Jump to content

nkenny

Member
  • Content Count

    1240
  • Joined

  • Last visited

  • Medals

Community Reputation

825 Excellent

About nkenny

  • Rank
    Master Gunnery Sergeant

core_pfieldgroups_3

  • Interests
    I live here, Really.

Profile Information

  • Gender
    Male
  • Location
    Oslo, Norway

Recent Profile Visitors

2386 profile views
  1. nkenny

    AI Facts & Myths Compilation List

    Emphasis and underlines my own. My answers are derived from extensive testing and working with the danger FSM for the LAMBS Danger.fsm mod. @Gunter Severloh 1. AI knowledge is not perfect and can be off by many hundred meters if the conditions are right! A single sniper shot, even if not suppressed, will rarely give the AI perfect knowledge of the shooters position. This can be tested by playing with LAMBS Danger.fsm and switching on the various debug modes. In vanilla it is easily demonstrated by looking at where reinforcing units with a GUARD waypoint choose to move. If the shooter makes noise or otherwise exposes himself, the AI's knowledge will quickly become more accurate. 2. That is an intriguing idea. There is per now no way to consistently direct the AI's attention--- aside from shooting, suppressing and flanking! Which TOTALLY works! 3. The answer is both! The relevant danger.fsm data is "Explosion" and "Bullet Close". The former reports position of impact, the latter the shooters position. There is a system of weighting within the FSM which decides which response is prioritised. It is not entirely clear by which process these individual events are triggered. Bullet close does seem to require some previous knowledge of the shooter, but it is not consistently the case that a single shooting event causes both and explosion and a "bullet close" (assuming the bullet whizzes past the listener). It gets more complicated. If a shooting event happens nearby-- another event is triggered "Fire" which reports the shooters location. Interestingly, the shooter need not be an enemy. Arma3 AI is pretty good. Whats lacking is some pathfinding, but more importantly, feedback of the AI's current state. -k
  2. nkenny

    How to Change Playable Character Avatar?

    It isn't an issue, it is a feature. The idea being your profile is your avatar. The equipment carried is secondary. It makes a lot of sense in multiplayer 😉 -k
  3. @TesACC As a leader of a community of that size, I think that your analysis of the problem is spot on, even as I don't agree with your conclusion. It is hard, if not impossible to justify adding GM and CSLA to our mod repository for the reasons you outlined. For all practical intents and purposes CUP and RHS offer adequate, comparable or superior assets. Assets which are inherently compatible with existing Arma3 universe. I think the underlying problem is this: Global Mobilisation is everything Arma3 should have been on release: specific in scope and having an inherently interesting scenario/setting. The cognitive dissonance arises because GM is fundamentally and thematically not compatible with Arma3. Example: If a 7.62 from a GM rifle performs radically different from vanilla, RHS or CUP weapons it is incompatible. If the vehicles added in GM do not fit out-of-the-box into the default A3 sandbox, they are incompatible. All of this translates into a confused message. Unfortunately, I don't think there is an easy solution. I will continue to buy and support Bohemia products, but I don't expect to see massive use of neither GM nor CSLA anytime in the future. -k
  4. To my knowledge there is no simple way to check individually-- beyond gradually adding/removing mods from the load order and manually checking performance. Performance is many things. Adding mods with a large number of assets will increase load time. Certain mods may have high quality or poorly optimised models (which means placing many of them on screen simultaneously is taxing for lower end systems). If you are using some sort of mission framework for your common scenarios, check these for bugs or script loops. Why don't you post a list of mods you are currently using?
  5. @sammael A quick fix is to do what redarmy suggests-- setting the group to <group> enableAttack false; Generally when I set up missions where I need the AI to remain static, I either use the garrison module or manually place AI with 'this disableAI "PATH";' set up. @redarmy It is funny you should mention it. EnableAttacks and the attack feature in itself is something we've been experimenting with quite a bit. I was considering making a video on the topic, but real life is real life. For the next release we remain committed to a seamless integration, but you will see enableAttack disabled at phased intervals or when specific tactical actions are made. This does increase cohesion quite a bit. The reason why not to simply disable it is because sometimes, sometimes it actually works really well. The attacking state is good at finding positions to engage from, and the sending out of troops is a nice way making the AI scout. It doesn't work so well for CQB-- hence, our tweaks. 🙂 @terox There are many ways to disable the AI. This ranges from the individual FSM and the group level "tactics". You can check our wiki (or my youtube channel) for more information, it is possible to do so by variable or checking an eden editor box or by Zeus interface. In short: <unit> setVariable ["lambs_danger_disableAI", true]; will disable the AI on an individual level. <group> setVariable ["lambs_danger_disableGroupAI", true]; will disable group level actions. Note that the individual actions trigger the group actions. Hence, if you wish to disable the AI for a spawned group-- do so on the individuals. -k
  6. @redarmy Artillery is due a rewrite. There have been been some reports of errors, though these have not been easy to debug. @Synchronized Ownership of vehicles is retained for some time. So if a unit leaves a vehicle, but that vehicle is still owned by that side. It will register as a low priority target for enemy forces. As Diwako states, adding a module to intelligently destroy potential enemy assets would be a massively invasive feature. Not to mention, it would demand a contextually intelligent AI. :) Update Progress is going well. 2.5 is very much in a playable state. It is based on an entirely new FSM and newfangled tactics engine. Provided we're happy with the performance, I hope to see a release before Christmas. -k
  7. Time to follow this thread a little closer again. First of all I would like to say thanks to everyone for the kind words. LAMBS Danger.fsm crossed the 50000 Steam Subscribers threshold a little while back. The team and I are happy and excited to see and hear about the games you play. We will be bringing out version 2.5 quite soonish. In fact, for those of that snoop our gitHub, will have seen quite a bit of activity recently. Version 2.5 will bring with it the much anticipated FSM rework and a considerable reorganisation and renaming of internal functions. The end result is a faster, more deadly AI; for developers, a streamlined and consistent environment. Two interesting features that will be coming with the next version are reworked group behaviours and a shared memory within the group. The former is already familiar. AI Groups will perform tactical manoeuvres, such as flanking or collective suppressive fire. With the next version, that feature has been updated, tightened and enhanced. The different tactical states are now more apparent and potent. You'll see group deliver area fire on known and expected locations and they will assault buildings with all the more rigor and aggression. Group actions are therefore much more integral to the AI's performance in a combat environment. Likewise the intelligence or evaluative process for group actions has been improved. There is also more consistency in AI using flares or manning static weapons. As to the former, a group now has a shared memory or consciousness of nearby enemies. When an AI soldier is in an assessment state, he will seek out enemies known to the group, but not directly threatening him. In other words, you will see the AI converge on known enemies or deliver sympathetic, prophylactic, suppressive fire when the enemy is not immediately within reach. The same system is used to make the AI more capable of clearing through buildings when enemies are suspected. Fun, dangerous, stuff. Thank you for playing :) Ken
  8. nkenny

    DUI - Squad Radar

    I speak from personal experience. The nametags are fantastic. :) -k
  9. Cool stuff! Literally. -k
  10. @mickeymen Have you checked the gitHub wiki? https://github.com/nk3nny/LambsDanger/wiki/Variables-and-functions ---- Here is the first of a three part series that covers the design philosophy of LAMBS Danger and how the module system works. Next video will consider Zeus and a more in depth look at the various modules. -k
  11. LAMBS Danger.fsm affects no weapon settings. With that said, the shared information may have other emergent properties. If anything the tendency of infantry to move and shoot in cqb may cause decreased accuracy. -k
  12. @LSValmont <unit> setVariable ["lambs_danger_disableAI", true]; -k
  13. LAMBS Danger fsm New release: Current version 2.4.4 *HOTFIX* Changelog Added context_menu actions support to the Zeus Enhanced mod Added configurable setting for ‘combat mode’ sharing distance Added patrol option to taskGarrison and taskCamp (default is disabled) Improved taskAssault (more reliable) Improved taskRush with dynamic cycle times Fixed random suppressive fire in taskCQB and improved house selection routines Fixed vehicles suppressing through terrain and high flying aircraft Fixed units will more reliable man static weapons in taskGarrison and taskCamp Fixed units will more reliably deploy static weapons from modded content Fixed units with LAMBS AI disabled will no longer be forced into combat mode by information sharing Fixed artillery registration issue Fixed in Zeus, moduleCQB and moduleAssault would always delete the logic even when asked not to. Fixed various issues in Eden and Zeus modules This version fixes and improves many small aspects of the mod related to task modules and group level behaviour. For players using the Zeus Enhanced (ZEN) mod, we’ve also connected the modules to the contextual menu for ease of use. The next move from the team will be an restructure of the naming scheme of many of the functions. As LAMBS Danger has proceeded the complexity of the project has increased. We are experimenting with future features. among these are: enhancements to artillery, a much requested reinforcement system, adding a post-combat AI, a dynamic close air support engine, and various other pieces. As always we are committed to keeping the mod from overriding mission makers intentions and hence creating a seamless improvement to the game. From the team diwako / joko / nkenny
  14. @mickeymen TaskGarrison is meant to be a long term garrison, not simply checking a building, or temporarily holding it. Because the waypoint is actually a handler for a scripted function and not a "way point" in the BIS sense, it should be treated as end state when activated through waypoints. There are technical and game design reasons for this. It is, however, possible to skip out of the garrisoned state. You can do that by using the taskReset module, or simply by running the taskReset function directly on the unit you wish to have abandon the garrison. That could be achieved within the trigger or statement you are using to have the garrisoned unit move again. The function to taskReset is [<group>] call lambs_wp_fnc_taskReset; Doing so comes with the limitation that this group will be fullly _reset_. Essentially the members of the group will join a freshly created group. -k
  15. @mickeymen 1. We allow the waypoint modules to be a little more invasive in terms of changing behaviours and combat modes. Best way to see which do what is to experiment a little. That said: When AI enter buildings, they are often limited to walking pace. This is a limitation of pathfinding. 2. In the same spirit as (1), the waypoint modules are not designed from the ground up to be easily linked into a chain of actions. These are better realised as "end states" or "fire and forget" modules. Easy ways in which to deploy the AI to a specific task. That doesn't mean it is impossible to string together different behaviours. The way to realise it is by using the taskReset module. Of course, if you are constructing a more advanced mission like that, I would advice creating your own scripted handler and calling the functions directly. @M1stb0rn As Joko says, only artillery units will be added to the artillery pool. Future versions will add CAS support 🙂 @Jimi Markkanen Not familiar with IFA3. By default it will only consider underbarrel grenade launchers. I suspect these weren't really around in WW2. hehe. Perhaps future version. -k
×