How exactly does centerspeed work?

[mannon]

Yeah I don't think it would be difficult to code the game to allow it. It's just a matter of turning the zero axis stickiness on and off more or less. The question is whether or not they did. Up to this point I could have sworn that weapons kicked across the axis, but now I'm becoming convinced that they do not. Still need more tests to confirm, though.

[psijaka]

Mar 2, 2011 8:11:55 GMT -5 Den said:

Okay. So I modified the Super Duper Pretty Cool Guy so that it kicked horizontal five degrees per second with zero centerspeed and up five degrees with five centerspeed.

Any speed from viewkick or CenterSpeed is completely neutralized on the next shot.
Any straight upwards/downwards movement will be nullified by a straight left viewkick.

Centerspeed independently handles the X and Y axis.
If it were to kick ten up and one hundred left, the path taken would not be a linear route to the center. It would hit the zero on the pitch axis while still having a way to go on the yaw.


The maximum possible Viewkick is limited to 100 (or ten degrees).
Once it hits the edge (and almost no stock weapon can actually do that), any speed going that way immediately goes to zero, centerspeed will begin pulling back immediately.

Image at 16:9 resolution, default 65° FOV

Note that in this image, the box is a perfect 260x260 square.

Well, horizontal viewkick causes the camera to tilt.
The further along it goes, the more the camera tilts.
It actually stops at that point 130 pixels to the left/right of the point of origin and despite stopping in place, continues to twist... to a max of ten degrees.

Course, even though the camera tilts, the path taken by the x-only viewkick is still a straight line.

Using this perfectly vertical door jamb as reference...


The horizontal viewkick hits the edge in two seconds (since the viewkick speed is 5), but it continues to turn for exactly another two seconds.
This tilt is also a part of what the CenterSpeed needs to counteract - the camera will stay on the edge until the centerspeed takes care of the part of the tilt.

Essentially, the x axis has twice the "area" it can travel than the vertical. While a few weapons (like a rapid fire Kiparis) can reach the horizontal edge, none will reach full tilt. So... none of that is really important.

Though it does tell that horizontal viewkick causes the camera to tilt. 5° at the edge, 10° possible maximum.

@den - Can I just clarify one thing; is this (in yellow) in response to a single shot?

[mannon]

Oh bloody hell I never even saw this post by Den for some reason! O,O lol

He answered the compound velocity thing a long time ago! Okay so at least we know that velocity does not compound, that's cool. The tilting is interesting. I wonder does it always tilt like this or does it change if you look down? Perhaps the recoil axes don't work quite the way the ones under your control do.

But from the fact that it's exactly half as much rotation in degrees I don't think it's affected by the pitch axis. Looks like they just added in some tilt as a visual thing.

From reading this I think it was from a single shot. He said that he had zero centerspeed so that it would just keep going. So only one shot would have been needed.

Wow... I can't believe I never saw this... hmm... This also shows that either the velocity|acceleration model is correct or if vkick is a displacement something other than centerspeed determines the initial velocity. Otherwise with a centerspeed of 0 it wouldn't kick at all.

For a displacement model to still be viable and still follow the recentering time rule precisely I'm not quite sure how that would work out. It's still possible, but in order to make the displacement model viable we're having to make it more and more complex to match our observations. While the velocity model seems to be chugging along fine.

I'm still kicking myself for not seeing this post earlier! lol

Maybe I should go back through the whole thread just in case.

[psijaka]

Yes, a very significant post. I didn't give it the attention it deserves because I was distracted by the M14 v FAL debate.

Pretty strong evidence for the velocity|acceleration model, I would say. Still 2 outstanding matters:

Is residual veloccity carried over to the next shot?

Can the gun cross the axes between shots?

Managed to do some video capture for the first time today (childrens TV!). So I will shortly try to capture some bursts, amongst other things.

[mannon]

Actually if you read it carefully Den appears to answer the first question.

Mar 2, 2011 8:11:55 GMT -5 Den said:

Any speed from viewkick or CenterSpeed is completely neutralized on the next shot.
Any straight upwards/downwards movement will be nullified by a straight left viewkick.

[psijaka]

Am I missing something here?

[mannon]

Den basically says that residual velocity is not carried over. Otherwise the view would continue listing up or down if the next viewkick was strait horizontal. Instead the up and down movement is cancelled and it simply kicks horizontally. At least that's how I read it. As near as I can tell Den's testing has answered the residual velocity question for us.

Now we just need to prove whether or not the view can cross the axis, something I believe it cannot do from my testing thus far. (Which surprised the hell out of me, honestly.)

[mannon]

Did some more testing just to increase confidence levels. I used a FAMAS with an ACOG in MW2 and fired 350 individual bursts. (5 complete lives worth of ammo with Scav Pro.)
Or if you want to check my math:
30 clip + 180 reserve = 210 rounds
210 / 3 shots per burst = 70 bursts per life
70 * 5 lives = 350 total bursts

The ACOG served dual purposes. First it reduces the center speed thus magnifying the effects of recoil in general. This could either help or hinder kicks across the center showing up in bullet decals, but I already tested without so I figured it'd be good to test it with it in case it might show some kicks that would otherwise recenter. The ACOG also has nice little vertical and horizontal lines.

This time around I decided to check for both X and Y axis kicks across center, approximately doubling my odds of potentially seeing a kick that goes across center.

In all 350 bursts the bursts would go up, down, left, and right... but always together. In other words every single burst would result in 3 bullet decals that were either within one of the four pie slices of the ACOG or were on the edges of one pie slice.

The really interesting bit turns out to be the Y axis. The FAMAS is quite capable of kicking down, but it has a strong tendency to kick up instead. This obviously means that the VKick is stronger in the up direction, but that there is some down kick too. Considering this fact whenever the gun kicks down on the first shot and doesn't recenter, it should have a very strong likelihood of generating an upward kick on the second shot that should be strong enough to push it past center if it's possible to do so. Interestingly I had several kicks that put 1 or even 2 shots below center. But I never even once had a shot below center and also a shot above center.

If it is possible for a kick to cross the center it must be extremely unlikely. Since the competing models either kick across center fairly often or never with no in between I believe that this is strong evidence that viewkick is literally incapable of kicking the view across the center without actually centering first.

While it sounds really gosh darn golly gee whiz strange I have to admit that this burst pattern actually matches my experience with using the FAMAS and M16, two of my favorite guns. I always thought that it was due to some residual velocity, but this does not seem to be the case. In fact in my testing it appears quite possible for the 3rd shot to be more on center than the 2nd in some cases.

Unless someone comes up with some mechanism other than this to explain the observed behavior or proves the hypothesis wrong I think I'm going to assume that the view cannot kick from one side of center to the other. It appears that upon reaching center during a kick the view sticks there. This behavior was predicted by the Velocity|Deceleration model, but does not necessarily prove it.

[aequinox]

Throw an ACOG on the HK21 and watch it recoil into all four pie slices.

[mannon]

Not possible in a 3 shot burst even if it could cross the center. ;p Plus I don't have BO. Anything greater than a 3 shot burst would have the possibility of recentering on the third shot and thus allowing a fourth shot to cross.

Basically the view only crosses the center whenever it manages to recenter first. Although, it only needs to recenter 1 dimensionally, not on both axes. Because of this behavior a burst of 3 shots turns out to be the magic number.

When you fire 3 shots there are only two significant viewkicks. The 3rd is purely visual. The first viewkick starts at the center and can go anywhere the weapon is capable of kicking. The second kick then has the opportunity to attempt to cross the center if it kicks back in the opposite direction of the first kick. But only if the view did not recenter in time.

Now it is very possible for the kicks to be in opposite directions. But if the view isn't recentered then a kick that goes towards center will simply stop when it hits center and not cross the center at all. If the view recenters on the second shot then it may allow the gun to actually kick in the opposite direction, but both of the first shots will be on the center for that axis.

Thus there is no combination of kicks or recenterings that would allow any weapon within 3 shots to cross center and put bullet decals into more than one pie slice.

It's a strikingly strait forward test, and I have yet to witness a single 3 shot burst from any weapon that crosses center after hundreds of bursts.

NOTE: Sniper weapons are somewhat of an exception, but it's not because recoil works any differently, it's simply because of the Idle sway throwing things off. The idle sway constantly changes the center position.

[mannon]

I have a new question which we probably should have asked a long time ago. Well actually I did think of it, but we had so many other questions it seemed like a minor issue.

We need to determine whether viewkick is a float or an integer. We know lots of variables in thee game such as health are integers. I think we really need to know if the function that randomly picks a viewkick within the weapon's viewkick parameters returns an integer. Mainly because if it is an integer then that would alter the probabilities, because there would be a very finite number of vkick values between the min and max values. It would be a small effect, but to some extent I think it would compount with the size of the burst.

I would also like to know if it's actually possible to get a completely null kick or if the viewkick is always non zero. I think it's probably possible to get zero, and if viewkick isn't an integer it likely would make very little difference anyway. But if it is an integer then that does alter the probabilities a bit.

Granted, these are rather subtle quirks that are unlikely to have a huge impact. But dammot... I want a perfect recoil model now that we're so close. ;p

[Den]

If I were to set viewkick to 1 and 0, I could get a speed of 0.4 or 0.9 or 0.2 or maybe even 0.281 or however specific the speed may get.

[mannon]

Awesome! Thanks Den.

Though, I was almost hoping it would be an integer just because it'd be easier on teh maths. ;p

We'll just have to treat it as all real numbers within the range though of course the variable couldn't actually represent all of them, that would be infinite. But a float could certainly represent any number that would have any significant difference in the recoil for our purposes and that's close enough.

I'm sure that's what everyone is already doing, anyway. At least this way we don't have to refit the models for integers vkicks.

[psijaka]

I have been running some new simulations based upon viewkick=velocity; centerspeed=acceleration, and am not allowing the gun to cross an axis between shots; eg. if the gun is only slightly to the right and gets a big kick to the left, the gun stops in the center rather than crosses from right to left. My previous simulations allowed this to happen.

The results are quite startling, to say the least. Guns with a strong bias upwards and to the right are not affected, but guns with high random recoil are, in an unexpected way.

I first ran the Stoner + ACOG, as it has lots of random recoil. I simulated 1000 three round bursts and calculated what percentage of times the gun fires it's 2nd, 3rd and 4th shots when the gun has fully recentered.

2nd shot - 7%
3rd shot - 16%
4th shot - 12%

(My previous method gave 7%, 6% 5%)

I was so astonished at the high proportion of 2nd and 3rd shots that I thought that I had made a blunder, but after a lot of checking and thought, have decided that this is indeed what happens if you cannot cross an axis between rounds.

The question is, what happens in the game? Is it possible for a gun to shift from right to left between rounds?

In all the testing that I have done, I have NEVER seen a gun cross the axis between rounds.

What are other bros thoughts on this? Have you ever seen a gun cross from right to left between rounds?

Once we know the answer, I will recalculate the Black Ops recoil chart and re-do all of the recoil plots.

[Den]

If I were to set a gun to fire left one and right one, whichever direction it goes first will be the direction it always goes.

That is to say if it is traveling left, a right-shot will only cause it to stop in place rather than go right. Any right-kick will not actually make the view go right until the view returns to the center axis.

And expectedly, with a Centerspeed of one, while the view is off to the left, the right kick will interrupt the Centerspeed and make it stop in place again, requiring the centerspeed to accelerate again.

[mannon]

I'm a little confused. If it kicks left then a right kick makes it simply stop then that would suggest that velocity is preserved. Buf if a small right kick can actually slow it down then it must overwrite the velocity as previous tests indicated?

As for crossing the axis, yeah... I've fired hundreds of 3 shots bursts specifically looking for shots to cross the axis and I just don't think it's possible.

With random viewkick it should be possible to get a small kick left and a strong one right to send it cross axis if it's remotely possible. The second kick would generally have to be stronger than the first to even reach the axis, but I've seen no crossing. In my tests I have seen a lot of recentering, though.

[psijaka]

Thanks for the replies.

Just done some more testing M60 big ammo; 3 round bursts. Did not see a single 3rd shot go the opposite side to the 2nd shot.

What I did notice was that a high proportion of 3rd shots seem to be centered horizontally; especially if the 2nd shot was off to one side. Maybe 3 round bursts are best with the Pig?

[themccannman]

why did treyarch go go to all the trouble of making such a complicated recoil system that isn't even really that realistic?

[n1gh7]

The recoil system hasn't probably changed since the engine was made. However many years ago that was.

[mannon]

And thus was not made by Treyarch. It's a modified Quake 3 engine and probably uses a lot of things that were already built into that. So we probably have IW to blame if not Id.

CoD games are also of questionable realism anyway. They are shooters, and not even tactical shooters at that.

[psijaka]

My stickied "Recoil Chart and Plots" thread has now been updated based upon the viewkick=velocity, centerspeed=acceleration model, with no crossing of the axes between rounds.

[themccannman]

I'm not concerned about the recoil system it seems to work fine, i'm just concerened that treyarch took the time to program all of these different sets of recoil for each gun (or maybe jsut 2 or 3) instead of working on fixing the ad look and sound in the game, i don't care what the game resolution is, or how many pixels there are, treyarch games look straight up ugly no matter high quality of a picture you make it.

[mannon]

They were pretty lazy when it comes to stats too considering most weapons are just very slight variations on others. And they didn't even do it in a logical way like in MW2 where weapons usually balance out. By that I mean that if a gun is more accurate it usually has something else working against it such as low damage or ROF. If a gun does high damage it usually has a lower ROF or more recoil, ect... BO stat variations seem random to me. For example the FAMAS is nearly exactly like the Enfield, even the same viewkick. They just increased the ROF and since it shoots faster they ALSO increased the centerspeed. O,o Statistically about the only thing the Enfield does better than the FAMAS is it has a slightly faster ADS time.

Even the M4 and ACR were better balanced than this. Yes the ACR was generally considered to be the better gun, but that really depends on usage. It simply has ridiculously good accuracy. The M4, however, actually does have a slightly better ROF, and it is still quiet accurate. I'd say it competes with the ACR at closer ranges. You have to go for headshots with the ACR to make up for it, and if you get hit then you'll flinch and miss your targets, so it's a risky weapon to use up close. (You're better off switching to a good close range secondary.)

Feels to me more like the only real reason to use the Enfield over the FAMAS is the slower ROF gives you a little advantage in recoil sometimes so you can probably get tighter groupings when burst firing. However, I think the FAMAS is more consistent when firing full auto which can be practical (with compensation) at longer ranges than I'd have thought.

Anyway... I have to agree that the Treyarch games do just seem to be less polished or less skillfully crafted in general, at least when compared to the IW ones. Then again it's probably not a skill thing but style. I think that they consciously make their games a little different than the IW ones to differentiate themselves as a company. I think that they are doing it because they think that this is their signature in the CoD franchise and I think they honestly believe that it's better than the IW flavor CoD games. ;p

[asasa]

On dual wield weapons, recoil increases when you fire both at once. Does it add the velocities of the two together?

[TheHawkNY]

I have a few questions:

(1) Centerspeed always pulls toward the point the first shot was fired towards, not the point the last shot was fired towards. Is that correct?

(2)

Mar 15, 2011 12:29:56 GMT -5 psijaka said:

a=centerspeed/2.5 (to give centerspeed/5=recentering per second). A deceleration, in other words.

Why is this the case? I don't understand where the 2.5 comes from, can you please explain?

(3) Which one of these is the case? (Each is done individually for horizontal and vertical)
(A) When each shot is fired, a random velocity between the min and max is generated.

(B) Or, assuming the min is negative and the max is negative:

• If the current position is negative, then a random velocity between the min and 0 is generated.
  
• If the current position is positive, then a random velocity between 0 and the max is generated.

(C) Or, a random velocity between the min and max is generated, and:

• if the current position is negative, then if the random number was positive it is set to 0.
  
• if the current position is positive, then if the random number was negative it is set to 0.

It seems to me like the belief is that (C) is the case, but I want to make sure.

I have more questions, but I'll start with those.

[eLantern]

Feb 17, 2011 18:45:00 GMT -5 psijaka said:

And what is the effect of the following Gunkick variables?

adsGunKickReducedKickBullets = 0
adsGunKickReducedKickPercent = 75
adsGunKickPitchMin = 5
adsGunKickPitchMax = 15
adsGunKickYawMin = -5
adsGunKickYawMax = 10
adsGunKickAccel = 800
adsGunKickSpeedMax = 2000
adsGunKickSpeedDecay = 32
adsGunKickStaticDecay = 40

--------------------------------------------------------------------------------------

Feb 17, 2011 20:33:55 GMT -5 mannon said:

As for the gunkick Den has shown that it is essentially irrelevant. The gunkick figures kick the gun in a various direction similar to the way the view is being kicked off target, but the recentering of gunkick is so fast that it is essentially zeroed out by firetime in pretty much every weapon. Den went so far as to give an AK47 zero idle or recoil, but left gunkick alone and it was a laserbeam across the map with negligible deviation off target.

---------------------------------------------------------------------------------------

Feb 17, 2011 21:46:50 GMT -5 n1gh7 said:

For further clarification, gunkick is how much the gun moves on the screen. Viewkick is like how much your whole body moves, and gunkick is how much the gun moves in your hands.

---------------------------------------------------------------------------------------

Feb 18, 2011 0:22:23 GMT -5 mannon said:

Good. I thought that gunkick was relative to the view rather than target, but didn't want to say it without any sort of real knowledge. (Though I assumed with viewkicks as large as we have and gunkicks as small as the numbers suggest as well as the fact that the gun could kick in the opposite direction it would put the gun WAY off camera if it wasn't view relative.)

View Kick throws you off from the direction you aimed at without actually altering that aim, so if you let it recenter it-self you'll be aiming at the same spot you were in the first place. But it's really more of an offset, because while the pitch axis does indeed get rotated by the yaw axis the yaw axis always remains perfectly vertical, (or horizontal... however you want to look at it... when you turn sideways the turn always keeps level to the ground never moving the view higher or lower). Gunkick is parobably a similar offset rotating around the same point, but used just to throw the gun off a little to make it more visually interesting.

---------------------------------------------------------------------------------------

Dec 3, 2011 3:17:43 GMT -5 Den said:

Actual Gunkick however causes the weapon to move away from the center of the screen and by association, moves where the shots are going.

---------------------------------------------------------------------------------------

So did you (psijaka) ever figure out how to account for GunKick within your Recoil Plot Charts?

I wonder about this because it appears that several weapons within MW3 seem to be pretty affected by their GunKick numbers, regardless to mannon's above comment about GunKick being essentially irrelevant.

IMO, the statements below leave questions about how accurate your formula is in figuring out the true recoil plots of the weapons, especially those which appear to be more significantly affected by GunKick...

Jan 8, 2012 18:50:18 GMT -5 mmacola said:

...every weapon has Gunkick, but most of them are really, really, (really) minimal.

Feb 11, 2012 18:01:02 GMT -5 8bitasplode said:

(The MW3) Thermal (Scope) is only a really helpful (attachment) for guns when most of their recoil is (caused by) GunKick.

Feb 24, 2012 14:47:44 GMT -5 Robospy said:

Den explained to us why thermal reduces recoil on LMGs, because of the "GunKick" which is like normal recoil but it doesn't move the screen, Thermal removes this, thus decreasing recoil. However, the only weapons with siginigant GunKicks are the L86, Mk46, PKP, G18, FMG9 and PM-9.

So apparently the Thermal completely removes GunKick, but I don't how that was proven. Also, I've read in a post somewhere that there's some evidence that the FAD's recoil is affected by a decent portion of GunKick and that by putting the Thermal Scope on it causes it become laser-like... making it more accurate than what the Kick Proficiency would provide it? ...I don't know, but I think that's another good question.

Anyways, in regards to GunKick having a real affect on a weapon's accuracy how does CenterSpeed play a role between both ViewKick and GunKick?

[Marvel4]

Mar 22, 2012 14:29:47 GMT -5 eLantern said:

So apparently the Thermal completely removes GunKick, but I don't how that was proven.

Easy. Modify a sniper rifle, reduce all view kick values to 0 and increase the gun kick values. With the ACOG, there is recoil. With the sniper scope, there isn't.

Anyways, in regards to GunKick having a real affect on a weapon's accuracy how does CenterSpeed play a role between both ViewKick and GunKick?

Centerspeed only affects view kick.

[eLantern]

Mar 22, 2012 15:16:19 GMT -5 Marvel4 said:

Mar 22, 2012 14:29:47 GMT -5 eLantern said:

So apparently the Thermal completely removes GunKick, but I don't how that was proven.

Easy. Modify a sniper rifle, reduce all view kick values to 0 and increase the gun kick values. With the ACOG, there is recoil. With the sniper scope, there isn't.

Awesome and this is because when looking down sight through an actual scope attachment (those which black-out peripheral vision - I believe these are all "held-breath" sights) the weapon model is removed which in-turn takes away the GunKick variable, correct?

Anyways, in regards to GunKick having a real affect on a weapon's accuracy how does CenterSpeed play a role between both ViewKick and GunKick?

Centerspeed only affects view kick.

Good to know.

BTW, can we then assume that the Thermal Scope, or any scope for that matter, must modify or change (typically worsen) a weapon's CenterSpeed?

Otherwise, wouldn't all weapons, regardless to how little GunKick factored into its overall recoil, see a reduction in their true recoil when equipping a scope attachment since they'd essentially be removing GunKick from the formula? Again, regardless to how small that GunKick influence was.

...or is it simply a matter of increased magnification by the scope causing a perception of increased recoil on the weapons which have only a minor amount of GunKick influence to begin with. This would mean that scope attachments provide all weapon's a reduction in real recoil even though the perception caused by magnification could persuade people to think otherwise, right?

[Megaqwerty]

Mar 22, 2012 17:16:50 GMT -5 eLantern said:

Awesome and this is because when looking down sight through an actual scope attachment (those which black-out peripheral vision - I believe these are all "held-breath" sights) the weapon model is removed which in-turn takes away the GunKick variable, correct?

Spot on. If you have no weapon model, you have no gun kick.

Mar 22, 2012 17:16:50 GMT -5 eLantern said:

BTW, can we then assume that the Thermal Scope, or any scope for that matter, must modify or change (typically worsen) a weapon's CenterSpeed?

Otherwise, wouldn't all weapons, regardless to how little GunKick factored into its overall recoil, see a reduction in their true recoil when equipping a scope attachment since they'd essentially be removing GunKick from the formula?

Other scopes (ex. RDS, ACOG) do not remove the weapon model when you ADS in (duh) and thus gun kick is not removed.

Also, scopes do not have to reduce centerspeed and there are in fact guns known to do this in Black Ops. It is speculated among bros that the ACOG does not reduce centerspeed on some weapons in MW3.

[eLantern]

Mar 22, 2012 17:35:41 GMT -5 Megaqwerty said:

Mar 22, 2012 17:16:50 GMT -5 eLantern said:

BTW, can we then assume that the Thermal Scope, or any scope for that matter, must modify or change (typically worsen) a weapon's CenterSpeed?

Otherwise, wouldn't all weapons, regardless to how little GunKick factored into its overall recoil, see a reduction in their true recoil when equipping a scope attachment since they'd essentially be removing GunKick from the formula?

Other scopes (ex. RDS, ACOG) do not remove the weapon model when you ADS in (duh) and thus gun kick is not removed.

When I say scopes I mean sights which remove the weapon model... held-breath / black-out peripheral vision

Mar 22, 2012 17:35:41 GMT -5 Megaqwerty said:

Also, scopes do not have to reduce centerspeed and there are in fact guns known to do this in Black Ops. It is speculated among bros that the ACOG does not reduce centerspeed on some weapons in MW3.

That's fine and dandy, I already knew that...
Again, I am focused on scopes, but the reason I was questioning whether or not scopes must have some effect on CenterSpeed had to deal with what I've heard previously... that the Thermal Scope actually increases recoil for most weapons.

Although, how could this be the case since the GunKick is removed and it's unlikely that the attachment would alter the ViewKick numbers, right? So that only leaves a changing or modifying of the CenterSpeed, which is not out of the realm of possible, since we've witnessed similar changes in previous CoD titles with all sorts of Sight attachments... just as you point out.

I then gave a possible answer which was that all weapons equipped with a scope actually receive an increase in accuracy, but because most weapons have a very minor amount of GunKick to begin with they won't have a noticeable accuracy increase, instead thanks to an increase in magnification the weapon is perceived to worsen in it's accuracy even though that's not actually the case.