming in "between the leaves" was in a blind spot. So--how could this be "simulated"?

Is it possible to tweak settings in SD's files so it will sometimes not pick up and incoming plane? Or would it be "acceptable" to create a blind spot by eliminating 10-20 degrees of it's covered arc? Say it scans from 0 to 340 degrees. Yes--planes would always be able to sneak in from the same area but it would certainly give us all a reason to dive during the day!

If you guys think it might work let me know. I think I can take a shot at that little task.

Peto

[tater]

http://mpgtext.net/subshare/410Better%20IJNAF%20Bomb%20Loads%20v%201.3.rar

Better IJNAF Bomb Loads version 1.3 BETA

Adds a new set of Bombs files, and alters the original.

The standard loads are possible:

AirTorpedo (not used in SH4)
Bomb100kg
Bomb250kg
Bomb500kg
AirDepthCharge (detonates at 25m +-1m now)

In addition, I have added:

AirDepthCharge_45m (detonates at 45m +-1m)
Bomb30kg
Bomb60kg
Bomb800kg

Lurker's (serbuto's?) BombDummy is also there.

All the bombs have been corrected in terms of damage. Don't get hit. A single big bomb will likely sink you, and it should, frankly. They might need to all be worse, subs are too strong, IMO. All the bombs have an AP of 25 right now. The DCs have an AP of 0.

The aircraft ranges are not changed, so I will need to make a compatible version for other plane mods I have done, this is just a test for now.

As always, no commercial use allowed, everyone else can do with it as they please.

[Peto]

Thanks tater! I'll try it out and let you know when I get sunk :shifty:!

I'm playing a career now but am changing my current command to "expendable" to see how the new visual/bomb mods work.

[Peto]

btw: I figure if anyone would know this, it would be you.

Is it possible to set a plane as an escort for a ship/convoy? That could be nasty :up:.

Thanks!

Peto

[tater]

You'd have to have a group spawning when you want it to spawn, then a plane, or flight of planes moving waypoints along with the convoy.

[Peto]

That's what I was afraid of. It'd be fine for single missions but a bit tedious to add to the campaign...

[tater]

Maybe not. Do the convoys that you want escorted with planes in a separate working layer. Make a single ship per group to start.

Make a copy of it when it's done, and change the unit to a plane in the text file.

You could then run the automated campaign editor on the aircraft layer, and have it zig zag in a big aircover type pattern. Say 50 km legs, at an angle off base course of 80 degrees, and a speed of 90. Merge the new ZZed layer with the ship layer, then run it and see if they move together. If they don;t, mess withthe speed and ZZ angles til they do.

Still a PITA, but possible.

tater

[Peto]

Alright. I promised myself I wouldn't do this again :shifty:. I scripted 100's of convoys for SH3 and I had to get new glasses :88). I think I will try doing this starting in 43 probably. If I can get it to work out it may be good addition to the game considering the subject of this thread...

Can you point me to the utility you use for adding zig-zag waypoints please? Getting back on the horse requires me to find my saddle and harness :lol:.

Cheers!

[tater]

http://www.subsim.com/radioroom/showthread.php?t=113918

I just did a test. It's gonna be insanely complex, the planes will need to backtrack.

Easier would be a zone system where they fly a pattern over an area, then have a new group in the new area. The convoy passes from one patrol area to another. Set the start times such that the air patrols exist only around the time when the convoy is there.

[lurker_hlb3]

http://mpgtext.net/subshare/410Better%20IJNAF%20Bomb%20Loads%20v%201.3.rar

Better IJNAF Bomb Loads version 1.3 BETA

Adds a new set of Bombs files, and alters the original.

The standard loads are possible:

AirTorpedo (not used in SH4)
Bomb100kg
Bomb250kg
Bomb500kg
AirDepthCharge (detonates at 25m +-1m now)

In addition, I have added:

AirDepthCharge_45m (detonates at 45m +-1m)
Bomb30kg
Bomb60kg
Bomb800kg

Lurker's (serbuto's?) BombDummy is also there.

All the bombs have been corrected in terms of damage. Don't get hit. A single big bomb will likely sink you, and it should, frankly. They might need to all be worse, subs are too strong, IMO. All the bombs have an AP of 25 right now. The DCs have an AP of 0.

The aircraft ranges are not changed, so I will need to make a compatible version for other plane mods I have done, this is just a test for now.

As always, no commercial use allowed, everyone else can do with it as they please.
FYI

For those who use RSRDC, there may be a problem when using this mod. There is a change in the weapons loadout for the B5N2Kate, i.e. from 1x500kg to 1x800kg. There are three layer that use the original 1x500kg loadout and with this change there is the "possibility" for a CTD.

[Peto]

That would be very doable and actually makes more sense. Air Patrol and escorts were rarely coordinated by the Japanese, mainly due to inter-service rivalry, bad (or no) radios and the "defensive" nature of ASW (as if the ground forces weren't defending :shifty: ). I was looking through a couple books yesterday and I might start out by recreating a couple historical convoys and giving them aircover as appropriate.

I've tried timing a B24 to escort a convoy in sh3 and learned it takes about 6 billion way-points so I like your idea much better about zone patrols.

[tater]

There were some plans in place by the IJN to patrol certain sea lanes (about 30km wide swath) between certain ports. I think something similar would work. Radiused waypoints to provide coverage, then looped waypoints to mix up the flight plans.

Looks like by 1944, most ASW squadrons had large numbers of radar equipped planes. They flew some at night as well. About 1/3 to 1/2 had MAD gear as well.

Wonder of a radar could be given the same negative surface factor and turned into a MAD so that the planes could have MAD and visual sensors. We'd want the range so short on the MAD that it would only have a max range of a few hundred meters. Planes would fly waypoints at 50m. and maybe detect to 100-150m depth. Course since their air dropped DCs only had 25m and 45m settings, they can't prosecute the target effectively. They will call in more planes/ships, however.

[tater]

BTW, I've read interviews with IJN ASW guys who said that with expert crews, and flying very low, they could detect as deep as 250m.

I'll have to try a new radar with a negative surface factor of maybe 6-7, and a max range of 300m

tater

[tater]

Lurker,

I was lazy and didn't want to add a new loadout. I will make a new version with both the old 500kg load, AND the 800kg in the name of compatibility.

tater

[Peto]

There were some plans in place by the IJN to patrol certain sea lanes (about 30km wide swath) between certain ports. I think something similar would work. Radiused waypoints to provide coverage, then looped waypoints to mix up the flight plans.

Looks like by 1944, most ASW squadrons had large numbers of radar equipped planes. They flew some at night as well. About 1/3 to 1/2 had MAD gear as well.

Wonder of a radar could be given the same negative surface factor and turned into a MAD so that the planes could have MAD and visual sensors. We'd want the range so short on the MAD that it would only have a max range of a few hundred meters. Planes would fly waypoints at 50m. and maybe detect to 100-150m depth. Course since their air dropped DCs only had 25m and 45m settings, they can't prosecute the target effectively. They will call in more planes/ships, however.

I was reading about this in T. Roscoe's book yesterday. They needed 80 planes to be able to patrol the corridor effectively. The MAD idea is good. The planes had to fly low and only could detect in a path about 150 meters wide (I think that's what I read). When a MAD contact occurred, a marker would be dropped automatically. After getting 4 "marks" they would attack in the middle of them. I doubt that much detail is doable (or nessecary). Honestly, I think the MAD will be tough to model mainly because the computer won't be working on "best guesstimate" properties like RL pilots had to. It would be nice to add a few radar equipped Emilys to fly night patrol as well (as you mentioned).

Going back to SD Radar for a moment: I was going to check out dat files et al with minitweaker last night but my tweaked files aren't giving me modifiable data at the moment. Would it be possible to put a small up angle on radar so it doesn't detect low flying planes until they're very close? MinHeight=2 perhaps?

I think we should keep in mind that the Japanese were never able to accomplish all these things for the simple reason that the planes were never available. Therefore, maybe there should always be gaps in the coverage.

[tater]

Regarding gaps, there are a few ways. One is to have the groups spawn every X hours, but not 100% of the time. So there might be a daylight patrol, 90% of the days in a given area.

Then the random group composition.

make different versions of the planes, a MAD version, one regular plane with, and one without the new visual sensor, etc. Have the group have the leader (which has a 100% probability) be a vanilla search plane, then have there be an X% chance of a MAD plane, a Y% chance of a radar plane, and so forth.

Some days nothing at all, other days planes, but not the "good" kinds, other days an ugly mess of planes :)

tater

[Ducimus]

Tater ill try your plane loadout beta tonight. Im in between patrols, so its a good time for me to plug it in and see just how much dead i become :88)

That said, planes ARE still too easy to avoid. Its that damn radar. Im wondering of applying enviormental dampeners to it via the sensors.cfg is an answer. Or, just simply lower its max radius, or raise the surface factor on it. Larger the surface factor, the closer the plane has to be in order for it to see it i beleive.

[Rockin Robbins]

Following closely to see if planes are spotting subs too well. Problems are non-variability in the opacity of the water, and apparently planes able to see subs from a mile away. In order to see a sub at periscope depth, you would have to have a very high angle due to the refractive index of water. Once you exceed the rafractive index, you can no longer see through the water, even if it is crystal clear, it merely reflects incident light like a mirror. A plane a mile away is WAY below the refractive index. All he can see at your distance is a reflection of the sky. What IS the refractive index of salt water? Can't be less than 30º, can it? Sorry to be the one throwing monkey wrenches but it IS important if we're trying to reflect (hehehehe) reality here.

A great thing about this mod is that it gets rid of the chickens running all day at periscope depth. Your radar is your friend here, and you better stay on the surface to see where the planes are if you want to avoid nasty surprises. Certainly TC at periscope depth is eliminated here. That's a good thing. Can't have any wartime vacationers avoiding conflict.:arrgh!:

[tater]

I posted about the critical angle above in the thread a ways, and I agree. The tricky bit is that since units can only have 1 visual node, if you shorten the visual range too much, they won't see you on the surface.

[hyperion2206]

Tater ill try your plane loadout beta tonight. Im in between patrols, so its a good time for me to plug it in and see just how much dead i become :88)

That said, planes ARE still too easy to avoid. Its that damn radar. Im wondering of applying enviormental dampeners to it via the sensors.cfg is an answer. Or, just simply lower its max radius, or raise the surface factor on it. Larger the surface factor, the closer the plane has to be in order for it to see it i beleive.

I think you're right: the SD is too powerful (at least early in the war). When I read the patrol logs of USS Gunnel I was surprised to learn that most planes were picked up by watch standers, not by SD radar.:o
Would be cool if one could recreate that for the game.:rock:

[Ducimus]

I honestly don't think refractive angle is a moddable. In reality, this is a hackjob on the AIs visual schema. We're making it do something it wasn't orginally intended to do. I'm just grateful it works at all. Truth be told, its sitting on a razor's edge of working and not working as it is right now. If you'll note, im into decimals on the variable i was working with.

[tater]

Have you done any controlled tests with weather? Might be interesting to see what the lateral offset max range is dead clear vs with any kind of chop.

tater

[Ducimus]

Nope, but im fairly confident that any fair amount of chop makes you undetectable visuallly. I keep retesting it in a career game, and im starting to wonder if its working. Unless they fly within 8,000 meters (maybe as few as 6) in clear/calm weather, they just keep right on going by.

edit: I should add that i think the biggest problem is, the speed of the aircraft in conjunction with detection time. Its universal for ALL visual nodes. Can't touch it without messing everything up. Detection time is too long, planes are too fast, so if you tighten up their visual node, they wont see squat.

[tater]

I forgot about that aspect.

I honestly think the lateral aspect should be pretty narrow, but it's gotta be a compromise based on what the code allows.

BTW, a % of the planes ALWAYS fly on by. I think that if they are not armed with bombs, they won't react, even if they would fire MGs at you on a 2d pass. Since the "basic" load for all is unarmed, some will fly by.

One thing I'm not sure of is what kind of checking the AI does before attacking. They will obviously shoot MG fire at the water above you, and I think they are dropping regular bombs on me which have zero chance of hitting as I'm submerged.

I assume the planes get a load picked randomly from the loadouts (including "basic").

A quick test case would be to throw a bomb or DC in the various "basic" loadouts and see if they all attack. Any that don't didn't spot you.

I guess a scripted test first. A plane with basic loadout right over you. If it never attacks, but does with a bombload, that's one case for sure. Course they might also be detecting the scope if you are watching them. :)

tater

[Ducimus]

>>and I think they are dropping regular bombs on me which have zero chance of hitting as I'm submerged.

I dont remember which ones i changed, but one of the first things i did was change the depth accuracy of bombs and such, back when TMaru was FTT. In testing, they manage to damage my flak gun and deck gun quite a bit, but not much more then that if im in a crash dive. If im at periscope depth, then i get a whammy for sure.

>>. Course they might also be detecting the scope if you are watching them.

I thought of that. When testing i reran with, and without periscope. your OOD at the table is a BIG HINT if your seen or enemy near by. he tenses up. Id just sit there at 8X TC, and watch his posture. Once he tenses up (and stays tense), you know a plane is incoming, at which point F12 to exterior view, and watch the fun.

[lurker_hlb3]

http://mpgtext.net/subshare/410Better%20IJNAF%20Bomb%20Loads%20v%201.3.rar

Better IJNAF Bomb Loads version 1.3 BETA

Adds a new set of Bombs files, and alters the original.

The standard loads are possible:

AirTorpedo (not used in SH4)
Bomb100kg
Bomb250kg
Bomb500kg
AirDepthCharge (detonates at 25m +-1m now)

In addition, I have added:

AirDepthCharge_45m (detonates at 45m +-1m)
Bomb30kg
Bomb60kg
Bomb800kg

Lurker's (serbuto's?) BombDummy is also there.

All the bombs have been corrected in terms of damage. Don't get hit. A single big bomb will likely sink you, and it should, frankly. They might need to all be worse, subs are too strong, IMO. All the bombs have an AP of 25 right now. The DCs have an AP of 0.

The aircraft ranges are not changed, so I will need to make a compatible version for other plane mods I have done, this is just a test for now.

As always, no commercial use allowed, everyone else can do with it as they please.
FYI

For those who use RSRDC, there may be a problem when using this mod. There is a change in the weapons loadout for the B5N2Kate, i.e. from 1x500kg to 1x800kg. There are three layer that use the original 1x500kg loadout and with this change there is the "possibility" for a CTD.

Conducted some controlled test with this mode and with Ducimus TEST_TMaru_AI_Visuals_1.04 mod, neither mod has any ail effects on RSRDC v102. The only issue is (a) aircraft will no longer strafe you if their weapons load is set to “basic” and B5N2 Kate that are configured with 1x500kg loadout will no longer attack targets. During testing G4M Betty with 3XDC loadout didn’t appear to work, however the H6K/H8K Flying boats with 4XDC loadouts will sink you. I’m planning to integrate these two outstanding modes into the next versions of RSRDC

[tater]

I'll check the betty. I really need to make a spreadsheet for doing the eqp files, all it takes is one number not properly serial and they don't work. I'll test later.

[tater]

Betty was a typo as I thought. one of the eqp entires had the same number as another.