SPREAD pattern. (hit distribution inside the cone)


(Glottis-3D) #1

Smooth said that now the hit distribution inside the spread-cone is totally random.
I thought, that this may be the reason, why spread (espesially in the MAX spread situation) feels so random.

So i thought that we could try a more ‘natural’ normal-distribution function, i.e. Gaussian function.
which looks like this:


the probability is axis Y, coordinates is axis X.

So, one of the methods to generate random coordinates for normal-distribution things(from uniformely random) is:

And with this kind of distribution. the hit area will transform like this:


This leads us to the situation, in other words, where the outer parts of the spread-cone are less probable than central area.

Not only this. Because pretty much a very lot of physical processes in the world are distributed according to this gaussian-law (like sand dropping from your hand on the floor, etc). i thing this distribution will feel more natural.

Why this even the issue? because when spread-cone goes as large as or bigger than hitbox, i get constant feeling that i hit absolutely zero. and i just dont like that in max spread all weapons are thus useless. they should be just ‘much less usefull’.


(spookify) #2

[QUOTE=krokodealer;490488]Smooth said that now the hit distribution inside the spread-cone is totally random.
I thought, that this may be the reason, why spread (espesially in the MAX spread situation) feels so random.

So i thought that we could try a more ‘natural’ normal-distribution function, i.e. Gaussian function.
which looks like this:


the probability is axis Y, coordinates is axis X.

So, one of the methods to generate random coordinates for normal-distribution things(from uniformely random) is:

And with this kind of distribution. the hit area will transform like this:


This leads us to the situation, in other words, where the outer parts of the spread-cone are less probable than central area.

Not only this. Because pretty much a very lot of physical processes in the world are distributed according to this gaussian-law (like sand dropping from your hand on the floor, etc). i thing this distribution will feel more natural.

Why this even the issue? because when spread-cone goes as large as or bigger than hitbox, i get constant feeling that i hit absolutely zero. and i just dont like that in max spread all weapons are thus useless. they should be just ‘much less usefull’.[/QUOTE]

Great Post Totally agree!

Even Tapping or Bursting doesn’t seem to affect the spread which is the most frustrating! I am not getting the feel or the hit I think I should be getting especially with my play style which I feel is too ETish and I hit max spread right away.

Sprint Spread - Hip Fire needs to be toned down…

Played last night for one full hour 2 v 2 and it felt like all my kills were luck…


(stealth6) #3

I think I understand your point, but could you have picked worse images? I guess you have to ignore all letters, numbers & colors to make some sense out of them.

Also on the first image I think you mean that the X-axis is offset not coordinates.


(Glottis-3D) #4

[QUOTE=stealth6;490490]I think I understand your point, but could you have picked worse images? I guess you have to ignore all letters, numbers & colors to make some sense out of them.

Also on the first image I think you mean that the X-axis is offset not coordinates.[/QUOTE]
sorry about images - took second from a wiki page, first from some math-rlated site.
X-axis is any measureable property - for us it is one coordinate of the ‘Random’ Hit on the wall (from example) or - yes - the offset from your crosshair.

and the numbers in the second image are:
in the top circle they stand for coordinates of the circle points (x,y) taken uniformely from 0 to 1.(i.e. the radius is sqrt(x^2+y^2)) In practice these (x,y) coordinates will be our uniformely random shot coordinates.
in the low circle these are new normal coordinates via Box-Muller formula

-upd-
in the wiki page the formul for our case goes in ‘Polar form’ section.


(Smooth) #5

What you want it just a tighter max spread basically, which we’re looking at for some (if not all) weapons.

A system where shots are weighted towards the centre of the spread, but where the odd shot can ‘escape’ and hit the very edge would actually introduce a more random feeling for if the shots were on average hitting the same area.

If that doesn’t make sense to you; if we were to weight bullets to the centre of the spread, to maintain a similar/average distribution pattern we’d have to introduce a higher maximum spread at the extremities. Yes more would be in the middle, but the odd occasional edge-bullet would feel very weird. It also introduces another (largely unneeded) variable to balance when we can achieve almost the same thing by just making the current cone smaller.

We’re also toying with have a more gradual spread increase over time, but this requires a few more changes to properly support. You won’t see these in the next update, but hopefully the one after that.


(Glottis-3D) #6

i totally undestand this, but thats how nature seem to work. And anyway those odd shots would be VERY rare like once in a million shots.

because, like in this picture for Gaussian:


hits that go 3 times the Sigma (standart deviation) are less than 0.1% of probability .

+++plus you can restrict extreme odd shots, add another ‘If…then…’ you guys are the programmers, not me :DDD


(BAMFana) #7

I don’t like overly randomized spread patterns regardless of what the distribution can be classified as. Individual fixed spread patterns for each gun are preferable because they remove any and all complaints about luck while still allowing the devs to fine tune the spread of each gun. Fixed spread patterns also introduce more depth to the game, because it allows the players to – through experience – learn to reliably compensate for the spread pattern of each gun and thus increase their effectivity in combat.

The argument – endemic of modern casualized games – that random spread patterns are more realistic and thus preferable makes little sense to me. Gameplay concerns should always trump realism concerns, and in this case there very good arguments in favor of non-realism (fixed spread) improving gameplay, and no good arguments for realism (randomized spread pattern) improving gameplay. While the angle of the skill curve is a concern, to prevent less skilled players from being dominated in every encounter, this sort of “random chance based” leveling of the skill curve is really not the answer.

If Extraction used fixed spread patterns, the complaints that the spread is too wide would immediately be invalidated, as it is only the player’s own fault for not adequately evaluating the situation (can I win this fight, given that the spread of my gun is so-and-so and given that I can compensate so-and-so?) before deciding to engage. Instead, with a randomized spread, players will always complain – and to a certain degree their complaints will be valid – that their fate is being decided by RNG rather than their own skill, because each player’s evaluation is now limited to an assessment of probabilities (can I win this fight more than 50% of the time, given that the spread of my gun is so-and-so?). This is of course a very abstract representation of actual gameplay, but it should make the implications clear.

Many modern game developers unfortunately, in my opinion, value realism too highly when considering the gameplay implications of certain mechanics. I think that chance elements interfering with player control (this is an important distinction; chance elements that affect the environment or other variables that are not directly related to a player’s control of his avatar are generally beneficial) are generally detrimental to gameplay quality, and can turn players off of a game rather than on. I don’t think casual players care whether the spread is random or not, but it can certainly frustrate skilled players as demonstrated by the repeated “omg spread” complaints from alpha/beta players, to the point where they find little enjoyment in a game.

[Edit: A bunch of changes to hopefully make my intentions a bit clearer. Read later posts in the thread to further clarify the point, as the terminology I used in this post can be especially confusing.]


(Glottis-3D) #8

i understand, that in shotguns you can have spread pattern, but how in a AR you can have fixed spread?
1st shot goes center
2nd shot goes always left for 1 degree
3rd shot goes always right 2 degrees??

no jokes intended, i literally dont know what is fixed spread pattern for an automatic gun


(spookify) #9

THIS!
This is just like BF3 that had 50 guns or whatever and each had its own spread, recoil and ROF. It took awhile to master each gun but at least it wasnt tht random and each gun could be mastered.

NOTE: Not all guns were mastered in BF3 at least for me. I could only handle or liked 10 of the 50ish guns!
Actauly now that I look at my stats I only mastered 5 haha!
http://bf3stats.com/stats_pc/spookify4

I think I would have used more if I did quit. I was working on 10K kills per gun… PLUS I was also banned from 90% of all metro servers so the game was insta dead to me when they increase the suppression.



(Smooth) #10

Spread exists in-game as a translation of a shooter’s aim, which is never going to be perfect. That’s why it:

[ul]
[li]Is less accurate from the hip and when ironsights[/li][li]Is less accurate when you’re running and jumping around than crouched and stationary[/li][li]Gets less accurate over time as you fire the weapon[/li][/ul]
We want to reward players who do take the time to stop an aim at their targets with some benefit to offset the massive downside of becoming a more static target. It’s also necessarily random (within limits) because otherwise the purpose of it’s existence is quickly defeated.

This introduces choice and depth into the combat that would not otherwise exist if the ultimate ‘best’ method was to ALWAYS run-and-gun. In the same way as having to ALWAYS stand still and iron-sight in other games also removes choice and depth.

EDIT: BF3 uses random recoil and spread (with min/max limits) almost exactly like us. We just have a much tighter spread and generally less recoil :stuck_out_tongue:

Pointing to this thread again: http://forums.warchest.com/showthread.php/38582-the-current-state-of-the-spread-system?p=483134&viewfull=1#post483134


(spookify) #11

[QUOTE=Smooth;490508]Spread exists in-game as a translation of a shooter’s aim, which is never going to be perfect. That’s why it:

[ul]
[li]Is less accurate from the hip and when ironsights[/li][li]Is less accurate when you’re running and jumping around than crouched and stationary[/li][li]Gets less accurate over time as you fire the weapon[/li][/ul]
We want to reward players who do take the time to stop an aim at their targets with some benefit to offset the massive downside of becoming a more static target.

This introduces choice and depth into the combat that would not otherwise exist if the ultimate ‘best’ method was to ALWAYS run-and-gun. In the same way as having to ALWAYS stand still and iron-sight in other games also removes choice and depth.

EDIT: BF3 uses random recoil and spread (with min/max limits) almost exactly like us. We just have a much tighter spread and generally less recoil :p[/QUOTE]

This sounds great on paper but in its current state it is not making the transition to the game. Hopefully your coming tweaks will makes me stop wanted to rip my eyes out.

Edit: Red Dot might help this however I am not a big IS fan…


(attack) #12

[QUOTE=Smooth;490508]

EDIT: BF3 uses random recoil and spread (with min/max limits) almost exactly like us. We just have a much tighter spread and generally less recoil :p[/QUOTE]
to compare bf3 with xt is lets say od.


(BAMFana) #13

[QUOTE=krokodealer;490504]i understand, that in shotguns you can have spread pattern, but how in a AR you can have fixed spread?
1st shot goes center
2nd shot goes always left for 1 degree
3rd shot goes always right 2 degrees??

no jokes intended, i literally dont know what is fixed spread pattern for an automatic gun[/QUOTE]

There are a few different ways to do it. I don’t know if any games have an absolutely fixed spread pattern that maps each bullet to the exact same trajectory each time, although that could potentially be pretty cool, but games that are favorable to competitive play generally have spread patterns that are more or less fixed (CS:GO example showing the combined effect of recoil and spread, although it isn’t entirely accurate since decals are client side), rather than entirely randomized spread within a cone of fire. I guess we could say that there is an idealized spread pattern template (a line that the bullets follow when you hold fire), and the bullets then deviate randomly from that line depending on differing variables (the effect of both recoil and spread have to be taken into consideration here). Personally I think a good way to do it would be to have the bullets follow the spread pattern line very accurately when not moving and crouched or ironsighted (probably want a different, tighter, pattern when ironsighted as well), and adding some random deviation when the player is moving while shooting.

Another point is the random variance attributed to the first bullet fired. In previous versions of Counter-Strike, for example, the first bullet fired would always hit dead center in the crosshairs. I think it makes sense to have this as the default behavior for certain guns in Extraction even when hipfiring (Arty’s semi-auto rifle comes to mind), although it might not be ideal to have it for all guns.

[QUOTE=Smooth;490508]We want to reward players who do take the time to stop an aim at their targets with some benefit to offset the massive downside of becoming a more static target. It’s also necessarily random (within limits) because otherwise the purpose of it’s existence is quickly defeated.

This introduces choice and depth into the combat that would not otherwise exist if the ultimate ‘best’ method was to ALWAYS run-and-gun. In the same way as having to ALWAYS stand still and iron-sight in other games also removes choice and depth.[/QUOTE]
I certainly agree with your goals, but, assuming that my understanding of how the spread currently works in Extraction and how you plan it will work in the future is somewhat correct, I don’t think your chosen path is ideal for accomplishing these goals without causing other problems. See my reply to kroko earlier in this post.

I don’t think that the “purpose of of it’s existence is quickly defeated” by limiting the random variance to a defined pattern. The combination of recoil (in Extraction’s case a forced movement of the player’s view) and spread pattern makes it physically impossible for a human to correct perfectly each time, even on a stationary target with 100 % accurate adherence to the recoil pattern.

[Edit: A bunch of changes to hopefully make my intentions a bit clearer. Read later posts in the thread to further clarify the point, as the terminology I used in this post can be especially confusing.]


(spookify) #14

[QUOTE=Smooth;490508]
EDIT: BF3 uses random recoil and spread (with min/max limits) almost exactly like us. We just have a much tighter spread and generally less recoil :p[/QUOTE]

But xT is a lot more run and gun and tracking players to kill then BF3. The running spread is to high in comparison. You give it the feel of ET but when you strafe you get nothing. I think the spread as you talk about works for BF3 because it takes half the shots to kill someone in BF3 then XT plus arent the XT hit boxes smaller then BF3?

So you are making everything harder, tighter and more random but not tweaking the other variables to makes up for this. THIS is why we the players are getting this lucky or random feel when we play.


(Smooth) #15

The game isn’t done, there’s a LOT of changes coming for things like this and hopefully they do make it feel much better :slight_smile:


(Smooth) #16

[QUOTE=spookify;490513]But xT is a lot more run and gun and tracking players to kill then BF3. The running spread is to high in comparison. You give it the feel of ET but when you strafe you get nothing. I think the spread as you talk about works for BF3 because it takes half the shots to kill someone in BF3 then XT plus arent the XT hit boxes smaller then BF3?

So you are making everything harder, tighter and more random but not tweaking the other variables to makes up for this. THIS is why we the players are getting this lucky or random feel when we play.[/QUOTE]

I’m saying the system is similar, not that the values or behaviour is the same.

See: http://forums.warchest.com/showthread.php/38582-the-current-state-of-the-spread-system?p=483134&viewfull=1#post483134


(Mustang) #17

@BAMFana:

You know you’re talking about recoil patterns and not spread patterns right?
Stop confusing things by mixing terms. :tongue:


(potty200) #18

The spread on the first 3-4 bullets seems fine. The increase in spread over a short time is too high and the MAX spread of every gun in the game is far too high. Using statistics from the last couple of cups, I would rather see more people with 45%+ Acc. People would be rewarded with skill and experience this way. The current state of spread is one of the key reasons I do not find the game fun to play.

In short. Lower max spread. increase time it takes to get to max spread. (For all guns)


(Glottis-3D) #19

ok, since you guys are already in the works on the spread system, lets just wait for it and test the crap out of it. :smiley:


(BAMFana) #20

[QUOTE=Mustang;490516]@BAMFana:

You know you’re talking about recoil patterns and not spread patterns right?
Stop confusing things by mixing terms. :tongue:[/QUOTE]

I’m actually talking about both, but I’ll be the first to admit that I’m not explaining it very well. I guess what I described as an “idealized line spread pattern” in my post would be better described as a recoil pattern, whereas the randomized deviation from that line is decided by the spread pattern. In any case, if the spread pattern is too wide, the recoil pattern becomes much less meaningful as the bullets will tend to spread out into an even distribution within a “cone of fire” rather than following the recoil pattern. Compare the CS:GO example I posted and the chart posted in the thread Smooth refers to in his post.