Print

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

[CapnHector]

Well, from having farmed Ordos a lot myself using Conquest last version, I actually think the AI handles the shield quite well for what it is. It's just that the shield is really bad. But if you give it hardened and stabilized shields and solar shielding and give your officer Field Modulation and max the flux stats, the AI can actually use the Conquest to tank somewhat effectively. This is helped a great deal by the ship being actually able to back out of trouble which is unusual for a capital and certainly helps the AI which occasionally makes very bad decisions.

the Conquest was included as the AI capital ship in the strongest fleets that could fight multiple Remnant Ordos without losing ships, under the assumption that the player pilots an Onslaught in a tanking role and directs the combat, and where strength is measured by killing speed in victorious combat.

+once speed is not the only metric, or you don't have a player tank things also change a lot.
Best case speed to kill middle-high strength remnant stacks does show something, but without the full context it leads to "AI conquest best, Onslaught bad" posts like the one above you.

Yeah I'll need to remember to give more context in the future. Although I do agree that I think the Conquest was a particularly strong ship in any context last version if you know how to build it - I did a playthrough where I just beelined to a Conquest and jumped into the late game - but that's not specifically just because it is good at farming Remnants.

[Lawrence Master-blaster]

Onslaught was already great under AI control, something is wrong with your setup.

AI is bad with weapons in hardpoints because it switches targets too often, a problem that is only compouded if the ship also has low maneouverability. My current fleet has four Sunders and even these have serious problems with staying on target - and they're destroyers with +50% maneouverability from the Helmsmanship skill. Onslaught is built around the two TPCs - without them you are literally better off with a Dominator - so it simply has no chance as an AI ship.

Then you have other factors like movement(AI can't really use burn drive to once the battle starts because it prioritizes facing towards the enemy while Maneouvering Jets work in every direction)

The entire test can be thrown out once you allow mixed fleets with frigate hunters to help out the slow capitals/cruisers, or if you care about not retrying the same battle 20 times.

So the first question is: why would I use an Onslaught with frigate support which increases its effective DP when I can just use a capital that does not need to be babysit instead? And use the DP I would otherwise "waste" on frigates to get even more capitals?

But I actually do run mixed fleets(currently 6 Glimmers, 4 Conquests and 4 Sunders) and I did try replacing Conquests with Onslaughts and the results were painful to watch. As for "not retrying the same battle 20 times" I don't even know what this is supposed to mean.

[TerranEmpire]

Just to quickly give my opinion here.

I think the design intention behind the Onslaught is to create the best CQB duelist in the game. Correct me @Alex if I'm wrong.
And I think that Onslaught is THE deadliest knife fighter. A well-built Onslaught can beat a Paragon or a Radiant or a Pegasus if it gets close and the opponent can't get out of range. Your job as a captain/commander of the fleet is to create a situation where you can maximize its advantage and minimize its weaknesses. But if you can create a situation that is advantageous for the Onslaught, nothing can escape it.

Obviously, this is true for a lot of other ships, it's just that for them the advantageous situation is completely different, and that's why we have different playstyles/preferences/opinions. Thx Alex for implementing such a great game.

TLDR Onslaught is good as is, but you need to know how to use it.

[Alex]

@Alex I re-ran the tests for you using the same saves, same layouts, identical everything vs. the same Ordo and Jangala station, but setting shield efficiency to 0.8, top speed to 35 and removing the missile hardpoints rotate flag.

...

Thank you so much! I really appreciate it. So, alright, the Pegasus will stay that way for the next hotfix, and at 50 DP. We'll see how it goes after that, of course, as with everything

I just checked the squall script, and actually it's 100+150.
Looks like alex just forgot to update the statcard when he changed from +200 to +150.

Yeah; already fixed in-dev for the next hotfix!

Im talking about base balance without skill stacking. Any ship will be much better with 3 s-mods and combat skills. Doom is totally wrecked in SIM battles by dominator at 25 DP points and dominator is the slowest ship with exposed rear armor.

The Doom is extremely not meant for head-on 1-1 fights, so this is pretty much expected. It's about supporting allied ships in different ways, and it's very good for shifting the focus of its support quickly.

(Edit: perhaps it's more accurate to say it's not meant for fights where it's forced to trade damage, rather than specifically 1-1s.)

- Sometimes they turn sideway while attempting to back-off from an engagement. They can do so when they still have shields and can even do so while they are still shooting (wasting their attack). Lost a few ships in fights they should have won but somehow they decided to expose their flank for no reason. It is much more apparent with maneuvrable ships but I seen capitals do so as well. Hammerheads are very prone to do this when backing off, but I also seen Wolfs do their same even when they were able to teleport away.

Hmm, I'll keep an eye out for that, thank you!

- Sometimes, as their try to get closer, ships forget to raise their shields when their are getting in range of their foes longest range beams or guns.

This is probably intentional; the AI is willing to take some hits on armor for a flux advantage. At times this'll look "wrong" and like unnecessary damage, but there's basically some amount of damage it doesn't care very much about.

- IR autolance has funky behavior when left auto-firing; It fire at random interval and doesn't always spend all it's available charges when it does.

It only fires a small portion of its charges at shields; the tooltip explains its behavior!

[Draba]

- IR autolance has funky behavior when left auto-firing; It fire at random interval and doesn't always spend all it's available charges when it does.

It only fires a small portion of its charges at shields; the tooltip explains its behavior!

That does make it much better than what I expected from the patch notes, any chance the thumper could get something similar?
I feel in the M ballistic slot there is too much competition for it in its current form, isn't very good against shields and by the time there would be a chance to shine it's usually out of charges.

I think the design intention behind the Onslaught is to create the best CQB duelist in the game. Correct me @Alex if I'm wrong.
And I think that Onslaught is THE deadliest knife fighter. A well-built Onslaught can beat a Paragon or a Radiant or a Pegasus if it gets close and the opponent can't get out of range. Your job as a captain/commander of the fleet is to create a situation where you can maximize its advantage and minimize its weaknesses. But if you can create a situation that is advantageous for the Onslaught, nothing can escape it.

Keep in mind it has 2 of the best energy weapons in the game (1000 range efficient hardflux, decent hit strength), +1 L and 5 M frontal ballistic turrets.
It can be built to shoot from a distance really well, sticking some unguided missiles in the 4 M turrets and going balls out with the bazillion armor is just stronger.
Actually likes it if a bunch of brilliants/novas/radiants jump in its face, doesn't care about squall/MIRV nerf. Does care about bad loadouts, see the post above yours genuinely comparing it to a Dominator.

[Grievous69]

I didn't see much talk about this but I'm really happy how s-mod bonuses and penalties turned out. I was very negative initially, thinking it would either not change much, or introduce too many wacky bonus stats leading to unfun decisions. But it somehow ended up in the perfect middle, being interesting enough to make you think a bit, but not overwhelmingly important where you're paralyzed until you checked everything out. I'm always glad to be shown wrong in a positive way.

That said it did increase the need for variable weapon stats, as now we have even more ways to make them misleading due to various buffs. Although I'm sure that'll happen somewhere along the way to 1.0.

[Draba]

I didn't see much talk about this but I'm really happy how s-mod bonuses and penalties turned out. I was very negative initially, thinking it would either not change much, or introduce too many wacky bonus stats leading to unfun decisions. But it somehow ended up in the perfect middle, being interesting enough to make you think a bit, but not overwhelmingly important where you're paralyzed until you checked everything out. I'm always glad to be shown wrong in a positive way.

That said it did increase the need for variable weapon stats, as now we have even more ways to make them misleading due to various buffs. Although I'm sure that'll happen somewhere along the way to 1.0.

Yeah, the 20% reload penalty on missile rax and 25% maneouverability on heavy armor is just enough to make you consider something else.
Armored weapon mounts opens options up a bit for ships low on slots.
Builtin mags might be a bit too strong in the sense that it could turn out to be kinda-sorta mandatory for burst weapons, but effective cost is also pretty high so might be fine.

[Vanshilar]

Missile hardpoints normally don't, but on the Pegasus they do, thanks to a MISSILE_HARDPOINTS_ROTATE flag in the csv. Which is specifically so that dumbfire missiles/torpedoes are more usable there, which is kind of a problem, so, uh. Very much self-inflicted.

Huh...how does that work? I console created myself a Pegasus and played around with it in the sim (about all I can do on this laptop) but can't seem to get the rear missiles to rotate, with Cyclones on them. This is trying both -RC8 and -RC9. Not sure if it being console-created instead of a "legitimate" one from a shop or something makes a difference, or if it doesn't work with the Cyclone or something.

Wasn't he entire "point" of LG ships is that they put energy mounts on ships that can't support them? If Executor is going to be its own thing then so should Brawler LG, Hammerhead LG, Falcon LG, Eagle LG...

It's more that currently, under the hood, the Executor is a skin of the Pegasus, which means "copy the base hull and then make the following changes" instead of being a base hull of its own. That means that changes to the Pegasus will automatically affect the Executor unless Alex remembers to "change it back" for the Executor, and also there are only a limited amount of types of changes you can make using the skin method. Making the Executor its own base hull means that their properties can go their own separate ways without Alex having to worry about keeping them compatible with each other via a skin. This is an under-the-hood change that won't affect us players directly, just Alex and modders.

Even though he optimizes his fleet carefully and searches for the best combinations using exacting tools to analyze damage output his playership Onslaught does a significant chunk of the total damage, for example one of his battle reports had his Onslaught XIV doing 34% of the damage done by his fleet and the rest of his fleet was 4 Conquests and 2 Gryphons with optimized layouts.

It was actually 3 Conquests and 2 Gryphons (note that my total fleet was 200 DP, not 240 DP), but yeah, the player-controlled Onslaught regularly did 30% or more of the overall fleet damage. When I was testing with a fleet of the more underpowered ships like the Falcon or the Eagle, it was somewhere over 40% of the overall fleet damage, and that was with a close to 240 DP fleet. That's why I tend to go with DPS numbers; at around 1200-1500 DPS a player-controlled Onslaught is worth around 3.5 AI Gryphons or a bit less than 2 AI Conquests, and those are pretty strong in their own right.

I think that's why in these discussions, it's always worth keeping in mind whether the forum poster is talking about the ship being player-controlled or under AI control. I'm not particularly swayed by reports of the Pegasus being able to burst kill a Radiant under player control because I do it all the time in the Onslaught spamming Proximity Charge Launchers, a tactic that the AI doesn't know how to use. It's certainly possible that when I get around to testing it myself in a week or two I'll agree that it's overpowered, but it's just something that I'm used to expecting the player-controlled flagship to be able to do.

Yeah, under AI control the Conquest was strongest capital last version according to Vanshilar, assuming player fields an Onslaught and does the tanking; whether it is still so in .96 remains to be seen because what made Conquest so strong was double Squall combined with large ballistics and the former were significantly (and deservedly imho) nerfed.

Well technically I never released results for the other capitals so I can't really say that the Conquest was the strongest capital under AI control, only that it was the strongest one that I'm aware of. However, there were no takers for "feel free to submit an Onslaught loadout that you think does really well against Ordos under AI control" and I did dabble in a number of other capitals, but never found anything as good as the Conquest. My take on the whole Onslaught vs Conquest debate though is that Onslaught is better for player, Conquest is better for AI.

I admit that makes it harder to gauge what the DP should be for ships (the fact that different ships may perform better under player control versus under AI control). I think it's readily apparent that many phase ships, especially [REDACTED BROCCOLI], are much more suitable for the player than for the AI.

It'll be fun to compare the different capital ships under player control and under AI control once I get back from my trip in a week or two. I think that'll put the DP estimates on firmer footing -- or at least to see their performance under a controlled set of conditions. Otherwise a lot of the discussion right now lacks context or is largely feelings-based (i.e. "I feel like...") rather than based on something more concrete.

I've been meaning to ask but when I see hullmods I noticed they had the term common as their type. Was their any plans to make rare or very experimental hullmods that players had to obtain?

I think a lot of mods add their own mod-specific hullmod type, so it's at least useful for that. No idea about vanilla though.

The Squalls were actually buffed in their anti-shield role, the damage increasing from 250 kinetic to 100+200 kinetic.

Actually that's just a description error, testing in sim shows that Squalls do 250 kinetic damage (i.e. 500 damage) to shields as they should. I already sent a bug report about it here.

- Sometimes they turn sideway while attempting to back-off from an engagement.

Yes that's been the AI for a long time I think, basically they're anticipating taking enemy fire on the sides (where their armor is still intact) instead of on the front (where the armor is likely gone already). Unfortunately it means that they miss with their hardpoints and/or can no longer pressure the enemy with the hardpoints, so I would rather they not, but that's known behavior.

- IR autolance has funky behavior when left auto-firing;

I actually quite like the IR Autolance behavior. It uses the new "USE_LESS_VS_SHIELDS" hint (the Mining Blaster is the only other weapon to have it). Basically when enemy shields are up, it'll only fire a few shots, just enough that it'll regen back to full capacity quickly. Then when enemy shields go down is when it spams its charges, which is what you want for a weapon that does very little to shields.

This is a pet peeve of mine, actual description would be "Strongest judged by a single, very narrow metric.

Yes, and that metric would be "how quickly it can kill off a sufficiently difficult enemy fleet" for which I chose double Ordos. Calling it narrow is like calling 0 to 60 mph a narrow measure of acceleration -- sure, it doesn't measure 0 to 50 mph or 0 to 100 mph, but it gives a pretty good intuitive metric of a car's acceleration.

Also not counting battles where something gets blown up ofc"

Ignoring fleets which are designed to get blown up (such as Derelict Operations fleets), if your fleet is getting blown up left and right then I don't think it's a good sustainable way to run a fleet. I don't find "hey I won but I lost half my fleet" to be very persuasive of a well-constructed fleet.

Not to mention, once a ship dies its contribution to the battle is exactly zero. Sure, you can send in another, but that ship is still contributing zero until it gets to the front lines, so a ship that doesn't die is going to contribute more by comparison. So basically, a fleet that doesn't die is going to do better than one that does. That rule is more to exclude fleets like say 200 Reaper-laden reckless Kites.

The entire test can be thrown out once you allow mixed fleets with frigate hunters to help out the slow capitals/cruisers,

No, you can use the same testing method. I started by testing via spamming the same ship just to establish a baseline for what each ship individually can do, since it's easier to evaluate a ship's performance in isolation without worrying about cross-ship interactions. In this case for testing these capital ships I'll likely be supporting them with Gryphons. (So for example, player Onslaught + 2 Pegasus + 3 Gryphons, or player Invictus + 2 Executor + 2 Gryphons, etc.).

or if you care about not retrying the same battle 20 times.

Funny thing, it's the underperforming ships like the Falcon and the Eagle that I had to try multiple tries, stronger ships like Conquest and Gryphon were pretty much nearly always completion and it's just a matter of how fast it was.

Conquest is certainly good, but saying it's objectively the strongest AI capital is just silly IMO.

If you have a stronger one you are more than welcome to propose a loadout for it. I don't think anybody has said that it's "objectively" anything because it's not defined as to what that even means, only that thus far it does the best as far as is known under a quantitative measurement of a fleet's ability to kill a sufficiently challenging enemy fleet.

Since AI is "dumb", can't plan far ahead, and mostly makes decisions based on ship positions and relative flux levels, ships with high speed and fast flux dissipation will always be easier for it. Conquest excels at those, with the only problem being the horrible shield, which is another thing AI can't manage well.

Actually, the winning strategy seems to be to dump so much stuff at the enemy fleet that they can't mount a proper offensive at your fleet. Basically, if your ships are sitting there exchanging fire back and forth, then you've done something wrong or your fleet simply doesn't have enough offensive power. That's why the highest-DPS fleets ended up being the ones that can do the most long-range damage, like the Conquest, the Gryphon, and the Atlas 2. But yes, it also covers up that the AI is "dumb" in terms of positioning and flux management when the enemy is too busy dying from far away.

The Ordos fleets in 0.96a are a lot more dynamic and will quite often suicidally zoom forward into your fleet, breaking up the relatively smooth and static battle lines, so it'll be interesting to see if this type of "long range glass cannon" strategy still works well in this update. I'm all for having different playstyles be successful, but this was the most successful one that I found in the last version. Each version seems to have a playstyle that's generally more successful (when I first started playing it was Drover spam) and this was the most successful one I found in the last version, so it'll be interesting to see via testing what the new meta might be in 0.96a.

That does make it much better than what I expected from the patch notes, any chance the thumper could get something similar?

I was going to post this exact thing in the Suggestions forum, I hope the Thumper can get the same "USE_LESS_VS_SHIELDS" hint that the IR Autolance and the Mining Blaster have.

[Alex]

... any chance the thumper could get something similar?
I feel in the M ballistic slot there is too much competition for it in its current form, isn't very good against shields and by the time there would be a chance to shine it's usually out of charges.

Hmm, I don't think that'd be a good idea - it *is* quite decent against shields; the high DPS and low flux cost make it so. For the autolance, it'd be an absolute waste, but for the Thumper it's more of a decision and it's something impactful enough to where e.g. you might keep the group on manual control. I realize it's a bit of a slippery slope but making weapon autofire "smart" is something I want to be *extremely* conservative with.

I didn't see much talk about this but I'm really happy how s-mod bonuses and penalties turned out. I was very negative initially, thinking it would either not change much, or introduce too many wacky bonus stats leading to unfun decisions. But it somehow ended up in the perfect middle, being interesting enough to make you think a bit, but not overwhelmingly important where you're paralyzed until you checked everything out. I'm always glad to be shown wrong in a positive way.

Hey, awesome, I'm really happy to hear it!

That said it did increase the need for variable weapon stats, as now we have even more ways to make them misleading due to various buffs. Although I'm sure that'll happen somewhere along the way to 1.0.

Yeah, I definitely get that. It's also, unfortunately, pretty complicated to do comprehensively. And e.g. for something like damage - how is that even going to cover scripted effects? It's not an *impossible* problem but I looked at it maybe 2 months ago, I think spending a couple of hours thinking about it, and the end result was backing away slowly.

Huh...how does that work? I console created myself a Pegasus and played around with it in the sim (about all I can do on this laptop) but can't seem to get the rear missiles to rotate, with Cyclones on them. This is trying both -RC8 and -RC9. Not sure if it being console-created instead of a "legitimate" one from a shop or something makes a difference, or if it doesn't work with the Cyclone or something.

Ah - looking at the code, the MISSILE_HARDPOINTS_ROTATE hint only affects mounts pointed within the front 180 degree arc of the ship. I think this was because the rear mounts on the Pegasus rotating looked untidy.

It's more that currently, under the hood, the Executor is a skin of the Pegasus, which means "copy the base hull and then make the following changes" instead of being a base hull of its own. That means that changes to the Pegasus will automatically affect the Executor unless Alex remembers to "change it back" for the Executor, and also there are only a limited amount of types of changes you can make using the skin method. Making the Executor its own base hull means that their properties can go their own separate ways without Alex having to worry about keeping them compatible with each other via a skin. This is an under-the-hood change that won't affect us players directly, just Alex and modders.

(Yep, exactly! Though in this case I'd ended up, for the sake of making the changes more quickly, just adding two new properties to .skin files for max speed and shield efficiency.)

[Draba]

... any chance the thumper could get something similar?
I feel in the M ballistic slot there is too much competition for it in its current form, isn't very good against shields and by the time there would be a chance to shine it's usually out of charges.

Hmm, I don't think that'd be a good idea - it *is* quite decent against shields; the high DPS and low flux cost make it so. For the autolance, it'd be an absolute waste, but for the Thumper it's more of a decision and it's something impactful enough to where e.g. you might keep the group on manual control. I realize it's a bit of a slippery slope but making weapon autofire "smart" is something I want to be *extremely* conservative with.

It's not a decision under AI control, AI just dumps thumper at every opportunity
Does noticeable damage against shields, but the ships that could consider thumpers usually have much better shield damage from other sources and want the burst on hull instead.
In practice arbalest does ~same shield damage as thumper burst for much less flux, and often similar hull damage as sustained thumper (higher hit strength).
A bit silly comparison all things considered (accuracy, RoF), but thumper is just not very good when given to the AI.
Builtin mags might help it out a bit, fingers crossed.

[Alex]

(I'd just say then that if this needs a change, the change should be in the ship AI and not adding an autofire hint to the Thumper, in this case; I think that's the more appropriate level for this type of decision.)

[Candesce]

Personally, one of the things I use Thumpers for is absolutely murdering frigates trying to harass bigger ships, where only getting 1.2 efficiency versus their shields doesn't matter much and absolutely mulching their hull in the short period where their shields go down and they realize they need to GTFO matters lots.

Also, even with Elite Ballistic Mastery, I don't think the toggle would do a great job of actually avoiding hitting shields; Thumper projectiles aren't THAT fast.

[Sly]

Since AI is "dumb", can't plan far ahead, and mostly makes decisions based on ship positions and relative flux levels, ships with high speed and fast flux dissipation will always be easier for it. Conquest excels at those, with the only problem being the horrible shield, which is another thing AI can't manage well.

Actually, the winning strategy seems to be to dump so much stuff at the enemy fleet that they can't mount a proper offensive at your fleet. Basically, if your ships are sitting there exchanging fire back and forth, then you've done something wrong or your fleet simply doesn't have enough offensive power. That's why the highest-DPS fleets ended up being the ones that can do the most long-range damage, like the Conquest, the Gryphon, and the Atlas 2. But yes, it also covers up that the AI is "dumb" in terms of positioning and flux management when the enemy is too busy dying from far away.

My experience has also been that this strategy has been successful in previous patches as well. Well, of course it is! If you have fire superiority at any given place in the battle, you'll naturally have the initiative.

I'd like to add that I've been experimenting with different tactics since the last update dropped, creating escort "wings" or "elements" of 3-4 ships: typically a slower ship with an escort of three, with a specific purpose in mind. To reference the strategy you mentioned, I deploy one, two, or three 'tough' and 'defensive' elements that can stand off against a larger part of the enemy fleet during the initial part of the engagement, but aren't very good at chasing down and destroying enemy ships. They are supported by much more fragile, but faster and more powerful "DPS/Fire Superiority" elements that are detached from the main body of my fleet. I'm seeing a great deal of success, with less micro-management.

This sounds like a totally normal thing to do, I imagine? The difference being that on paper, my lean-running fleet (depending on what point of the game I'm in) is outnumbered and outgunned at virtually every stage of the game. Like you and Amoebka have said, the AI is somewhat "dumb" with position and flux management, and it's a simple matter to peel their forces away from one another by stretching the line of battle out, and then using the "divide and conquer" strategy. I've had a lot of success with this so far against pirates, pathers, faction fleets, and small-to-medium sized ordos.

Sprinkle in a small spattering of fast, long-range, but ultimately only annoying hulls and weapons and the enemy formation breaks apart reliably, and melts away sections at a time.

Of course, there is a limit to just how far this can go. No matter how clever I might think it is, being severely outclassed is still fatal.

An addendum is that if you have one or two phase ships in your fleet, whether piloted by the player or not, the new AI behavior for clustering around one another as they "circle the wagons" cannot be underestimated for exploitation. They will voluntarily block one another's firing arcs in a large blob to better protect themselves against a phase ship.

I guess all I really wanted to say was, "hey, I see the same thing and I think that's spot on. Here's a few cool things you can do with that in mind!"

[Alex]

An addendum is that if you have one or two phase ships in your fleet, whether piloted by the player or not, the new AI behavior for clustering around one another as they "circle the wagons" cannot be underestimated for exploitation. They will voluntarily block one another's firing arcs in a large blob to better protect themselves against a phase ship.

They should only do that when there isn't much else other than phase ships nearby, though - is that not what you're seeing?

[Doctorhealsgood]

I have been messing around with the missile autoloader hullmod a bit but... Could it be a little bit more lenient? If you have more than 1 small missile slots more often than not chances are that you won't get a greater bang from your buck than by just picking EMR. And they cost the same.