Recoil Control Mechanic in Black Ops III

[probaddie]

In most patch notes released for Black Ops III, some weapons have had their "ability to control recoil" either reduced or increased. No notion of "control" has ever been known to be part of the recoil model in Call of Duty, and so this phrasing indicates that a change to the recoil model has occurred in Black Ops III. Indeed, that is the case: Marvel4 and I have discovered a new recoil mechanic and the variables that govern its behavior. In short, this mechanic causes the player's recoil to either be dampened or exacerbated depending on the direction they are turning at the moment they fire a shot. (I recommend reading my recoil post - just the basics - before continuing.)


The mechanic affects both ViewKick and GunKick - but not equally, as we'll explain soon - by scaling the drawn velocities by a factor depending on whether the player is looking into or away from a kick, or not at all. In the latter case, no scaling is ever applied. In the previous two cases, the scaling is given by two weapon statistics called kickOpposedInputScalar and kickAlignedInputScalar. Henceforth, they will just be called opposedScalar and alignedScalar. They are both decimal (floating-point) numbers that are positive and tend to be near 1.0 for most weapons. The mechanic, like almost all other recoil mechanics, is applied independently to the yaw and pitch coordinates. How these are applied to a given dimension can be broken down into three primary cases:

• If the player is not turning (with respect to the given dimension, either yaw or pitch), no scaling is applied.
• If the player is turning in the positive direction and receives a positive kick, or is turning in the negative direction and receives a negative kick, the kick is scaled by the alignedScalar.
• If the player's turning direction and received kick are of opposite sign, the kick is scaled by the opposedScalar.

In plain English: when you are "fighting" the kick, your kick is scaled by opposedScalar, and when you're giving in to the kick, it is scaled by alignedScalar (with one exception noted below). In the above cases, "positive" and "negative" are with respect to the ViewKick coordinate system (positive is left/up, negative is down/right). Looking up or down to the 85 degree pitch limit to lock your view does not affect the mechanic; as long as look input is provided, one of the two scalars will be applied. Also, any change in view due to aim assist does not count as "turning" for the purposes of this mechanic.

Interestingly, this mechanic is technically bugged. Note that the mechanic uses ViewKick-based coordinates in determining which scalar to use. This causes a problem for GunKick, as with GunKick the pitch coordinates (down is positive, up is negative) are the exact opposite of ViewKick. When a player is looking up (positive look direction) and receives a downward (positive) kick, the player is fighting the recoil and we expect the opposedScalar to be applied to the given kick. However, because the signs of these two quantities match, the alignedScalar is applied instead. Put simply, when the player is fighting a vertical GunKick kick, the game actually applies the alignedScalar scalar, and vice-versa.

The discovery of this mechanic sheds light on the meaning of the patch notes with regard to "recoil control." To achieve higher "control", the opposedScalar and the alignedScalar can be decreased.

Edit: Changed the meaning of "diverge" and "converge" to be relative to the kick rather than the original point of aim/center of the viewport. Changed "rate" to "scale" in the variable names, as nothing is changing over time. [Not relevant.] Also, a bunch of other corrections.

Edit: Replaced variable names with the real names.

[Deleted]

Very interesting information you and Marvel4 have recovered. I too have wondered what this "recoil control" statistic meant. Thanks for shedding light on this mechanic and hopefully Marvel4 can provide you with hard numbers so we can see this in effect.

[haoz]

convergeRate and divergeRate are near 1 but not necessarily greater or less than 1? i.e. convergeRate could be less than 1 for better recoil stability, but some guns could have a convergeRate greater than 1 making recoil harder to control?

The recoil in this game feels like it has a lot of variance compared to previous games. The VMP and KN-44 can suddenly jerk at times. I assume this cannot be a function of this recoilControl mechanic, but I wonder if other aspects of the recoil mechanic has changed. I can see that minimum magnitude in this game is higher than in AW or Ghosts, so maybe that's sufficient to change the behavior of recoil.

[mannon]

Wish we had recoil plots. Recoil plots are tasty. mmmmm... data...

[probaddie]

Mar 4, 2016 14:50:39 GMT -5 mannon said:

Wish we had recoil plots. Recoil plots are tasty. mmmmm... data...

I'm getting there....

[probaddie]

Mar 4, 2016 12:54:54 GMT -5 haoz said:

convergeRate and divergeRate are near 1 but not necessarily greater or less than 1? i.e. convergeRate could be less than 1 for better recoil stability, but some guns could have a convergeRate greater than 1 making recoil harder to control?

The recoil in this game feels like it has a lot of variance compared to previous games. The VMP and KN-44 can suddenly jerk at times. I assume this cannot be a function of this recoilControl mechanic, but I wonder if other aspects of the recoil mechanic has changed. I can see that minimum magnitude in this game is higher than in AW or Ghosts, so maybe that's sufficient to change the behavior of recoil.

Now that Marvel4 has these numbers in his raw data spreadsheet, I can tell you that the divergeScale is always less than or equal to one and that convergeScale is greater than or equal to one. (Note that we inverted the meaning of "converge" and "diverge" since I opened this thread. And the Pharo is a major exception, with a convergeScale of 0.15.) Thus, fighting to control your weapon never makes your recoil worse, and giving in to it can never make it better (excepting vertical GunKick, which is the exact opposite).

I've received no indication from Marvel4 that new recoil variables are at play in the weapon files. The only comment I can make is that many weapons in this game have significant adsSpread. The handguns, shotguns, SMGs and LMGs all have very significant adsSpread, which would introduce some variance into the player's accuracy. Perhaps that is what is responsible for what you are experiencing.

[mannon]

Mar 4, 2016 16:32:37 GMT -5 probaddie said:

I'm getting there....

Awesome!

[broth3r]

Nice find, but I'm curious as to the reason behind it. Drift0r was speculating this addition may be to increase predictability and/or add another tuning variable, which might be valid.

But I have another theory - or hope, at least. By buffing weapon control like that, there is in effect a buff to accuracy against stationary targets, but a nerf against mobile ones. Unless the shooter is lucky his target is moving into the recoil, this will make their weapon kick more as they drag their aim to track them; and the more erratic, the worst. I want to believe that was a deliberate effort to boost the survivability of mobile players. Of course, that effect is minute at best - and far, faaaaar from the likes of Titanfall where I felt far safer the faster I moved, where here it's a death wish at worst. But still... I want to believe.

[omega8trigun]

I can't remember. Gunkick is the visual recoil your gun shows when firing aka the way the gun moves in response to recoil, and viewkick is the angle the view moves upwards in response to recoil, correct?

So the issue with the bug is that if you fight recoil, your viewkick will decrease properly but your gunkick will increase instead of decreasing like it should? So that would mean you still get the benefits in terms of recoil dampening, but your gun is just moving in the wrong way?

[mannon]

Viewkick moves both your "camera" view and the gun with it. Gunkick moves the gun on your screen, pushing aim off center. Both are 2 dimensional, though mostly vertical due to weapon stats.

[haoz]

Mar 7, 2016 18:35:16 GMT -5 mannon said:

Viewkick moves both your "camera" view and the gun with it. Gunkick moves the gun on your screen, pushing aim off center. Both are 2 dimensional, though mostly vertical due to weapon stats.

And note gunkick is not visual recoil. Visual recoil is just how the sights bounce around. Gunkick is actual movement of the position of the aiming point in the viewport.

[omega8trigun]

Mar 7, 2016 18:43:28 GMT -5 haoz said:

Mar 7, 2016 18:35:16 GMT -5 mannon said:

Viewkick moves both your "camera" view and the gun with it. Gunkick moves the gun on your screen, pushing aim off center. Both are 2 dimensional, though mostly vertical due to weapon stats.

And note gunkick is not visual recoil. Visual recoil is just how the sights bounce around. Gunkick is actual movement of the position of the aiming point in the viewport.

That's what I needed clarified. Thankyou. So then this bug is pretty bad.

[haoz]

Mar 7, 2016 19:20:30 GMT -5 omega8trigun said:

Mar 7, 2016 18:43:28 GMT -5 haoz said:

And note gunkick is not visual recoil. Visual recoil is just how the sights bounce around. Gunkick is actual movement of the position of the aiming point in the viewport.

That's what I needed clarified. Thankyou. So then this bug is pretty bad.

Not really. My understanding is that in the past few cods, there has not been a lot of gun kick. As the spreadsheet shows, a typical gunkick range is -10 to 10. The minimum magnitude viewkick is 10 or greater for guns in blops3.

I think this recoil mechanic is interesting technically, but I don't think it affects gameplay significantly. 

[esupin]

So if you ADS and strafe while not moving your aim, you don't suffer any recoil penalties (except maybe autoaim pulling your crosshairs a little bit)?

[probaddie]

Mar 8, 2016 0:32:40 GMT -5 esupin said:

So if you ADS and strafe while not moving your aim, you don't suffer any recoil penalties (except maybe autoaim pulling your crosshairs a little bit)?

Nope. In the given case, you're neither being helped nor punished by this mechanic, even when your aim is being pulled purely by aim assist.

[probaddie]

Mar 7, 2016 21:55:02 GMT -5 haoz said:

Mar 7, 2016 19:20:30 GMT -5 omega8trigun said:

That's what I needed clarified. Thankyou. So then this bug is pretty bad.

Not really. My understanding is that in the past few cods, there has not been a lot of gun kick. As the spreadsheet shows, a typical gunkick range is -10 to 10. The minimum magnitude viewkick is 10 or greater for guns in blops3.

I think this recoil mechanic is interesting technically, but I don't think it affects gameplay significantly. 

I'm not so sure. Note that the Grip confers only a 7.5% reduction to ViewKick, whereas the benefit from this mechanic can be as high as a 50% reduction (M8A7) and the penalty as high as a 30% increase (most handguns) in ViewKick and GunKick.

[tflame]

Kudos on the decode! It's mainly intended to increase the skill ceiling of gunplay by increasing the reward for using look controls to cancel out recoil. This includes making it more possible to hit stationary enemies at a distance behind cover (or "using mind bullets" as the kids call it) by fighting the upwards recoil. Different guns get different rewards, generally the higher skill cap guns get more reward.

Several of the patches reduced the amount of the "Alignedinputscalar", which is the punishment for looking into the recoil. People ended up finding it more random and sporadic, perhaps overly punishing. Hence the "Enhanced ability to control recoil".

[haoz]

Mar 8, 2016 2:44:41 GMT -5 probaddie said:

Mar 7, 2016 21:55:02 GMT -5 haoz said:

Not really. My understanding is that in the past few cods, there has not been a lot of gun kick. As the spreadsheet shows, a typical gunkick range is -10 to 10. The minimum magnitude viewkick is 10 or greater for guns in blops3.

I think this recoil mechanic is interesting technically, but I don't think it affects gameplay significantly. 

I'm not so sure. Note that the Grip confers only a 7.5% reduction to ViewKick, whereas the benefit from this mechanic can be as high as a 50% reduction (M8A7) and the penalty as high as a 30% increase (most handguns) in ViewKick and GunKick.

If it's the same scaling to both viewkick and gunkick, it seems that the viewkick scaling would dominate right? I do agree that the developers should notice that their code is doing something unintentional if it is the case.

Practically, with the M8 as a burst weapon, I'm not sure if recoil control is practical without using a mouse for input. Not every burst goes upward in the same consistent amount. I also don't see how buffing the ICR opposing kick scaling benefits the gun. Currently, the gun has somewhat uniform and low recoil: up, left, and right. It is impossible to compensate for. I think only the guns that recoil with a directional tendency are helped by this.

[probaddie]

Mar 8, 2016 21:38:11 GMT -5 haoz said:

Mar 8, 2016 2:44:41 GMT -5 probaddie said:

I'm not so sure. Note that the Grip confers only a 7.5% reduction to ViewKick, whereas the benefit from this mechanic can be as high as a 50% reduction (M8A7) and the penalty as high as a 30% increase (most handguns) in ViewKick and GunKick.

If it's the same scaling to both viewkick and gunkick, it seems that the viewkick scaling would dominate right? I do agree that the developers should notice that their code is doing something unintentional if it is the case.

Practically, with the M8 as a burst weapon, I'm not sure if recoil control is practical without using a mouse for input. Not every burst goes upward in the same consistent amount. I also don't see how buffing the ICR opposing kick scaling benefits the gun. Currently, the gun has somewhat uniform and low recoil: up, left, and right. It is impossible to compensate for. I think only the guns that recoil with a directional tendency are helped by this.

I take your point about feasibility of recoil control on console. Having only played on PC, I can only say that, with recoil as low as it's been in recent games, it's certainly possible for skilled players on PC to account for it, even in cases where there is no clear directional bias. I only meant to point out the relatively high scalars being used with this mechanic compared to the grip attachment.

[mannon]

Used to seem you were really punished for firing continuously and it was really in your best interests to fire in bursts instead unless you were in some CQB hipfire fight with a full mag. Does this negate that old wisdom? I have noticed that I tend to just keep firing more than I used to, where in the past I would burst or even pop individual shots off. I figured I was just getting lazy, but maybe it's not such a bad thing after all?

Having said that, if the effect is less when you're aiming at moving targets would that reverse it again and make it better to burst fire moving targets, but go full auto on stationary ones? Well... I guess we won't really have definitive numbers on that, and each weapon could be different... but I thought I'd ask.

[lustindarkness]

I find all of this very interesting. It seems to me that some very interesting mechanics were put into this game and they worked hard on it. Just wish they had worked just as hard on some other things too.

Thank you guys for your hard work to decipher all this.

[illram]

So grip isn't totally worthless then? I.e. it can make a fairly significant change in gunkick (non visual) depending on the gun?