PDA

View Full Version : Very Interesting Visibility Results


MarkShot
01-18-08, 05:02 PM
This was inspired by another thread discussing tracking in bad weather.

* Back Ground *

Testing done with SH3/GWX 1.03 and 16km option.

I set up a custom mission with stationary ships: my sub and tankers about ever 500km diagonally.

In Mission Editor, you can vary:

- Time
- Wind Speed
- Precipitation
- Clouds
- Fog

* Findings *

Note: Crew spotting is defined as meaning the game engine plots contacts on the nav map and calls out "ship spotted".

(1) During daytime, the maximum range which your watch crew is able to spot ships is not impacted by wind, rain, clouds, and fog. Basically, the crew is able to see to about 6km.

(2) During daytime, the angle off your bow that the crew is able to spot ships is quite dependent on wind, rain, clouds, and fog. So, distance is not impacted, but the angle off the center line of the keel is very much impacted. In very clear conditions, the crew spots contacts easily 60 degrees off center. In very poor conditions, the crew has trouble with anything beyond 10 degrees off center.

(3) In daytime, there is a huge variance between what the AI can see and what the human player can see. In the worst weather, the human player can mainly spot only out to about 500-600 meters, where as the AI can still spot out 6km.

(4) It looks like the speed of spotting may be impacted by environmental conditions. More testing would be needed.

(5) I have definitely AI spotted ships further than 6km in actual game play. It would appear that another untested variable has to do with contact's motion or speed.

(6) The range the player is able to spot ships in clear conditions appears to greatly exceed the capabilities of the AI. It appears in poor conditions that the exact opposite holds; the AI is able to spot to a much greater distance.

* Discussion *

It was very interesting to find out that player spotting and AI spotting are entirely different. It also interesting to see that there is some form of peripheral vision affect modeled for AI spotting which probably doesn't make any logical sense, since a watch crew should be scanning in all directions.

---

Perhaps, some of you heavy duty modders can explain some of this to me, since it is all kind of counter intuitive.

(1) Why does angle off of bow matter?

(2) Why is AI spotting the same distance wise?

(3) Why does the AI spot better than the player?

Finally, what implications are there when one is actually playing the game and on patrol?

---

One thing which I take away from this is that in good conditions when expecting contact at any moment, you should be on the bridge looking yourself. However, in bad conditions, you are better off depending on the crew. (Of course, the other option is to use sonar in bad conditions which has much greater range than eyeballs.)

MarkShot
01-18-08, 05:32 PM
Yep, I haven't been testing counter detection here, but mainly detection by ownship/crew.

KeptinCranky
01-18-08, 05:42 PM
Umm, were these ships totally stationary, that is engines off (no smoke) or with engines running but speed set to 0 knots (smoke)?

MarkShot
01-18-08, 05:47 PM
0kts; no smoke.

I have no idea via PROPERTIES how to stoke the engines. Their course was set to 90 degrees of my surfaced sub. So, in general there was very large profile.

Now, another variable despite real world physics of occlussion may have been how the game resolves detects objects on very similar LOBs but a different distances. I hadn't considered that.

MarkShot
01-18-08, 05:50 PM
Set up looked like this:

http://mywebpages.comcast.net/markshot/tempimages/sh3me.jpg

Sailor Steve
01-19-08, 04:07 PM
I don't know about GWX 1.03, but in 2.0, in the worst weather possible, my crew called for a ship spotted at short range. Up on the deck I saw a destroyer, 400m at 90°, going in the opposite way. I just kept on my course and the DD didn't attack, she just went on. Have to say I was sailing my beloved type II, in such case I guess size does matter :hmm:
Smaller is better? Who'd of thought?

WolfOfCampscapel
01-19-08, 06:51 PM
My watch crew just spotted a destroyer... at 8800. Since I use stock 8km visibility, they're definitely better than me :p

its a viic
01-19-08, 08:50 PM
I don't know about GWX 1.03, but in 2.0, in the worst weather possible, my crew called for a ship spotted at short range. Up on the deck I saw a destroyer, 400m at 90°, going in the opposite way. I just kept on my course and the DD didn't attack, she just went on. Have to say I was sailing my beloved type II, in such case I guess size does matter :hmm:
Smaller is better? Who'd of thought?

Its not the size of the sub that matters but in how you... drive it :sunny: :sunny: :sunny:

MarkShot
01-20-08, 08:12 PM
I have definitely confirmed that the AI has peripheral vision issue.

I got a sound contact on a lone merchant. So, I set off in its direction.

I come to the bridge to have a look for myself. There it is a fairly large merchant about 6km 70 degrees starboard. Now, the watch crew is reporting nothing. I immediately order a turn towards the merchant. As soon as the front of the sub comes around, "Ship spotted!".

Here is the thing I don't get. Why angle of the bow have any relevance to the watch crew? Clearly there is must be a technical reason why it is so modeled, but I cannot possibly thing of one.

Of course, in real life, when you have single pair of eyes, you will in most cases spot better forwards. But there are five on the bridge. They each should be scanning their own sector.

Why hasn't anyone taught the crew in SH3 how to stand a watch?

Thanks.

The image shows exactly the behavior which I am describing.

http://mywebpages.comcast.net/markshot/tempimages/periph.jpg

Brag
01-20-08, 08:37 PM
In RL, at night you don't see a thing in front of you, only your peripheral vision works. :know:

MarkShot
01-20-08, 08:43 PM
This is in the day time. No disrespect, but that doesn't really explain these observations.

MarkShot
01-21-08, 09:33 AM
(a) No one is aware of angles affecting AI spotting? {possible being in systems I often find things which others miss}

(b) No one really cares?

(c) I have failed to explain it clearly?

(d) No one has anything relevant to add?

Sorry, I find so many techniques under discussions here: TDC settings, RPMs, depths, ... It seems odd that something as significant of how the AI sees the world would be of no interest to anyone.

Hilis Hatki
01-21-08, 09:52 AM
Brag: "In RL, at night you don't see a thing in front of you, only your peripheral vision works."

MarkShot: "This is in the day time. No disrespect, but that doesn't really explain these observations."


I remember a post, here or at ubi with pics even, within the first month of sh3's release back in 05 that ... when the player look straight at a ship at night he could not see it yet as he panned the view to the side it become more visible on the screen. This person was saying how unrealistic this was. Dont know if it wae explained to him:D . Like when star gazing.



A ship on a constant bearing from the sub should be harder to spot than a ship on a changing bearing.


No smoke??
Is the sim modeled to spot smoke and/or ship differently.

"peripheral vision affect modeled for AI spotting" maybe its coded too much to the direction of the boat instead of to the crew.

OR maybe

The officer of the watch has the front viewing sector. Does where a "watch qualified" crew member is on the deck make a difference, in sim? in your testing?

Is there more eyes looking forward than to the sides?

MarkShot
01-21-08, 09:57 AM
Interesting point about the bearing rate. However, the results seem to show that bow on low bearing rate is easier for the AI to spot, than acute bow high bearing rate. ???

MarkShot
01-21-08, 10:07 AM
I was using a mod with GWX 1.03 which fixes sonar range that you always get max performance.

However, I thought Kpt. Lehmann said nothing was changed sensor-wise between 1.03 and 2.00.

Hilis Hatki
01-21-08, 01:27 PM
:D
This thread made me watch the watchers:D

on a VIIc

..| ....|
-W- ..c-
..|

..|
.-c

-c ...c-
.|.....|


The watch officer looks 360.
All others have a 90 fov.

There seems to be a small dead zone on 90 starboard.
The 360 viewer should switch place with the middle watch?:o

bigboywooly
01-21-08, 02:00 PM
Am not sure the AI " sees " smoke
Maybe its wakes
Maybe its sound

Try running with the noise meter on
Not exactly a reliable tool but

If you run surfaced parallel to a convoy at a distance of your choice - around 3\4 or 5km
Now run flank and the meter turns red indicating you have a high CHANCE of being detected
Drop speed to slow and the meter will turn green indicating little or no chance of being detected
Now the visual distance hasnt changed so why less chance of the AI spotting you ?
Either the wakes OR the sound get factored into the game or both

Your results are interesting
Carry on :up:

Hilis Hatki
01-23-08, 02:14 PM
Hey MarkShot,

Any new results?

It would be interesting to know if the code takes into account which crew member is standing where on the deck or if the deckwatch is taken into account as a "whole" when it come to spotting contacts.


today;
Weather clear, wind 7m/s ,sunset

I spotted smoke at 10 degrees.

w/o "no visual contact"

I locked on target with UZO.

w/o "visual at 8700 m".

I changed stations.

w/o "no visual contact"

MarkShot
01-23-08, 04:29 PM
Sorry, nothing new to report. Switched games at the moment. My U-Boat is once again hard to spot in heavy seas in the darkness of night:

http://mywebpages.comcast.net/markshot/tempimages/invis01.jpg

http://mywebpages.comcast.net/markshot/tempimages/invis02.jpg

MarkShot
01-23-08, 04:33 PM
Oh, this was interesting, but not viz; sonar.

Your SO, seems to lose tracking capability if you switch out of 1:1 TC. Also, he loses a contact as it heads into your baffles, but you can easily follow it yourself.

Additionally, you may be aware of a contact in the rear quarter that the SO cannot pick up. However, if you put the sonar head on it and request the SO to follow, he will then acquire the contact.

MarkShot
01-23-08, 04:41 PM
Look at this! Not one, but two convoys! :) I wonder if I can assemble two wolf packs to close in on both of these? I've never tried that before.

http://mywebpages.comcast.net/markshot/tempimages/invis03.jpg

MarkShot
01-23-08, 05:47 PM
There we go. We are inside the screen and will be coming to PD to shoot !

Real technique, but you cannot do this in SH3/GWX, since at 1-2kts it takes way too long to dive and come back up again.

http://mywebpages.comcast.net/markshot/tempimages/invis05.jpg

VipertheSniper
01-24-08, 07:54 AM
There we go. We are inside the screen and will be coming to PD to shoot !

Real technique, but you cannot do this in SH3/GWX, since at 1-2kts it takes way too long to dive and come back up again.

http://mywebpages.comcast.net/markshot/tempimages/invis05.jpg

In the late years I always dive very deep, and once inside the screen blow ballast, of course when you're reaching 25m you have to order PD. But the method works pretty well

MarkShot
01-24-08, 10:35 AM
Yes, it was recently discussed that blowing ballast doesn't seem to draw attention in SH3.

Maybe I will give that a try. Thanks.

In other sims if you need to get to PD quickly from being on the surface, you might order a crash dive followed by a PD, but that doesn't seem to work in SH3. You keep going down. So, I was worried that blowing ballasts and trying to level off would result in broaching.

Now, I have already completed my patrol in AOD and am in SHCE. Will be back to SH3/GWX soon. I like playing all three of these games, since each offers differents options and experiences.

http://mywebpages.comcast.net/markshot/tempimages/here1.jpg

Pisces
01-24-08, 01:02 PM
Hmm, did you just hijack your own thread to two other games? Which one is it about?
Sh3/GWX, Sh1 or AOD or whatever. I'm not downing those older games. I just lost visual contact on the subject here, if you catch my drift. ;)

MarkShot
01-24-08, 02:36 PM
Okay, now I will tie it back to the original subject in deft manner.

SH3 was the very first subsim to give the player two means to perceive and respond to the world. In SH3, you may respond either via the crews (AI) perceptions of the world and how they plot it on the map, or you may respond to what you see with your own two eyes in the 3D and hear at the hydrophone station. Granted that these two earlier classics presented bridge, UZO/TBT, and scope views, but what you and crew saw were functionally equivalent. In terms of sound, there was no listening to the hydrophones and surface swishing noises.

Now, SH3 is not first game to offer a 3D world. SH2 did, but what you and the AI saw was very similar to the close match seen in AOD/SHCE. SC/DW also offer a 3D world. However, with SHOW TRUTH off, you only see what you think you will see anyway. However, SC/DW did introduce (or reinforce) the concept of the player being able to perceive the world by virtue of working the stations.

So, in this regard, SH3 is quite unique. Psychologically when I started writing this thread, I realize I really wasn't prepared for this dualistic nature of the world that distinguishes between your own perceptions and that of the crew. AOD and SHCE are much simpler in that regard. I am still trying to figure out what this new behavior means to me as an SH3 player. When I should be depending on my crew and when my crew should be depending on me?

Well, are we now back on topic? :)