Bullet velocity - inaccurate ?

[maturin 12]

Isn't the Vintorez a marksman rife? It's shooting like a pistol.

[InqWiper 0]

Isn't the Vintorez a marksman rife? It's shooting like a pistol.

It uses a heavy subsonic 9x39mm SP-5 cartridge. The bullet is very effective at penetrating body armor. For this purpose it is equipped with a hardened steel or tungsten tip to penetrate a 6 mm (0.2 in) high-density steel plate at 100 m; a 2 mm (0.08 in) steel plate or a standard army helmet can be fully penetrated at 500 m, however the rifle is typically employed under 400 m.

Not really made for long range sniping I guess, but I would use it over a pistol :)

[max power 21]

No, you don't do that in this kind of context. You have formulas (which you deduce from newtonian mechanics, experiments and dimensional analysis) and you have the parameters, then you do the math and get a result. In this case there's a model of drag that is widely used and accepted for its validity. I have simulated the trajectories for a set of parameters that are natural. My bullets aren't of 20m calibre, and I don't shoot under water. A null hypothesis is completely irrelevant here. Sure, I have tried small deviations to these parameters and as expected there are small deviations to the flight time, it's only when you have grenade sized bullets that you get flight times of over one sec. But I'm not going to test all parameters including the completely unplausible ones.

So if this simulation, using normal parameters, supports the idea that the bullets fly to slow, then it's likely also good evidence that the in-game bullets fly to slow. BIS probably use some simplified numeric integration (not euler backwards) in order to save CPU-time, but then a lot of people have a lot of CPUs.

Nope. Gathering evidence to support a theory is the wrong way to go about it. You don't need to test all possibilities. You just test the one, since this is a simple test of speed. If the bullet behaves as predicted consistently, then it's not possible that it doesn't, therefore it satisfies the null hypothesis test.

I suppose if you really were into it, you would do the test with high and low fps to see if there was a difference, and then you would do a test to make sure that fps doesn't somehow influence the test in some other way (like retard the script you're using to measure the length of time the bullet lives).

[Spokesperson 0]

Nope. Gathering evidence to support a theory is the wrong way to go about it. You don't need to test all possibilities. You just test the one, since this is a simple test of speed. If the bullet behaves as predicted consistently, then it's not possible that it doesn't, therefore it satisfies the null hypothesis test.

No, you don't need to test all possibilities. But you can mathematically prove that flight times will vary in certain intervals within certain parameter intervals and that's enough. The null hypothesis is not relevant here, I'm not doing statistics here. I'm doing a mathematical deduction.

I'm not using Arma to do computations because Arma is what we suspect to be wrong. I've used Matlab to see if I can reproduce Arma-like results and I can't with normal parameters. The simulations are backed up by an analytic calculation as well. The bullet flight-times are too high.

[max power 21]

Okay, I think I understand what you're doing. You were seeing what it should be and messing around with bullet weight. I thought you were trying to say something about how the game handles stuff.

Do these mods also alter drag or just initial velocity? I dont think the issue at the moment is initial velocity, but drag.

I did some testing on several weapons and will post the results below. Slowing down time to 0.1 seemed to result in the bullet hitting its target almost 0.14 seconds later so all tests were performed at normal time. The tests only measure XY velocity not Z. If a bullet is dropping straight down it will return 0 speed.

Press show spoiler to show results:

Bizon shot at 500 meters (9x18):

Initial velocity: 349.219m/s

Final velocity(no not 0 you funny guy you, the one before that): 88.0545m/s

Distance travelled: 499.062m

Flight time: 3.095s

Average velocity: 161.248m/s

Velocity maintained: 25.2147%

Bizon SD shot at 500 meters (9x18 SD):

Initial velocity: 319.515m/s

Final velocity: 124.031m/s

Distance travelled: 499.259m

Flight time: 2.63098s

Average velocity: 189.762m/s

Velocity maintained: 38.8185%

MP5A5 shot at 500 meters (9x19):

Initial velocity: 399.815m/s

Final velocity: 187.829m/s

Distance travelled: 499.191m

Flight time: 1.891s

Average velocity: 263.983m/s

Velocity maintained: 46.9890%

MP5A5SD6 shot at 500 meters (9x19 SD):

Initial velocity: 319.727m/s

Final velocity: 165.713m/s

Distance travelled: 500.115m

Flight time: 2.246s

Average velocity: 222.669m/s

Velocity maintained: 51.8295%

M1911 shot at 500 meters (.45 ACP):

Initial velocity: 259.608m/s

Final velocity: 157.856m/s

Distance travelled: 499.624m

Flight time: 2.51599s

Average velocity: 198.579m/s

Velocity maintained: 60.8055%

AK74 shot at 500 meters (5.45x39):

Initial velocity: 899.994m/s

Final velocity: 446.901m/s

Distance travelled: 498.838m

Flight time: 0.804001s

Average velocity: 620.445m/s

Velocity maintained: 49.6560%

M16 shot at 500 meters (5.56x45):

Initial velocity: 929.997m/s

Final velocity: 453.482m/s

Distance travelled: 498.799m

Flight time: 0.852s

Average velocity: 585.445m/s

Velocity maintained: 48.7617%

M4A1SD shot at 500 meters (5.56x45 SD):

Initial velocity: 319.856m/s

Final velocity: 237.372m/s

Distance travelled: 500.088m

Flight time: 1.84399m

Average velocity: 271.198m/s

Velocity maintained: 74.2121%

AKM shot at 500 meters (7.62x39):

Initial velocity: 709.963m/s

Final velocity: 268.571m/s

Distance travelled: 498.939m

Flight time: 1.21001s

Average velocity: 412.344m/s

Velocity maintained: 37.8433%

Vintorez shot at 500 meters (9x39):

Initial velocity: 299.783m/s

Final velocity: 182.342m/s

Distance travelled: 499.604m

Flight time: 2.183s

Average velocity: 228.862m/s

Velocity maintained: 60.8247%

DMR shot at 500 meters (7.62x51):

Initial velocity: 899.975m/s

Final velocity: 568.629m/s

Distance travelled: 499.447m

Flight time: 0.705017s

Average velocity: 708.418m/s

Velocity maintained: 63.1828%

Dragunov shot at 500 meters (7.62x54R):

Initial velocity: 869.997m/s

Final velocity: 540.838m/s

Distance travelled: 498.858m

Flight time: 0.739998s

Average velocity: 674.135m/s

Velocity maintained: 62.1655%

Saiga shot at 500 meters (12 gauge slug):

Initial velocity: 387.556m/s

Final velocity: 29.0898m/s

Distance travelled: 499.732m

Flight time: 6.074s

Average velocity: 82.274m/s

Velocity maintained: %7.5060

M107 shot at 500 meters (12.7x99):

Initial velocity: 849.986m/s

Final velocity: 663.827m/s

Distance travelled: 499.213m

Flight time: 0.678009s

Average velocity: 736.292m/s

Velocity maintained: 78.0986%

KSVK shot at 500 meters (12.7x108):

Initial velocity: 849.981m/s

Final velocity: 645.956m/s

Distance travelled: 498.806m

Flight time: 0.708008s

Average velocity: 704.52m/s

Velocity maintained: 75.9965%

Interresting to see the quite huge difference between normal bizon and bizonSD. Also obvious sillyness in AKM and Saiga testings.

Those mods make it so the ballistic arcs and muzzle velocity match their real world counterparts. Unless there is something weird going on with gravity, the ballistic arc should be wrong if the drag model is wrong, no?

Some of those tests are really out of whack. I don't know if you could automate a larger sample of tests to make sure none of these are anomalies caused by some computer related hoju?

Edited January 13, 2010 by Max Power

[InqWiper 0]

Okay, I think I understand what you're doing. You were seeing what it should be and messing around with bullet weight. I thought you were trying to say something about how the game handles stuff.

Those mods make it so the ballistic arcs and muzzle velocity match their real world counterparts. Unless there is something weird going on with gravity, the ballistic arc should be wrong if the drag model is wrong, no?

Some of those tests are really out of whack. I don't know if you could automate a larger sample of tests to make sure none of these are anomalies caused by some computer related hoju?

When doing the testing I fired the weapons several times until I reached pretty much the same value several times in a row. The differences between 2 shots were never really big though, it was like 0.05 seconds flux max in bullet travel time.

[max power 21]

Thanks for being so thorough!

Well, I was looking at the initial velocity numbers and comparing them with the muzzle velocities of their real life counter parts, and with the exception of the sd ammo which seems to initialize a little fast, they all seem very close.

[cartier90 0]

But it's not the initial velocities in question , but the way they slow down so much too create a overall flight time that's too slow.

[max power 21]

Well, I was confirming that the initial velocities weren't the problem. We might find that the ballistic arc is all wrong. I wonder what the expected final velocity of a 5.56mm NATO would be after 1 second of flight time.

I'm also pretty sure it's a drag problem but you never know. I wonder what these tests would find with the Fincuan's ballistic mod. I wonder what config entries he changes and in what way he changes them.

[cartier90 0]

I appreciate everones effort here to solve this conundrum. It was only a hunch that the flight time was too long - I don't have use of my comp At the mo so look forward to playing again soon with perhaps a solution ? If poss.

[Spokesperson 0]

We seem to agree that something is wrong. Next poster creates a ticket ;)

[maturin 12]

what about range for larger weapons? I found myself unable to reach some trucks with the Shilka's 20mm cannon at 2000 meters, even if I arced it a good ten degrees.

Does anyone have any input about this? I don't really know what I'm talking about, so it's not worth a thread, but according to wikipedia the effective range of the bushmaster is 3000 meters. So shouldn't the 25mm cannon on the Cobra be able to reach targets at 1500-2500 meters without lobbing rounds like a catapult?

[fincuan 0]

I'm also pretty sure it's a drag problem but you never know. I wonder what these tests would find with the Fincuan's ballistic mod. I wonder what config entries he changes and in what way he changes them.

99.9% of actual numbers I used are tested and calculated by NWD, and according to him the resulting trajectories correspond very well to what they are in real life. He achieved this by modifying both initial speed and drag. I only ran a few quick tests which seemed to confim the same values still apply in Arma2.

The small arms ballistics simulation seem very good to me, and I doubt the flighttimes are off by more than a few milliseconds provided that FPS remains at a playable range. Even if they were we are talking about so short periods of time that something like stopwatch and "I feel" measurements don't count. Also something which no-one seems to have mentioned yet: The initial speed might be 900 m/s, but the bullet starts slowing down the moment it leaves the barrel, and by the time it gets to 450 m it'll be travelling at a much much lower speed. The bullet also travels a longer distance than 450 m because it doesn't fly straight, it arcs a bit.

Edited January 16, 2010 by Fincuan

[Spokesperson 0]

Also something which no-one seems to have mentioned yet: The initial speed might be 900 m/s, but the bullet starts slowing down the moment it leaves the barrel, and by the time it gets to 450 m it'll be travelling at a much much lower speed. The bullet also travels a longer distance than 450 m because it doesn't fly straight, it arcs a bit.

No, if you read carefully, that has been taken into account.

[Spokesperson 0]

Ticket created

http://dev-heaven.net/issues/show/8219

[fincuan 0]

Well in my opinion you rushed the ticket a bit. There might or might not be an issue, but not much credible ingame data. So far I found exactly one set of ingame measurements I would trust, those of InqWiper, and even those do not agree with my own results. An example: In my tests ~0.622 sec to 450m with 5.56 ball, using ACE which has basically the same NWD-values as my mod.

Edited January 17, 2010 by Fincuan

[cartier90 0]

I believe they tested on vanilla. Thanks spokesman .

Edited January 17, 2010 by cartier90

[Spokesperson 0]

Yes I rushed it and there's no edit button. :p

The best thing would be if someone could post a set of data with trajectories taken from Arma2. I.e measurements of position and time every 0.001 sec for example. Out of that it would be possible to make a graph and then compare it to what it should be.

There should also be more clarity about the parameters that were used in game (+ vanilla only).

Maybe the results get different with non-constant C_d (might be used in game), but the ticket is mainly a suggestion that they look closer at the ballistics, because there might be an error.

[cartier90 0]

It was assigned quickly with Guba believing it to be important - good sign.

[Spokesperson 0]

When you do the tests, shoot straight (not up or down), and in vanilla. Otherwise you won't have an initial velocity of say 900m/s in the direction of the "target". (if you don't you have to compensate for that, or the flight times will be higher than they should) Shoot above the sea and measure the flight time while the bullet is in air after 500m. (You would have to shoot from some levitating platform). And disable wind.

[Evil_Echo 11]

I looked at the ACE ticket about this too.

You folks basically have the notion correct - the game physics is off.

The problem is that we ( ACE ) can fit the curve to match range or flight time but not both. Range seems more important in most cases. Although flight time does factor in when timing a shot at a moving target.

[fincuan 0]

Here's a few sets of stock Arma2 1.05 ingame ballistics observations. I only ran it for a few weapons, because by all means this should revel if there are some general problem with the physics. My test were conducted on Utes in the default "editor conditions". After firing the bullet was "reset" to flying completely horizontally with its original initial speed, at which point recording was started. An FSM was used to output the relevant data to arma2.rpt each frame. I'll see if I can get the testbed to a releasable form later on, but here's the data.

Anyone got reliable real data to compare these with?

M16A4

Edited January 18, 2010 by Fincuan

[fincuan 0]

DMR

[fincuan 0]

AK74

[fincuan 0]

AKM

At least we see why there aren't many 7.62x39 sniper rifles around :)

Edited January 18, 2010 by Fincuan