Jump to content
You need to play a total of 50 battles to post in this section.
wilkatis_LV

Dispersion - it's worse than it seems

28 comments in this topic

Recommended Posts

[ADRIA]
[ADRIA]
Players
4,914 posts
7,006 battles

Everything written here is based on LittleWhiteMouses work of mapping ships dispersion patterns, thanks her for that :cap_like:

 

Disclaimer: this is not me trying to say that something she has done / shown / written is wrong. Quite the opposite - I'm aiming to add more information to what's already known

 

I'm sure most if not all of you are familiar with LWMs format: 15km, locked onto a Fuso, something like 180 shells mapped, the ship firing is to the right side of the image (not below it). Just a reminder of how that looks:

Spoiler

Vanguard:

dispersion2-1024x576.jpg

 

PEF:

efdUvDg.png

 

Bourgogne:

2K7qC2G.png

 

WV41:

VY1YTLt.png

 

Nice pictures, sure, but I wanted to put some numbers behind them. Lets go with PEFs one as the example here simply bcuz that's 1st one I chose to check, but after testing the same points on multiple others I can say that this isn't just some magical exception

 

We can compare the width of Fuso with the width of the dispersion area in the image by simply measuring them:

JDX5sOG.png

 

From that we get their size ratio to be x5.355

 

Fusos beam is listed as 28.65m, so with the x2 ship size scaling (or was it x2.61? - whoever knows leave it in comments, I'll probably edit the wrong one out)  that would be 57.3m (or 74.8m) in game.

This gives us horizontal dispersion of 307m (or 401m) :fish_book:

 

Where's the problem? German BBs horizontal dispersion is calculated as Range[km] * 9.8 + 66 = H.Disp[m]

In case you're curious, list of "how you calculate h.disp." for all ships:

Spoiler

7NK3ocx.png

 

So from that we get 15 * 9.8 + 66 = 213m

What we have practically vs what we "should" have had is x1.44 (or x1.88) larger :cap_fainting:

 

This means that dispersion area, as bad it might have been expected to be already by the numbers given to us... is actually quite a bit worse. Noticeably worse.

When asked about it LWMs answer was:

On 12/31/2018 at 5:57 PM, LittleWhiteMouse said:

What I suspect is that the horizontal dispersion value is something similar to what World of Tanks used -- where the spread described where 50% of the shells would land.

Which certainly makes sense. Quite sad, but makes sense. 

 

So now when you know your chances of hitting the target are even worse than you might have previously expected - happy hunting :Smile_veryhappy:

 

P.S. I wonder how this affects CV drop ellipses for bombs and rockets? :cap_hmm: Today - about an hour ago from me posting this - Fara uploaded video showing the reworked Hoshos gameplay, and there he managed to slingshot an AP bomb way out of the ellipse... :cap_yes:

  • Cool 14

Share this post


Link to post
Share on other sites
[ICI]
[ICI]
Players
724 posts
4,096 battles

That is some good data. 

Share this post


Link to post
Share on other sites
[DREAD]
Players
6,071 posts
5,294 battles

Hmm didn’t they say in one of the “How it works” videos that all shots land inside the ellipse? 

 

Are you sure the multiplier (x2) is even applied in the target ellipse size? I mean just because ships are twice the size it doesn’t mean that everything is twice the size. Same for time btw. there are also different time compression factors as well if I remember correctly (something like for shots/shells and another one for ship movement)

Share this post


Link to post
Share on other sites
[ADRIA]
[ADRIA]
Players
4,914 posts
7,006 battles
40 minutes ago, Darth_Glorious said:

In port horizontal dispersion is the horizontal standard deviation :

https://en.wikipedia.org/wiki/Normal_distribution#/media/File:Empirical_Rule.PNG

In port your horizontal dispersion is that same value you can calculate by whichever formula fits your ship, just at your listed max range.

 

For example - that GK in your signature.

  • Base range 20.6km -> 268m listed max h.disp. (with acc in 3rd slot 249m)
  • Range mod 23.9km -> 300m listed max h.disp. (with acc in 3rd slot 279m)
  • Legendary 18.95km -> 252m listed max h.disp. (with acc in 3rd slot 234m)
  • Legendary + Range 21.98km -> 281m listed max h.disp. (with acc in 3rd slot 262m)

They may differ by +/- 1m due to rounding.

 

 

The "standard deviation" part of accuracy is sigma, and that's not shown in-game

 

 

35 minutes ago, 1MajorKoenig said:

Hmm didn’t they say in one of the “How it works” videos that all shots land inside the ellipse?

I'm pretty sure they did.

 

*(One quick search later)*

They kind-of mention it at 1:14, but it's definitely stated at 4:05 (video should start at this)

 

Tbh not even surprised :cap_old: In the video about aircraft they also stated that "all aircraft see 11km" which is not true as floatplane spotters actually see 15.

 

44 minutes ago, 1MajorKoenig said:

Are you sure the multiplier (x2) is even applied in the target ellipse size? I mean just because ships are twice the size it doesn’t mean that everything is twice the size.

It's not. The only thing increased is the ship - but it needs to be increased as it's in-game and that's where I'm taking the size from.

 

Effectively look at it as 53-point-something (or 70-something) meter wide ship compared to the width of dispersion as it is.

 

And it couldn't really be the "you're not locked so it's 2x dispersion size" - 1st of all she is locked, and 2nd - neither 1.44 nor 1.88 equal to 2, as close as the 2nd gets to it.

  • Cool 1

Share this post


Link to post
Share on other sites
Beta Tester
760 posts
11,590 battles

You search for truth in a game where scaling has been twisted to the point that a 100m long DD can't fit in the 2km gap between two ships? Go outside ask why the sun shines, it has been here from the begining, will likely stay that way.

  • Funny 2

Share this post


Link to post
Share on other sites
[TNG]
Players
17 posts
1,769 battles

This doesn't change anything really. The proportions are still the same, or am I wrong? It's still just mindblowing to me that WtfGaming doesn't just give us the stats and formulas.

Share this post


Link to post
Share on other sites
[YARRR]
Beta Tester
8,723 posts
14,669 battles
4 hours ago, wilkatis_LV said:

and there he managed to slingshot an AP bomb way out of the ellipse... :cap_yes:

 

Aircraft aim is independent of the actual drop circle, leading to extremely wonky behavior when attacking over obstacles as Fara did.

Is ready for release comrade!

  • Funny 2

Share this post


Link to post
Share on other sites
Players
4,085 posts
6,476 battles
4 hours ago, 1MajorKoenig said:

Are you sure the multiplier (x2) is even applied in the target ellipse size? I mean just because ships are twice the size it doesn’t mean that everything is twice the size.

 

3 hours ago, wilkatis_LV said:

It's not. The only thing increased is the ship - but it needs to be increased as it's in-game and that's where I'm taking the size from.

 

Thats what I was thinking about aswell. But its the same thing here as with other values that are shown or sometimes not shown to us: We can only guess if those numbers given are the real values, that are used for the mechanics or if they are just "make-up". I mean, asking the question from the other side: How do we prove, the numbers are correct?

 

3 hours ago, IronJew said:

This doesn't change anything really. The proportions are still the same, or am I wrong? It's still just mindblowing to me that WtfGaming doesn't just give us the stats and formulas. 

 

It doesnt change anyway in that sense, true. Its a little bit like with the fire chance, that is shown for each ship. We have ships, that tell you, you´ll have a 42% fire chance. Then you can put some flags on to add 1%/2%. So you might think "meh, its not worth it, ill start a fire with every second hit anyway". But the real fire chance is something completly different, since we have hidden factors. And there are modules ofc, that can lower the fire chance of someone shooting at you. So what is your real fire chance? You will never be able to tell for a specific engagement. Like you say: stats and formulas are hidden. Why? Who knows. When I look at all the constant, unneccsairy chages, that WG is applying (ribbons change, position of ribbons change, the way ribbons are shown changes, Hydro gets unified aka changed for all ships, concealment gets unified aka changed for all ships, radar ranges get changed, gunbloom when not spotted gets changed to 3 sek then back to 20 sek - while the video explaining it now wrong still remains online.... my god there is plenty of more that I cant come up with now) - It looks like a strategy to confuse people.

Share this post


Link to post
Share on other sites
[DREAD]
Players
6,071 posts
5,294 battles
7 hours ago, wilkatis_LV said:

In port your horizontal dispersion is that same value you can calculate by whichever formula fits your ship, just at your listed max range.

 

For example - that GK in your signature.

  • Base range 20.6km -> 268m listed max h.disp. (with acc in 3rd slot 249m)
  • Range mod 23.9km -> 300m listed max h.disp. (with acc in 3rd slot 279m)
  • Legendary 18.95km -> 252m listed max h.disp. (with acc in 3rd slot 234m)
  • Legendary + Range 21.98km -> 281m listed max h.disp. (with acc in 3rd slot 262m)

They may differ by +/- 1m due to rounding.

 

 

The "standard deviation" part of accuracy is sigma, and that's not shown in-game

 

 

I'm pretty sure they did.

 

*(One quick search later)*

They kind-of mention it at 1:14, but it's definitely stated at 4:05 (video should start at this)

 

Tbh not even surprised :cap_old: In the video about aircraft they also stated that "all aircraft see 11km" which is not true as floatplane spotters actually see 15.

 

It's not. The only thing increased is the ship - but it needs to be increased as it's in-game and that's where I'm taking the size from.

 

Effectively look at it as 53-point-something (or 70-something) meter wide ship compared to the width of dispersion as it is.

 

And it couldn't really be the "you're not locked so it's 2x dispersion size" - 1st of all she is locked, and 2nd - neither 1.44 nor 1.88 equal to 2, as close as the 2nd gets to it.

 

Gotcha. It COULD still be the other way round though - that the same scale as for ships is applied

 

That would give for your picture at 15km:

 

x = 166/31 x 28.65m = 153.x m

 

Could that be?

 

 

EDIT: that would be max dispersion of the ellipse though. Doesn’t mean any of the 180 shells need to land right there in the corners of course but they could.

  • Cool 1

Share this post


Link to post
Share on other sites
Players
1,373 posts
5,380 battles

The Wows Wiki confirms that the port dispersion figure is the horizontal distance where at least 50% of your shells will land, so your actual "widest to widest" shell group WILL be wider than the shell dispersion given in Port. 

 

That explains the difference in the OP's group size vs port figure. 

 

http://wiki.wargaming.net/en/Ship:Gunnery_%26_Armor_Penetration#Shell_Dispersion

  • Cool 1

Share this post


Link to post
Share on other sites
[TNG]
Players
17 posts
1,769 battles
Vor 9 Stunden, ForlornSailor sagte:

It doesnt change anyway in that sense, true. Its a little bit like with the fire chance, that is shown for each ship. We have ships, that tell you, you´ll have a 42% fire chance. Then you can put some flags on to add 1%/2%. So you might think "meh, its not worth it, ill start a fire with every second hit anyway". But the real fire chance is something completly different, since we have hidden factors. And there are modules ofc, that can lower the fire chance of someone shooting at you. So what is your real fire chance? You will never be able to tell for a specific engagement. Like you say: stats and formulas are hidden. Why? Who knows. When I look at all the constant, unneccsairy chages, that WG is applying (ribbons change, position of ribbons change, the way ribbons are shown changes, Hydro gets unified aka changed for all ships, concealment gets unified aka changed for all ships, radar ranges get changed, gunbloom when not spotted gets changed to 3 sek then back to 20 sek - while the video explaining it now wrong still remains online.... my god there is plenty of more that I cant come up with now) - It looks like a strategy to confuse people.

 

You are completely right about the first part of your paragraph, I assume the Aiming module is a little more effective than it seemed.

 

About the second part -- The Russians are good at confusing everyone, including themselves.

Share this post


Link to post
Share on other sites
[DREAD]
Players
6,071 posts
5,294 battles
1 hour ago, IanH755 said:

The Wows Wiki confirms that the port dispersion figure is the horizontal distance where at least 50% of your shells will land, so your actual "widest to widest" shell group WILL be wider than the shell dispersion given in Port. 

 

That explains the difference in the OP's group size vs port figure. 

 

http://wiki.wargaming.net/en/Ship:Gunnery_%26_Armor_Penetration#Shell_Dispersion

 

Thanks for the link. However - it seem like written by players as it says “confirmed by experiments by players”. Sounds like one - maybe sound but still only one - of the options? I mean is there any confirmation from WG? I remember a very good thread on dispersion by Sub_Octavian on the NA forum a while ago. I have to find that one 

Share this post


Link to post
Share on other sites
Players
4,085 posts
6,476 battles

@wilkatis_LV just thinking... The value given for maximum dispersion - it should be the one when shooting unlocked? Now the dispersion pattern on the screen is obviously locked, so you´d except a tighter, better dispersion. Maybe that would explain the discrepancy? If you´d go and shoot unlocked, then maybe you´ll get the numbers you´d expect to see?

Share this post


Link to post
Share on other sites
[ADRIA]
[ADRIA]
Players
4,914 posts
7,006 battles
5 hours ago, IanH755 said:

The Wows Wiki confirms that the port dispersion figure is the horizontal distance where at least 50% of your shells will land, so your actual "widest to widest" shell group WILL be wider than the shell dispersion given in Port. 

 

That explains the difference in the OP's group size vs port figure. 

 

http://wiki.wargaming.net/en/Ship:Gunnery_%26_Armor_Penetration#Shell_Dispersion

That actually is a very very new line there :Smile_veryhappy:

 

The whole Gunnery page seems to be heavily updated lately. Guess someone finally re-started updating wiki?

 

4 hours ago, 1MajorKoenig said:

is there any confirmation from WG?

There never is. At least I haven't seen that happening even once

 

2 hours ago, ForlornSailor said:

@wilkatis_LV just thinking... The value given for maximum dispersion - it should be the one when shooting unlocked? Now the dispersion pattern on the screen is obviously locked, so you´d except a tighter, better dispersion. Maybe that would explain the discrepancy? If you´d go and shoot unlocked, then maybe you´ll get the numbers you´d expect to see?

No, that's dispersion when you are locked as you are supposed to be locked every time you shoot. That's the "natural" state of it. Being "unlocked" is seen as a penalty on top of that

 

Also we see way wider numbers than expected despite being locked on target. Since LWM is locked on the target they should be the tighter version, and even if the 213 was "unlocked" -  we should have gone below it not way above :cap_yes:

Share this post


Link to post
Share on other sites
Players
4,085 posts
6,476 battles
23 minutes ago, wilkatis_LV said:

Also we see way wider numbers than expected despite being locked on target. Since LWM is locked on the target they should be the tighter version, and even if the 213 was "unlocked" -  we should have gone below it not way above :cap_yes:

 

Uh yea, you are right, I was thinking in the wrong direction. But your answer brought me to a new idea. You have calculated the width between the 2 shells, that have landed the farest from each other, right? And got 401 m. Now the dispersion in game says 213 m. This should be counted from the middle to where I aim, so right there, where the Fuso is on that picture or exactly in between the 401 m. That means, it can spray 213 m in both directions from there, giving us a, so to say, dispersion pattern that strechtes over 426 m? If I havent missed something, I´d be willing to accept those numbers, because you prolly havent seen the worst dispersion with 180 or so shells? So if youd keep on shooting you might see worse dispersion, getting closer to that 426 / 213 m maximum dispersion?

Share this post


Link to post
Share on other sites
[ADRIA]
[ADRIA]
Players
4,914 posts
7,006 battles
8 minutes ago, ForlornSailor said:

Uh yea, you are right, I was thinking in the wrong direction. But your answer brought me to a new idea. You have calculated the width between the 2 shells, that have landed the farest from each other, right? And got 401 m. Now the dispersion in game says 213 m. This should be counted from the middle to where I aim, so right there, where the Fuso is on that picture or exactly in between the 401 m. That means, it can spray 213 m in both directions from there, giving us a, so to say, dispersion pattern that strechtes over 426 m? If I havent missed something, I´d be willing to accept those numbers, because you prolly havent seen the worst dispersion with 180 or so shells? So if youd keep on shooting you might see worse dispersion, getting closer to that 426 / 213 m maximum dispersion?

It's 401m if the x2.61 scaling is correct, but I think it was the x2. And this case would fit only the x2.61 sizes

 

I did consider this initially, but every reference I've read / heard / seen talks about h.disp. as "width" a.k.a. full size of it not just the length of each of the 2 short radii of the ellipse.

Of course, as 1MajorKoenig already mentioned shells don't have to go to the edge simply because they can, and the x1.88 expected size is quite close to the x2 :cap_yes:

 

But yeah, I'd put that as the 2nd best theory (closely) right after the "dispersion given is only for 50% of shells", and that only because it has been mentioned that WG used to do (still do? no idea) this in their other titles (WoT being the example here)

Share this post


Link to post
Share on other sites
[DREAD]
Players
6,071 posts
5,294 battles
9 minutes ago, wilkatis_LV said:

It's 401m if the x2.61 scaling is correct, but I think it was the x2. And this case would fit only the x2.61 sizes

 

I did consider this initially, but every reference I've read / heard / seen talks about h.disp. as "width" a.k.a. full size of it not just the length of each of the 2 short radii of the ellipse.

Of course, as 1MajorKoenig already mentioned shells don't have to go to the edge simply because they can, and the x1.88 expected size is quite close to the x2 :cap_yes:

 

But yeah, I'd put that as the 2nd best theory (closely) right after the "dispersion given is only for 50% of shells", and that only because it has been mentioned that WG used to do (still do? no idea) this in their other titles (WoT being the example here)

 

...and we have theory no.3:

 

- same scale for ellipse as for ships -> meaning the dispersion here actually happens in 166/31x28.x m = 153m

 

- max vertical dispersion at 15km is calculated as 213 m (main gun module in use?)

 

- potentially we just haven’t seen any hits outside 153m in this sample but hits between 153m and 213m might still happen. Sigma would be driving the distribution of shells towards the center anyway 

 

 

While I have seen the WIkI page I somewhat doubt the concept of standard dispersion. It would be fundamentally wrong to speak about MAX dispersion in that case 

  • Cool 1

Share this post


Link to post
Share on other sites
[ADRIA]
[ADRIA]
Players
4,914 posts
7,006 battles
1 hour ago, 1MajorKoenig said:

- max vertical dispersion at 15km is calculated as 213 m (main gun module in use?)

That's actually a good point

 

LWM uses it for tests, right? That would take the dispersion down to 198m

 

But yeah, that would be the 3rd way how this same point could work with the same outcome

Share this post


Link to post
Share on other sites
[DREAD]
Players
6,071 posts
5,294 battles

I am curious now - is there any way we get to know which one is the correct one? 

 

Didn’t follow the discussion on NA - did they come to a conclusion?

 

Maybe ask Sub_Octavian?

Share this post


Link to post
Share on other sites
Players
4,085 posts
6,476 battles
34 minutes ago, 1MajorKoenig said:

I am curious now - is there any way we get to know which one is the correct one? 

 

I think its very hard to find out, which one it is. With test you will hardly get a 100% correct match, since I think the absolute worst dispersion - which you would need to match - will occure really rare?

But much like you im currious myself.

Share this post


Link to post
Share on other sites
[ADRIA]
[ADRIA]
Players
4,914 posts
7,006 battles
6 hours ago, 1MajorKoenig said:

I am curious now - is there any way we get to know which one is the correct one? 

Probably not. But we can always tag the usual suspects like @MrConway and @Crysantos so they can maybe poke the devs a little bit

(Conway, Crysantos - check the opening post, that's what we're interested in)

 

6 hours ago, 1MajorKoenig said:

Didn’t follow the discussion on NA - did they come to a conclusion?

Was there a NA discussion aswell? That LWM quote came from PMs :Smile_veryhappy:

  • Cool 1

Share this post


Link to post
Share on other sites
[WG-EU]
WG Staff, Alpha Tester
2,646 posts
2,413 battles
14 hours ago, wilkatis_LV said:

Probably not. But we can always tag the usual suspects like @MrConway and @Crysantos so they can maybe poke the devs a little bit

(Conway, Crysantos - check the opening post, that's what we're interested in)

 

Was there a NA discussion aswell? That LWM quote came from PMs :Smile_veryhappy:

 

It is very early for math.

 

I have no reason to believe that the information given in the wiki (that the value means the area that 50% of the fired shells will land in) is incorrect. LWMs experimentation also seems to support this.

 

We're planning to have more guests from the Dev team on the stream throughout the year, I'll see if we can maybe do a game mechanics deep-dive if I can find a willing victim ;)

  • Cool 2

Share this post


Link to post
Share on other sites
Players
12 posts
1,107 battles
17 minutes ago, MrConway said:

We're planning to have more guests from the Dev team on the stream throughout the year, I'll see if we can maybe do a game mechanics deep-dive if I can find a willing victim ;)

 

Yes please, that'd be interesting to watch.

Share this post


Link to post
Share on other sites

  • Recently Browsing   0 members

    No registered users viewing this page.

×