Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[Morrokain]

In theory (and in the dev build ), the AI shouldn't be autofiring high-flux weapons to that degree. If it is, it's a problem, but solving it by making drastic changes to how the AI uses a large fraction of weapons ... doesn't seem like the right move.

Yeah I agree with this overall. To be fair, I don't think most people are arguing for a *large* fraction of weapons to be changed - depending upon your definition of large. It is more just that there are use cases here that exist in vanilla weapons that would be an overall benefit to vanilla if that makes sense. A large fraction of weapons would probably be going too far with it - if things like the Heavy Blaster and Gauss Cannon are any indication.

From what I've seen the heavy kinetic damage + high impulse of multiple Atropos torpedoes is more what causes overloads than the AI overrunning it's flux - though an argument could certainly be made that it could be the combination of those two things (inefficient flux use being the second) that gets the AI in trouble in the first place - so there is definitely a gray area to consider.

(And of course the diversity of design spectrum that has already been mentioned. Vanilla comes first! But, well, mods are also a pretty big factor too considering the number and quality of them that currently exist. Not trying to be pushy here, btw, in case it seems that way because text != tone. I'm just getting my overall thoughts out on the subject. Anyway, the number of responses already indicates you take these things seriously and I think I can speak for all of us when I say we really appreciate it.)

[Alex]

From what I've seen the heavy kinetic damage + high impulse of multiple Atropos torpedoes is more what causes overloads than the AI overrunning it's flux - though an argument could certainly be made that it could be the combination of those two things (inefficient flux use being the second) that gets the AI in trouble in the first place - so there is definitely a gray area to consider.

Yeah, that makes sense. It seems easy to underestimate just how quickly a bunch of kinetic damage can drive up flux, and in the case of Broadswords, it seems like firing off an HB shot or two and not makes a difference of a few seconds, which isn't too much.

Still, the HB is definitely a tricky weapon for the AI to use! It's just... how it is, I suppose, with it being a bit of an outlier in terms of flux use and damage and so on. It really benefits from more situational awareness and forward planning. So it'll naturally be less good in AI hands in some situations, of which "vs *some* (but not all) fighters" is one, but not the sum total. So it probably doesn't make too much sense to over-focus on fixing that specific one.

(Actually, talking about this now, I'm remembering spending a bunch of time tuning the "whether to put a group on autofire" logic specifically for these kinds of situations! That the HB is better even in some situations, on a heavily over-fluxed loadout, seems like a win.)

Even something like Hellbores - they'll miss fighters a lot, but they're pretty effective vs, say, incoming Piranhas or Flash bombers, no? Due to the shots passing through missiles. And that's a cool moment, to see a hellbore shell carve a path through a cloud of bombs. Plus it's flux-cheap! I understand what you mean about it costing zero-flux bonus, though; I'm sure you're right in it being detrimental for that reason in that situation. But that seems like a rather minor issue overall; I don't know that it's worth the cost of "adding a new rule the player has to remember about the Hellbore so they're not confused" and "the Hellbore is also *less* useful in some (possibly/maybe smaller) number of situations".

Not trying to be pushy here, btw, in case it seems that way because text != tone. I'm just getting my overall thoughts out on the subject.

No worries, not taking anything here in a bad way at all!

[RustyCabbage]

Not sure if your dev build, or the mission I'm using to test, or some mod I have enabled is messing things up, but I can't reproduce this at all. Removing the Heavy Blasters (leaving the excess OP empty) from the Attack variant Medusa (the one you're using as an example) made the Medusa significantly more aggressive, to the point where it actually would reliably attack the Condor.

The effect is even more drastic if you compare the results against the 2x Broadsword Condor, where removing the HBs allows the Medusa to win almost unscathed versus losing over half its hull with them still equipped (or in unlucky cases, even dying).

edit: I have the same result with Broadsword Condor on vanilla using the Coral Nebula mission. Heavy blasters shooting at fighters is a major minus, at least in this example.

Hmm - you didn't spend the OP freed up by removing the blasters, did you? I suspect in-dev changes to flux management and autofire factor in here, too.

It does do better without blasters vs the 2xBroadsword Condor, though - for me, it takes damage in both cases, and the fight seems like it could go either way (in my quick test, it barely beat the Condor with the HBs, and had about half hull left without HBs). A variant with Pulse Lasers instead of Heavy Blasters does better than either of these, winning consistently and easily while only taking a sliver of hull damage. And it's still a very over-fluxed variant, 2x flux generation vs dissipation - around 1200 vs 600.

Thanks for double checking; I left the 24 OP unused, yes. I think your point about how the pulse laser performs significantly better indicates the importance of a rule for specific weapons - it's not so much an issue of how over-fluxed the ship variant is as it is specific weapons having detrimental effects when targetting fighters.

I don't think it's very many weapons either, imo it's only really problematic with the Heavy Blaster, the Gauss and the Tachyon Lance - generally inefficient but high burst and in the latter cases high refire delay.

[Morrokain]

Yeah, that makes sense. It seems easy to underestimate just how quickly a bunch of kinetic damage can drive up flux, and in the case of Broadswords, it seems like firing off an HB shot or two and not makes a difference of a few seconds, which isn't too much.

Still, the HB is definitely a tricky weapon for the AI to use! It's just... how it is, I suppose, with it being a bit of an outlier in terms of flux use and damage and so on. It really benefits from more situational awareness and forward planning. So it'll naturally be less good in AI hands in some situations, of which "vs *some* (but not all) fighters" is one, but not the sum total. So it probably doesn't make too much sense to over-focus on fixing that specific one.

(Actually, talking about this now, I'm remembering spending a bunch of time tuning the "whether to put a group on autofire" logic specifically for these kinds of situations! That the HB is better even in some situations, on a heavily over-fluxed loadout, seems like a win.)

Even something like Hellbores - they'll miss fighters a lot, but they're pretty effective vs, say, incoming Piranhas or Flash bombers, no? Due to the shots passing through missiles. And that's a cool moment, to see a hellbore shell carve a path through a cloud of bombs. Plus it's flux-cheap! I understand what you mean about it costing zero-flux bonus, though; I'm sure you're right in it being detrimental for that reason in that situation. But that seems like a rather minor issue overall; I don't know that it's worth the cost of "adding a new rule the player has to remember about the Hellbore so they're not confused" and "the Hellbore is also *less* useful in some (possibly/maybe smaller) number of situations".

You know, speaking of this, can you remember why the AI needs to put a STRIKE hinted weapon off autofire and fire it manually? - When the player putting such a weapon on autofire doesn't actually make it used on fighters? I'm curious about it.

(I'm more leaning towards including the "don't fire at fighters" hint, btw. The idea of calling it out explicitly is really selling me on it; could see maybe using it for a few vanilla weapons, even.)

Missed this! Yay! 

I hope it happens. My mod really needs this for several weapons. Basically using any weapon that generates flux that isn't specifically designed to be used for fighters should not be used on fighters. (it's a DnD based mod so weapon types are more synergistic and role based and less multipurpose like how vanilla's are)

STRIKE,USE_VS_FRIGATES certainly helps, but it's not always fullproof in the case of missiles - especially when combined with DO_NOT_CONSERVE which is necessary to ensure the AI uses the full clip. I remember seeing non-missile weapons sometimes fire at fighters too with that hint... but tbh that could have been an earlier version or back when my station modules still had the periodic missile reload mod - which iirc changes autofire AI. It doesn't happen often.

One thing that is a downside is the no-autofire behavior I mentioned above. Too many weapon groups of that category and the AI won't switch between them and use them effectively.

[Alex]

Thanks for double checking; I left the 24 OP unused, yes. I think your point about how the pulse laser performs significantly better indicates the importance of a rule for specific weapons - it's not so much an issue of how over-fluxed the ship variant is as it is specific weapons having detrimental effects when targetting fighters.

I don't think it's very many weapons either, imo it's only really problematic with the Heavy Blaster, the Gauss and the Tachyon Lance - generally inefficient but high burst and in the latter cases high refire delay.

Hmm - I think it's rather the opposite! It shows that even in a fairly constrained scenario (Medusa, 2x HB) it can be good to fire, or not. And that even in the "bad" case, it's pretty close! I don't think that replacing a dynamic algorithm that occasionally gets it wrong with a static rule that will *also get it wrong* is a step in the right direction. At best it's a lateral move.

(And static rules tend to produce more apparent, annoying problems. Consider a case of a Medusa with *just* 2x HB - should it still not fire them? It really should. So now that nice, clear, static rule has an exception. What about if it has one PD Laser and 2x HB? What about if it has 2 PD Lasers? What if it has Safety Overrides, drastically changing the flux calculation? It just doesn't work.

I mean, you could still go the "this weapon can't target fighters because its targeting systems aren't capable of it" route and have that be fine - it won't fire, because it "can't". But just as far as helping the AI perform better across the board, it's not going to do that.)

You know, speaking of this, can you remember why the AI needs to put a STRIKE hinted weapon off autofire and fire it manually? - When the player putting such a weapon on autofire doesn't actually make it used on fighters? I'm curious about it.

Putting STRIKE weapons on autofire seems like a bad idea, so the AI... doesn't. It's not specifically about fighters; STRIKE makes the AI take more care with aiming the weapon at other targets (which is why adding it can make it hesitant to fire, since it's looking for a higher-percentage shot).

(I'm more leaning towards including the "don't fire at fighters" hint, btw. The idea of calling it out explicitly is really selling me on it; could see maybe using it for a few vanilla weapons, even.)

Missed this! Yay! 

I hope it happens. My mod really needs this for several weapons. Basically using any weapon that generates flux that isn't specifically designed to be used for fighters should not be used on fighters. (it's a DnD based mod so weapon types are more synergistic and role based and less multipurpose like how vanilla's are)

To be honest, I kind of forgot about Gauss/Hellbore piercing missiles; I probably wouldn't want to use this tag on them in vanilla. But for mods, yeah, it could be handy. My concern is it'd be overused by mods - i.e. put on weapons where equivalent-ish vanilla weapons don't do this...

STRIKE,USE_VS_FRIGATES certainly helps, but it's not always fullproof in the case of missiles - especially when combined with DO_NOT_CONSERVE which is necessary to ensure the AI uses the full clip. I remember seeing non-missile weapons sometimes fire at fighters too with that hint... but tbh that could have been an earlier version or back when my station modules still had the periodic missile reload mod - which iirc changes autofire AI. It doesn't happen often.

Hmm? Yeah, probably to do with the missile reload. As a rule, AI won't fire non-ANTI_FTR missiles at fighters unless it's in panic mode.

[Morrokain]

Putting STRIKE weapons on autofire seems like a bad idea, so the AI... doesn't. It's not specifically about fighters; STRIKE makes the AI take more care with aiming the weapon at other targets (which is why adding it can make it hesitant to fire, since it's looking for a higher-percentage shot).

Ehhh, that doesn't seem to be the reason. The weapons I'm talking about are small, fast projectile weapons or even instant strike beams against capitals. They do fire, but not efficiently in regards to their clip/regen.

(Working on reproducing this. I have one case - but the weapon not being used is HE and the shields are up on the enemy vessel. So I'm not sure if that is a good example. That being said, other HE strike weapons are used on shields at times.)

To be honest, I kind of forgot about Gauss/Hellbore piercing missiles; I probably wouldn't want to use this tag on them in vanilla. But for mods, yeah, it could be handy. My concern is it'd be overused by mods - i.e. put on weapons where equivalent-ish vanilla weapons don't do this...

Why is that a concern? It would/certainly should be reported to the mod author- not against vanilla. If the modder chooses to do this then they would take on the burden of any subsequent bug reports (and again stat card/description explanations make a huge difference there.) If they get annoyed by it then they can just not use the hint.

STRIKE,USE_VS_FRIGATES certainly helps, but it's not always fullproof in the case of missiles - especially when combined with DO_NOT_CONSERVE which is necessary to ensure the AI uses the full clip. I remember seeing non-missile weapons sometimes fire at fighters too with that hint... but tbh that could have been an earlier version or back when my station modules still had the periodic missile reload mod - which iirc changes autofire AI. It doesn't happen often.

Hmm? Yeah, probably to do with the missile reload. As a rule, AI won't fire non-ANTI_FTR missiles at fighters unless it's in panic mode.

They don't have that hullmod anymore and in this case I'm talking about ships. And this is not the case in my experience regarding the Atropos - specifically. That is, unless "overwhelming" (as in just numerous not actually threatening) fighters cause panic mode for some reason.

Evidence:

Spoiler

Screens of Weapon definition:

Start:

End:


Video Evidence: (Towards the end.)

[close]

[FooF]

I was under the assumption that the AI was more loose with flux (and firing high-flux weapons) than it actually is so I am more than happy to concede the point, though I appreciate the testing with the Medusa as the HB is kind of the poster-child for what I was getting at.

[Alex]

Ehhh, that doesn't seem to be the reason. The weapons I'm talking about are small, fast projectile weapons or even instant strike beams against capitals. They do fire, but not efficiently in regards to their clip/regen.

(Working on reproducing this. I have one case - but the weapon not being used is HE and the shields are up on the enemy vessel. So I'm not sure if that is a good example. That being said, other HE strike weapons are used on shields at times.)

Ah - STRIKE has some baked-in assumptions, probably, then. Unless it also has USE_VS_FRIGATES? But anyway, that's getting pretty off-topic for the thread. My apologies for contributing to that, myself.

Why is that a concern? It would/certainly should be reported to the mod author- not against vanilla. If the modder chooses to do this then they would take on the burden of any subsequent bug reports (and again stat card/description explanations make a huge difference there.) If they get annoyed by it then they can just not use the hint.

Hmm, yeah.

(Says video is private.)

I was under the assumption that the AI was more loose with flux (and firing high-flux weapons) than it actually is so I am more than happy to concede the point, though I appreciate the testing with the Medusa as the HB is kind of the poster-child for what I was getting at.

Your assumption is probably right given the currently-out state of the AI! And, yeah, Medusa + 2xHB is basically... maybe not the most overfluxed possible combination in vanilla, but it's up there. And especially as something where the disparity between player and AI control is larger. Which is something else I was trying to address while making the changes; so, yeah, hopefully that's in a better place in general, just as far as the AI using high-flux weapons more competently, vs fighters and otherwise.

(IIRC, the measuring stick was having the Attack Medusa be able to beat a stock Hammerhead - the one with Heavy Mortars -in a 1vs1, or some such... I think it used to consistently overflux itself and ether take chip damage or get harpooned. Now, it wins that fight consistently with minimal hull damage...

Ah, and that reminds me - I think I fixed (or at least mitigated a lot) the "turrets turn towards target, fire too early, first volley misses" issue, since that came up with HBs and the Medusa. To be perfectly honest, not 100% if this is already in the currently-out release, but I think it's only in the dev version. Forgot to add this to the patch notes I've been keeping, though.)

[Morrokain]

(Says video is private.)

Whoops! Fixed that sorry!

Unless it also has USE_VS_FRIGATES? But anyway, that's getting pretty off-topic for the thread. My apologies for contributing to that, myself.

(It does, yeah. But agreed this is off-topic. I'll make a new post if I can get a solid case for it. My fault!)

[TaLaR]

Passthrough weapons (Hellbore, Gauss, Plasma) on high level character are all awesome against fighters. Enemy fleet concentrates multiple carriers on you -> backpedal while firing -> most fighters die before they reach you (rest is handled by PD and allied interceptor screen) -> enemy carriers are at low replenishment and open for attack.

HVD, Mauler, etc also become a lot more accurate with character skills (faster projectiles, less recoil, better target leading...).

I haven't tested Plasma yet but that is what I've heard. I already said Gauss was pretty good... though I didn't see any passthrough there against Daggers. Helbore?? That is not my experience at all after several hours of testing.

Yes, it is without character skills but that is the point! Not all of your AI ships can have officers so they should be left out of the equation completely. I don't want behavior that is only good for 1/3 of my allied ships! 

*EDIT* Another counterpoint to that - it would make those skills more mandatory than they really need to be.

After testing a bit more:
- Is Gauss actually passthrough? It can pass missiles/bombs, but now that I've checked even max skills Gauss can't pass through a Talon. Gauss is still decent to use vs fighters even if it doesn't penetrate though (never stopped by missiles/bombs, usually fully destroys fighters leaving no debree to obstruct more shots).
- In comparison even skill-less Plasma easily passes through a Talon, but leaves debree intact.
- Shields always prevent passthrough, so each shield stops at least 1 projectile. Not all fighters are shielded though.

Building fleet around 10 officers is already staple of current meta. Anything else is wildly sub-optimal, with sole exception of Spark Drover spam.

(to edit) It would no less stupid for max skilled ships to refuse to use their advantage (by simply not firing).

[Alex]

(Yeah, Gauss/Hellbore is just missile passthrough.)

[Morrokain]

Building fleet around 10 officers is already staple of current meta. Anything else is wildly sub-optimal, with sole exception of Spark Drover spam.

(to edit) It would no less stupid for max skilled ships to refuse to use their advantage (by simply not firing).

I disagree on it being a wasted advantage if the ships are saving flux and using their other weapons effectively. You shouldn't need an anti-ship weapon (what non anti-fighter/PD large weapons ideally should be) to deal with fighters in the first place imo. Hellbore definitely qualifies as anti-ship in the era where shields weren't a thing. Gauss is the reaction to said shielded ships. To me, if that were the case and these larger weapons are needed to fend off fighters, there is a problem there in terms of overall fighter balance. That means, inherently, that things that don't have these weapons (so anything less than a capital in 95% of cases) are at a strict disadvantage against them.

If meta is based on officers, it really shouldn't be. Skill changes will hopefully address this a bit. I don't want really anything other than campaign difficulty designed around officers making a ship/weapon more powerful. If weapons are designed around that, the weapon is essentially broken in my eyes - one way or the other. There are soo many problems with that sort of design - and they hit hardest on the side of the enemy AI fleets. Officered ships aren't guaranteed to match weapons with hullmods or skills of any kind. The exception is carriers/non-carriers and even then only in category not in specific benefit. So if a weapon needs these skills to be effective at something, that is a losing gamble the majority of the time.

Weapons should ideally work within their role regardless of skills or even hullmods. Same goes for fighters. If a fighter is only useful with skills, it's a bad fighter and should be looked at - especially if it is high tier. These sorts of things are bonuses and as such they should not be design considerations for what works or does not work, but rather they are useful tools to get past a higher difficulty at the later stages of the game.

I also don't want to limit AI builds to whether the ship supports an officer or not. This is especially true because officers are less often found on smaller ships both with player fleets and AI fleets. Smaller ships suffer the most anyway due to limited PPT and weapons. It is already a little painful to have to limit builds based upon the AI limitations as it is. Though I definitely understand the considerations and difficulty of balancing, when I see things like "this isn't an AI-friendly ship/weapon" I grimace a bit. It's understandable to be sure, but ideal? No.

The point is to make changes trending away from this rather than towards it. That's how I see it, anyway. I'm also not advocating that all ship weapons only target ships, etc, I definitely agree there is room for multipurpose/multirole weapons and that is fine. But so is there room (and necessity in a couple cases) for the opposite- a more strict role that the weapon excels at over multipurpose weapons performing the same one.

[Tartiflette]

Fair enough, yeah.

Edit: added "NO_TURN_RATE_BOOST_WHEN_IDLE" weapon hint.

That plus a ignore fighters tag? Whooooo this is great, thanks a lot. Combined with a smarter autofire decision AI, it should help a LOT when dealing with ships supported by fighters.

[Tartiflette]

Relevant to the talk about weapons focussing too much on fighters:
https://twitter.com/NiaTahl/status/1283538572021301253

Turns out there was an oversight and in some (frequent) circumstances a ship would focus on the closest target regardless of its size, causing over-aggressive target switching to fighters.

[Hiruma Kai]

Relevant to the talk about weapons focussing too much on fighters:
https://twitter.com/NiaTahl/status/1283538572021301253

Turns out there was an oversight that in some (frequent) circumstances a ship would focus on the closest target regardless of its size, causing over-aggressive target switching to fighters.

That is very interesting.  I wonder how much that fix will change the AI behavior in heavy fighter spam situations.  Certainly, it should make improve the odds for the gunship in 1 on 1 gunship versus carrier situations.  It does make me wish we could test some of the AI modifications and fixes Alex has waiting for the next release, just to see the effects.  AI is such a huge factor in this kind of testing.

Anyways, I haven't forgotten the Condor testing I still wanted to do, but been a bit busy this week and will hopefully have some time this weekend.