Jump to content

nkenny

Member
  • Content Count

    1276
  • Joined

  • Last visited

  • Medals

Everything posted by nkenny

  1. Have tried to do a search. Am I the only one having trouble getting the alternative syntax for checkAIFeature to work? I am trying to check if a unit has "MOVE" or "PATH" AI features enabled. https://community.bistudio.com/wiki/checkAIFeature -k
  2. This post outlines some of the updates and improvements. With are descriptions of two new mods, and the latter part considers the future and potential of FSM editing. Feel free to add your voice to the technical or feature request discussion. LAMBS Danger has been updated. - Improved civilian behavior.fsm - Fixed all known bugs - Added server key This is the last update before rewriting the FSM from ground up. Links in Original post. LAMBS RPG7 updated to encompass additional weapons as per requests: - Added support for RPG32 and MAAWS - Added server key @LSValmont Added LAMBS Turrets. I like the idea. This is a first pass experiment which explores some of the config options available. It works by tweaking the accuracy of most Coax and mounted machine guns. It also more strictly forces the AI to remain shooting until the burst is complete. Some of the turreted behaviour appears to be hard coded-- especially in how targets are quickly determined dead and cycled. The effect on game play is present, but more for longer ranges. I am still exploring options to deal with close in fighting. The mod is available on my Github or through Steam Workshop. --- LAMBS Ork AI Make boyz more stompier. This is an AI extension for the Warhammer 40k mod There is Only War mod for Arma3. It adds a number of features related to ORK behaviour and AI. Features - Adds custom FSM for Ork behaviour - Makes Ork prefer close combat (and will enter buildings) - Ork vehicles will hammer suspected enemy positions - Adds Mob formation - Adds information sharing - Adds WAAAGH mode - Adds Panic state - Adds routines for looting - Adds occasional suppression DAKKA-DAKKA fire. Mob mentality Orks disregard all formations in favour of the mob. The mob is strong, tight and fast. Praise Mork and Gork Orks that experience personal kills may experience the true blessing of Waaagh. A waaaghing ork becomes permanently immune to suppression, gains increased skills and experience, and heals all wounds. If one ork has it, more do; Waaagh! spreads to all group members within 20 meters. Panic Orks in big mobs are resistant to panic. Once their numbers dwindle, things change. The panic stage makes the ork (and nearby ones) static and hesitant to grab for the glory that is war. --- Steam Workshop --- Technology It is the technical aspects of this mod which were the prime motivator for doing this mod. It features a custom danger.fsm and customised formation.fsm -- based of the civilian formationC.fsm. In many ways it is a showcase for how and what AI modifications can affect in Arma3. Bearing in mind that many of the core AI features are opaque at best of times-- or beholden to an arcane formula of weapon and unit configs. One of the chief difficulties in working with AI is that many of the features are hard to grasp or debug. Especially the base soldier fsmFormation,contains links to functions which I have yet to find. Leaving me to suspect they are related to engine features. Incidentally according to the biki it, fsmFormation, cannot be edited-- which I have done successfully here. Within these difficulties are an interesting promise! Not only is the danger.fsm very powerful. The fact that fsmFormation can be edited makes it possible to write some custom behaviours for vehicle crew. Perhaps it will be possible to remove the more self destructive WW2 bomber style helicopter attack patterns or make devilishly shifty tanks. For mods and conversions that explore specific settings, like TIOW, fsm editing has powerful potential, because it is possible to customise behaviour suitable to each faction. Next step is to explore a bit. Then rewrite the core danger.fsm to allow for more clearly defined stages and states. The ORK AI linked above being a template for that. - Investigating - Assaulting (also in CQC) - Skirmishing or patrolling - Panicked or hiding - Looting or rearming Link to overview picture of current Ork danger fsm : LINK -k
  3. @Tankbuster It doesn't 🙂 Only units that are in a 'danger' mode are affected. Units which are 'stopped' are not given movement orders either. I studiously avoid altering formations or preexisting waypoints. I never touch enableAI/disableAI features either. The only change to look out for is that by default, the 'DIAMOND' and 'FILE' formations trigger a CQB mode. The CQB mode makes the AI clear building positions in a organised manner-- until all known enemies are killed.
  4. @vafana I'll look into it. Especially the RPG42 is a good candidate. I am afraid that if the AI start shooting MAAWs at helicopters they will quickly become annoying.
  5. @db2 Good catch! Throwing an error message? Haven't seen it on my end. @Lucasam I haven't thought about that, but yes. Put on headless clients and servers and it will work 🙂
  6. Thank you for your kind words. I am living out of a backpack for the next few days, which curtails my ability to respond with full detail (and links) The BIKI posts proved very useful when developing the mod! Thank you for yours and others work. I delved into all I could find on the topic. It was interesting to Find a fsm discussion from 2004-2005, predating the ability to setVariable on units (groups, buildings and logics for that matter), which suggested that such a feature would be ideal for use with the danger.fsm-- precisely the technique which I have adopted. The variables I track are: LastAction, lastGesture and isPanicked, on individual units. And isCQB on the group. (No variable is written or checked until needed) -- One of the key decisions was to recognise that the danger.fam is only part of the whole story. The AI seems to be governed by at least three other subsystems which are not immediately available: basic pathfinding, bounding manoeuvres (shooting and covering-- the formations class in CfgFSMs), and what I assume is some group level tactical decision-making AI. In this respect I tried to focus on features that expanded the available REACTIONs to danger, rather than proactive, tactical modes of behaviour. One such exception is the CQB mode -- toggled by File or Delta formation. I suspect this mode would ideally be handled elsewhere, but functions well enough here. -- With this in mind I leave many aspects of AI unchanged, instead relying on the built in features of pathfinding and target priorities. I generally use knowsAbout as a simple check for knowledge, but getHideFrom when more accuracy is required. The former command apparently being part of a series of Hidden and Hide commands developed simultaneously with the danger.fsm, but never fully completed. I try not to toggle AI features, because my solution is intended to be universal and not overrule existing scripts. I have unfortunately not gotten enableAIFeature to return useful information (in accordance with what is posted on the wiki)-- it would be super useful. Because the danger.fsm is only concerned with immediate survival, I have not written any detailed analytic system for the AI. Instead simple checks of vehicle types and available weapons are performed. I suspect that AI healing and reviving is better handled on another level of fsm, and have therefore not explored it. I will definitely check that module. There is a reason why I keep Arma2 still installed. 🙂 -- The current actual implementation of finding safe positions is comparably dumb-- but early in development. I simply find nearby building positions (or failing that a position located away from the threat). I then rely on the (reasonably good) threat and movement evaluations present in vanilla. Haha, indeed turning vehicles is still sorted by doMove shenanigans. Thanks Nathalie. -- I will post a new picture of the fsm flow chart once I return to my PC. For now there should be a comparison on the Steam workshop 🙂 -k
  7. Closely related to the above mod are two other co developments. The LAMBS Suppression mod and LAMBS RPG7 tweak. Both extend AI in interesting manners. LAMBS Suppression increases the AI's susceptibility to suppression. The AI will remain suppressed for a longer time. This in turn has a great effect on game play. Firefights are extended while retaining the dangers of assaulting prepared troops. For a closer look please check the Github or look to the Steam Workshop. LAMBS RPG7 is a simple mod which makes the AI fire the RPG7 more indiscriminately at infantry, vehicle and air targets. As I have argued exhaustively elsewhere, this is a natural extensions of Arma3 small arms tactics. The AI is configured in such a manner that it may prefer the RPG to the carried rifle. Links to Github and Steam Workshop. Thank you for reading, Ken
  8. nkenny

    AI Discussion (dev branch)

    Just noticed community wiki has been updated with this: https://community.bistudio.com/wiki/Arma_3_CfgBrains_Config_Reference Thanks Bohemia! -k
  9. The campaign suggests that the LDF contingent represents an elite force. I do not think that the presence of sidearms detract from the realism any more than do universal issue of personal radios, optics, or any other technological amenity-- in the context of a 2035 setting. Extrapolating from contemporary institutions suggest even more strongly that an increasingly professionalized force are equipment-wise far removed from the large conscript armies of the cold war era. -k
  10. Sounds like you are setting up custom scenarios with the mission editor. Good on you. The best way to accomplish what you want is to write a script and link it with the scenario. To do this you will need basic understanding of your mission file folder structure. Your missions are generally contained in your /Documents/Arma3/<yournameHere>/Missions/<your mission name> file path. 1. Go ahead and make a mission. find the central town and place down a marker. Double click it and give it the name 'CENTER'. 2. Next you will need a few spawning locations. I suggest you place down two or three vehicles suitable for the faction; like some trucks or MRAPS. We will use these vehicles as spawn points for the assault force. On each vehicle we will run the script which will spawn in units! First however, we write the script. 3. the script. Create a .sqf file. Just open notepad and change the file name to unitSpawn.sqf. Obviously, place this file in your mission folder (right into the root) Paste the following into the .sqf file and save it: // Spawning a continuous stream of assaulters // by nkenny /* DESIGN Assign a vehicle as a spawning point. While the vehicle remains alive, spawn attackers. Order the attackers to move and attack the 'CENTER' marker. Wait until the attackers are all dead-- then repeat. ARGUMENTS 0, OBJECT -- Vehicle that is used as spawning object 1, SIDE -- Side of spawned unit (WEST, EAST, RESISTANCE) 2, COUNT -- Number of units in the spawned group */ // init _vehicle = param [0]; _side = param [1]; _count = param [2]; // the magic while {alive _vehicle} do { // create _grp = [getpos _vehicle,_side,_count] call bis_fnc_spawnGroup; // order _wp = _grp addWaypoint [getmarkerPos "center",100]; _wp setWaypointType "SAD"; // backup _wp2 = _grp addWaypoint [getMarkerPos "center",50]; _wp2 setWaypointType "GUARD"; // wait for it! waitUntil {{alive _x} count units _grp < 1}; sleep 30; }; 4. Now it is time to revist the vehicles you just placed. Copy the following into their Init (Initialization) fields when you double click on the vehicle. For the WEST aligned vehicles: 0 = [this,WEST,4] execVM "unitSpawn.sqf"; For the EAST aligned vehicles: 0 = [this,EAST,4] execVM "unitSpawn.sqf"; 5. Playtest There is always room for testing and improvement. One could add more variety (and control) to the units spawned as well as more powerful checks to prevent AI from popping into existence right in front of a player. That said. As an easy basis. It gets the job done. -k
  11. nkenny

    Pete's Texture Templates

    @IndeedPete This has been a fantastic resource for our community. Thanks a million! -k
  12. I think BWMods Pzf3 is carried in an offset manner. Is the question concerning creating a new launcher, or tweaking all existing ones? -k
  13. As much as I think a black P07 would be nice, I doubt Bohemia will add it to replace another reskin they added specially for this DLC. What exactly is the problem with the .45? -k
  14. @killzone_kid Dev branch 145895 17.07.19 Thought it would be added to stable.
  15. * Added: enableAI / disableAI "RADIOPROTOCOL" This does not seem to be the case in Arma stable. At least it is throwing an error.
  16. nkenny

    Arma 3 DLC - CONTACT

    @Glow Yes, indeed. However, I have yet to see an implementation of procedural textures which match even the most basic and simple texturing. Certainly powerful, and a great stop gap. Hardly a consistently applied design principle. -k
  17. nkenny

    Arma 3 DLC - CONTACT

    For that we have CUP, RHS, 3cB Factions, Veterans Mod, and any number of other expansive content additions. All I want are more vanilla assets available in solid greens. edit: Every single Olive Drab or Coyote brown comes with an exponential area of use. As ably demonstrated by Zukov above. -k
  18. nkenny

    Arma 3 DLC - CONTACT

    #thisIsWhatHappensWhenYouExpandYourCoreGameWithSensibleOliveDrabReskinsInsteadOfFactionLockedSpeculativeCamouflagePatterns Give us solid green vehicles! -k
  19. nkenny

    How important are Unit Websites these days?

    In nopryl.no we use our forums quite extensively. Indeed, it is due a major update in a few days. We're an older community, having existed since Operation Flashpoint days and of medium size-- around 20-30 attendants on our regular Tuesday sessions. Our forums are used to register and organise events, post mission briefings and after-action reports and both tactical and practical Arma guides. We like having the forum backlog. Discord comes with an impressive array of convenient features. But as an older community we like the tangible weight of having our own servers, rather than being committed fully to an ethereal cloud. Because most of our sessions are planned in advance, and our size and constitution dictates that pick-up games are not always happening, the main point of contact contintues to be our forums and Team Speak. For communities both smaller and larger-- or those organised more along pick-up play. I can easily see webpages and forums being of lesser importance. -k
  20. nkenny

    Arma 3 could have been BIS at its peak

    I suspect the reason why Bohemia are branching out to games which do not always turn into commercial powerhouses is because they are a creative group of people whom enjoy creating games. I'd rather have a passion for games than clinical ones. The sheer plethora of community made content-- which shows no signs of stopping-- should be as clear indication as any that Arma is not suffering a death spiral. Which is not to say that there are not lessons to be learned in transition from Arma3 to any new title. Arma3 is a splendid game in many ways, but still woefully underdeveloped-- or riddled with poor or inconsistent design decisions in others. Some of which have only, really, become visible after-the-fact. Precisely because Bohemia is balancing on an edge of games development for the love of games and profitability. -k
  21. nkenny

    AI Discussion (dev branch)

    @oukej Thank you for the detailed description. I have been experimenting with various values, but am left with a question. Does suppression level affect more than AI accuracy? It seems to me like it also pushes the AI to move (or avoid) a position where suppression is received. If that is the case: is this behaviour accessible somewhere? --- For people interested in suppression values, I have made a small steam workshop mod which alters minimum suppression time from 1 to 5 seconds and maximum from 10 to 30 seconds. There are some interesting effects, but the nature and complexity of fire fights in ARMA make it hard to quantify. The effect can be explored in any fire fight, but *shameless* here is a tiny mission with some debug information. If you are like me, and of the 'lets set up a firefight in the editor'-persuasion you can easily paste in the following code for some more (limited) feedback: Cheers, nkenny
  22. nkenny

    The effect of weapon on AI accuracy

    Intro So I have continued to explore the effect of optics.This led to two discoveries. I realised that 500 meters was a breaking point, config wise for weapon fire. Theoretically one should see a sharp distinction, between fire with and without optics. This was supported by looking at the Time outs or failed rounds. At 400 meters the Type 2 sniper optic came out, barely ahead. Figures In short I repeated the test according to all previous specifications, but at 500 meters. The test was done on Development version 1.95.145895. Each full cycle took about 30 minutes. - Full source here as usual. Conclusions & findings At 500 meters more nuance becomes visible. In the case of the MXM there is little difference in killing efficiency. Save that by pure numbers, (1) Ranged combat optics, like the HARM, actually decrease efficiency. Rate of fire decreases for very little increase in killing power. In the case of (2) sniper optics, there is also very little gain, though overall greater consistency is achieved. This is seen by fewer time outs, but overall more targets left alive on each cycle. A curious discovery is that at 500 meters ironsighted MXM shooters remained standing 8 out of 10 times. All other shooters preferred prone stance. The overall winner is ironsights. Higher overall rate of fire pays of in the specific environment created in the test scenario. Perhaps against moving targets, or targets with shorter periods of visibility numbers will be skewed more in favour of magnified optics. If my methods are correct, I can only conclude that the MXM is poorly configured in regards to optics. This may be generalised to affect other weapons. (While remembering that external ballistics of the fired round is of great importance). Secondary experiment I proposed that optics allowed weapons to engage targets at further ranges. Running full 30 minute tests was time prohibitive, so I ran much faster 10 cycle attempts. The test was set to 800 meters. This led to the following finding (RAW DATA): The raw data is fairly easy to read. Only the Sniper Optic equipped MXM performed with any consistency, actually scoring 16 hits. Second place is again Ironsights which scored 4 hits. Loser is RCO equipped MXM which failed to fire a single round. While this supports my general finding, that optics enable accuracy at longer ranges, the strong showing of the ironsighted MXM rifle suggests that some config tweaks are in order. -k
  23. http://www.rollingthunder.it/
  24. nkenny

    AI Discussion (dev branch)

    I've explored the effect here. I speculate that the fire modes made available by optics extend the range which the AI can attack. Within 400 meters the effect of optics is negligible. -k
×