Black Ops Recoil Plots + MW3, COD4 links

[psijaka]

COD4 recoil plots can be found at denkirson.proboards.com/index.cgi?action=display&board=general&thread=3645&page=1
MW3 recoil plots can be found at denkirson.proboards.com/index.cgi?board=general&action=display&thread=3256

---

This post contains information on the recoil characteristics of the guns in Black Ops, and is divided into 3 parts:

1 - Recoil Chart showing the effects of Rate of Fire, Recenter Speed and Kick on a range of full auto guns.

2 - Recoil Plots for all full auto and burst fire guns.

3 - Kill Time Probabilities showing the chance of getting a kill at a given time when using various guns. This is work in progress and more plots can be found from page 7 onwards.

---

1 - Recoil Chart showing the effects of Rate of Fire, Recenter Speed and Kick on a range of full auto guns.

The three columns on the right shows the likelihood of the gun fully recovering from the recoil of the previous rounds, resulting in that shot going where the first one went.





Method:

The calculations assume that viewkick is a velocity applied to the gun, and that centerspeed is acceleration applied to the gun to return to the starting point.

I have assumed that recoil occurs as a random linear distribution between the limits for vertical and horizontal kick as defined in Den's weapon damage chart.

The centerspeed figure is applied to this distribution; recentering occurring when kick <= centerspeed / 5 x fire time. The recentering is applied to the vertical and horizontal kick separately (there has been some discussion about this on this forum).

To produce the figures, I simulated the firing of 10,000 three round bursts, calculating where the gun would be after recentering is applied, at the point at which the next round is fired. This is a rather inelegant use of a spreadsheet, and I am sure that my laptop breaks out in a sweat when I log on! I may look at writing some code in future.

The probability of the gun recentering before the second shot has also been determined by calculation (results from the 2 methods match), but the situation gets too complex for my brain for subsequent shots.

Please also refer to asasa's "How exactly does centerspeed work?" thread, where the theory is discussed in detail. denkirson.proboards.com/index.cgi?board=general&action=display&thread=2461 Also take a look at jester's plots on page 4 of this thread.

---

2 - Selected Recoil Plots

Recoil plots for 100 simulated 10 round bursts. The 2nd and 3rd shots are in green (the first is always on target).

SMGs without attachments



Automatic rifles without attachments.



LMGs & CZ75

---

3 - Kill Time Probabilities showing the chance of getting a kill at a given time when using various guns. This is work in progress and more plots can be found from page 7 onwards.

These plots are derived from the simulation of 5000 bursts of up to 17 rounds. The following assumptions are made:

-Aim does not shift from the centre of the target (unless otherwise stated)
-The target does not move
-No headshots; no shooting through cover
-A target size of 1.0m high x 0.4m wide is used throughout, representing the torso

The lines between the shots are there for clarity only; do not read times or kill probabilities from these, use the markers instead.

The following ranges are covered:

Long - 70m - all weapons on minimum damage. Longshotting range, and the maximum on Den's charts.

Medium - 35m - The maximum range at which most ARs get full damage unsilenced. All silenced ARs and the M60 get minimum damage. The mid point on Den's charts.

Short - 17.5m - Most SMGs get full damage unsilenced and minimum damage silenced. M60 a 2 hit kill. 25% of the way up Den's charts.

Close - 7.5m - Most SMGs get full damage when silenced, and the Skorpion starts to become dangerous (3HK). A lot of encounters take place around 7.5m, but many people will be moving fast and firing from the hip.


These are the only guns that give a 50% chance of a kill within 1 second of full auto fire.


LMGs, Assault Rifles and the Skorpion included in this full auto plot.

EDIT - have removed the "burst firing at 35m" plot, due to error in calculating the M16 and G11 burst delay time.


Now before you flame me abut undervaluing the Spectre, remember that the plot above is with no recoil compensation. The Spectre's performance changes completely in the hands of a skilled player - example to follow soon. Note that the AK74 and MP5 have a virtually identical response.


2 methods of recoil compensation are shown above:

1/ offsetting aim by aiming low (middle of the bottom half of the torso), or by aiming low left (middle of the bottom left quarter of the torso).

2/ shifting aim whilst firing. The figures are degrees per second (vertically, or vertically and horizontally in the case of the Spectre). Best results at this range appear to be downwards at 4 deg/sec and left at 2 deg/sec.

Note that recoil compensation is not shown for the Skorpion as it would be counterproductive.

I suspect that even this plot probably does not show the true strength of the Spectre, as I cannot simulate a skilled player dynamically shifting aim in response to variations in recoil.


A plot comparing a range of guns from different classes at short range, with 'ADS in' time and recoil compensation applied. A lot to take in here, I know, so to improve clarity I have started at 0.3 sec. Similar plot at close range below.



At this range we see the true beastliness of the biased recoil weapons in the hands of a skilled player. Even the lowly Enfield is deadly. Also of note is the AK74 RF grip, and, of course, the Skorpion.


At last we see the true strength of the Spectre. But at this range the Skorpion is a 3SK and is absolutely deadly.


More Kill Time Probability plots can be found from page 7 onwards; including burst firing and single shotting (page 8).

If anyone would like me to produce recoil plots or kill time probabilitiy plots of guns/attachments not so far covered, just ask. No promises, but I will try to respond to sensible requests.

EDIT
12 Dec 2012 - Disruption as photobucket bandwidth limit smashed!
20 Sept 2011 - medium range burst fire pl;ot removed due to error.
15 Sept 2011 - New medium range plots added, including burst fire
14 Sept 2011 - Close range with ADS and recoil compensation added
13 Sept 2011 - Short range with ADS and recoil compensation added
12 Sept 2011 - long range full auto kill time plot added
11 Sept 2011 - Short & close range SMG kill time plots added - with recoil compensation
10 Sept 2011 - Short range SMG kill time plot added
09 Sept 2011 - Kill Time probability plots added to OP. Medium range ARs & LMGs..
11 May 2011 - major update. Chart and plots changed to viewkick=velocity; centerspeed=acceleration and to not allow crossing of axes between rounds.
07 May 2011 - warning about effect of crossing axis between rounds posted.
17 Mar 2011 - warning about validity displacement method posted.
28 Feb 2011 - Assault Rifle ACOG recoil plots added
24 Feb 2011 - Recoil plots replaced with versions having correct aspect ratio
23 Feb 2011 - Shotguns added to chart
22 Feb 2011 - Snipers and Secondaries added to recoil chart
22 Feb 2011 - AK74 Attachments added; correct aspect ratio
21 Feb 2011 - SMG, LMG, CZ75 plots added
21 Feb 2011 - AR recoil plots added and removed - error discovered. Corrected and reposted!
21 Feb 2011 - M14/FAL added; inc max recenter time
16 Feb 2011 - add SMG ACOG and CZ75 full auto
15 Feb 2011 - correct Kiparis recenter speed, add AR/LMG ACOG/IR scope data

and finally, You are my bro, bro. There, saying that felt surprisingly good!

[phale]

Excellent work man! This is awesome!

Thoughts on the numbers:
LOL @ Uzi, ROFL @ RPK
AK74u isn't that much more accurate than MP5 it seems...
M60 Grip is the second most accurate weapon in the game? CHUG CHUG CHUG
M16 is actually more accurate than G11... not to mention even more accurate with ACOG.

Which reminds me, update the chart with ACOG and stuff.
Also, PM63 has 1600 CenterSpeed, while Kiparis has 1500.

[psijaka]

Thanks for the feedback, phale

Will correct the chart for the Kiparis.

I will be adding more data; specifically ACOG and IR scope. Any other suggestions?

The results did turn up some surprises, as well as confirming some intuitive hunches about certain weapons.

The M60 grip is a favourite of mine, and I suspected that the slow rate of fire did wonders in controlling the recoil.

The M16 v G11 comparison was an eye opener though. My casual glance at the stats led me to believe that the G11 was way ahead of the M16 due to much higher recenter speed. I must look at M16 + ACOG & IR scope.

The Skorpion was a surprise too; this gun really is like a laser. More to it than meets the eye, and on the basis of these calculations, definately does not need a buff.

[arcanine2009]

Nice chart!

Are you gonna add the semi-auto ARs, shotguns and sniper rifles? Or is it too complex?

I'm very interested to see M14 w/ and w/out grip Vs FN Fal, as well as Dragunov Vs PSG1 in terms of accuracy in leading w/ 2nd shots, etc.

[psijaka]

Thanks for tthe feedback arcanine

I intend taking a look at all of the semi autos, although these need handling differently as you do not know when the next shot will be. What I may do is work out at what time the gun is guaranteed to have recovered from the kick of the previous round. This should be relatively easy, but I want to fill in the gaps in the full auto chart first (ACOG, IR scope etc). And the CZ75 full auto!

[arcanine2009]

Yeah, I was thinking about that too.

In terms of overall recoil is there any other wayto tell which gun is more accurate/ better than the other in that regard (FN Fal Vs M14 and Drag Vs PSG1) when leading w/ second or more shots?

[psijaka]

I would say the one that recovers the fastest. Depends upon how quickly you can loose off accurate shots.

[psijaka]

Have updated the Recoil Chart to include results for Automatic rifle/LMG ACOG/IR scopes. Feedback welcome.

[phale]

I like contrasting the M16 IR and RPK ACOG stats

[asasa]

Nice chart!

Makes weapons with centered recoil appear better than they really are, though.

[psijaka]

Feb 15, 2011 20:21:21 GMT -5 asasa said:

Nice chart!

Makes weapons with centered recoil appear better than they really are, though.

Thanks for the feedback asasa. And thanks again for putting me on to the recenterspeed/5 relationship; this was the breakthrough for me; the rest was just playing with numbers, Lots of numbers!

I agree that recoil is only one aspect of a gun's worth, and that at medium ranges skilled players can compensate for recoil that has a bias upwards and to the right (Famas, Galil, Enfield, MAC11, Spectre etc). However, for serious longshotting, nothing less than absolute accuracy will do.

These guns also seem to have a relatively high chance of the second shot being on target, but after that, the cumulative effect of the recoil becomes very significant, and the chances of subsequent shots being on target diminishes rapidly. Emphasises the importance of burst firing.

IMO the chart does quite clearly emphasise the superiority of the Famas over it's cousin, the Aug (although I am sure that it has it's devotees). I was trying to resist the temptation of sugesting nerfs on this thread, but a reduction in the recenterspeed of the Famas springs to mind.......

[psijaka]

Feb 15, 2011 20:17:03 GMT -5 phale said:

I like contrasting the M16 IR and RPK ACOG stats

I knew that ACOG increased recoil, but just did not know how badly. Whenever I have tried using one I have always given up in frustration after a few games; just what is the point.

The calcultions reveal that there are 2 guns that perform well with the ACOG. I must revisit the M16 ACOG.

And the Enfield. This gun, unlocked early, should perhaps not be discarded quite so readily. It also has SMG ADS intime (without ACOG), and one of the highest "second shot" accuracy figures in the game.

And as for the RPK + ACOG - Why would anyone bother using this gun? Just touch the trigger and you will be shooting birds out of the sky!

Figures are rounded to the nearest percent, so that 0% means less than 0.5%, by the way. There is still a chance, albeit negligible.

[mannon]

The M16 appears to be really odd in that it actually gets less recoil with an ACOG because the ACOC centerspeed is actually an improvement! O,O Hence... less recoil with ACOG. And the IR is even better.

As near as I can tell actually the M14+Grip and Skorpion+Grip seem to have the highest chances of recentering completely on the second shot in their class. (Skorpion is actually over 50%!!) I exclude sniper rifles and shotguns because many of them always recenter due to long firetimes, and I didn't bother looking at LMG's. I just thought I'd mention.

[Mr.Bob Loblaw]

Nice work psijaka! Just so I'm clear, the '% on Target' section deals with the probability of the 2nd/3rd/4th bullet decals (as seen if I was shooting at a cement wall, for example) overlapping/touching the previous bullet's decal? I guess I don't fully understand the definition of what a 'Target' is in this context.

Also, anyone know if it would it be a fair assumption to make that these calculations don't change at all across the CoD series? If that is the case, there can be good potential here for the next installment; hopefully after a few days after release (MW3?) a similar chart can be produced.

Even better, a simulator that plugs in the centerspeed calcs, similar to what this guy was close to pulling off.

[psijaka]

Feb 16, 2011 7:01:56 GMT -5 Mr.Bob Loblaw said:

Nice work psijaka! Just so I'm clear, the '% on Target' section deals with the probability of the 2nd/3rd/4th bullet decals (as seen if I was shooting at a cement wall, for example) overlapping/touching the previous bullet's decal? I guess I don't fully understand the definition of what a 'Target' is in this context.

Also, anyone know if it would it be a fair assumption to make that these calculations don't change at all across the CoD series? If that is the case, there can be good potential here for the next installment; hopefully after a few days after release (MW3?) a similar chart can be produced.

Even better, a simulator that plugs in the centerspeed calcs, similar to what this guy was close to pulling off.

Thanks for the feedback. The chart seems to be well received so far.

My definition of "on target" is when the gun has completely recovered from the kick of the previous rounds. So the 2nd bullet should follow the same trajectory as the first.

I would assume that the calculation method is the same across the COD series, and now that I have the method sorted, it would be easy enough to produce a new table for MW3 once the necessary data is available.

I have been thinking of writing a javascript simulator to produce the tabulated results. The recoil patterns link you posted would also be a good idea; and this could be done with a spreadsheet easily enough (say a 20 round burst for each gun).

But I would like to go as far as I can with the basic chart before I do this; I still have some combinations of SMG with ACOG to cover (and the CZ75 full auto). And then I may look at worst case recentering times for the semi autos.

Lots to think about!

[imrlybord7]

I would argue that this makes the AUG look quite a bit better than the FAMAS. These guns take too many bullets to kill to just look at the chance of the second shot being perfectly on target. The AUG not only beats the FAMAS for every subsequent shot, it also has its accuracy degrade at a much slower rate.

Also, it's funny that so much effort has been put into calculating this stuff. For CoD4 and WaW we had all of the info to do this but never bothered because guns weren't clones of each other where recoil ended up being the primary method of differentiation between similar guns.

[psijaka]

Have added SMG ACOGs amd CZ75 auto

[kevdood83]

Feb 16, 2011 9:13:17 GMT -5 psijaka said:

My definition of "on target' is when the gun has completely recovered from the kick of the previous rounds. So the 2nd bullet should follow the same trajectory as the first.

I'm still confused by what "on target" means...

You mean you fired off the whole clip and tracked the first 4 bullet holes? and those that overlapped were "on target?"

Or did you pop one round, wait, pop a 2nd round... etc etc?


Also, what does the RC/Round column mean?

Nice chart though! (Once I understand it.)

[mannon]

No he's using a spreadsheet and a brute force simulation of tons upon tons of shots fired. On target means literally EXACTLY on target. In other words the centerspeed completely cancelled out the viewkick.

RC stands for recenter I do believe. It indicates how much viewkick that gun can completely compensate for between shots fired based on centerspeed and firetime.

[kevdood83]

Feb 16, 2011 16:16:57 GMT -5 mannon said:

No he's using a spreadsheet and a brute force simulation of tons upon tons of shots fired. On target means literally EXACTLY on target. In other words the centerspeed completely cancelled out the viewkick.

RC stands for recenter I do believe. It indicates how much viewkick that gun can completely compensate for between shots fired based on centerspeed and firetime.

Ah ok, so it sims as if you were shooting a full clip, tons and tons of times. And over all that data it shows the percentage of time the centerspeed completely cancelled out the viewkick for the 2nd, 3rd, and 4th rounds to hit EXACTLY where the 1st one did.

So theoretically in that data, the 2nd round might have missed while the 4th hit in one case... then in the next case the 2nd round hit and 3rd and 4th missed.... then in the case after that on the 3rd hit...etc etc. After simming 100 clips we get the overall percentage for the 2nd, 3rd, and 4th rounds hitting.

Correct?

[psijaka]

Feb 16, 2011 16:09:12 GMT -5 kevdood83 said:

Feb 16, 2011 9:13:17 GMT -5 psijaka said:

My definition of "on target' is when the gun has completely recovered from the kick of the previous rounds. So the 2nd bullet should follow the same trajectory as the first.

I'm still confused by what "on target" means...

You mean you fired off the whole clip and tracked the first 4 bullet holes? and those that overlapped were "on target?"

Or did you pop one round, wait, pop a 2nd round... etc etc?


Also, what does the RC/Round column mean?

Nice chart though! (Once I understand it.)

mannon has answered your question of what "on target" means (thanks, mannon)

The way I produced this data was to simulate the firing of 50,000 three round bursts, and looking at the percentage of times that the gun fully recovers from the effects of recoil (Viewkick) before the 2nd, 3rd, and 4th shot (if it were being fired) were fired. Thus the "on target" bullet would go where the preceding bullet went (assuming that the aim had not been deliberately changed).

RC/round means the amount of recentering that can occur between successive rounds, based upon recenter speed/5 and the time between rounds.

Hope that this helps.

[psijaka]

Have been playing with my recoil spreadsheet and have simulated firing a small number of 10 round bursts. The results can be displayed as below:
EDIT - the simulation used to create these images contains an error


This is the result of simulating 6 ten round bursts from each of the 2 guns, so it is just a snapshot of what might happen.

The scale is in 100 viewkick unit squares, from -400 to +400, both horizontal and vertical.

Is this information actually of any use?

[Mr.Bob Loblaw]

Feb 16, 2011 18:58:11 GMT -5 psijaka said:

Is this information actually of any use?

I for one think so. If correct, then hey, we have that much better of an understanding of recoil patterns for any CoD AR/SMG/LMG, right out of the box.

[kevdood83]

That fire spread chart is great!

A chart of each guns spread in3 or 4 round burst would be awesome. 10 bursts maybe.. like in the one you posted above.

[phale]

Feb 16, 2011 18:58:11 GMT -5 psijaka said:

Have been playing with my recoil spreadsheet and have simulated firing a small number of 10 round bursts. The results can be displayed as below:



This is the result of simulating 6 ten round bursts from each of the 2 guns, so it is just a snapshot of what might happen.

The scale is in 100 viewkick unit squares, from -400 to +400, both horizontal and vertical.

Is this information actually of any use?

That looks beautiful. This would be great for being able to simulate recoil for all the guns and attachments without having the limitations in the game. I think that the next big step would be to somehow incorporate range into the visuals, so we can have a better understanding of how the recoil affects gameplay. For instance, get images of player models at different ranges and scale them correctly with the recoil simulation.

Your work is excellent and another important breakthrough in our understanding of CoD. Which reminds me, if you have the time, could you do the same for CoD4 and WaW guns?

[psijaka]

Thanks for all the positive replies, bros!

I need to think carefully where I go from here. I think that my method is sound, but am not yet 100% sure. So I would like to wait a while to see if anyone can spot any errors.

I really would like people to try to understand how I have achieved these results, and to verivy that I am on the right track. mannon (on the "how does centerspeed work" thread) has produced exactly the same results as me for the 2nd shot probabilities, using a mathematical method. He (and I) assume that the kick of a gun is distributed randomly within a rectangle defined by the 4 kick figures that are published for each gun. So we can work out the likelihood that the gun will have kick in either direction, that is greater than the recentering that is applied before the next round (recenter/5*time between rounds). The either is important; if either horizontal or vertical kick > recentering, then the gun will not recenter in time.We have come up with this method independently, and we get the same results exactly for the "second shot".

The simulator I use fires 50,000 three round bursts, applying randomly generated vertical and horizontal kick to each round. The simulated figures from the 2nd shot match closely figures derived mathematically, so I am confident that the simulator works correctly for the 3rd and 4th shot.

But am I right in assuming that kick is randomly distributed within a rectangle?

And am I right in assuming that centerspeed works on horizontal and vertical kick independently?

Evidence posted on the "how does centerspeed works" thread suggests that this is how things work; and it certainly makes sense from an "in game" programming point of view to keep things as simple as possible; there is a lot going on already and why be so precise about trying to make the kick and the guns response to it that accurate?

Next for me to consider is how to display the information graphically. The Famas v Aug image is only based upon 6 ten round bursts for each gun, and if I repeat the simulation, I often get an image that looks significantly different. If I were to increase the number of bursts (a lot of work) the image may get too cluttered to understand. Catch 22. Note that the image suggests that the Aug usually kicks to the left a bit; this is not the case as the left and right kick figures are the same size.

I need to think about this and I will post some selected images representing recoil in different styles to see what you think. A lot of work and it isn't just going to happen overnight. I do have a life!

[MastaQ]

Wow. Even the Uzi with ACOG + Rapid Fire is more accurate than the regular RPK after the second shot. That's pretty pathetic.

Great chart.

[psijaka]

OK here goes...

recoil plots of 100 ten round bursts for the Famas and Aug:
EDIT - the simulation used to create these images contains an error


1 square = 100 ViewKick units

[psijaka]

Feb 17, 2011 3:12:16 GMT -5 MastaQ said:

Wow. Even the Uzi with ACOG + Rapid Fire is more accurate than the regular RPK after the second shot. That's pretty pathetic.

Great chart.

Thanks for the feedback mastaq. Some pretty grim figures there!

I really think that I have got to grips with this now. Just hope that my method is sound!

[kevdood83]

Feb 17, 2011 3:17:17 GMT -5 psijaka said:

OK here goes...

recoil plots of 100 ten round bursts for the Famas and Aug:



1 square = 100 ViewKick units

That looks good, I only have a couple of thoughts on it.

I was kind of thinking the same thing you already said... too much data might make the chart for each gun look too cluttered.

Also, 10 round burst might be too much and create a lot of variance with guns that kick more wildly... leading to inconclusive data.


....Perhaps try 100 bursts of 6 rounds? Might that give a more clear idea visually?