nkenny

@mickeymen Testing of optics. Repeat of experiments according to the parameters listed above, save for the inclusion of a 400 meter stage. At the 400m stage the AI stance is set to "AUTO". Optics background The AI have three types of Optics configured. None, OpticType 1 ~ RCO, MRCO combat optics, and OpticType 2 ~ SOS, LPRD sniper optics. These optics add extra fire modes to the AI. These fire modes come with unique settings for engagement range, rate of fire and dispersion (generally dispersion is the same for all ranges however). Data Source Source data Conclusions Within 400 meters the effect of optics is negligible. The core weapon class is more important. That said optics add to the maximum range by with which a weapon can engage. This testing scheme does not explore that to further detail, but looking at the MXM config the following becomes apparent. MXM, single * aiRateOfFire = 2 * aiRateOfFireDistance = 500 * MaxRange, 450 * MidRange, 350 * MinRange, 120 MXM, single medium Optics 1 * aiRateOfFire = 6 * aiRateOfFireDistance = 700 * MaxRange, 750 * MidRange, 450 * MinRange, 120 MXM, single far Optics 2 * aiRateOfFire = 8 * aiRateOfFireDistance = 900 * MaxRange, 900 * MidRange, 550 * MinRange, 200 Optics should therefore extend the maximum range which the AI will engage. This also accounts for why there is no immediate increase of potency. Even when an optics fire mode is used, the weapon is still limited by external ballistics-- and rate of fire in these longer range modes are generally slower. By vanilla configuration, Arma2 and Arma3 has tended to give Ranged Combat Optics to squad leaders. I would speculate that the reason for this is to get the squad to engage targets at greater range. Interestingly, given other variables, this may be counterproductive. Ideally the AI squad leader should be trying to move his element as close as possible where a larger number of squad weapons are accurate and able to engage. edit: To clarify. Giving squad leader a short ranged weapon-- like an SMG, might not be a bad idea. -k

@mickeymen Sure, I can do that. It is relatively easy to set up with lane system. As each weapon is individually configured, I suspect that optics will affect the performance characteristics of each weapon in a unique manner. -k

@JB47394 Nah. Keep ranting. Preferably into a bug report.

bis_fnc_taskPatrol is not bad. [group,position,range] call bis_fnc_taskPatrol; If you are running CBA there are even more powerful solutions.

Create a script which assigns move waypoints to a random selection predetermined locations. Add a Cycle WP to have them repeat it. Break the cycle on a trigger or variable. Then update all units with a Guard WP on their current location. Now the AI will react somewhat dynamically to your cleverly preplaced 'Guarded By' triggers.

Check the link I posted. You would be better served with a MOVE wp

https://community.bistudio.com/wiki/Waypoint:Guard I'm having some trouble understanding what you are trying to achieve with your script. Note that adding guardPoints creates a hierarchy of importance (starting with the first) which gives some explanation why your group won't move. Again: to what end are you using dynamic Guard waypoints? If you want the unit to dynamically hold and fortify positions, there are better options. The best use of Guard WP and guardPoints is to create a network. A single GuardPoint (by script or 'Guarded By' X trigger) surrounded by one or more units with Guard WP. When the guardpoint is undefended, a fresh unit will move to it. Additionally, free units will move to respond to spotted units in the area. -k

@FoxFort Same. Been refreshing the forum all day long. -k

Fun missions. Great production for such a short term project. The terrain is beautiful, to the point and distinct. I look forward to similar mission projects in the future 🙂 -k edit: also second that question. What mods supply the headgear in the pictures?

@tortuosit I had the same impression a few patches back, i.e., two-three weeks ago. Was not in a position to make a proper test of it.

This is simply fantastic. -k

Perhaps the easier solution is to change the task. If the Titan launcher is not meant to be fired during the mission-- any stand in intel or prototype document may take its place. An alternate option may be that the special Titan launcher is a Static weapon. Perhaps this will ease your implementation? Speaking as a reasonably veteran mission designer, if the technology is holding you back-- think outside the box. It is better to keep producing content than rub your head uselessly against the same problem. -k

The best and most important mod of 2019. Epic in scope and facial hair. -k

I really appreciate the new, generic additions to the game. The quality of the added assets is tremendous and positively dripping with potential. In fact, I suspect that I am not alone in being more excited about the editions to the sandbox than I am exploring the single player content. I hope it doesn't go unnoticed that the dimensional qualities of generic assets multiply their usefulness. (1) A single olive drab retexture lets an asset used across as a stand in for any faction on nearly any terrain. The reskinned FV-720 Mora is beautiful in its Livonian livery, but had it been generic green, it could have been used everywhere. The same is true for uniforms and other trappings. Given the expanse of Arma3 assets, this is potentially an endless task, but a few choice retextures expand the range where those particular assets can be used infinitely. (2) Another dimensional aspect is technology. Especially older, legacy weapons and equipment add to the simulations depth. When Apex introduced the AKM and AKS and RPG7, it suddenly became possible to actually simulate more primitively equipped foes. This feeds back to the authenticity of the modern weapons; the MX rifle is truly experienced as a 2035 rifle only when contrasted to an older, inferior one. Notably, because this is a 2035 setting, one does not need an entire library of obsolete weapons. The AKM can be a stand in for any AK-variant. I can illustrate these aspects with two examples: (a) Consider the Mk.30 HMG. With thermal and magnified optics it is a tough weapon to use as a mission maker. In the hands of a player it quickly achieves sniper like accuracy and the AI is not far behind, particularly when employed against helicopters. The modern HMG is simply too good for the majority of mission types. If it was possible to disconnect the M2 HMG from the technical truck to create a tripod variant, mission making possibilities would expand immensely. (b) Closely related is the deadliness of high tech weapons. This is particularly true of anti-aircraft weapon systems. This is primarily a AI concern, but there are no ways of discouraging helicopters from entering air space beyond a binary, "I see you, you are killed" responses. There are no soft air countermeasures. The static (or man portable) AA Launchers are both extremely potent. A config level change which would allow the RPG7 to target air craft (and preferably infantry), would immediately open new, very interesting scenario possibilities. Or in a mode similar to what is suggested in [a], creating a tripod mounted Minigun based on the Qilin assets, would introduce a fitting for 2035, static ZSU-23-2 equivalent. Good for discouraging both helicopters and UAVs. Obviously playing modded Arma3 will alleviate these concerns. For all my suggestions covered in [1] and [2] have been met by the community. But mods are not always available and these low level expansions to the core game expand the horizon of potential scenarios in a huge manner. If anything, take it as critical viewpoint for future games. -k

The function responsible for the change in hand animation does not appear to be running on AI forces. The simplest way is to equip accordingly: correspond the attached grip with the underlying classnames. In the case of the HK416: rhs_weap_hk416d145_grip, rhs_weap_hk416d145_grip2, rhs_weap_hk416d145_grip3 -k

When I read this I had to try. My testing can confirms it. Vehicle mounted units gain suppression if a bullet strikes a ground surface (including buildings) near them. There is no difference between static weapons, open topped vehicles or armoured vehicles in this regard. In other words: It is possible to suppress the crew of a tank by hitting the ground next to the tank. Hitting the tank itself will have no effect. The extent to which suppression affects crew accuracy or decision making processes is unknown to me. -k

@lonewolf96 Next DLC Swedish Foreign Legion

if (isServer) then { _units = units group this; _house = nearestBuilding this; _house = _house buildingPos -1; _house = _house call BIS_fnc_arrayShuffle; if (count _units > count _house) then {_units resize (count _house);}; { _x disableAI "PATH"; _x setUnitPos selectRandom ["UP","UP","MIDDLE"]; _x setPos (_house select _forEachIndex); _x addEventHandler["Fired",{params ["_unit"];_unit enableAI "PATH";_unit setUnitPos "AUTO";_unit removeEventHandler ["Fired",_thisEventHandler];}]; } foreach _units; }; There are many ways to accomplish what you want. Pasting the above code into the leader of a units init field should work. It will pick the closest building and move the group there. *Untested* -k --- edit: I realise now that what I responded with was not entirely what you were asking for. There are however some dynamically spawning mission systems out there-- EOS like @Tobur linked is one. I have also made some, but they generally rely on mods to a lesser or greater extent. -k

@ISparkle I avoided the up-armoured CSAT troops, because I wanted to test the very baseline protection levels for each faction. Fundamentally I agree with your point. However the solution I would want is to add some generic types of armour which would also happen to fit the Gorka uniforms closely. That way a single retexture multiplies in usefulness. With that said, it bears mentioning that the Spetsnaz faction is currently the best armoured recon troops in the game. -k

Don't underestimate armoured CSAT armour suit. Let me give some numbers. The data is derived from a single static shooter equipped with a 6.5 Katiba Rifle. At a distance of 150M are four inactive targets. The principal measurements are time to kill all four targets and accuracy (shots fired and hits). Each permutation was tested in 50 cycles. The numbers presented test a number of things. First a baseline, that is facing standing targets. Second is facing prone targets. Then follows armoured targets in the order of CSAT, NATO and AAF. Unfortunately the AAF tests were run with only the Lite Carrier Rig, so numbers for that faction are not entirely accurate. I'm looking primarily at the number of hits and time spent for each testing revolution. So. Looking at the figures. If the targeted fire-team drops prone, their chances of survival increase quite substantially. The average time increases from 12 to 22 seconds-- almost doubling. The number of hits registered, interestingly, also increases to 374 (a 34% increase from the baseline). If the team remains standing, but is equipped with basic CSAT armoured uniform, i.e., the Spetsnaz equivalent, the survival time again leaps ahead: 27 seconds (15 second increase). Almost half a minute more to find and kill the shooter. The number of hits necessary is now 614, a 125% increase. NATO armour is the most powerful basic armour tested in this run. The basic plate carrier and enhanced combat helmet add the best survival potential: 34 seconds and a whooping 773 hits necessary~ a 176% increase from naked troops. Finally the lightly armoured AAF trooper is still not left behind. With the lite carrier the average time increases to 23 seconds (11 second increase) and requires 585 hits (109% increase) --- There are some factors to bear in mind. This test is done in a white room. No cover, no motion, no incoming enemy fire. In a realistic setting, even 3 seconds of response time is substantial. For very little weight CSAT fatigues greatly improves survivability. Especially when combined with a robust helmet. Finally, the Russian Spetsnaz troops are recon troops. In comparison NATO recon troops are without armour at all. -k

Strictly speaking the Contact-DLC Gorkas have the same level of protection as the CSAT fatigues at half the weight. There is also a new unique camo pattern, Tundra, for the Russians. Quite advanced in other words. -k

Some information in regards to AI suppression: 1. Suppression is measured 0 to 1. 1 is max suppression. 0 is the default state. 2. Suppression reduces effective AI accuracy. This has been tested in some detail here here. 3. The time an AI is suppressed is affected only by Skill and not rank. A max skilled soldier will remain suppressed for less than 1 second A 50% skilled soldier will remain suppressed for around 7 seconds A 20% skilled soldier will remain suppressed for 11 or more seconds. The chart reads like this: edit: Suppression here indicates how much time in seconds it takes a soldier to go from Full to no suppression. 4. Suppression seems unaffected by caliber. It was my belief that higher caliber weapons would increase Suppression values. However in testing with a tank I did not find this to be the case. 120mm overhead had the same effect as a 7.62x51mm. Driving the tank closer to the AI changed nothing. Explosives seem to have no special suppressive qualities. edit: Further testing does seem to indicate some difference between near hits by 5.56x45 and 9.3x64. However the brief period of suppression means that this makes for little if any difference in practical terms. Testing suppression yourself One way of testing suppression values is by pasting the following code into your own mission. (NB: works best for smaller ops, and only in scenarios were EAST is the enemy) Another is simply by playing the small scenario I've put together for that purpose: LINK All tests performed on the latest Devbranch, 145791

According to Wikipedia: "The Livonian language is spoken by fewer than 100 individuals as a second language, and is understood to be fast approaching extinction."(Livonia) Last native speaker died in 2013. I suspect it would be rough to get voice actors. -k

Updated experimental suite by nkenny Introduction Having a bit of time on my hands I have rebuilt my experimental weapons testing suite. The previous testing facility was calibrated to demonstrate the large difference between different mods configuration of the same weapon. With the ability to generate more numbers more easily, comes the opportunity to perform more refined tests between otherwise similar weapon systems. Experiment I ran the experiments using no mods and the latest Development Release candidate 1.95.145769. The tests were run on a fresh, default profile with no tweaks made to AI skill or settings. The goal was to test every assault rifle present in vanilla Arma 3 (with DLCs) to create a point of comparison and reference. Description The new testing suite is built around ten walled lanes. Each lane is reused for testing cycles where shooter and targets are spawned and despawned. The shooter is equipped with a weapon and six magazines. The shooter begins facing four targets~ a fire team. The targets are unarmoured and unarmed. Each target is presented on a line with five meter intervals. The distance between shooter and targets can be configured. All actors are forced to remain static for the duration of the cycle. Each cycle runs until all targets are dead or the shooter runs out of time (three minutes). Typically running out of time means the shooter has expended all ammunition. Once either condition is fulfilled, the lane is cleared-- shooter and targets despawned-- and a new cycle is initiated. The results of each cycle is recorded. The end of the exercise has the user manually pasting the data into a larger file. Measurements Each weapon was tested across four ranges: 25, 100, 200 and 300 meters. Each testing cycle ran 100 times, with a reset and reload of the mission between weapon systems. The record contains figures for each round, as well as an overall average of the current figures: Weapon classname Accuracy, hits divided by shots fired Hits, both total and per cycle Shots, both total and per cycle Time spent, measured in seconds and minutes Cycle Success or Time out. (In the case of Time out, the number of survivors is recorded) Expectations Assault weapons to perform in a homogeneous manner with a few outliers. Data There is a massive amount of data. For persons interested in each test cycle in detail I provide a link to the raw data in the references. To give an example of how this data looks, and is read, check the SPOILER tag in the references section. The data presented is the 6.5 MX rifle. First spoiler contains the data, second provides a screenshot of the new VR suite. Below that is a summary of all weapons organised by range categories. Source: link Findings The test revealed clear trends in weapon configuration. As a rule of thumb the Average Time measurement is the best indicator of efficiency. The number reflects the time in seconds necessary for the AI to kill four unarmoured soldiers at the given range. (1) This confirms and earlier finding. The AI will in perfect conditions kill four soldiers with considerable speed. Cover, concealment, body armour and suppression will all mitigate these numbers. But being spotted is very dangerous. (2) A surprise finding is that accuracy is at the highest at the 100 Meter mark. At 25 Meters weapons are configured for automatic fire, and this has a toll on marksmanship. In fact some weapons suffered Time outs at close range. (3) As noted in earlier tests. When presented with four targets the AI will not focus on a single enemy. The AI will shift fire, seemingly randomly, to cover all four. As discussed elsewhere, while this compromises the timing, this is not a flaw in game or simulator terms. (4) I will now go through each range category and list, roughly, the best and worst. At 25M the best weapons are the AKM, CAR95 (CTAR) and mk20. The Katiba trails very closely behind, losing only a little to accuracy. The AK12 is also interesting, because it has a kill time similar of the top weapons, but does at a cost of ammunition-- 600 rounds more than its older relative the AKM. The worst weapon of the bunch is the MX rifle which suffers the greatest numbers of timeouts and longest time (nearly double!) At 100M the stand out weapons are the SPAR-16, MX-rifle and Promet (MSBS65). Throughout the test the TRG21 will closely match the SPAR-16 in performance. Already the AKM is outmatched and will continue to suffer. At 200M the MX and Promet remain top dogs, with the 5.56 rifles SPAR-16 and TRG21 offering equal time to kill. The AK12 and Katiba and are starting to suffer. With the CAR-95 falling closer to the AKM in time to kill. At 300M the MX simply dominates in time to kill. The Katiba has a time to kill of 66 seconds, but Time outs betray inconsistent performance. Interestingly the SPAR-16 is now starting to flex its superior range to the TRG21, though I expect the 5.56 caliber is making both weapons suffer. The Mk20 is now beyond effective range. Low accuracy, low time to kill, and many timeouts. The CAR-95 is notable for tremendous high rate of fire: 5433 rounds. The AKM proved unable to engage targets beyond 250 meters. Conclusion The large number of weapons and wealth of data makes for a potentially confusing dataset to work with. I will here present a way of understanding the numbers: Weapon pairs. Finally I will discuss some aspects the numbers do not suggest. The Mk20 and TRG21 The case of the Mk20 and TRG21. These weapons provide a good starting point to understanding how weapons are configured. Notably the weapons feature very different performance characteristics: the Mk20 is an excellent CQB weapon-- high rate of fire and accuracy at 0-100 Meters, whereas the TRG21 comes into beyond this range. It is notable that I can find no simulation or realism reason why the weapons perform as they do. As far as I can tell both are reasonably modern bullpup rifles slinging 5.56x45 caliber bullets. There is no self explanatory reason why one should be better at CQB than the other! It seems to be a decision made purely for GAME DESIGN reasons*. Bohemia wanted AAF forces to be equipped with a close quarters weapon, and handled a mixture of both to the guerilla counterparts. This is not necessarily poor game design. Having weapons with strongly configured differences makes for more colourful scenarios. It also makes AI behaviour more consistent. Weapon pairs What is also interesting is that the weapons pair mentality is repeated! Consider the default CSAT and NATO weapons: the Katiba and MX. These weapons reflect the characteristics of the Mk20 and TRG21, only to a greater degree: The MX rifle is an excellent (if not the best) long range assault rifle. It is amongst the worst at 25 Meters or less, which is where the Katiba dominates. Had the testing targets featured armour, I expect the powerful 6.5 rounds to have skewed results more. Another weapons pair is present in the Apex DLC. The SPAR-16 and CAR-95 (CTAR). Here the difference is more nuanced, but still present. The SPAR-16 is a superior ranged weapon, whereas the CAR-95 is better at closer ranges. NATO has the better long range weapons, CSAT has the better close in weapon. An interesting development is that the CAR-95 has such a high effective rate of fire. At 300 Meters, where it ostensibly is the worse choice, the weapon will still put out a very large amount of bullets (5433, almost 2000 more than the next in line)-- the spray of bullets still matters in range, as the suppressive effect has considerable effect on AI accuracy. Of course not every weapon is organised so easily as NATO = long range, CSAT = short range, there are a few odd ones out. Amongst them is the new Promet assault rifle. I expected this to be ahead of the pack in all categories. Certainly in player hands its high mechanical rate of fire and good iron sights can make it a favourite-- but barring any changes to AI configuration, it actually features a very distinct band of effectiveness: at 100 to 200 meters it operates at its peak**. Another set is the AKM and AK12. The AKM is clearly intended as a low tier weapon, best suitable for guerilla forces. The AK12 is an upgrade in all ways. Whereas the AKM loses effectiveness at 50 meters, the AK12 remains relevant out to 200M, whereafter it stops being in the game. Like the CAR-95, the high rate of fire will play into suppressing the AI. The high rate of fire also introduces some volatility, it oscillates between fast kill times and and simply missing. Best weapons (a) What to read from this: Going by numbers alone, the MX rifle is by far the superior assault rifle in Arma3. Aside from the 25M range category it is either in the lead or completely dominates. At 300M there is no real contest. (b) The SPAR-16 is close behind. It performs consistently across all ranges, being near the top or middle at all ranges. This reflects the SPAR-16 being in many ways simply a better TRG21, held back by its 5.56 caliber. (c) That does not mean that the Katiba is a slouch. It is amongst the best CQB rifles and remains relevant out to longer ranges. At 300 meters it featured the second best time to kill. As I will get to later, the Katiba also exists within a different ecology of small arms than does the MX. Missunderstandings What not to read from this. The best AI weapons of each range category need not reflect player ability or preference. The AI is bound by different rules than players are. The idiosyncrasies of AI handling have little bearing on players experience of the weapons. Especially concerning the practical application of full-auto fire in a high ping environment-- or when applied actively to shoot through cover. It is also important to remember that each weapon exists within different ecologies of equipment. Optics play a major part in improving weapon accuracy-- all of these tests have been performed using no attachments. The availability of armour on both own and enemy troops and the presence of support weapons within the squad all greatly affect the overall fire power expressed by a squad. Logistical challenges, such as the weight of weapons and ammunition, or the benefits of full ammunition compatibility, are also not accounted for. In fact the presence of inferior or lower tiered weapons only add to the simulation. That the older AKM is an inferior rifle (aside from CQB) only adds meaning to the superiority of the ultra modern MX-series. Final words Contrary to the expectation that all weapons to perform largely identical-- subject only to the whims of caliber, I found strong thematic tendencies. CSAT weapons are geared towards close combat, whereas NATO weapons are better at longer ranges. The weapon configuration show cleverness and intelligence: Weapons perform intuitively, with only a few unexpected hard limits (250M for the AKM) or uncommunicated contrasts (the Mk20 is a CQB rifle whereas the TRG21 is long range rifle). It is also clear that the assault weapons have been considered not isolated but in context to available equipment. Future tests 1. In the future I want to test consistency of Global Mobilization weapons to their Vanilla counterparts. 2. An accuracy and killing power test comparing Standing and Prone, and the effect of various default armour classes. 3. Some glaring omissions are: the AKS in 5.45-- which handles more like an SMG in any case. The Type-115 and the new carbine version of the AK12. These may be explored in the future*** 4. A look at the light machine guns and marksman rifles present in Arma3 References - This document in Goggle Docs format, may be easier to read - Original thread on nopryl.no, further numbers, weapons and experiments are here - Experimental mission, links to mission for your own testing - Raw data, to draw your own conclusions MX Rifle data at 200M & New VR course -k * In fairness the Mk20 iron sights are very suggestive that this weapon is better at closer ranges. The TRG21 in contrast has excellent adjustable sights. ** It is possible the Promet rifle should be paired, contrasted to the new AK12 carbine. *** As a matter of fact these experiments have already been performed and are available in the source material. Both in RAW data form and the linked Google document. I chose to keep them away from this analysis because these rifles are in the case of the AKS, much closer to SMGs, the AK12U, essentially the same as the AK12, and the Type-115; a special, very limited use Assault Rifle.

That matches my own testing. AI does not prefer players to other targets. Though I have not created an experimental suite to test that specific thing, there have been no indications of such behaviour. I would attribute this to: 1. Confirmation bias. The game stops when the player dies. Making it easy to overstate the importance of one death. 2. Player perception differs from AI perception. Players will underestimate the AI's ability to look through cover and listen to sounds. Additionally, players are louder than AI. By which I mean to say a player is likely to engage aggressively with automatic fire and explosives. 3. Players occupy roles of importance. Squad leaders, machinegunners, Anti-tank weapons, armoured vehicles and helicopters. These all make the player a more threatening target. (In line with 1 and 2) -k