AI Discussion (dev branch)

[en3x 209]

25 minutes ago, The Man Without Qualities said:

i was laughing my ass off reading this! You are not talking about BIS quality assurance, right?

The first real bug tracker was set up and maintained by the community before BIS started this half maintained thingy we have right now?

 

I believe that Bohemias QA are hard at work, despite community tracker.

 

Evidence to this for example is latest update. Here is some fixes that they caught

 

Spoiler

• Fixed: The RscFrame, RscLine and RscCheckbox classes were not exported by BIS_fnc_exportGUIBaseClasses (https://forums.bistudio.com/topic/202717-bis_fnc_exportguibaseclasses-missing-classes)
• Fixed: The door of Taru's Transport Pod was not working correctly
• Fixed: It was possible to respawn before the countdown was finished in End Game scenarios
• Fixed: Formatting of texts in the DLC preview menu was incorrect (texts ended after the '&' symbol)
• Fixed: Corrected scenario endings for Seize Edoris and Seize Feres when time runs out
• Fixed: Script error when in the Respawn Screen and one of the loadouts / roles was removed
• Fixed: When watching an End Game scenario as a Spectator and the last phase was reached, the Spectator could see the Intel as a side task until the Schematics were picked up for the first time
• Fixed: The "FOB" text (used in Spectator's header) was not localized
• Fixed: Players could disappear from a Spectator's list of entities while dead in some cases
• Fixed: Spectators could see invisible entities in their map view in some cases
• Fixed: The FOB could be established even when the area was not secured in the End Game Kavala scenario
• Fixed: The Kuma and Mora could behave like submarines
• Fixed: Incorrect placement of two metal boxes in the vehicle obstacle course in the MP Bootcamp scenario
• Fixed: The Mk17 Holosight (black variant) had incorrect zeroing
• Fixed: NATO bipods would heat up in Thermal Imaging
• Fixed: NATO bipods had an incorrect material assigned
• Fixed: The RHIB boat would rise up too easily after spamming the A and D keys
• Fixed: The pilot proxy of the Xi'an was misaligned in exterior view
• Fixed: Download lines were connected to players in vehicles even though they could not download in End Game scenarios
• Fixed: Hidden Video Settings could still be adjusted
• Fixed: The Friendly Fire module was not working correctly after loading a save in SP when applied on a specific entity
• Fixed: BIS_fnc_inTrigger did not return the distance to the border when an area was passed as an array without a height specified
• Fixed: The BIS_fnc_inTrigger function returned an incorrect distance to the border when used with ellipse shapes
• Fixed: Incorrect shadows being cast on the UAV Terminal
• Fixed: Visual lighting on the straps of the UAV Backpack was incorrect
• Fixed: Error in the Respawn Screen when updating the position list
• Fixed: Faction Showcases were optimized
• Fixed: The MQ-12 Falcon had very low depression for its camera; now it should be the same as other UAVs (https://feedback.bistudio.com/T119472)
• Fixed: Bipods attached to the AK-12 would clip with hands of characters (https://feedback.bistudio.com/T120289)
• Fixed: When spectating an End Game match, Spectators would not be notified about instant Intel pickups, only download-able ones
• Fixed: If you mysteriously manage to exit your Time Trial vehicle, you will now be disqualified
• Fixed: Script error when the interface size or the resolution where changed from the Video Options without restarting the game
• Fixed: The AR-2 Darter could get some damage to its rotors when taking off while using the selectPlayer command
• Fixed: Character collisions in static weapons were incorrect
• Fixed: The Marshall APC in the Resurgent West scenario could get stuck in some cases
• Fixed: The Targeting Pod of the Y-32 Xi'an VTOL was not aligned properly (https://feedback.bistudio.com/T124063)
• Fixed: Secondary explosions on vehicles were triggered when the vehicle had its simulation disabled
• Fixed: The position of the Pawnee helicopter in Showcase NATO was incorrect
• Fixed: Titan AA missiles were losing track of their target after being launched in some cases
• Fixed: Some sensor properties used incorrect units
• Fixed: AT Titans dropped too much after soft-launches from vehicles
• Fixed: The roll indicator in the Neophron plane was incorrect (https://feedback.bistudio.com/T124603)
• Fixed: Exiting the game was sometimes triggering a save
• Fixed: Targets were incorrectly warned about Xi'an locking
• Fixed: The Feedback link in the Pause Menu was not working correctly
• Fixed: Targets were incorrectly warned about DAGR and Scalpel locking
• Fixed: Incorrectly deducted number of CM for the Blackfish
• Fixed: The spectator camera would warp when following fast moving vehicles
• Fixed: Error after removing respawn inventories during a scenario after respawn
• Fixed: Correct "Countermeasures" hint for Showcase Helicopters
• Fixed: Using the Respawn Module could result in a collision of the new vehicle with an old wreck
• Fixed: Inconsistent targeting camera sensitivities utilizing the new maxMouseX/YRotSpeed properties (https://forums.bistudio.com/forums/topic/200467-jets-sensor-overhaul-radars-irs-lazors-pgms/?do=findComment&comment=3186688)
• Fixed: When the respawn position was not set to 'Where it was destroyed', players could set an incorrect respawn mode in the Respawn Module
• Fixed: AA Titan would miss targets too often

 

I fail to see what is so laughable about that. If community set first bug tracker doesn't change that.

[road runner 4344]

Haters are always gonna hate!!

[inlesco 233]

What if... someone simply posted a list the most relevant (in their opinion or whatever) AI problems at the moment with links to the feedback tracker or any useful Youtube videos for reference and the developers, in that case, could effectively reply to such a constructive and well-written post in a straightforward manner, addressing each problem one by one (the list should be a numerical list, of course, for easier backtracking)?

Of course, the devs may choose not to respond to the issues listed, but at least we'd be able to have an elaborate list of existing AI problems that any community member may suggest edits for and contribute accordingly, so it's easy to track problems for both the community and devs in the long term.

Unfortunately, I can't really dedicate much time for Arma lately, but I'm hopeful that someone will give in 3-5 hours to craft such a list sometime, post it here and have a meaningful discussion in which both parties can participate equally. 

Thoughts?

P.S. In fact, I could make a private or public Git repo for A3 AI issue tracking, so people can watch or star it, make pull requests for updates, etc., so everyone can contribute.

[Antilochos 106]

1 hour ago, inlesco said:

What if... someone simply posted a list the most relevant (in their opinion or whatever) AI problems at the moment with links to the feedback tracker or any useful Youtube videos for reference and the developers, in that case, could effectively reply to such a constructive and well-written post in a straightforward manner, addressing each problem one by one (the list should be a numerical list, of course, for easier backtracking)?

Of course, the devs may choose not to respond to the issues listed, but at least we'd be able to have an elaborate list of existing AI problems that any community member may suggest edits for and contribute accordingly, so it's easy to track problems for both the community and devs in the long term.

Unfortunately, I can't really dedicate much time for Arma lately, but I'm hopeful that someone will give in 3-5 hours to craft such a list sometime, post it here and have a meaningful discussion in which both parties can participate equally. 

Thoughts?

P.S. In fact, I could make a private or public Git repo for A3 AI issue tracking, so people can watch or star it, make pull requests for updates, etc., so everyone can contribute.

Something BI could do themselves to, right? I mean, that is what I do with my work; keep up a list what I have to finish...

[otrebla_snake_ita 2]

2 hours ago, inlesco said:

What if... someone simply posted a list the most relevant (in their opinion or whatever) AI problems at the moment with links to the feedback tracker or any useful Youtube videos for reference and the developers, in that case, could effectively reply to such a constructive and well-written post in a straightforward manner, addressing each problem one by one (the list should be a numerical list, of course, for easier backtracking)?

Of course, the devs may choose not to respond to the issues listed, but at least we'd be able to have an elaborate list of existing AI problems that any community member may suggest edits for and contribute accordingly, so it's easy to track problems for both the community and devs in the long term.

Unfortunately, I can't really dedicate much time for Arma lately, but I'm hopeful that someone will give in 3-5 hours to craft such a list sometime, post it here and have a meaningful discussion in which both parties can participate equally. 

Thoughts?

P.S. In fact, I could make a private or public Git repo for A3 AI issue tracking, so people can watch or star it, make pull requests for updates, etc., so everyone can contribute.

 

Maybe it could be cool, but many people already took a lot of time to report bugs to the tracker (and maybe searching for the repro steps). That's why I think BIS should do something like this by themselves. I mean, I like them very much, but that's their job, we can help them by reporting bugs and problems, but they should find a way to collect and solve them in an efficient way since they are the developers!

 

Honestly I fear there are some structural problems that can be solved only with a huge work on the game engine itself (aka "Arma 4"): I see the AI changes happen very slowly and problems are never solved 100%...Or maybe they just need more people working on AI!
I don't know and of course I hope to be proved wrong. 

[The Man Without Qualities 110]

23 hours ago, en3x said:

 

I believe that Bohemias QA are hard at work, despite community tracker.

 

Evidence to this for example is latest update. Here is some fixes that they caught

 

  Reveal hidden contents

• Fixed: The RscFrame, RscLine and RscCheckbox classes were not exported by BIS_fnc_exportGUIBaseClasses (https://forums.bistudio.com/topic/202717-bis_fnc_exportguibaseclasses-missing-classes)
• Fixed: The door of Taru's Transport Pod was not working correctly
• Fixed: It was possible to respawn before the countdown was finished in End Game scenarios
• Fixed: Formatting of texts in the DLC preview menu was incorrect (texts ended after the '&' symbol)
• Fixed: Corrected scenario endings for Seize Edoris and Seize Feres when time runs out
• Fixed: Script error when in the Respawn Screen and one of the loadouts / roles was removed
• Fixed: When watching an End Game scenario as a Spectator and the last phase was reached, the Spectator could see the Intel as a side task until the Schematics were picked up for the first time
• Fixed: The "FOB" text (used in Spectator's header) was not localized
• Fixed: Players could disappear from a Spectator's list of entities while dead in some cases
• Fixed: Spectators could see invisible entities in their map view in some cases
• Fixed: The FOB could be established even when the area was not secured in the End Game Kavala scenario
• Fixed: The Kuma and Mora could behave like submarines
• Fixed: Incorrect placement of two metal boxes in the vehicle obstacle course in the MP Bootcamp scenario
• Fixed: The Mk17 Holosight (black variant) had incorrect zeroing
• Fixed: NATO bipods would heat up in Thermal Imaging
• Fixed: NATO bipods had an incorrect material assigned
• Fixed: The RHIB boat would rise up too easily after spamming the A and D keys
• Fixed: The pilot proxy of the Xi'an was misaligned in exterior view
• Fixed: Download lines were connected to players in vehicles even though they could not download in End Game scenarios
• Fixed: Hidden Video Settings could still be adjusted
• Fixed: The Friendly Fire module was not working correctly after loading a save in SP when applied on a specific entity
• Fixed: BIS_fnc_inTrigger did not return the distance to the border when an area was passed as an array without a height specified
• Fixed: The BIS_fnc_inTrigger function returned an incorrect distance to the border when used with ellipse shapes
• Fixed: Incorrect shadows being cast on the UAV Terminal
• Fixed: Visual lighting on the straps of the UAV Backpack was incorrect
• Fixed: Error in the Respawn Screen when updating the position list
• Fixed: Faction Showcases were optimized
• Fixed: The MQ-12 Falcon had very low depression for its camera; now it should be the same as other UAVs (https://feedback.bistudio.com/T119472)
• Fixed: Bipods attached to the AK-12 would clip with hands of characters (https://feedback.bistudio.com/T120289)
• Fixed: When spectating an End Game match, Spectators would not be notified about instant Intel pickups, only download-able ones
• Fixed: If you mysteriously manage to exit your Time Trial vehicle, you will now be disqualified
• Fixed: Script error when the interface size or the resolution where changed from the Video Options without restarting the game
• Fixed: The AR-2 Darter could get some damage to its rotors when taking off while using the selectPlayer command
• Fixed: Character collisions in static weapons were incorrect
• Fixed: The Marshall APC in the Resurgent West scenario could get stuck in some cases
• Fixed: The Targeting Pod of the Y-32 Xi'an VTOL was not aligned properly (https://feedback.bistudio.com/T124063)
• Fixed: Secondary explosions on vehicles were triggered when the vehicle had its simulation disabled
• Fixed: The position of the Pawnee helicopter in Showcase NATO was incorrect
• Fixed: Titan AA missiles were losing track of their target after being launched in some cases
• Fixed: Some sensor properties used incorrect units
• Fixed: AT Titans dropped too much after soft-launches from vehicles
• Fixed: The roll indicator in the Neophron plane was incorrect (https://feedback.bistudio.com/T124603)
• Fixed: Exiting the game was sometimes triggering a save
• Fixed: Targets were incorrectly warned about Xi'an locking
• Fixed: The Feedback link in the Pause Menu was not working correctly
• Fixed: Targets were incorrectly warned about DAGR and Scalpel locking
• Fixed: Incorrectly deducted number of CM for the Blackfish
• Fixed: The spectator camera would warp when following fast moving vehicles
• Fixed: Error after removing respawn inventories during a scenario after respawn
• Fixed: Correct "Countermeasures" hint for Showcase Helicopters
• Fixed: Using the Respawn Module could result in a collision of the new vehicle with an old wreck
• Fixed: Inconsistent targeting camera sensitivities utilizing the new maxMouseX/YRotSpeed properties (https://forums.bistudio.com/forums/topic/200467-jets-sensor-overhaul-radars-irs-lazors-pgms/?do=findComment&comment=3186688)
• Fixed: When the respawn position was not set to 'Where it was destroyed', players could set an incorrect respawn mode in the Respawn Module
• Fixed: AA Titan would miss targets too often

 

I fail to see what is so laughable about that. If community set first bug tracker doesn't change that.

You BELIEVE a lot but KNOW nothing.

I KNOW that BIS Q&A judging by the results / release quality / speed of fault fixing is far from being professional.

 

And when I read the bug fixing list you refer to...shame-shame-shame.

Tons of careless mistakes by BIS devs - as usual. Clever that the same guys are celebrated as heroes for fixing their own faults - must apply that to my own profession.

 

Quote

NATO bipods would heat up in Thermal Imaging

What is really needs is someone going over the entire material (regardless if static or mobile) and check among others the thermal behavior.

The outcome would be an endless change log.

 

en3x, there are tons of tickets out there dealing with utterly wrong configured map objects resulting into tiny bushes making a tank flip over or tiny boxes stopping an MBT from full speed to zero without moving an inch.

Careless coding - no QA or they simply don't care.

You meanwhile must remember which bush you can crush and which you must avoid - something an AI can't - the best AI routine must fail if it is feed with bullshit parameter.

[en3x 209]

During my years of research and youtube videos containing knowledge, one of them being Star citizen (wealth of knowledge there, lots of programmer interviews and good questions)

I learned that what sometimes seem like simple bug, may be actually really difficult to fix in reality. In another words what seem in our eyes non professional bug fixing may be

giant undertaking that requires lots of time and lots of manpower. Another thing to keep in mind is that no game can truly fix every bug possible. Mainly because developers have

2 choices - will they do fixes but not implement new features or will they do some bug fixing and implement/code in new game systems. The first choice doesn't make sense because

you can't just keep fixing game without adding new features (not commercially viable). Another thing with game development is that code changes and new bugs come up wheter

they fix bugs or change code. So bugs never stops coming.

 

real virtuality 4 engine is incredibly sophisticated but also shows age in some areas. Engine is coded using different sets of systems that are interconnected between each other,

I like to picture is as giant web.

[Antilochos 106]

13 minutes ago, en3x said:

During my years of research and youtube videos containing knowledge, one of them being Star citizen (wealth of knowledge there, lots of programmer interviews and good questions)

I learned that what sometimes seem like simple bug, may be actually really difficult to fix in reality. In another words what seem in our eyes non professional bug fixing may be

giant undertaking that requires lots of time and lots of manpower. Another thing to keep in mind is that no game can truly fix every bug possible. Mainly because developers have

2 choices - will they do fixes but not implement new features or will they do some bug fixing and implement/code in new game systems. The first choice doesn't make sense because

you can't just keep fixing game without adding new features (not commercially viable). Another thing with game development is that code changes and new bugs come up wheter

they fix bugs or change code. So bugs never stops coming.

 

real virtuality 4 engine is incredibly sophisticated but also shows age in some areas. Engine is coded using different sets of systems that are interconnected between each other,

I like to picture is as giant web.

Then like I suggested before; why does BI not tell us how it is? If it's to complicated cq being a giant web to fix it, just tell us that. We aren't complete idiots and we might even be reasonable.

 

Trust me, even if they tell us to fuck off with the problems we have with the game, that will be better than not saying anything at all, leaving the players guessing if it will be fixed or not (and frustrated that they're kept in the unknown for long).

[The Man Without Qualities 110]

2 hours ago, en3x said:

During my years of research and youtube videos containing knowledge, one of them being Star citizen (wealth of knowledge there, lots of programmer interviews and good questions)

I learned that what sometimes seem like simple bug, may be actually really difficult to fix in reality. In another words what seem in our eyes non professional bug fixing may be

giant undertaking that requires lots of time and lots of manpower. Another thing to keep in mind is that no game can truly fix every bug possible. Mainly because developers have

2 choices - will they do fixes but not implement new features or will they do some bug fixing and implement/code in new game systems. The first choice doesn't make sense because

you can't just keep fixing game without adding new features (not commercially viable). Another thing with game development is that code changes and new bugs come up wheter

they fix bugs or change code. So bugs never stops coming.

 

real virtuality 4 engine is incredibly sophisticated but also shows age in some areas. Engine is coded using different sets of systems that are interconnected between each other,

I like to picture is as giant web.

Thank you, haven't had so much fun since weeks.

Years of research - youtube videos...okeeeey.

You are right, some tiny changes to the user might require in the end dumping parts or the entire engine or the game design or whatever.

And it might take time.

 

Man, we talking about some bugs which are purely config! (undestructible bushes).

Pathfinding in 3D, especially several layers might require a total different concept.

But 15+ years are gone, and if you never start - changes never come.

 

BTT: Within the boundaries of current engine as is it would be interesting if BIS could now replace the bloody HCs by a more advanced server side software which is able to utilize (wet dream!) 192GB RAM, 8 cores of CPU and maybe things like NVIDIA TESLA if it would be from any use for AI calcs at server side.

If even player side AI would be moved to server which is capable and far more predictable in performance, I assume the overall MP performance involving AI should be much better.

Nuff resources for AI would at least prevent the big difference in capabilities depending on client side performance.

 

[Vasily.B 529]

16 hours ago, The Man Without Qualities said:

Thank you, haven't had so much fun since weeks.

Years of research - youtube videos...okeeeey.

You are right, some tiny changes to the user might require in the end dumping parts or the entire engine or the game design or whatever.

And it might take time.

 

Man, we talking about some bugs which are purely config! (undestructible bushes).

Pathfinding in 3D, especially several layers might require a total different concept.

But 15+ years are gone, and if you never start - changes never come.

 

BTT: Within the boundaries of current engine as is it would be interesting if BIS could now replace the bloody HCs by a more advanced server side software which is able to utilize (wet dream!) 192GB RAM, 8 cores of CPU and maybe things like NVIDIA TESLA if it would be from any use for AI calcs at server side.

If even player side AI would be moved to server which is capable and far more predictable in performance, I assume the overall MP performance involving AI should be much better.

Nuff resources for AI would at least prevent the big difference in capabilities depending on client side performance.

 

THIS + as i readed :
 

Quote

NATO bipods would heat up in Thermal Imaging

I couldnt stop laughing. Why they didnt fixed tank barrel by the way? Why its taking heat while using coaxial mg? Why i reported it 1,5 year ago and its not fixed? Pure config case, but yeah, call me hater.

[road runner 4344]

51 minutes ago, Vasily.B said:

Pure config case, but yeah, call me hater.

Seeing as you seem to know how to fix it, why not release a "hotfix" yourself?

[Vasily.B 529]

2 hours ago, road runner said:

Seeing as you seem to know how to fix it, why not release a "hotfix" yourself?

Why not to fix whole game for developers?

[R0adki11 3949]

24 minutes ago, Vasily.B said:

Why not to fix whole game for developers?

Perhaps a better idea would to think before your post, as at present your in familiar territory of posting flame-baiting comments. You of all people should understand the forum rules by now.

 

And with that lets turn to the topic in hand which is about the AI.

[The Man Without Qualities 110]

3 hours ago, road runner said:

Seeing as you seem to know how to fix it, why not release a "hotfix" yourself?

 

Sure, it is normal that the customer is fixing the product.

Ever heard in a garage "if you know that the engine is broken why you don't fix it yourself?"

[joostsidy 685]

One more post that is NOT a discussion about AI, and you will leave me no choice but to.... - threatening silence -....... unfollow this thread... :.-( ;-)

[pr9inichek 133]

@razazel

On 04.01.2017 at 1:17 PM, razazel said:

Having trouble with all airplanes on all airfields or only some or some combinations?

Yeah. Sometimes...

 

In Altis we have 6 airfields:

1) airlfield;

2) Altis Intl. Airport;

3) AAC Airfield;

4) Molos Airlfield;

5) Almyra;

6) Feres Airfield.

 

Player can take-off & take-on from most of them.

As examples:

airfield

 

 

 

AI don't know about this field and Feres Airfield, in my opinion u must to teach their.

 

A-143 Buzzard take-off like drunk pilot:

 

 

Carrier:

 

 

Bonus:

 

 

 

[road runner 4344]

7 hours ago, The Man Without Qualities said:

 

Sure, it is normal that the customer is fixing the product.

Ever heard in a garage "if you know that the engine is broken why you don't fix it yourself?"

And you've obviously never heard that to make your car look better, or to increase it's performance, you add these alterations yourself.

I'm willing to bet you are not a simple vanilla game user, and use plenty of addons......to make the vanilla game more appealing, which makes your post moot, and slightly  hypocritical

The AI issue has been thrashed out add naseuim, there's countless AI enhancement addons that make these little motards somewhat more controllable, I've also brought up several AI issues, but unless BIS employ a specific AI coder, it's just never going to get better, and at present I don't believe they have such a person.

Vasily.B made a comment that it's a simple config fix, if he actually believes this, he would be revered amongst the forum members as being the bloke who finally  fixed the AI. by producing this fix.....but he's not, he's known for the exact opposite, he has many valid points as already mentioned by me, but his manner, and his hyperbole statements ..are how he's known on the forums.

 

You don't make a claim that the AI can be fixed with a simple config change, and don't back these claims up, by producing that fix.... it's hyperbole at its finest.

As I've often said in these discussions, AI that climb ladders, and don't walk off roof's after a helo insertion is what I'd personally love, but till that happens, then I try to avoid placing AI in such places and predicaments.

The AI is borked, but it doesn't make this a game stopper, and if its a game stopper for you personally, maybe ARMA 3 isn't for you? Plenty of other people have used whatever workaround suits their gaming need/style.

[The Man Without Qualities 110]

 

 

3 hours ago, road runner said:

You don't make a claim that the AI can be fixed with a simple config change, and don't back these claims up, by producing that fix.... it's hyperbole at its finest.

 

On 28.5.2017 at 11:39 PM, The Man Without Qualities said:

Man, we talking about some bugs which are purely config! (undestructible bushes).

Yepp, like I wrote, some AI misbehavior is not caused by AI logic itself, it is caused by wrong input. If AI is considering bushes as "no obstacle" but it is like a hedgehog, Ai must fail.

So it is not an issue of the AI routine dev, it is an issue of the map maker and object maker.

If that is hyperbole then - Yes, I am!

 

3 hours ago, road runner said:

The AI issue has been thrashed out add naseuim, there's countless AI enhancement addons that make these little motards somewhat more controllable, I've also brought up several AI issues, but unless BIS employ a specific AI coder, it's just never going to get better, and at present I don't believe they have such a person.

 

BIS spend a lot of time into AI, I am sure they had and have skilled staff there. The problem is that they do not finish and polish and constantly maintain. Usually the last 5% of effort making 95% of the overall impression :-)

If you see the example when AI routines were implemented where AI search obstacles for cover when in battle. Unfortunately they sneaked to close to objects when firing ATs or AAs, resulting into tons of suicide cus AI fired AT right into the obstacle they used for cover. So all the work was wasted for the moment because it was not properly tested and it went public like that. It is meanwhile fixed.

 

3 hours ago, road runner said:

And you've obviously never heard that to make your car look better, or to increase it's performance, you add these alterations yourself.

 

No, I prefer to leave that to guys which should have the knowledge. I also drink wine without knowing exactly how to make it. And I complain instead of doing sugar or whatever into it if I do not like.

 

3 hours ago, road runner said:

As I've often said in these discussions, AI that climb ladders, and don't walk off roof's after a helo insertion is what I'd personally love, but till that happens, then I try to avoid placing AI in such places and predicaments.

The AI is borked, but it doesn't make this a game stopper, and if its a game stopper for you personally, maybe ARMA 3 isn't for you? Plenty of other people have used whatever workaround suits their gaming need/style.

That is personal preference and tolerance. Good for you that you can deal with it, I cannot, maybe others too.

Please accept that some individuals stick to the method that one should only add things which work.

When I play CTIs, I am 20-60% of the time busy to look after my AI driven vehicles if they stuck somewhere, if they lay flipped over, cracked their wheels...

If so, I have to drive there, flip back/repair/jump to driver position to clear situation.

Fine for you, not for me.

But that is part of an more specific AI related thread already.

[fn_Quiksilver 1636]

1 hour ago, The Man Without Qualities said:

 

 

BIS spend a lot of time into AI, I am sure they had and have skilled staff there.

 

When was last time BI has employed a specialist AI programmer (someone who has gone to school specifically for AI programming/Game AI programming)?

 

AFAIK I have only seen oukej and klamacz working on AI, and while they have done some great work, I don't think either has specialized career in AI programming or Game AI programming.

 

as to the driving 'overhaul' a year ago ...

 

what we thought we were getting

 

 

 

 

 

 

what we actually got

 

 

 

 

 

 

The one caveat is the new "setDriveOnPath" command is very powerful, unfortunately it requires a lot of SQF scripting to be useful in anything beyond highly scripted cutscenes, and is still subject to vehicles inability to reverse.

[teabagginpeople 398]

11 minutes ago, fn_Quiksilver said:

 

When was last time BI has employed a specialist AI programmer (someone who has gone to school specifically for AI programming/Game AI programming)?

 

AFAIK I have only seen oukej and klamacz working on AI, and while they have done some great work, I don't think either has specialized career in AI programming or Game AI programming.

 quicksilver trying to lose people jobs tha fcukkk. haha

 

what makes you think they're not skilled or have not learned about the engine in what they're doing with AI.with regards AI and their limitations within armas sand box nature and it's multitude of working cogs. it makes it a big thing to tackle. I mean they get tanks to do as they wish. an unbreakable wall pops up in the mix flips the tanks. wait why did it flip the tank? the tank pysx are not happening? I have seen players get stuck on terrain.

 

my point is ,it's a damn rabbit hole. for the most part I see AI doing the stuff they are supposed to. and even sometimes stuff you do a double take and think wtf is that a player? then like some pointed out they get caught in the cogs.

 

 

 

 

[fn_Quiksilver 1636]

14 minutes ago, teabagginpeople said:

 quicksilver trying to lose people jobs tha fcukkk. haha

 

what makes you think they're not skilled or have not learned about the engine in what they're doing with AI.with regards AI and their limitations within armas sand box nature and it's multitude of working cogs. it makes it a big thing to tackle. I mean they get tanks to do as they wish. an unbreakable wall pops up in the mix flips the tanks. wait why did it flip the tank? the tank pysx are not happening? I have seen players get stuck on terrain.

 

my point is ,it's a damn rabbit hole. for the most part I see AI doing the stuff they are supposed to. and even sometimes stuff you do a double take and think wtf is that a player? then like some pointed out they get caught in the cogs.

 

 

 

 

 

we are talking about history now ;) A3 main development cycle is over, there is a small stay-behind team to work on some DLC, but most resources divested to other project(s) I am guessing.

 

My gripe is that AI is a huge part of ArmA, but was grossly under-resourced during main A3 development arc.

 

If no noise or constructive criticism is leveled at BI for this (IMO) mistake, why should there be any changes for their next title?

 

I think my criticisms are fair, and I both criticize and compliment BI where I think due. 

 

Two main criticisms:

 

- Under-resourced AI development for entire ArmA 3 development arc. 

- Broken and wholly unreliable AI wheeled-vehicle driving for 1 year, worse than the merely unreliable prior state.

 

Some compliments:

 

- Great scripting command additions like "forgetTarget" and "targets" and "doSuppressiveFire" and "setDriveOnPath" (forceFollowRoad is great in theory but broken in practice)

- AI forward-firing helicopter AI fixed!!!

- Fixing wing AI is good since Jets DLC!

- AI infantry will fire at fast(er) moving helicopters now!!!

 

 

Also this has nothing to do with any particular employees. oukej and klamacz IMO did a great job with the resources and time they had, but AI was not a priority for management, and so they were often under-resource and/or pulled away to work on other projects. For instance oukej pulled to work on Sensors and Jets DLC, and klamacz pulled to work on Tanks DLC / some physics simulation stuff. Not saying those are bad projects, I love the sensors stuff and Jets DLC, but it once again shows AI shuffled to the bottom of the priorities.

[teabagginpeople 398]

2 minutes ago, fn_Quiksilver said:

 

we are talking about history now ;) A3 main development cycle is over, there is a small stay-behind team to work on some DLC, but most resources divested to other project(s) I am guessing.

 

My gripe is that AI is a huge part of ArmA, but was grossly under-resourced during main A3 development arc.

 

If no noise or constructive criticism is leveled at BI for this (IMO) mistake, why should there be any changes for their next title?

 

I think my criticisms are fair, and I both criticize and compliment BI where I think due. 

 

Two main criticisms:

 

- Under-resourced AI development for entire ArmA 3 development arc

- Broken and wholly unreliable AI wheeled-vehicle driving for 1 year, worse than the merely unreliable prior state.

 

Some compliments:

 

- Great scripting command additions like "forgetTarget" and "targets" and "doSuppressiveFire"

- AI forward-firing helicopter AI fixed!!!

- Fixing wing AI is good since Jets DLC!

- AI infantry will fire at fast(er) moving helicopters now!!!

I get your points quicksilver. I would say the AI moves along with most things. seems snails pace at times but it gets its turn :D

 

a lot of the lighting has been jacked for coming up on one year now shadows at night.  

the annoying popping of bushes on altis and some clutter.

 

just two examples to show you it's not just AI that can get overlooked in long term. and these are not attacks. I like to be realistic in my approach. for the most part so far the AI runs. now pumping a lot of money into AI is not going to keep the lights on if you understand me. it would be wise to do it alongside money makers. example jets dlc and the fix you mentioned. 

 

is there not a  Dr who also works on AI?

 

 

[fn_Quiksilver 1636]

1 minute ago, teabagginpeople said:

I get your points quicksilver. I would say the AI moves along with most things. seems snails pace at times but it gets its turn :D

 

a lot of the lighting has been jacked for coming up on one year now shadows at night.  

the annoying popping of bushes on altis and some clutter.

 

just two examples to show you it's not just AI that can get overlooked in long term. and these are not attacks. I like to be realistic in my approach. for the most part so far the AI runs. now pumping a lot of money into AI is not going to keep the lights on if you understand me. it would be wise to do it alongside money makers. example jets dlc and the fix you mentioned. 

 

is there not a  Dr who also works on AI?

 

 

 

overlooked for the entire development arc, 2012-2017, not 12 month ;)

 

[teabagginpeople 398]

27 minutes ago, fn_Quiksilver said:

 

overlooked for the entire development arc, 2012-2017, not 12 month ;)

 

 haha but that cannot be 2012 -17 it just cannot ,I cannot think off the top of my head but there had to be  AI improveme.....  oh wait

 

 never mind thankfully a fine upstanding Gentleman or lady provided some evidence to the contrary

 

- Great scripting command additions like "forgetTarget" and "targets" and "doSuppressiveFire" and "setDriveOnPath" (forceFollowRoad is great in theory but broken in practice)

- AI forward-firing helicopter AI fixed!!!

- Fixing wing AI is good since Jets DLC!

- AI infantry will fire at fast(er) moving helic

;D

[twakkie 57]

This thread really shows the worst of the BI forums. Time to stop following this.