Annoying, unbelievable tank spotting

[c3k]

Camo: imagine if it had effect in game. Mods would be critical or simply be disallowed.

[Oddball_E8]

Camo: imagine if it had effect in game. Mods would be critical or simply be disallowed.

Well if camo was implemented in the game, it would be on a behind-the-scenes level.

In other words, troops/vehicles would have a binary deciding if they had a camo bonus or not. No modding would change that (although it might confuse players if their modded camoed shermans don't get any camo bonus).

[c3k]

Okay 2 troop types modelled.

1. Feldgrau

2. Some flektarn variety

Game gives 2 a bonus.

Modder makes uniform 1 look like multicam and uniform 2 blaze orange.

Extreme example, but you see the issue?

[agusto]

Modder makes uniform 1 look like multicam and uniform 2 blaze orange.

With a large, red crosshair sewed on the chest!

[A Canadian Cat]

Camo: imagine if it had effect in game. Mods would be critical or simply be disallowed.

There is considerable debate amongst soldiers if camo has any effect in real life.

By that I mean comparing pattern A vs pattern B vs pattern C. I do *not* mean comparing red overalls vs camo.

[c3k]

^^^

Absolutely. And do not construe my camo example to be anything other than an observation about possible effects of implementing a change.

[ASL Veteran]

I do *not* mean comparing red overalls vs camo.

I believe British soldiers used the Red camo pattern at Waterloo if I'm not mistaken.

[A Canadian Cat]

Hee, hee. But those were not overalls

[agusto]

There is considerable debate amongst soldiers if camo has any effect in real life.

Find the soldier:

http://img.gawkerassets.com/img/17o8s3e4os5h5jpg/original.jpg

http://soldiersystems.net/blog1/wp-content/uploads/2012/05/wheres-waldo-1.jpg

http://www.worldaffairsboard.com/attachments/ground-warfare/14094d1239514330-photos-excellent-camouflage-soldiers-afv-aaa-etc-acu-sofa.jpg

Now imagine what it would be like if they d were just normal green uniforms.

[Michael Emrys]

I believe British soldiers used the Red camo pattern at Waterloo if I'm not mistaken.

That was to conceal the post-wounding bleeding from fellow soldiers. So, debatable whether could be considered camo.

Michael

[John Kettler]

ASL Veteran,

And some lucky ones wore the better thought out green pattern!

http://en.wikipedia.org/wiki/Rifle_Brigade_(Prince_Consort's_Own)

Regards,

John Kettler

[Oddball_E8]

Okay 2 troop types modelled.

1. Feldgrau

2. Some flektarn variety

Game gives 2 a bonus.

Modder makes uniform 1 look like multicam and uniform 2 blaze orange.

Extreme example, but you see the issue?

Yes, the issue is that type 1 will never get camo bonus and 2 will always get camo bonus, no matter what the modders do.

That is what i said in my post. So modding will not matter and does not need to be banned.

[Michael Emrys]

http://en.wikipedia.org/wiki/Rifle_Brigade_(Prince_Consort's_Own)

Fixed the link for you, John.

Michael

[akd]

Find the soldier:

http://img.gawkerassets.com/img/17o8s3e4os5h5jpg/original.jpg

http://soldiersystems.net/blog1/wp-content/uploads/2012/05/wheres-waldo-1.jpg

http://www.worldaffairsboard.com/attachments/ground-warfare/14094d1239514330-photos-excellent-camouflage-soldiers-afv-aaa-etc-acu-sofa.jpg

Now imagine what it would be like if they d were just normal green uniforms.

They would definitely not blend in as well with the floral print couch, but then what if they had to hide on a green couch? As for for the others, totally uknown. Might have been the same to the human eye.

[John Kettler]

agusto,

Nice pics, shall have to send that last one to my brother, George, retired from the U.S. Army.

Michael Emrys,

No idea what went wrong there (certainly didn't Copy/Paste the disambiguation page; why would I?), but thanks!

Regards,

John Kettler

[c3k]

Wading in...

The odd problem with Tiger spotting that Steve alluded to, linked in #54 upstream, was my discovery. I did not know the cause, only the result. I ran 1,000's of iterations. I submitted the results, they looked into it, found the cause, and fixed it.

Now, the first question remains, "Is there a problem?" There is definitely the OPINION that there is, but we need more than that.

Tank vision is modeled. Maybe not accurately, but it is modeled. The vision devices are modeled: if the crewman has a vision block, that is different than a periscope, which is different than a rotating periscope which is different than a cupola with 360^ vision blocks. Each one has a different field of view. That is modeled. Test it. I have.

Depending on which position is occupied, the tank gets different spotting abilities. E.g., if no one is manning the 360^ view cupola, the tank is likely blind to the rear. (Depending on other vision devices.) Test it. I have.

Next, the cover/concealment issue. It is also conflated with the spotting interval issue. As well, the background issue.

What are these? Well, we know that the game only checks for LOS to units at certain (random) intervals during the turn. So a unit may be invisible to the tank (for example) if the unit moves during the interval between spotting checks. The interval is random but on the order of 6 seconds, or so. It starts at different times for different units and has a variability. Hence "random".

Admittedly, it is hard to reconcile the OP observation with expectations. What was the difficulty level? How many times did he recompile the turn? Of those times, how many times was the hunter team spotted and killed?

Overall, I agree: my "gut" sense is that tanks spot infantry too easily. Now, let's try to quantify that. That's where it gets harder. Much harder.

Ken

[A Canadian Cat]

Wading in...

<snip>

Overall, I agree: my "gut" sense is that tanks spot infantry too easily. Now, let's try to quantify that. That's where it gets harder. Much harder.

Thank for the comments. Any thoughts on the test I created and reported on in post #66? Is that on the right track? What would be needed - apart from lots more iterations?

My gut says the same. I can see a path to getting some repeatable numbers of some kind. My biggest worry is: what help is that when I do not know what I should be expecting? How do we figure out if what we are seeing in any given test is what we should be seeing?

[c3k]

First, ENSURE that every single unit in the test is identical. All the same stats, all the same weapons, all the same orders, all the same command status, etc.

ENSURE that no result will affect another unit. E.g., lane 2 team being killed does not have a morale penalty felt by lane 28. (Units in the same formation share morale, instaneously, without regard to LOS, etc.)

Next, create a spreadsheet. You'll need it.

Finally, run it so you have several hundred iterations. 30 lanes? 10 runs will give 300 results.

Annotated the time and distance for the first spot of each unit. E.g., Lane 7 spotted at 37m and 15 seconds after passing through the bocage gap. Run some basic maths on the results. Standard devs, etc.

That produces a baseline.

Now, change one variable.

Perhaps kill off one tank crewmember. Or, change the tank type to one with no cupola vision blocks. Or, add another antitank team from the opposite side.

Retest. Compare to baseline.

Change one more variable. Etc.

After 10 tests you'll have 3,000 iterations. (Each test is 10 iterations of 30 lanes.)

That will provide actionable data. Of course, some of it may be garbage if you've changed more than one thing at a time.

It takes time, trial and error, persistance, and plenty of cold beer.

Ken

[Oddball_E8]

First, ENSURE that every single unit in the test is identical. All the same stats, all the same weapons, all the same orders, all the same command status, etc.

ENSURE that no result will affect another unit. E.g., lane 2 team being killed does not have a morale penalty felt by lane 28. (Units in the same formation share morale, instaneously, without regard to LOS, etc.)

Next, create a spreadsheet. You'll need it.

Finally, run it so you have several hundred iterations. 30 lanes? 10 runs will give 300 results.

Annotated the time and distance for the first spot of each unit. E.g., Lane 7 spotted at 37m and 15 seconds after passing through the bocage gap. Run some basic maths on the results. Standard devs, etc.

That produces a baseline.

Now, change one variable.

Perhaps kill off one tank crewmember. Or, change the tank type to one with no cupola vision blocks. Or, add another antitank team from the opposite side.

Retest. Compare to baseline.

Change one more variable. Etc.

After 10 tests you'll have 3,000 iterations. (Each test is 10 iterations of 30 lanes.)

That will provide actionable data. Of course, some of it may be garbage if you've changed more than one thing at a time.

It takes time, trial and error, persistance, and plenty of cold beer.

Ken

Unfortunately most players will not have the tenacity to do that, but they will readily stand firmly by their own "experienced" gut feelings as if it was a fact written in stone.

EDIT: I should add that I personally think that tank spotting behaviour is pretty well modeled and therefore will not perform a test on this scale.

[A Canadian Cat]

Unfortunately most players will not have the tenacity to do that, but they will readily stand firmly by their own "experienced" gut feelings as if it was a fact written in stone.

Indeed. And I for one would rather have some data to back up my gut or change it:

If it disagrees with experiment, it's wrong. - Richard Feynman

Having said that I am not sure if I have the fortitude to do what @c3k is suggesting. It must be because I don't like beer:)

[c3k]

LOL. I know. I tried to prove a specific hypothesis I had. It took me a week to get just the map right. It involved a spiderweb type setup. A spotting tank would be in the center with enemy tanks placed around the radius. Hub and spoke. I started with about 2 battalions (~100 tanks) and dismounted all the crews. I ran them through a gauntlet of single shot snipers until all the TC's were killed. Then I used however many crews had 4 survivors, but not the TC. I sat them for ~20 minutes to recover morale, remanned their tanks, then moved the tanks into the center of the web, one at a time.

That was my 4 man baseline. (From a 5 man crew.) (I didn't want to run the test with the same crew. That would, possibly, bias all the results. I wanted several dozen different crews, but all with their TC killed off.)

Then I did it again to get 3 man crews.

Then I repeated it all with a different model of tank. (Viewing devices were different.)

Oh, and I also changed the radius of the spokes.

I did all this well before BN was released. I did not prove my hypothesis.

Do I want to do another test like this because someone is complaining about one result (which may just be an outlier)? Not at all.

[Michael Emrys]

It's reassuring to note that there are people who are almost as lazy as I am. I think being a go-getter is highly overrated.

Michael

[poesel]

Tank vision is modeled. Maybe not accurately, but it is modeled. The vision devices are modeled: if the crewman has a vision block, that is different than a periscope, which is different than a rotating periscope which is different than a cupola with 360^ vision blocks. Each one has a different field of view. That is modeled. Test it. I have.

I don't think anyone disputes that. It only looks like these guys in the tanks work like robots until they freak out and bail. Ever observing all that is possible, near instant reaction to threats and very effective replacement of taken out crew members.

What are these? Well, we know that the game only checks for LOS to units at certain (random) intervals during the turn. So a unit may be invisible to the tank (for example) if the unit moves during the interval between spotting checks. The interval is random but on the order of 6 seconds, or so. It starts at different times for different units and has a variability. Hence "random".

AFAIK the intervals are not random but the start of the interval is random per team. Clever way to spread out the load for calculations.

But the worst case with this system is that a team may be sitting nearly 6 seconds directly under an enemy tank without noticing it. Corner case - maybe. But for close assaults on tank it actually happens quite often.

Overall, I agree: my "gut" sense is that tanks spot infantry too easily. Now, let's try to quantify that. That's where it gets harder. Much harder.

For a good test you first need a hypothesis. What is the expected outcome for, say, this situation:

1) buttoned M4A1 under fire from 20mm gun, non-firing panzerschreck team in high grass 50m to the back

2) same, team 50m to one side

3) same as 1), no 20mm

4) same as 2), no 20mm

5-8) same as 1-4) but low grass

Everyone, regular, +0, flat ground, sunny midday

So how many seconds (or never) until the tank spots the team? What do you think?

[Vanir Ausf B]

For a good test you first need a hypothesis.

What is definitely needed is something to compare the results to so that there is some basis to make a judgement. Otherwise the interpretation of the results will be largely subjective. For example, I keep coming back to your own testing of tanks spotting tanks vs. spotting infantry. I think almost everyone would agree that a Sherman tank sitting in an open field should be more easily spotted than a squad of infantry in the same field, but your test suggests this is not so in the game.

[John Kettler]

Vanir Ausf B,

All that roaring, snorting and track squealing (was around a late model Sherman on concrete about a month ago) would, you'd think, make it practically impossible for infantry to fail to notice a tank, particularly an approaching one. Bit of a difference in presented area, too!

Radial engine Sherman

GAA V8 for a Sherman If you thought the radial was loud. The AT team might go deaf!

Regards,

John Kettler