Jump to content
Bohemia Interactive Forums

royaltyinexile

Former Developer
 View Profile  See their activity

  • Content Count

    778

  • Joined

  • Last visited

  • Medals

Posts posted by royaltyinexile

  2. Posted

    The rain is back, same as the old one. Then why did you remove that in the first place? Let me guess, "It is under development but it's not a priority".

    There are certainly some strange attitudes to be found in the dev branch discussion. :)

    I added some further details back here:

    There has been some small improvements to near and distant rain textures and new sounds. Overall, they're fairly modest, as our guys didn't have much scope for bigger improvements without more significant underlying changes, which we wouldn't want to risk at this stage, but it's now in a better state for Launch.

    Best,

    RiE

  3. Posted · Edited by RoyaltyinExile

    will rain effects be making a return before the game releases?

    Yes, I'd expect it to be re-enabled on dev branch very soon.â„¢

    There has been some small improvements to near and distant rain textures and new sounds. Overall, they're fairly modest, as our guys didn't have much scope for bigger improvements without more significant underlying changes, which we wouldn't want to risk at this stage, but it's now in a better state for Launch.

    I dont think I have had a single crash yet. I dont remember ever having one.

    Glad to hear that :yay:

    Best,

    RiE

  4. Posted · Edited by RoyaltyinExile
    noise controlled

    June 7th, 2012. Before the Great Re-imaginingâ„¢ in October 2012. It is now an UnTruth.

    Come on, now, let's be very clear: both Joris and I worked on TKOH and we both really like the improved flight model that RTD integration offers. Joris alone made TKOH's time trials mode (and an entire free DLC) because of that! :)

    Unlike some other statements made at that time, I can still stand by what was stated. If there were a reasonable way to have had this extra feature in game on launch, it'd have be there.

    Making a wild promise based on unclear knowledge or experience was a mistake; rushing in a feature in to the detriment of the game and the feature itself would be a failure. That's the "great re-imagining" I'd hope you might recognise.

    Best,

    RiE

  5. Posted

    Will there be any beta patches or dev branch on release that will include those changes? Or we have to wait for example 1.01?

    The Dev Branch already contains fixes and changes, the point being made is that not all of these things are guaranteed to make it into 1.00.

    It's a selective process. Fixes in dev branch that we can merge with a good deal of confidence, or can afford the time to test properly, or simply have to be included because they're critical, will be included.

    Or, put another way, all fixes are equal, but some fixes are more equal than others. :cool:

    Best,

    RiE

  6. Posted

    Sounds like a winner!

    However the AI is still homing in on you like deadly terminators within a few seconds even if you just fire a couple of shots from concealed position (and they had their backs towards me at ~300 m range).

    I will see if I can create a repro showing it off later tonight. Looking forward to test above mentioned changes...

    /KC

    I'd be interested in seeing the Repro you come up with! I find the AI locating targets from sound to be a bit better now, we made some changes particularly with player movement that made the AI less able to immediately zero in on you.

    I updated the original AI CQB repro, and I find the behaviour quite nice. You could easily change it so the player only shoots (rather than reveals himself to the enemy and shoots).

    Best,

    RiE

  10. Posted · Edited by RoyaltyinExile

    Change log notes, v.09136

    I don't think the exe sneaked out into the dev branch wild yet. Soonâ„¢.

    :cool:

    AI prefer to stop and shoot, rather than running to cover, if enemy is close

    Surprise Target

    • Dealing with the situation where AI spots the player in close range while already en route to cover

    • AI is now more likely to stop and engage the threat

    • Decision is linked to distance to target (currently 35m)

  11. Posted · Edited by RoyaltyinExile

    rev 109110: AI prefer to stop and shoot, rather than running to cover, if enemy is close.

    Hm, I don't think this splendid work has made it out into the wilds of the dev branch yet, but it's progressing promisingly under Dr. Hladik's careful watch and QA's noble care.

    The focus of the work (in collaboration with Klamacz) is upon killing the issues identified when AI are suddenly confronted with threats in CQB; involving:

    • Surprise Target - Dealing with the AI spotting the player while it is already en route to cover; making the AI more likely to stop and engage; linking this decision to distance to target (will engage at close range, will prefer cover at medium-long range)
    • Rotation Speed- increasing speed AI is able to turn; pegging this increase to accuracy, so when AI turn quickly, accuracy is lowered

    Obviously, things need to be properly tested so we don't destroy current, desired behaviour, but it should be unleashed on dev branch if initial validation goes well.

    Best,

    RiE

  12. Posted · Edited by RoyaltyinExile

    Change log notes, v.09067

    AI leading targets accuracy improved

    • All AI which uses some sort of bullets (including tank shells and grenades) are now able to lead moving targets better.

    • This ability is still limited by weapon config (max lead speed) and skill of AI.

    • Typical testcases are: running perpendicular to AI, evading Irfit GMG firing from 500m, etc.

    AI leading rockets improved

    • Accuracy of AI shooting unguided rockets was improved

    • Now they should be able to lead moving targets whatever is the direction of movement

    • Splendid disclaimer: Please be careful though, RPG missile had some config problems last time I checked

    AI helicopter gunner aiming improved

    • There was a problem reported about AI gunner in Comanche or Kayman unable to hit static infantry (overshooting). This was fixed.

    Some other most splendid changes (related to AI behaviour in CQB) are in splendid progress, but I don't think the exe sneaked out into the dev branch wild yet. Soonâ„¢.

  13. Posted · Edited by RoyaltyinExile

    In the most recent video they mention the fact that it is in BIS interest to release Altis to the dev branch prior to launch yet with the final patch from what I can tell it only patched to the same as the dev branches previous patch [...] I am just curious why things are said in videos and in other situations only to be proven completely wrong by patching and release information

    This is the last default branch update; there are further updates (drones, Altis, AI) planned for the development branch, which are yet to be deployed.

    This was announced in the stream, so nothing is 'proven completely wrong', so far as I can tell. :cool:

    Best,

    RiE

  14. Posted · Edited by RoyaltyinExile

    Latest Work:

    Worked on AI recoil control, dependent on skill

    • AI are now able to consider and control the recoil of the weapon

    • This ability is dependent on the skill, meaning that less skilled soldiers will tend to control it less, where elite soldier (skill 1.0) should control it all the time

    • This behaviour is more evident at closer ranges, when AI is standing and his recoil makes most of the shots to go over target's head

    Hearing (AI): speed makes a bigger difference

    • Now the speed of movement and fatigue (breathing effect) makes you louder and more likely to be heard by AI.

    :cool:

  16. Posted · Edited by RoyaltyinExile

    the attitude of no repro no fix, does not really bother me, I’m not trying to sell my product, but it should really bother you, or BIS.

    While I'm not sure how - following a request by two devs to provide the repro mission, in a thread designed to discuss AI, including AI programmers - that you've reached the conclusion it doesn't bother us, but ok.

    I think the bit that you're missing is that AI is also a product of the data, not just of the behaviour. If the data is broken, it breaks the AI, but the programmers don't fix the data. The programmers only try to fix the AI and shout at other people to fix their broken data. They shout a lot louder if they've spent some time profiling AI behaviour only to find out it's because of a mis-configured asset. They go very silent otherwise ;)

    For example, the data things that affect the AI in your case are, for example, at very least least: The configuration of the weapon the player is firing; The configuration of the environment objects between the player and the AI; The time of day.

    When I tried to reproduce the behavior you've shown I got a different reaction.

    Repro 1

    Most often the AI team split up and sent two guys to investigate the shot, eventually killing me.

    JAC_R_1.jpg

    In a second repro, I move the position of the enemies back when the shot is triggered.

    Repro 2

    More often than not, the AI don't react, which could be a product of the changes made to AI behaviour or it could be due to badly configured weapon properties.

    JAC_R_2.jpg

    Of course I've probably set up the mission in a number of different ways to you, so we're not testing the same situation, really. However, when I provide the repro, we can then both see if we can reproduce the behaviour given the same set of preconditions.

    The point is, on my first repro, I didn't have a bug, so the programmers can't fix it; it's only when I change some conditions that I can see a potential problem, and that's how repros - not videos - work.

    It appears to me that this issue would either be a product of a recent 'fix' to AI audibility, or misconfigured values for how loud the weapon is. Both of which are easier to tweak, test, and balance against the original repro, which I can now shove under a programmer's nose.

    Best,

    RiE

  18. Posted · Edited by RoyaltyinExile

    quotes like this from devs making the game REALLY scare me. a repo of the issue is provided in a VIDEO !!!!!

    As other replies have already outlined, a video isn't a repro.

    Well I disagree about that, asking for a repro mission makes a lot of sense - after all, you want to fix the issue, not try to guess the prerequisites to get to the issue.
    there are 100s variables which may change in the game ... w/o proving your test mission it's highly unlikely to get 1:1 setup ...

    A repro is a verifiable set of steps based on certain preconditions, often with an observation or set of notes.

    This is a place for discussing the changes we're making to the AI, not for ejaculating views about how products are developed or how repros should be made.

    As I've said before, this thread stays clean or this thread gets ignored. Please stay on topic. :)

    Best,

    RiE

  19. Posted

    Lost the AI reaction to unseen gunfire and grenade explosions, close by. This seemed to have been rectified last time, but now its gone back to how it was, would be nice if a dev commented on the progress of basic AI reactions.

    If you can provide the repro mission or link to the tracker where the steps are reproducible, programmers are far more likely to do something about it. Basically: no repro, no fix. :)

    Oh, and as a side note, there haven't been too many AI updates this week as our programmers needed to address some problems that popped up in regards to UAVs.

    Best,

    RiE

  20. Posted · Edited by RoyaltyinExile

    The fact that you scrape something means that you have not a clear vision of what you are supposed to do.

    We've tried to make clear in the blog; the project was indeed going in a very different direction to being with; in many ways, it was quite experimental.

    A change was made, and Arma 3 ended up with new project leadership, alongside a new sandbox design lead, playable content lead, audio lead, two new programming leads, mastering lead, and marketing communications lead. Perhaps more importantly, the rest of the team doubled in size, adding many great new designers, programmers and more.

    The project pivoted towards what we would expect from an Arma sequel. It was during this period of transition that several unexpected external events also hit us, which, as we've in the past explained, made the situation much more difficult. But now the studios are unified is pursuing one singular vision of Arma 3 and the team's long, hard work has resulted in a consistently positive reception in the alpha and beta testing phases. No, not everything is perfect, no, not everything that everyone would like will make it into the launch, but, most certainly yes, the team is in a strong position to deliver Arma 3 now and support it in the future. I see team members busting their arses daily to get things right.

    We believe the Arma 3 launch is - set beside comparable products out there - a huge and rewarding package of content together with a legitimate set of engine advancements. Due to the Arma 3 Alpha and Beta, with your support, the game will be our most solid release, by far, and we're committed to supporting the platform long term. Next week, Joris will make further announcements and clarifications about our future content plans, which answer some of the outstanding questions.

    Best,

    RiE

  21. Posted · Edited by DnA
    Failure to quote

    Do you mean the complexity of the proposed new rain system is deemed to be too much work to tackle at the moment? How complex is this new rain system intended to be?

    Sorry, I should have made that more clear, there is no new rain system, just some analysis done in spare time that outlines the cool future potential of the feature area (I spent six years in Glasgow, I have to like rain).

    The actual work that is going into it now is about updating what we can, like sounds/textures, which we hope can help in some way.

    Plainly, I don't see the point of fatigue if all it does is prevent you from sprinting. The effect/penalty is just too small.

    To add some extra info to your statement, fatigue is also linked to weapon sway.

    This means that if you are more fatigued, you are likely to be less accurate. On top of that, when you reach an exhausted state (when fatigue is 100%), the speed of your actions, like movement, reloading, swimming, are limited (to 80%). There are currently some technical problems in terms of making the effect any more punitive than 80%; however, work continues in this area.

    In regard to the persisting capacity issue, I agree that it also needs some further work. While there are some limitations in terms of the original implementation, our new designer, Radko, hopes to make further improvements, particularly in terms of the capacity, which still seems unreasonably large.

    Best,

    RiE

  22. Posted

    They didn't like the old rain and want to do a full overhaul eventually, but due to time restraints they plan to put in a version with some minor improvements first.

    Yes, that's accurate. Rain is being worked on at the moment. We're trying to make some improvements from the data side, given the absence of program time.

    It looks like any improvements we can afford to make now will be rather modest, but I hope it'll be better than the state we first saw at the launch of the Alpha.

    My analysis for a full rain overhaul scares the programmers, and its implementation is rather lower priority than other features that need some love.

    Best,

    RiE

  23. Posted

    Well - it seems there are missing a few unfluences. For example the stance of the shooting person and if he is near cover/in a house (Obviously the current system handles someone shooting out of a 2cm big gap of a window the same as someone laying on plain field). Also the likeliness of detection should be dependable of the weapon optics of the searching AI.

    It's affected by two separate things. The reveal of position is the general area the shot is fired from. This means the AI will start looking for a threat in this location. This is what the fix affected; it made the area a bit less precise and it now also factors in range, too.

    After that, the unit's visibility (which is derived from a bunch of factors, including stance, weapon fired, the cover/clutter he's in) determines the chance/speed that the AI will actually 'find' and target the threat.

    Best,

    RiE

  24. Posted

    Another one that doesn't understand how games are developed. Jets = config guys + artists. Showcases = mission designers.

    Thank you for that. It pleases me endlessly to know that there is still some sheltered corner of the internet where reason and sanity hide. :)

    So [...] that other direction
    We've had to put in a lot of effort

    Best,

    RiE

×