What can be done to make AI-Turret's actually hit stuff?

    Joined
    Dec 13, 2014
    Messages
    65
    Reaction score
    80
    Right now its seems that AI-Turret's (mostly cannon) can only hit big and slow-moving ships no matter what value is entered for the AI's accuracy. This is occurs because of the AI's inability to lead the target. Increasing the AI's accuracy inside the server.cfg does not seem to increase the ability to lead the target, or it's just not working due to lag at high speeds.

    Either way, it defeats the purpose of turrets to act as a defense against smaller, faster ships.

    While i agree that 100% AI-accuracy would be a bad thing, at the moment we have a situation where every shot either hits (big and slow) or misses (small/med and fast).

    I was thinking of a solution without solely relying on spray and pray ..which i think was stated to be not viable because of the calculations that would occur and furthermore would render slow firing sniper cannons virtually useless.

    Maybe one solution would be, to make turrets more inaccurate the faster they are turning. This way, slow firing turrets could hit far targets but would never be able to hit close fighters that are speeding past it, no matter how fast it can turn. To hit those pesky fighters, a big ship would have to rely on lots of small, fast firing turrets.

    To adjust the accuracy of those small turrets, the ship could be deliberately fitted with less mass enhancers forcing them to turn a bit slower. This Idea would of course require mass enhancers ..or the lack of them.. to have a greater impact, on even the smallest of turrets.

    But that's just my little idea how to tackle this big problem. Can somebody think of another, maybe even better solution?
     
    Joined
    Nov 1, 2014
    Messages
    317
    Reaction score
    98
    This is a separate issue, but the first thing that comes to mind to me as far as turrets actually hitting stuff is do a line-of-sight calculation. My planet homebase is being swarmed by pirates, but all my turrets are shooting at the ground while trying to shoot pirates on the other side of the planet, meanwhile ignoring the pirates they have a clear shot on.

    Are you talking about PvP? Turrets seem fine PvE but I've never tried PvP so it wouldn't surprise me if they have trouble hitting human piloted ships.
     
    Joined
    Jul 15, 2014
    Messages
    500
    Reaction score
    104
    Honestly, I'd say turrets are already accurate enough. As stated above, they're fine at hitting AI controlled ships for the most part. They have some difficulty with very small fast ships but those also tend to be the most vulnerable to lock on missiles. They're not as good against player controlled ships, but that's just something you as a player need to deal with. Buffing turrets accuracy honestly risks creating a serious imbalance, and removing piloting skill from the equation more then it already it.
     

    Master_Artificer

    Press F to pay respects
    Joined
    Feb 17, 2015
    Messages
    1,588
    Reaction score
    611
    How many turrets is generally considered a good cut off limit (before anti missile turrets) so i dont accidentally cause slideshows when I try to fight someone?
     
    Joined
    Jan 31, 2015
    Messages
    1,689
    Reaction score
    1,197
    How many turrets is generally considered a good cut off limit (before anti missile turrets) so i dont accidentally cause slideshows when I try to fight someone?
    This is a good question and I'm not sure if there's a concrete answer, but I'd love to know as well! I try to keep mine under 20 non-PD turrets because that just feels sensible, but I haven't actually stress-tested it. I'm guessing it varies from server to server and depends on other ships in the vicinity as well as features like stations, planets and asteroids nearby so I just try to keep mine low, cover all the vectors I can, and up their size if I need more firepower.
     

    jayman38

    Precentor-Primus, pro-tempore
    Joined
    Jul 13, 2014
    Messages
    2,514
    Reaction score
    780
    I've generally heard that 40 is a good rule-of-thumb maximum turret count. If they are evenly distributed, only about half (20) will be firing at the same time. If you distribute that evenly between attack and AMS turrets, you are talking about 10 of each type on either side. Besides, after placing and setting 40 turrets, the average human is pretty much done with turrets for a while.:p
     
    • Like
    Reactions: MacThule
    Joined
    Jul 1, 2013
    Messages
    84
    Reaction score
    19
    I've generally heard that 40 is a good rule-of-thumb maximum turret count. If they are evenly distributed, only about half (20) will be firing at the same time. If you distribute that evenly between attack and AMS turrets, you are talking about 10 of each type on either side. Besides, after placing and setting 40 turrets, the average human is pretty much done with turrets for a while.:p
    i turret it till it crashes,
     
    Joined
    Jan 1, 2015
    Messages
    922
    Reaction score
    289
    The most turrets I've ever put on a ship was 60. That was a two million plus block battleship. I never noticed any issues from that quantity. The ship also had 48 docked power reactors, without issues.
     

    Raisinbat

    Raging Troll
    Joined
    Dec 29, 2014
    Messages
    459
    Reaction score
    269
    Honestly, I'd say turrets are already accurate enough. As stated above, they're fine at hitting AI controlled ships for the most part. They have some difficulty with very small fast ships but those also tend to be the most vulnerable to lock on missiles.
    This is bullshit. A cannon turret is only accurate against high-speed player ships about 700m out; they cannot hit further out at all regardless of accuracy because they don't lead. They're only useful against ai ships because ai is bugged and can't move; if the AI would move like it's supposed to instead of sitting still they also wouldn't work against ai. You shouldnt ballance the game by making the ai too stupid to play.

    Those lock-ons are equally effective against large ships that are slow; how is that balanced?

    Buffing turrets accuracy honestly risks creating a serious imbalance, and removing piloting skill from the equation more then it already it.
    It's allready here, just use beams.

    While i agree that 100% AI-accuracy would be a bad thing.
    This is not achievable because the ships you're shooting at can change direction; Even if every shot is lead with perfect accuracy against a ship with high thrust the direction can change between you firing and the shot hitting.

    Maybe one solution would be, to make turrets more inaccurate the faster they are turning. This way, slow firing turrets could hit far targets but would never be able to hit close fighters that are speeding past it, no matter how fast it can turn. To hit those pesky fighters, a big ship would have to rely on lots of small, fast firing turrets.
    Won't this affect small turrets and big turrets equally? I don't see any reason small turrets would be more suitable for this than big turrets.

    Fast firing also isn't a benefit unless a single shot from the big turret is complete overkill; a 4 million damage round from a can/beam turret wouldn't be any more effective at killing fighters than a 1million round, but 1000 rounds at 1000 damage aren't any better than 1 shot at 1million (assuming fighter has 1mil HP) hell they're arguably worse since 1hit kill negates shield regen, and the turret size i usually see they're too weak to penetrate armor unless the turrets are 200mass or above. Only advantage to more bullets is hull penetration but that 1mil cannon can shoot several bullets at the same time too. Fast firing is good for manual leading, there's no advantage to it for the AI.

    Can somebody think of another, maybe even better solution?

    Turrets need to be able to lead their targets, it's ridiculous that they can't. Adjusting accuracy with leading will still make them miss their target, and perfect leading doesn't mean every shot will land. They already have perfect aim with beams, and beams do just as much damage as cannons (Where is my beam damage penalty with range???) so why does leveling the field matter?
     
    Joined
    Dec 13, 2014
    Messages
    65
    Reaction score
    80
    This is not achievable because the ships you're shooting at can change direction; Even if every shot is lead with perfect accuracy against a ship with high thrust the direction can change between you firing and the shot hitting.
    Yea, i guess you are right. I think i was referring to 100% accuracy with AI-Ship behavior in mind. Even with the ability to always hit a ship that does not try to evade, AI would be less effective than a Human (at least up close) because it cannot identify potential weak points and just chooses blocks at random.

    Won't this affect small turrets and big turrets equally? I don't see any reason small turrets would be more suitable for this than big turrets.

    Fast firing also isn't a benefit unless a single shot from the big turret is complete overkill; a 4 million damage round from a can/beam turret wouldn't be any more effective at killing fighters than a 1million round, but 1000 rounds at 1000 damage aren't any better than 1 shot at 1million (assuming fighter has 1mil HP) hell they're arguably worse since 1hit kill negates shield regen, and the turret size i usually see they're too weak to penetrate armor unless the turrets are 200mass or above. Only advantage to more bullets is hull penetration but that 1mil cannon can shoot several bullets at the same time too. Fast firing is good for manual leading, there's no advantage to it for the AI.
    Again, i suppose that's true. I guess with my proposal both the big slow firing and the many small fast firing turrets would miss the fighter up close.
    But in case the AI would be able to perfectly lead the target, i feel we need some drawback so that fighters or drones that wan't to get in close using speed, won't just be shredded. You are correct that at a certain distance, the ability to perfectly lead wouldn't translate into 100% accuracy. For close encounters however this would be almost the case. The only way i could think of to change this, would be to lower accuracy/leading depending on the rotation speed of a turret, almost like the accuracy penalty you suffer, while running or jumping inside an egoshooter game. This approach could also be used to prevent beam-turrets to always hit a fast target up close.

    Turrets need to be able to lead their targets, it's ridiculous that they can't. Adjusting accuracy with leading will still make them miss their target, and perfect leading doesn't mean every shot will land. They already have perfect aim with beams, and beams do just as much damage as cannons (Where is my beam damage penalty with range???) so why does leveling the field matter?
    Isn't this inability to lead the target kinda new? Maybe.. just maybe.. they removed it only to reintroduce it as an ability AI-crewmembers might be good at :p

    Oh.. and beam damage penalty sounds like something we need!
     
    Last edited:

    Raisinbat

    Raging Troll
    Joined
    Dec 29, 2014
    Messages
    459
    Reaction score
    269
    But in case the AI would be able to perfectly lead the target, i feel we need some drawback so that fighters or drones that wan't to get in close using speed, won't just be shredded. You are correct that at a certain distance, the ability to perfectly lead wouldn't translate into 100% accuracy. For close encounters however this would be almost the case. The only way i could think of to change this, would be to lower accuracy/leading depending on the rotation speed of a turret, almost like the accuracy penalty you suffer, while running or jumping inside an egoshooter game. This approach could also be used to prevent beam-turrets to always hit a fast target up close.
    It's a bit of an issue to tell how badly they would get slaughtered, especially since fighter ai is so bad, but i think people tend to exaggerate this. One of the first ships i made was a 30k mass ship with 100+ lock on turrets. It's not a very efficient ship by current standards, but 100 fighters could take it down in ai battles BEFORE the armor update and i can't imagine it doing any better now.

    Mass counts for fighters too and i don't think taking down 50.000 mass ships with 5000 mass worth of fighters is a good outcome for anyone. Would be nice having more specialization so some weapons are good vs fighters and some vs capitals, but i think that requires a large overhaul of the weapon, damage and power systems to do properly.
     
    Joined
    Jan 1, 2015
    Messages
    922
    Reaction score
    289
    Would be nice having more specialization so some weapons are good vs fighters and some vs capitals, but i think that requires a large overhaul of the weapon, damage and power systems to do properly.
    We already have that IMO. A beam/cannon weapon can slice and dice fighters and drones with extreme effectiveness, assuming it was built large enough in the first place to crack advanced armor blocks with each tick. Such a weapon however is of very little use versus capitals, unless said capital was itself designed around short range weaponry, and that would be a HUGE mistake on the part of whomever built the capital, again IMO.
     

    Raisinbat

    Raging Troll
    Joined
    Dec 29, 2014
    Messages
    459
    Reaction score
    269
    We already have that IMO. A beam/cannon weapon can slice and dice fighters and drones with extreme effectiveness, assuming it was built large enough in the first place to crack advanced armor blocks with each tick. Such a weapon however is of very little use versus capitals, unless said capital was itself designed around short range weaponry, and that would be a HUGE mistake on the part of whomever built the capital, again IMO.
    How is that less effective against capitals? You can put long range weapons on drones just as well, and then short range weapons don't work against them.
     
    Joined
    Jan 1, 2015
    Messages
    922
    Reaction score
    289
    Yes, you can put long range weapons on drones. However such long range weapons either are quite weak DPS wise or require large batteries to store power, which is less than ideal for small ships. Most small ships will use shorter range, rapid fire weaponry that can make ideal use of high power but low storage. Though truth, I am mostly concerned these days with coming up with effective anti-torpedo defenses, so I was overly myopic in referring solely to beams. My ships nowadays have a mix of beam and cannon turrets for defense versus fighters and torpedoes (plus absurd quantities of PD).

    My offensive firepower on my own 'capitals' tends to be extreme range, a mix of primarily missile/beam and a bit of cannon/beam thrown in for flavor. If engaged by another capital, my tactic is to maintain my range. If they try to get close, I fly retrograde so the attacker has to fly into my fire while trying to close. My ships, including my capitals, tend to have both high thrust and some overdrive, so even if I can be caught, they are going to be subject to a LOT of fire before they do.
     

    AtraUnam

    Maiden of crashes
    Joined
    Oct 15, 2013
    Messages
    1,113
    Reaction score
    802
    Ok hearing some statements that turrets do not lead. Turrets DO lead with the following exceptions that may cause them not to lead.
    • Dumbfire missiles to not lead
    • Putting two different weapon types on a single turret may break leading for one or both of those weapons
    • The AI accuracy is a linear not a percentage, AI theoretically is 99% accurate at whatever range you set the accuracy too.
     
    Joined
    Mar 2, 2014
    Messages
    1,293
    Reaction score
    229
    I'm pretty sure the AI can still lead the target. It just seems to aim for a point slightly different from the calculated one. Also, I don't think turret accuracy should be artificially lowered for bigger turrets. This would be unexpected behaviour, which had to be learned and memorized. The game is complicated enough. A better solution would be to reduce the firing arc of turret weaons. This would have a way bigger impact on slower turrets than fast ones.

    Fast firing also isn't a benefit unless a single shot from the big turret is complete overkill; a 4 million damage round from a can/beam turret wouldn't be any more effective at killing fighters than a 1million round, but 1000 rounds at 1000 damage aren't any better than 1 shot at 1million (assuming fighter has 1mil HP) hell they're arguably worse since 1hit kill negates shield regen, and the turret size i usually see they're too weak to penetrate armor unless the turrets are 200mass or above. Only advantage to more bullets is hull penetration but that 1mil cannon can shoot several bullets at the same time too. Fast firing is good for manual leading, there's no advantage to it for the AI.
    That's not true (except for shields). The more damage per shot, the more damage gets wasted.
     

    Raisinbat

    Raging Troll
    Joined
    Dec 29, 2014
    Messages
    459
    Reaction score
    269
    I'm pretty sure the AI can still lead the target
    No. No it can't. Circlestrafe an ai ship and no matter how many turrets it's got it will never. ever. hit you.

    A better solution would be to reduce the firing arc of turret weaons. This would have a way bigger impact on slower turrets than fast ones.
    Removing firing arcs would be wonderful, but i don't think this is all that relevant except at <100m range.

    That's not true (except for shields). The more damage per shot, the more damage gets wasted.
    This is true, but how many blocks you break doesn't matter if its overheating. Oneshotting a ship with 1 bullet or 1000 bullets take the exact same amount of damage; the 1000 bullets just leaves less ship remaining. Point is that big turrets, as long as they do less damage than instant kill, are no worse for shooting down fighters than small turrets. Also, repeating that a large, slow firing weapon can fire many bullets at once.
     
    Joined
    Mar 2, 2014
    Messages
    1,293
    Reaction score
    229
    No. No it can't. Circlestrafe an ai ship and no matter how many turrets it's got it will never. ever. hit you.
    AI can predict linear motion, circle strafing is too advanced. In other words: you're too good.

    This is true, but how many blocks you break doesn't matter if its overheating. Oneshotting a ship with 1 bullet or 1000 bullets take the exact same amount of damage; the 1000 bullets just leaves less ship remaining. Point is that big turrets, as long as they do less damage than instant kill, are no worse for shooting down fighters than small turrets. Also, repeating that a large, slow firing weapon can fire many bullets at once.
    How many fighters can be one-shot?
     
    Joined
    Dec 13, 2014
    Messages
    65
    Reaction score
    80
    Putting two different weapon types on a single turret may break leading for one or both of those weapons
    Wow okay, I guess you might have just answered what I needed to know. Most of my turrets use 4 barrels with sperate cannon/beam computers and a decreasing beam module count for each group. I did this to force the AI not to shoot them all at once but instead fire it with slight delay to make the shots hit in different spots. This went fine for a while but seems to be broken now.
    I will try to change the setup an see what happens!

    But doing this will once again make multiple barrels on any AI controlled turret a waste of energy without any benefit. I would like to see every output hit a different block randomly, like missiles already do. The way it is now is utterly useless.