Advanced Ballistics - Drag model revamp & Spin drift correction #5566

ulteq · 2017-09-28T10:18:53Z

Moved away from using the drag tables from the GNU exterior ballistics library
The drag functions are now based off this data from JBM Ballistics: http://www.jbmballistics.com/ballistics/downloads/text/
The differences are minor, but some players might still appreciate the additional authenticity
Fixed an error in the spin drift calculation (introduced with Advanced Ballistics - Performance improvement #4708)
Fixed an error in the code that removes the vanilla drag

Many thanks to @dedmen, @brainslush and @Laid3acK for the extensive help.

- Moved away from using the drag tables from the GNU exterior ballistics library - The drag functions are now based off this data from JBM Ballistics: http://www.jbmballistics.com/ballistics/downloads/text/ - The differences are minor, but some players might still appreciate the additional authenticity

jonpas

I trust what you are doing.

Neviothr · 2017-09-28T15:36:08Z

addons/atragmx/functions/fnc_calculate_distance_at_velocity.sqf

@@ -29,7 +29,8 @@ private _distance = 0;
 while {_velocity > _thresholdVelocity} do {
    private _bc = GVAR(targetSolutionInput) select 14;
    private _dragModel = GVAR(targetSolutionInput) select 15;
-    private _drag = parseNumber(("ace_advanced_ballistics" callExtension format["retard:%1:%2:%3", _dragModel, _bc, _velocity]));
+    private _temperature = GVAR(targetSolutionInput) select 5;
+    private _drag = parseNumber(("ace_advanced_ballistics" callExtension format["retard:%1:%2:%3:%4", _dragModel, _bc, _velocity, _temperature]));


Whitespace after parseNumber and format.

Neviothr · 2017-09-28T15:36:19Z

addons/atragmx/functions/fnc_calculate_solution.sqf

@@ -118,7 +118,7 @@ while {_TOF < 15 && (_bulletPos select 1) < _targetRange} do {
    _trueSpeed = vectorMagnitude _trueVelocity;

    if (missionNamespace getVariable [QEGVAR(advanced_ballistics,enabled), false]) then {
-        private _drag = parseNumber(("ace_advanced_ballistics" callExtension format["retard:%1:%2:%3", _dragModel, _bc, _trueSpeed]));
+        private _drag = parseNumber(("ace_advanced_ballistics" callExtension format["retard:%1:%2:%3:%4", _dragModel, _bc, _trueSpeed, _temperature]));


Neviothr · 2017-09-28T15:36:28Z

addons/rangecard/functions/fnc_calculateRangeCard.sqf

@@ -103,7 +103,7 @@ while {_TOF < 6 && (_bulletPos select 1) < _targetRange} do {
    _trueSpeed = vectorMagnitude _trueVelocity;

    if (_useABConfig) then {
-        private _drag = parseNumber(("ace_advanced_ballistics" callExtension format["retard:%1:%2:%3", _dragModel, _bc, _trueSpeed]));
+        private _drag = parseNumber(("ace_advanced_ballistics" callExtension format["retard:%1:%2:%3:%4", _dragModel, _bc, _trueSpeed, _temperature]));


ulteq · 2017-09-28T16:28:16Z

@Neviothr, can we fix those style issues later, in a different pull request?

I trust what you are doing.

@jonpas, please do not merge it yet.

WIP

…ation - The error was introduced with this PR (#4708)

dedmen · 2017-09-30T15:18:51Z

extensions/advanced_ballistics/AdvancedBallistics.cpp

-    double A = -1;
-    double M = -1;
+double calculateRetard(int DragFunction, double DragCoefficient, double Velocity, double Mach) {
+    std::vector<double> machNumbers = {};


Performance optimization would be to have these tables be built at compile-time.
Or making them global variables and just selecting a pointer to them here.
Because this way you have a memory allocation and memcpy everytime a drag function is choosen.

dedmen · 2017-09-30T15:19:38Z

extensions/advanced_ballistics/AdvancedBallistics.cpp

+    for (int i = 0; i < machNumbers.size(); i++) {
+        if (machNumbers[i] >= m) {
+            int previousIdx = std::max(0, i - 1);
+            double lcd = dragCoefficients[previousIdx];


longer variable names would be neat here. I have no idea what lcd/lmn/cd mean.

Agreed. Even though all we do here is a linear interpolation between the data points.

dedmen · 2017-09-30T15:20:20Z

extensions/advanced_ballistics/AdvancedBallistics.cpp

@@ -309,7 +249,7 @@ double calculateZeroAngle(double zeroRange, double muzzleVelocity, double boreHe

            v = std::sqrt(vx*vx + vy*vy);

-            double retard = calculateRetard(dragModel, ballisticCoefficient, v);
+            double retard = calculateRetard(dragModel, ballisticCoefficient, v, SPEED_OF_SOUND(15));


You are calculating this constant.. at runtime. Everytime the function is called. Why?

RELEVANT VARIABLE NAME

@dedmen, I expect it to be optimized away by the compiler.

It won't be here. because std::sqrt is not constexpr.

GCC 7.2.0 is smart enough. I will test VS 2017 tomorrow.

Visual Studio 2017 produces the exact same assembly code in both cases:

#define SPEED_OF_SOUND(t) (331.3 + std::sqrt(1 + t / 273.15f)) ... double r = calcR(dragModel, ballisticCoefficient, v, SPEED_OF_SOUND(15));

#define SPEED_OF_SOUND(t) (331.3 + std::sqrt(1 + t / 273.15f)) ... const float c = SPEED_OF_SOUND(15); double r = calcR(dragModel, ballisticCoefficient, v, c);

moving it into a const float doesn't change that std::sqrt is not constexpr before C++17.
Actually.. Not even C++17 has constexpr sqrt.
Does that assembly code that it produces calculate the Speed of sound. Or does it use a constant? If it calculates then it does exactly what I said

Just did a release compile. And yes it is calling sqrt at runtime.

I will just replace it with 340.275.

dedmen · 2017-10-02T07:01:08Z

extensions/advanced_ballistics/AdvancedBallistics.cpp

+const std::vector<double> machNumbersG8 = { 0.00, 0.05, 0.10, 0.15, 0.20, 0.25, 0.30, 0.35, 0.40, 0.45, 0.50, 0.55, 0.60, 0.65, 0.70, 0.75, 0.80, 0.825, 0.85, 0.875, 0.90, 0.925, 0.95, 0.975, 1.0, 1.025, 1.05, 1.075, 1.10, 1.125, 1.15, 1.20, 1.25, 1.30, 1.35, 1.40, 1.45, 1.50, 1.55, 1.60, 1.65, 1.70, 1.75, 1.80, 1.85, 1.90, 1.95, 2.00, 2.05, 2.10, 2.15, 2.20, 2.25, 2.30, 2.35, 2.40, 2.45, 2.50, 2.60, 2.70, 2.80, 2.90, 3.00, 3.10, 3.20, 3.30, 3.40, 3.50, 3.60, 3.70, 3.80, 3.90, 4.00, 4.20, 4.40, 4.60, 4.80, 5.00 };
+
+const std::vector<std::vector<double>> dragCoefficients = { {}, dragCoefficientsG1, dragCoefficientsG2, {}, {}, dragCoefficientsG5, dragCoefficientsG6, dragCoefficientsG7, dragCoefficientsG8, {} };
+const std::vector<std::vector<double>> machNumbers = { {}, machNumbersG1, machNumbersG2, {},{}, machNumbersG5, machNumbersG6, machNumbersG7, machNumbersG8,{} };


Sorry I completly forgot that. std::array is constexpr and would be a little better here.
You could then also use dragCoefficients.count() in line 137 instead of the hardcoded number. In case we ever get a G9 or smth.
(dragCoefficients.count() with vector is a bad idea though. Because that would be evaluated at runtime.)
It wouldn't make a noticeable runtime difference I think. It would safe some time when the extension is called first time and a little memory.

Premature optimization is the root of all evil.

- Small performance improvement

ulteq · 2017-10-03T19:33:35Z

Ready to use, if you ask me.

dedmen · 2017-10-03T20:25:18Z

addons/advanced_ballistics/functions/fnc_handleFired.sqf

@@ -20,7 +20,7 @@
 TRACE_10("firedEH:",_unit, _weapon, _muzzle, _mode, _ammo, _magazine, _projectile, _vehicle, _gunner, _turret);

 // Parameterization
-private ["_abort", "_AmmoCacheEntry", "_WeaponCacheEntry", "_opticsName", "_opticType", "_bulletTraceVisible", "_temperature", "_barometricPressure", "_bulletMass", "_bulletLength", "_muzzleVelocity", "_muzzleVelocityShift", "_bulletVelocity", "_bulletLength", "_barrelTwist", "_stabilityFactor", "_aceTimeSecond", "_barrelVelocityShift", "_ammoTemperatureVelocityShift"];
+private ["_abort", "_AmmoCacheEntry", "_WeaponCacheEntry", "_opticsName", "_opticType", "_bulletTraceVisible", "_temperature", "_barometricPressure", "_bulletMass", "_bulletLength", "_muzzleVelocity", "_muzzleVelocityShift", "_bulletVelocity", "_bulletLength", "_barrelTwist", "_stabilityFactor", "_barrelVelocityShift", "_ammoTemperatureVelocityShift"];


private keyword would be faster.

private _abort = true;
is evaluated at compile-time.
private ARRAY is runtime.
Also private ARRAY assigns nil to that variable. and later when you initialize it you assign again. so using private ARRAY you assign a value to the variable one extra time.

We can try to fix that in a separate PR. Feel free to make a GitHub issue for it.

But I think we're using private ARRAY in many places.

fun-fact. using private keyword is also faster than not using private at all.
using private ARRAY however is slower than not privatizing variables. on average.

Yeah.. I'll make a PR that changes it everywhere where appropriate.

Sounds good.

private ARRAY is only in use because some components haven't gotten the love of cleanup yet since private VAR = VALUE was introduced.

dedmen · 2017-10-06T16:56:18Z

addons/advanced_ballistics/functions/fnc_handleFirePFH.sqf

@@ -33,7 +31,7 @@ private _aceTimeSecond = floor CBA_missionTime;
            drop ["\A3\data_f\ParticleEffects\Universal\Refract","","Billboard",1,0.1,getPos _bullet,[0,0,0],0,1.275,1,0,[0.02*_caliber,0.01*_caliber],[[0,0,0,0.65],[0,0,0,0.2]],[1,0],0,0,"","",""];
        };

-        _bullet setVelocity (_bulletVelocity vectorAdd (parseSimpleArray ("ace_advanced_ballistics" callExtension format["simulate:%1:%2:%3:%4:%5:%6:%7", _index, _bulletVelocity, _bulletPosition, ACE_wind, ASLToATL(_bulletPosition) select 2, _aceTimeSecond, CBA_missionTime - _aceTimeSecond])));
+        _bullet setVelocity (_bulletVelocity vectorAdd (parseSimpleArray ("ace_advanced_ballistics" callExtension format["simulate:%1:%2:%3:%4:%5:%6:%7", _index, _bulletVelocity, _bulletPosition, ACE_wind, ASLToATL(_bulletPosition) select 2, CBA_missionTime toFixed 6])));


Minor error. format still expects 7 elements but you only give 6. Doesn't cause any runtime error and the extension doesn't care. So not a problem.

ThiagoGattiCarvalho · 2017-10-24T13:29:34Z

I predict bullet drops in Excel using Point-Mass formulas and my results are the same as Litz Advanced Ballistics software.

Good thing is that I can turn it into a Monte Carlo simulation by adding randomness to muzzle velocity. If I also add controlled randomness to range estimation and consecutive drop error (simulating AI) then, I have a WEZ simulation, just like Litz.

So, my suggestion is the inclusion a new ACE parameter:
ace_muzzleVelocityVariation = 10; // meaning muzzle velocity could be - 10 to + 10 m/s according to a normal distribution.

Anyway,

My key parameter for trajectory simulation are:

Air density calculation from temperature and humidity, like Litz' Advanced Ballistics Poit-Mass Solver does. I still struggle to achieve this.
Speed of sound calculation for the given weather (which is always changing).
Bullet data, like you already have.
A reference table relating mach speed and CD - Coefficient of Drag, which you already have
A good interpolation method for reading the above table.
Very very important, to the simulation interval. For 1200 FPS I calculate accurate numbers but, it is too much resource consuming, which may lead to the next topic.
A good gravity constant, which may be different than 9.81 for the sake of bullet drop and FPS optimization.

This is an example

t [s] | t lag [s] | vx [m/s] | vy [m/s] | vz [m/s] | v [m/s] | M | CD | FD [N] | ax [m/s2] | ay [m/s2] | az [m/s2] | x [m] | y [m] | z [m]

0.0000 | 0.0000 | 834.45 | 11.78 | 11.01 | 834.60 | 2.45 | 0.2725 | 0.000760 | -529.40 | -17.28 | -6.99 | 0.00 | 0.00 | 0.00
0.0008 | 0.0000 | 834.01 | 11.77 | 11.01 | 834.16 | 2.45 | 0.2726 | 0.000760 | -528.98 | -17.27 | -6.98 | 0.70 | 0.01 | 0.00
0.0017 | 0.0000 | 833.56 | 11.75 | 11.01 | 833.72 | 2.45 | 0.2727 | 0.000761 | -528.55 | -17.26 | -6.98 | 1.39 | 0.02 | 0.00
0.0025 | 0.0000 | 833.12 | 11.74 | 11.01 | 833.28 | 2.45 | 0.2727 | 0.000761 | -528.13 | -17.25 | -6.98 | 2.08 | 0.03 | 0.00
0.0033 | 0.0000 | 832.68 | 11.73 | 11.01 | 832.84 | 2.45 | 0.2728 | 0.000761 | -527.70 | -17.24 | -6.98 | 2.78 | 0.04 | 0.00
0.0042 | 0.0000 | 832.24 | 11.71 | 11.01 | 832.40 | 2.45 | 0.2729 | 0.000761 | -527.28 | -17.23 | -6.98 | 3.47 | 0.05 | 0.00
0.0050 | 0.0000 | 831.80 | 11.70 | 11.01 | 831.96 | 2.44 | 0.2729 | 0.000761 | -526.86 | -17.22 | -6.97 | 4.17 | 0.06 | 0.00
0.0058 | 0.0000 | 831.37 | 11.68 | 11.01 | 831.52 | 2.44 | 0.2730 | 0.000762 | -526.43 | -17.20 | -6.97 | 4.86 | 0.07 | 0.00
0.0067 | 0.0000 | 830.93 | 11.67 | 11.01 | 831.08 | 2.44 | 0.2731 | 0.000762 | -526.01 | -17.19 | -6.97 | 5.55 | 0.08 | 0.00
0.0075 | 0.0000 | 830.49 | 11.65 | 11.01 | 830.64 | 2.44 | 0.2731 | 0.000762 | -525.59 | -17.18 | -6.97 | 6.24 | 0.09 | 0.00
0.0083 | 0.0000 | 830.05 | 11.64 | 11.01 | 830.21 | 2.44 | 0.2732 | 0.000762 | -525.17 | -17.17 | -6.97 | 6.94 | 0.10 | 0.00
0.0092 | 0.0000 | 829.61 | 11.63 | 11.01 | 829.77 | 2.44 | 0.2733 | 0.000762 | -524.74 | -17.16 | -6.96 | 7.63 | 0.11 | 0.00

ThiagoGattiCarvalho · 2017-10-24T13:30:47Z

I mean, it is important to predict exact results in Excel, because it means we are in control here.

Another important feature I think would be nice to share is the calculation of air friction.

You can take all the values of my mod if you want. BTW, why don't you add all the ammo to ACE? I have even made the icons. Just wait one week and I will release an update.

I have simulated all ammo, one-by-one in Litz Point-Mass Solver and then took note of TOF to reach 408 m/s, which is the start of the transonic region. Then I have used Excel's solver on my drop calculator to tell me which BI AirFriction value makes that bullet reach 408 m/s in the noted TOF. This way I know exactly the realistic AirFriction/Drag value required to decelerate the bullet.

dedmen · 2017-10-24T13:38:37Z

Right now our SQF code is a bigger performance problem than the ballistics calculation inside the extension.

ThiagoGattiCarvalho · 2017-10-24T13:42:21Z

You can try to increase the simulation step and the gravity factor to counter balance. <http://www.avg.com/email-signature?utm_medium=email&utm_source=link&utm_campaign=sig-email&utm_content=webmail> Virus-free. www.avg.com <http://www.avg.com/email-signature?utm_medium=email&utm_source=link&utm_campaign=sig-email&utm_content=webmail> <#DAB4FAD8-2DD7-40BB-A1B8-4E2AA1F9FDF2>

…

On Tue, Oct 24, 2017 at 3:38 PM, Dedmen Miller ***@***.***> wrote: Right now our SQF code is a bigger performance problem than the ballistics calculation inside the extension. — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#5566 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AMAZnL9P3OqAIFQtlmb6-cpnLd0BV2CDks5svehhgaJpZM4PnCoU> .

--

ulteq · 2017-10-24T14:11:10Z

Another important feature I think would be nice to share is the calculation of air friction.

@QuickDagger, our airFriction values are generated automatically, based on the advanced ballistics drag model.

why don't you add all the ammo to ACE

We would also needs guns that are able to shoot it. I think it makes more sense to deliver the ammo classes as part of 3rd party weapon packs. At least until magazine groups are working.

Just wait one week and I will release an update.

Your ammo will work perfectly fine with ACE3, if you add those config entries: https://ace3mod.com/wiki/framework/advanced-ballistics-framework.html

If you have any questions or suggestions, please consider joining the discussion on: https://ace3public.slack.com/messages/C04CLGZCK/

ulteq added 5 commits September 27, 2017 15:48

The Mach number is now calculated in relation to the air temperature.

7b2cc57

Improved speed of sound calculation accuracy.

e8f5532

Advanced Ballistics - DLL update

f338faf

Advanced Ballistics - Added drag function reference (JBM Ballistics)

681e50e

ulteq added the kind/enhancement Release Notes: **IMPROVED:** label Sep 28, 2017

ulteq added this to the 3.11.0 milestone Sep 28, 2017

jonpas previously approved these changes Sep 28, 2017

View reviewed changes

Neviothr reviewed Sep 28, 2017

View reviewed changes

jonpas added the status/WIP label Sep 28, 2017

Advanced Ballistics - Fixed calculation error in the spin drift simul…

79ab583

…ation - The error was introduced with this PR (#4708)

ulteq changed the title ~~Advanced Ballistics - Drag model revamp~~ Advanced Ballistics - Drag model revamp & Spin drift correction Sep 29, 2017

dedmen reviewed Sep 30, 2017

View reviewed changes

ulteq added 4 commits September 30, 2017 20:20

More descriptive variable names

0c62f09

Minor performance optimizations

b307f5f

Fixed some minor issues

2779a4e

DLL rebuild

12c1967

dedmen approved these changes Oct 2, 2017

View reviewed changes

ulteq added 6 commits October 3, 2017 10:00

Utilize new 'toFixed' script command

333f2d0

- Small performance improvement

Fixed a typo

6e1189c

Use correct reference speed for the drag compensation

8b8d4a3

Updated all 'airFriction' values to match the new drag model

82f5552

'Default' atmosphere now equals the ICAO standard atmosphere

ba52cd3

Update reference humidity to meet the ICAO standard

15e33b2

ulteq removed the status/WIP label Oct 3, 2017

dedmen approved these changes Oct 3, 2017

View reviewed changes

dedmen reviewed Oct 3, 2017

View reviewed changes

jonpas approved these changes Oct 4, 2017

View reviewed changes

jonpas mentioned this pull request Oct 4, 2017

Short R3F compat update according with the new AB drag models #5587

Merged

dedmen reviewed Oct 6, 2017

View reviewed changes

PabstMirror approved these changes Oct 6, 2017

View reviewed changes

PabstMirror merged commit be482ea into master Oct 6, 2017

PabstMirror deleted the advanced-ballistics-drag-model-revamp branch October 6, 2017 18:05

dedmen mentioned this pull request Oct 6, 2017

Refactor private ARRAY to private keyword #5598

Merged

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Advanced Ballistics - Drag model revamp & Spin drift correction #5566

Advanced Ballistics - Drag model revamp & Spin drift correction #5566

ulteq commented Sep 28, 2017 •

edited

Loading

jonpas left a comment

Neviothr Sep 28, 2017

Neviothr Sep 28, 2017

Neviothr Sep 28, 2017

ulteq commented Sep 28, 2017 •

edited

Loading

dedmen Sep 30, 2017

dedmen Sep 30, 2017

ulteq Sep 30, 2017

dedmen Sep 30, 2017

commy2 Sep 30, 2017

ulteq Sep 30, 2017 •

edited

Loading

dedmen Sep 30, 2017

ulteq Sep 30, 2017 •

edited

Loading

ulteq Oct 1, 2017 •

edited

Loading

dedmen Oct 1, 2017

dedmen Oct 1, 2017

ulteq Oct 1, 2017 •

edited

Loading

dedmen Oct 2, 2017

ulteq Oct 2, 2017

ulteq commented Oct 3, 2017

dedmen Oct 3, 2017

ulteq Oct 3, 2017

dedmen Oct 3, 2017

ulteq Oct 3, 2017 •

edited

Loading

dedmen Oct 3, 2017

dedmen Oct 3, 2017

ulteq Oct 3, 2017

jonpas Oct 4, 2017

dedmen Oct 6, 2017

ThiagoGattiCarvalho commented Oct 24, 2017

ThiagoGattiCarvalho commented Oct 24, 2017 •

edited

Loading

dedmen commented Oct 24, 2017

ThiagoGattiCarvalho commented Oct 24, 2017 via email

ulteq commented Oct 24, 2017 •

edited

Loading

Advanced Ballistics - Drag model revamp & Spin drift correction #5566

Advanced Ballistics - Drag model revamp & Spin drift correction #5566

Conversation

ulteq commented Sep 28, 2017 • edited Loading

jonpas left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

ulteq commented Sep 28, 2017 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

ulteq Sep 30, 2017 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

ulteq Sep 30, 2017 • edited Loading

Choose a reason for hiding this comment

ulteq Oct 1, 2017 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

ulteq Oct 1, 2017 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

ulteq commented Oct 3, 2017

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

ulteq Oct 3, 2017 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

ThiagoGattiCarvalho commented Oct 24, 2017

ThiagoGattiCarvalho commented Oct 24, 2017 • edited Loading

dedmen commented Oct 24, 2017

ThiagoGattiCarvalho commented Oct 24, 2017 via email

ulteq commented Oct 24, 2017 • edited Loading

ulteq commented Sep 28, 2017 •

edited

Loading

ulteq commented Sep 28, 2017 •

edited

Loading

ulteq Sep 30, 2017 •

edited

Loading

ulteq Sep 30, 2017 •

edited

Loading

ulteq Oct 1, 2017 •

edited

Loading

ulteq Oct 1, 2017 •

edited

Loading

ulteq Oct 3, 2017 •

edited

Loading

ThiagoGattiCarvalho commented Oct 24, 2017 •

edited

Loading

ulteq commented Oct 24, 2017 •

edited

Loading