Jump to content
Bohemia Interactive Forums

Fireball

Member
 View Profile  See their activity

  • Content Count

    341

  • Joined

  • Last visited

  • Medals

  • Medals

Posts posted by Fireball

  1. Posted · Edited by Fireball 

    _rate = sleepTime/averageRespawnTime;
while {(random 1) > _rate} do {sleep sleepTime;};

    And the smaller sleepTime is compared to averageRespawnTime, the less predictable it is, but the more it will be affected by possible script lag.

    Then, you can have averageRespawnTime as a parameter in the lobby.

    So sleepTime may be 300 and it would wait randomly 300 seconds until, e.g. with 1800 averageRespawnTime, (random 1) would drop below 0.16. Well this way it may happen that they respawn instantly or almost never.

    I want it to wait at least once sleepTime and then beginn with the random wait sequence; but alas it would only happen in 5 min steps - not so random and chances are high that vehicles respawn very seldom.

    Thanks for your input but I'll stick to my method.

    What I want to know: do the average times to respawn look reasonable to you?

  3. Posted

    Why randomize the respawn timer? This is for friendly vehicles, yes?

    Sorry, I was unclear, it's about the OPFOR vehicle patrols (fixed in my post). I want to avoid the Swiss Watch[tm] effect; "ahh, 10 min passed, another patrol coming in, get ready", hence the random component.

    ---------- Post added at 08:15 AM ---------- Previous post was at 08:01 AM ----------

    Is there a method to have the mission automatically set up to my requirements though? Or do I need to set it up the same way each time?

    This should be set in the server config, as someone outlined in an earlier post:

    http://forums.bistudio.com/showthread.php?p=1886769&highlight=server+config#post1886769

  4. Posted · Edited by Fireball

    First of all, sorry for not replying here for a long time, I didn't get anymore notifications of the thread, for some reason.

    is it possible to reduce enemy AI's RPG? I think it is too much overpowered the insurgent, kill the long gun firefight immersion, especially (so annoying) when they started shooting your base with those damn RPGs. but, this is still one of my fav MP missions. Thanks for your effort, Fireball :)

    That's an easy thing. Currently it has the the same AT infantry unit 3 times in the eastInfClasses list, I'll replace one of them with an AR unit and there we have only 2/3 of the previous AT unit amount. I also thought it was a bit too much as it is now.

    Then a question to everyone:

    I'm currently making the OPFOR AI vehicle patrols respawning; what dynamic respawn time would you imagine?

    Minimum respawn time is 300 secs and my thought was then an additional random of 900 secs by default.

    I think I'll offer a choice for random respawn additional time of 300, 900, 1200, 1800 seconds.

    Any other ideas?

  7. Posted

    You have deleted my Text in the CIT, could you please at least add the ideas and conversation here in this Topic?!

    There were some interesting statements by Suma, what AI still can't do and what would help improving other annoying AI-Issues aswell ;-)

    Here is your text:

    Well if the AI does not know what is providing cover in their environment, what was the sense of implementing a half-working suppression-function then?

    I mean, what is so hard to program for you, that when AI is receiving heavy fire, that they simply spread a bit out, get the "f-word" down on their bellies *immediately* and do search for some cover behind trees or stones or whatever, from which they then return fire (what they already CAN DO with the bounding-overwatch feature) ? (example given in open terrain).

    In Cities they simply should avoid to pointlessly run directly into the direction of the shooter(s) that suppresses them, instead run around the next house an return fire (if it makes sense) by leaning around a corner.

    Their current non-reaction to incoming fire is simply not acceptable. You advertised the product with that capabilities during pre-release press-events and whatnot.

    Please do something, even if something simple like i described it 2 sentences above.

    And here Suma's answer to you:

    In the context of this topic it looks like a rant. This topic asked that units stay in cover when suppressed. You ask the very opposite, that the units spread when fired at. Currently units are unable to distinguish if they are in cover (suppressed) or in the open (being fired at), therefore they are unable to decide effectively what to do next.

    All other chatter which I edited out was effectively redundant.

    ---------- Post added at 05:04 PM ---------- Previous post was at 04:46 PM ----------

    As far I can tell, to effect "suppression", the crucial issue is to identify suppression by AI. Just going down and move along the floor is not always the best idea, specially if - hopefully any time soon - the artificial damage reduction for prone units is removed.

    Maybe you could add up simple elements for a check if a unit is suppressed:

    - Behind cover (house, tree, wall)

    - Enemy contact

    - Fire coming from that enemy in a certain rate/intervall towards the AI unit

    Disadvantage with the simplistic method:

    - We assume that the AI is in full cover, so it stays put there; if this doesn't hold truth in many cases, we'll have upset players thinking the AI is stupid if it doesn't move away from the unprotected position

    This would be key for a successful suppression detection.

  8. Posted · Edited by Fireball

    The CIT ticket brought up an idea though, how AI general reaction to suppressive fire maybe could be improved "the cheap way", without more processing power than Danger mode currently uses. As you write, in Danger mode, AI currently _does_ make use of cover somewhat, e.g. trees, houses and walls. So, while the source of fire is known they could just find the next cover of any sort (let's say, within 50-100m) and hide of the opposite side of the known enemy contact/fire, thus the visibility needn't be checked expensively for terrain cover.

    So if a squad is surprised by enemy fire on semi-/open field, this could also be used to improve #6471 even more; have them sprint to the next cover of ANY sort, hide and have them *pinned* if suppressive fire continues into their direction.

    This might improve AI behaviour significantly and gives the impression of fear.

    ---------- Post added at 07:54 AM ---------- Previous post was at 07:47 AM ----------

    Comment by twisted:

    This "cheap way" seems pretty damn good. And it would help with making the AI react in a more human way to incoming fire (what volume does it take to be defined suppressing fire?).

    On the issue AI knowing which cover would provide perfect cover - I think that humans wouldn't know if cover is perfect or not. the instinct is just hide behind something in the opposite position from which the direction sounds of shooting are coming from. Certain kinds of cover would take priority over others instinctively as well as well as their proximity.

    is cover perfect? it is if no AI is getting shot. this does raise a bigger issue on AI battlefield awareness that I really hope Arma3 causes a revolution in. like noticing if the AI right next to them has just been shot or not - which would be a good indicator of whether cover really is safe or it's time to GTFO.

  9. Posted

    I am looking for some or one co-op mission.

    And this is what I am looking for:

    • Random missions.

    • Mission that adapt it self after how many players there is.
      • 1 player: The mission adapt it self to play with 1 player.
      • 2 player: The mission adapt it self to play with 2 player.
      • 3 player: The mission adapt it self to play with 3 player.
        and so on....

    Does anyone know any missions like this?

    Try Insurgency.

  10. Posted

    Sorry about mixing things up. If the truth be known I was pretty mixed after being hit by a 40-ton lorry in a traffic accident.

    Oh, sorry to hear that - hope you're basically alright!

    A better label would be most welcome. How about "Blufor/Opfor spawn ratio" ?

    Well, what was explained by BOTA:49 already exists as a separate setting by the suggested name, which decides how many OPFOR human players may join, depending on the current amount of BLUFOR players.

  11. Posted

    Also, in the parameters page IIRC there's an option called Enemy AI numbers with several values (50%, 100%, 150% and COOP). What does the COOP value mean?

    The other things have been correctly explained, but here you probably mixed two settings up in your head:

    "Enemy AI numbers" has no "COOP" setting, that would be OPFOR : BLUFOR ratio (and there are no percentages, but...ratios!).

    If you ask about "Enemy AI numbers" - this is the most misunderstood setting and I think I shall make an explanation of it, since every server admin seems to yank these numbers up to 200% at every occasion, probably because they were play testing the mission alone or with two players and thinking that with more players it would get boring with so "few enemies".

    BUT, in fact, the "Enemy AI number" scales with the amount of players. 100 % mean that for every player in enemy territory , 4 enemies spawn somewhere outside of each players view cone (if there's space). 150% means then 6 enemy AIs per BLUFOR player and 200%, 8 enemies.

    So let's do the math; in a larger town like Zargabad, with 30 players in the middle of it, with setting at 200%, it'll be packed with up to 240 enemy AIs crawling!

  12. Posted

    The reason I posted the truncated crash report is that the crash looked like a one-timer that would be very difficult to reproduce. Not worth it to send dumps to Dwarden.

    Post every dump, specially if you get both .mdmp and .bidmp files! Suma has been able to fix very naughty crashes by a single good dump, even if it wasn't reproducible for him.

    ---------- Post added at 01:08 AM ---------- Previous post was at 12:50 AM ----------

    Ah yeah, here are my findings to the current BETA.

    First the positive, it never happened to me that the game was running Warfare BE for 15h+ without memory leaks!! I started it in the morning at 8 am and when I came back in the evening at 7 pm, one of the AI parties fighting won the match. :D Then I restarted for another round of Warfare BE and it was still running by now.

    BUT the sound failed unfortunately, it just went all silent. I can't tell if it was after 11 h or, by now, 15 h of running the game, but it doesn't matter.

    In the morning it worked and as I restarted the game now, it started working again.

  15. Posted

    By any slim chance that this great MP mission can have AI fill-ins just like the A&S promode for those of us in the other part of the world that can't join too many servers due to high ping time??

    There's a slim chance, if you add a feature ticket on my project space. :rolleyes:

    But anyhow, I'll think about it - it shouldn't be a problem that you kind of "request reinforcements" up to your squad size as a TL. If somebody human joins in the same squad, one AI would be removed for them. They'd be joined as AI into your squad.

    I think Insurgency 0.70 had this already in, but it was subsequently removed - not sure why.

  16. Posted

    Hi Fireball

    Left4Dead Maximum number of players in MP is just 8.

    but thank you Fireball for an excelent example of why not to use Ragdoll and yet another example of why it don't work in MP.

    I've counted up many great examples before L4D, but you chose to ignore it - you only want to see what you want to see.

  19. Posted · Edited by Fireball

    Hello everyone!

    Finally a new version out, after a good month of break and taking my time in testing the Ambient Civillian modules (ALICE2/SILVIE) along with some ambient bomber script.

    Few notes to ALICE2/SILVIE and the suicide bombers:

    - Disabling Ambient Civilians in the mission parameters will also disable ambient bombers

    - The defusal kit for ambient/suicide bombers is your gun; you got 1 second from the "Allah akbar" outcry to kill the bomber

    - On host servers (i.e. player hosts the mission) it's advisable to disable Ambient Civilian unless your gear is "da shizzle"

    - Fallujah has incompatible buildings with ALICE2, so no civilians will be spawned at all, but it has city centres, thus you'll have some traffic at least

    - Mana has compatible buildings with ALICE2 but no city centres, thus neither will work

    Credits to:

    - Reezo's IEDdetect (only used ambient bomber script)

    - Wolffy.au Module Improvement Projecte

    Other key features:

    - 30 player slots

    - 6 HMMVs

    - PMC weapons

    - Unflip vehicles through radio

    - A-10 air strike through controllable AI

    Further features and fixes are seen in the Roadmap.

    Current version: 0.80

    - Fireball

    --------

    Project Details

    Mission available at

    http://dev-heaven.net/projects/insurgency/files

    For improvements and fixes, please see the roadmap at:

    http://dev-heaven.net/projects/insurgency/roadmap

    Please feel free to post your suggestions here or, better even, file a ticket on DevHeaven in the project issue tracker.

  21. Posted

    Hi PuFu

    Because it leads to Objects being in different places on each machine. So for instance A body you can hide behind as player A is not there for Player C it is perhaps the other side of a wall.

    That's just plain wrong - ragdolls have nothing to do with body position.

    First of all you must assume physics work for everyone roughly the same, so it's really only about the body position which might be synced and maybe a body will be "prodded" down somewhere from the network code, but you won't notice.

    I don't know why you just ignored my post, I've counted up so many old and new shooters, which work perfectly fine with Ragdolls (R6 Vegas 2, GRAW1+2) and forgot one of the latest few most prominent ones (still):

    - BFBC2

    - CoD: Black Ops

    So why do you claim it doesn't work fine, if everyone's using them? Stop talking about something you have no idea about.

  22. Posted · Edited by Fireball

    Hi all

    And still not one can point to a single MP game with more than a handfull of entities that uses Ragdoll physics.

    Not one.

    Did any of you wonder why?

    Kind Regards walker

    - Rainbow 6 Vegas (2)

    - Ghost Recon Advanced Warfare (GRAW) 1+2 (both PhysX and absolutely awesome!)

    Special case for cineasts:

    - Brink (it's absolutely gamey, but ragdolls are awesome!)

    None of above-mentioned MP FPS employing ragdoll physics have funny exagerated ragdolls, hell not even Brink which I call cineastic - it always adds and never takes.

    Please cut the whining and be glad about PhysX - specially when I remember GRAW1+2 - damn, that was awesome with the physics!

    EDIT: Your argument ragdolls have to be synced across MP is nil btw. - only the flying arc or approximate position where the body lies has to be the same for everyone; the rest doesn't need to be the same, e.g. the stance.

    A good demo how ragdolls in Vegas 2 are used:

    The player is sort of lame, but you can see nicely how the dead enemies lump around when they fall down.

    As you can see ragdolls have been used for years past in MP games with good results.

  23. Posted

    -In the simple first aid module, units will bleed out after time, soldiers can still stop bleeding, and medics can still 'heal' incap/agony. Consecutive shots will make bleeding faster.

    This is how it is in Insurgency right now, but scripted - not with modules. I'd be glad already if those FA/AFA modules would work fine in MP.

    The scripted version works out pretty good, but there is desync sometimes and sometimes you can't heal someone - actually some unintentional realism, which could be very well built in into such a system - sometimes someone is wounded fatally and you can't treat him on the field.

    If the advanced first aid system would be something along ACE2 simulates I'd be pretty fine with it - I find it already "complicated" enough for a game but still fun. I think (pseudo-)realism like you describe should not be in games, rather in VBS (where it's probably rather fully realistic) - but that's only my humbly opinion.

    I'm rather on the bandwagon for gameplay > realism in this regard.

  24. Posted

    I've prepared an AI flown user-controlled CAS air strike:

    http://dev-heaven.net/issues/18566

    There is also a video for those interested how it works then.

    I'll offer this with an option for the old CAS where the player hops in and also an option to disable it entirely.

    Additionally there will be an option to set flying time; AI-flown A-10 will always last double the time of the player-flown CAS, due to the fact that time starts counting for the AI-flown air strike as soon you've called it.

    Tell me what you think!

  25. Posted · Edited by Fireball

    Currently trying to incorporate your IEDdetector script into Insurgency, but I can't get to happen it frequently enough.

    I currently run the ambient bombers script with this param:

    nul0 = [] execVM "common\server\IEDdetect\IEDdetect_init.sqf";
/* 400 - is the ambient radius from the player
2 - is the minimum amount of civilians near the player
70 - is the % of presence of a triggerman among them
10 - is the scan interval time
"house" - is the object classtype of what could potentially host triggermen bombs (there is always a 50% chance it will be a suicide bomber)
0 - is the beeping (0=no beep, 1=beep, 2=random)
100 - is the area to scan for suitable objects (e.g. "car") and attach the bomb to one of those, randomly. This is also the scan area for the triggerman to trigger the bomb
30 - is the area to scan for hostiles that can instill fear in the triggerman (read above for how fear works).
50 - is the area to scan for possible victims that will make the triggerman touch the bomb off.
65 - is the % for the triggerman to be a suicide bomber instead
WEST - is the side to consider enemy of the triggerman */
nul0 = [150,2,70,10,"house",0,100,30,25,65,WEST] execVM "common\server\IEDdetect\IEDdetect_ambientBombers.sqf";

    Maybe "house" as a base object class is not optimal, but then again I didn't want to make it obvious that only cars explode.

    Any ideas as to why I almost never encounter those suicide bombers?

×