Black Ops Recoil Plots + MW3, COD4 links

[mannon]

I think 10 rounds is a bit much unless you're wanting to compare good the weapons are at spraying, but that's a tactic with limited use. (Granted, it is limited to use by the weapons that are good at it or otherwise by situation.)

On the other hand if you are burst firing a weapon you would only care about the recoil pattern up to N rounds. Unfortunately I'm not really sure there's a good single figure for N. I guess mebbe look at on average how many hits it takes to kill with most weapons (3 to 4?) and on average how many misses will be incurred before the kill (no idea, 2 to 3 for most peeps?) then add those together, (5 to 7, mebbe split the difference at 6?).

It's hard to say because you're going to fire in longer or shorter bursts for different weapons based on your experience with them and their characteristics.

As for clutter. I had theorized on doing some more in depth recoil analysis for MW2 using screen capped images to locate all the bullet decals. It was unworkable due to hardware issues, though. (My borrowed capture thingy wasn't even HD and was HORRIBLE quality... aka it was a Dazzle, and I don't even have it now.) Anyway I was planning on actually just acquiring the X and Y coordinates from the pics so that I could do them over and over again, and then compile them in a more meaningful manner.

One thing I considered doing was basically taking each set of hits and applying them as a translucent additive layer in my graphic editing software and then flattening it. This way a single hit would be dim, but visible, and overlapping hits would get brighter.

I also considered using different colors for different shots, aka red for 2nd shot, green for 3rd, and blue for 4th. Unfortunately that only gets you out to the 4th shot. (1st shot is always centered, though so I was thinking just a white dot for it.) With each shot a different color they each have their own independent channel so it should all stack and collapse properly when using additive layers to produce a final full color image that gives an interesting map of the first four shots. Sadly there's no real way to go beyond four shots without either losing some fidelity in the data or coping out and using a second image. (Though a second image could map the 5th, 6th, and 7th shots without any loss in fidelity.) One could perhaps simply combine shots and thus do something like red for 2nd and 3rd shot, green for 3rd and 4th, and blue for 5th and 6th. That would get you up to a significant burst. Though it would lose a tad of fidelity I don't think it would be so bad. As for how translucent to make each individual hit that is simply determined by how much overlap you have per channel or how many layers you use, though I'd pick a certain amount to be standard for all layers so that everything can be compared directly.

Anyway, that also all sounds like a crap ton of work, though you could probably script it. But I thought it might give you some ideas. You could, for example, just run two or so of the above images and combine them as 50% additive layers. Shots overlapping in the same test run wouldn't build, but that's okay, it's more of a guide anyway since we're not, (yet) converting view kick units into actual degrees or anything.

[psijaka]

I have been thinking along the same lines as you, mannon, and here are 2 images I produced earlier where the 2nd and 3rd shot are one bright colour, and shots 4-10 another colour.

I'm only creating the images from a spreadsheet scatter plot, and copying into Paint to produce the .gif files, so I have not been editing the images. Not sure I will either; a lot of work.

[EDIT] - I think that a 3 round burst is a good choice, especially as this is what you get with the M16/G11.[/EDIT]

Anyway; here they are:

EDIT - the simulation used to create these images contains an error.


Feedback welcome!

[psijaka]

@ mannon

I am sure that we could assume that ViewKick units are proportional to degrees, as the angles we are talking about are very small. Any errors would probably be insignificant; if you are far enough out for the errors to be significant, then you have missed by a mile!

[psijaka]

Feb 16, 2011 13:56:54 GMT -5 imrlybord7 said:

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.

Sorry, missed this post, imrlybord7

You have a point about the subsequent shots being less likely with the Famas; interesting how different people interpret the data.

I think that the recoil makes a huge difference in how a gun "handles". I personally do not get on with the guns that pull up and to the right, but others swear by them (Famas is the most popular gun, according to Treyarch). AK47 for me!

[mannon]

Mmm... pretty.

There's probably an easy way to script collecting GIF's as layers and all that, but even so I'm not sure it's really necessary.

Then again if you could export as a vector graphic then each point would be represented despite overlap. You'd then just represent each point as a dot of a given color and translucency, but meh. Also more work since you'd almost certainly have to import it into a vector graphics program and fiddle with it. ;p

I'm fairly certain that ViewKick units are degrees/time, so yes they should be proportional to degrees. We just need to know what the time number is. But even without an exact match we can still compare one weapon to another.

[kevdood83]

Feb 17, 2011 13:01:32 GMT -5 psijaka said:

I have been thinking along the same lines as you, mannon, and here are 2 images I produced earlier where the 2nd and 3rd shot are one bright colour, and shots 4-10 another colour.

I'm only creating the images from a spreadsheet scatter plot, and copying into Paint to produce the .gif files, so I have not been editing the images. Not sure I will either; a lot of work.

[EDIT] - I think that a 3 round burst is a good choice, especially as this is what you get with the M16/G11.[/EDIT]

Anyway; here they are:



Feedback welcome!

This is something like what I was talking about, I really like the way this looks much better.

This creates a much more clear image, and since most skilled players arent shooting in bursts of 10, having bursts of 3 in a different color really helps to paint the picture of what kick behaviour to expect.

This is great Psijaka, Im really looking forward to the finished project.

[psijaka]

I will be taking a long hard look at my methods as a post on the "recenterspeed" threat has made me doubt their validity. Please treat the 3rd and 4th shot figures with caution until I resolve this. I will also not be creating a gallery of simulated 10 round burst plots just yet. But I will continue to look at how to display the data.

[mannon]

I think it's still an invaluable generalization, even if it's not quite a fully fleshed simulation.

[arcanine2009]

With the Famas having a top right recoil pattern, is it safe to say that the way to combat this is...

Shoot a little bottom left from the target?

Same thing applies to combating recoil patterns of all guns?

[mannon]

If you want to connect on the target on your 2nd and later shots, mebbe. But your first shot is always on target (barring the railgun nerf and 0.2 second hipfire to ADS firing delay while moving instead of standing still).

Generally I think it's better to put your first shot on target then compensate as you fire, though you do need to realize that you will be off target when you stop firing and have to decompensate. This works best with weapons that kick vertically as it is much easier to compensate in one dimension than two.

Though you certainly may wish to aim at the bottom left part of your target much the way people will aim guns with high vertical recoil at the gonads and let it kick it-self up to headshot. But I still think you want your first shot to actually hit your target.

[psijaka]

Feb 17, 2011 19:14:35 GMT -5 mannon said:

I think it's still an invaluable generalization, even if it's not quite a fully fleshed simulation.

Thanks for the encouragement; I am beginning to have my doubts.

I don't think the method I use will be too far "off the mark", although it is becoming obvious that there is more to it than meets the eye.

Once we do get to the bottom of it, I will be able to plug the new method into the spreadsheet and get some new figures.

[psijaka]

Feb 17, 2011 23:56:46 GMT -5 arcanine2009 said:

With the Famas having a top right recoil pattern, is it safe to say that the way to combat this is...

Shoot a little bottom left from the target?

Same thing applies to combating recoil patterns of all guns?

I would always shoot on target, usually at the body. That way then at least the first shot is on target, and there is a significant chance that the second will be as well.

The answer to handling recoil at longer range is to fire in bursts.

[mannon]

We might not be able to adapt your method to scale down to actual degrees of offset, but it is still a very handy way to compare weapons to each other. I think it is still a good generalization. More uncompensated viewkick on average means more off target, regardless of how we model it, so we're still going to get good results even if they aren't perfect.

That's my 2 cents worth anyway.

Not that I don't want to really nail it all down solid, too. heh But we're working on that. I keep trying to think of a way to model this thing even on a single axis and the complexity keeps causing me to rethink it.

[djgump35]

I love seeing such statistics, makes me wonder though if I fully understand the application.

Reading the chart, wouldn't the skorpion with grip be the most accurate multi-shot gun?

I am pretty sure that was the consensus answer before, but now it is clearly spelled out
with a lot of statistical data here. I think that gun with hardened pro, would be pretty good with a decent reaction time.

[mannon]

Yeah a Skorp with grip has an ungodly narrow vertical kick. It actually recenters vertically on the second shot at a rate of 97.6% of shots fired! O,O Even the horizontal recentering on second shot comes out at 57.6%. So a little better than half the time, (56.22%) the gun will be a laser focused on a point, and the rest of the time it's still basically a laser focused on a horizontal line.

It's just too bad that it's horizontal and not vertical or it would really rape people. Though, sadly headshots aren't particularly useful with it at min or max damage ranges. It could be great for aiming into a crowd of enemies and pinging all their heads, but the damage numbers mean that won't necessarily add up to fewer hits to kill, so it might be better to aim shoulder height or so where the targets are wider.

I haven't really used it though, so I can't speak from practical experience, just the numbers.

[mannon]

Something else to consider... While reducing firetime can potentially increase or decrease the position off target when the second shot fires, the same is not true of ViewKick, which is the actual random variable.

Centerspeed and FireTime will remain constant while we generate random viewkicks within the weapon's range. That's good news. It means that when you're mapping VKick 0 to X recenters and X to Y results in a steady increase in position off target. An increase in VKick never results in a second shot landing closer to target. Okay so it may not be directly proportional in a linear fashion, but you don't have to worry about anything flipping it around on you and you do know that a greater leftover VKick does mean greater distance from zero... at least we know it does for any given weapon.

We can't really say the same weapon vs weapon, though. In fact it would basically scale inversely to centerspeed. (In other words the same VKick produces a smaller kick with a higher centerspeed, hence the same leftover VKick should also represent less in terms of position... I think.)

[psijaka]

Feb 18, 2011 5:59:59 GMT -5 djgump35 said:

I love seeing such statistics, makes me wonder though if I fully understand the application.

Reading the chart, wouldn't the skorpion with grip be the most accurate multi-shot gun?

I am pretty sure that was the consensus answer before, but now it is clearly spelled out
with a lot of statistical data here. I think that gun with hardened pro, would be pretty good with a decent reaction time.

The Skorpion with grip is the most accurate multi shot gun. Unfortunately the damage at anything other than very close range make it a 5 hit kill, so this negates the high accuracy.

Maybe just as well, or we would be seeing "Nerf the Skorpion" threads!

I definately wouldn't go for Hardened Pro with the Skorp; it has poor bullet penetration so why bother? Sleight of Hand is a good choice, as you will be reloading often with a mag of only 20.

[djgump35]

I just thought with Hardened pro, reducing the flinch it would make cqc almost automatic.

I could see the ammo thing, perhaps warlord with extended, or sleight of hand with scavenger.

[mannon]

I think for CQC it probably doesn't need the grip, the damn thing is just really accurate already. Grip just magnifies that.

[arcanine2009]

OP, I just got the center speed times for semi-autos and sniper rifles..

M14:
w/out grip: 5 X 80/1400 = .285
w/ grip: 5 X 80/1400+200 = .25

FN FAL:
5 X 70/1400 = .25


>Sniper Rifles

WA2000:
5X65/825 = .39

Dragunov:
5X65/500 = .65

L96A1
5X100/500 = 1 <-- also note that L96A1's RPM is 60 RPM also, or 1 round per second..

PSG1
5X95/100 = .475


What do you guys think? Does M14 w/ grip = FNAL in recoil (not counting kick direction), and does this mean PSG1 is better at leading second shots then Dragunov?

[mannon]

Looking at the PSG1 and Druganov they are quite different. It looks like a bit of a lottery. The PSG1 certainly can recover from it's maximum kick faster than the Drug. However it not only will always kick up and right it will always kick too far to recover during it's firetime giving it a 0% chance of recentering. The Druganov on the other hand has a 48.76% chance of recentering horizontally and 28.36% odds of recentering vertically by the second shot. It still kicks a bit more right than left, but it's not all that unbalanced.

That all means that the PSG-1 may sometimes recenter faster than a Druganov, but will never recenter before you could fire the second shot, while a Druganov might get a fortunate kick and be recentered already before you can even fire or at least before the PSG-1 would. And there lies the issue. It's not as consistent. The PSG-1 kicks like a mule, but it's also one of the most consistent kicks in the game with only a variance of 5 VKU's in either direction.

Then again that's a wide visual kick that may make it harder to visually acquire your next target.

I would really have to get some hands on experience with them. Personally, I doubt I would like the PSG1 with that big ass kick. I hated the way the Barret kicked. Even the Intervention just felt less jarring even though it couldn't fire as fast. Though it was actually easier to land second shots with the Barret since coming back from the recoil was more predictable than the Intervention's idle. I just hated it, though. (Then again once I got the WA2K I pretty much never touched them again. They are really only good for SP quickscoping. For any real sniping there's a better weapon in either the WA2K or M21 EBR.)

See the centerspeed topic for full stats on all weapons.

[imrlybord7]

If you want to compare on recoil alone play Hardcore. You will find that the Dragunov destroys the PSG in that regard.

[psijaka]

Will include semi autos in the recoil table soon

I will be displaying % 2nd shots on target in minimum fire time and also recovery time from worst case kick (how long you have to wait before you can be certain that the gun has fully recovered)

[psijaka]

Have added M14 and FAL to the chart.

2nd shot accuracy is somewhat hypothertical with semi auto weapons but is included for completeness.

Of more importance is the maximum time the gun would take to recenter in response to worst case kick.

Is this useful?

If so will work on Snipers and Handguns next.

[psijaka]

Plots of 100 simulated 10 round bursts posted below recoil chart, for Automatic Rifles.

The method ofdetermining exactly where the gun is if it is not recentered before the next round is fired, is a bit dodgey, but the plots do give a good indication of how a weapon behaves if you let rip on full auto.

Please give feedback; are the images useful; Colours/size OK etc?

Note- please ignore the larger images of Famas/Aug that I posted earlier; the spreadsheet contained an error and the recoil pattern will not be as shown. In particular, the Famas always recovers from downwards or leftwards kick, so all the dots should be in the top right quarter of the diagram.

[mannon]

Nice. I like the colorly colorfulness. ;p Actually I'm serious, I think the colors help. Glad you're sticking with it.

[psijaka]

I played around with the colours quite a bit, and looked at assigning different colours to 2nd, 3rd, 4th shot etc, but the end result is just a mess! Sticking to just 2 colours for shots 2-3 and 4-10 gives a much clearer picture.

Have now posted SMGs and LMGs. Just look at the Skorpion!

Noticed that I have got the aspect ratio slightly wrong, but will live with it for now.

Not sure how far to go with attachments etc. Maybe just a few examples to show the effect of ACOG, IR, Grip and Rapid fire on a selected few.....

And a chamber of horrors! (Uzi RF ACOG)

[mannon]

Very nice. You can actually see the cross pattern developing in most of these where shots close enough to 0 on one axis recenter making shots more likely to land directly vertical and horizontal than in the diagonal directions.

[psijaka]

Yes you can see the cross pattern along the axes. These images are "hot off the press" after correcting the simulation error I discovered, and I hadn't appreciated the significance of them.

Note that they are not present in the larger, earlier (wrong) Famas/Aug plots.

[asasa]

This is awesome. Great job!