Fractal Softworks Forum

Please login or register.

Login with username, password and session length
Advanced search  

News:

Starsector 0.97a is out! (02/02/24); New blog post: Simulator Enhancements (03/13/24)

Show Posts

This section allows you to view all posts made by this member. Note that you can only see posts made in areas you currently have access to.

Messages - Vanshilar

Pages: [1] 2 3 ... 39
1
General Discussion / Re: should we just nerf the Onslaught?
« on: March 24, 2024, 12:00:56 AM »
At the moment, you can deter an NPC Conquest with a *** Dominator.

How the factions use the ships have little bearing on how well the player can use the ship in the player's fleet. Each faction has only a limited subset of the weapons, hullmods, etc. that the player can access.

The fact that you can make Conquests somewhat viable, you alone, and then everyone else is struggling. But the Onslaught is just universally good should explain to you what kind of balance situation we are in, at least based on community's viewing. I mean, even in your "Conquest is good, guys" video, you're still using an Onslaught as your main tanking capital ship, cause it's just that stupidly broken.

No. I only started using the Conquest about a year and a half ago; in fact it was CapnHector's Conquest Appreciation Thread that spurred me to really try it out. So I'm pretty late to the party in terms of figuring out how to use Conquests.

I use the Onslaught as the player-controlled flagship because it's fairly idiot-tolerant (it can handle some punishment for when I mess up, since I mess up quite often) and because I never learned how to pilot a broadside ship (Conquest), nor any of the other player favorites for flagships. Back when my flagship was a Medusa people also claimed that using Conquests only worked because I was using the Medusa too.

You keep saying the Onslaught is too good but have presented little evidence to support it, other than some vague "everyone knows".

ps. Also, did you measure the viability of the Conquest based on its ability to clear out Ordos as a DPS metric? As in what gives you the fastest resolution of combat??? That just sounds like you broke the game in multiple ways and then found like a strategy that sole winning grace is the fact that it kills enemies the fastest. But why would that matter if under all normal scenarios it's about killing an enemy at all?

Eh, we're discussing how ships perform in combat. The only victory condition in combat is to kill all the enemy ships (or cause them to retreat). How else would you measure their effectiveness in combat other than how good they are at killing enemy ships?

That's like saying Tempest is stronger than a Monitor, cause Tempest will kill Shepherds faster. I mean, you win either way...

Yes, if all you can choose from is a Tempest or a Monitor, it's better to have the Tempest, because it can actually kill something. When doing any sort of comparison, you need to have some sort of a specific, measurable metric by which to compare them over. There's no useful comparison if the result is "everything can do it".

This is also why it's better to have a sufficiently difficult enemy fleet to test against. The whole point is to have a test that can differentiate between different ships, weapons, etc., and thus the test can't be too easy to accomplish. There's little point to a participation trophy in comparisons.

Vanshilar, is the Conquest loadout from that video the one you use currently, or did you update it?

Not exactly but pretty close. Nowadays I'm testing different things out with the medium energy. IR Autolance looks good but trying out different things to see if they might be better.

I don't know how did you manage to misread that as "I don't know what you're talking about" - twice - but since apparently you did, let me clearly say it again:

If you were aware of it, then why did you say 1) you've never seen anyone else do it and 2) use it to support the claim that someone should be "extremely careful" about using it as a basis for balance changes? What is this extreme care that you think should be applied to this that doesn't need to be applied elsewhere?

Yes, achieving a local numerical superiority by using burst weapons in the initial stages of the fight is optimal and most effective, but very few fleets could do it in 0.96 and even less can in 0.97(against double+ Ordo)

Still wrong. When I mentioned testing over a dozen ships in 0.96a, including the Eagle, all of them used this strategy. So either the Eagle is some super duper overpowered ship or this strategy can be used a lot more readily than you claim. When I talked about using Manticores with Escort Package when 0.97a was released, it was using this strategy. You can even see in those attached screenshots how my fleet is surrounding the incoming enemy ships. It doesn't take some super duper secret fleet composition to make it work; I posted a video years ago of doing this using flagship Medusa and 22 LP Brawlers here:


(back then, with LP Brawlers costing 5 DP, or 4 DP with Support Doctrine, my fleet was 100 DP vs the enemy's 240 DP on the battlefield at once). I posted a video even earlier than that where 3 Odysseys, 3 Apogees, and 3 Hyperions naturally did this under complete AI control after I manually spread them out to capture objectives and get them into a line, i.e. after I spread them out, I gave no more commands and did not control any ships, so they were purely under AI control after that point:



Basically this can readily be done by a large number of fleets, using a large number of different playstyles and fleet philosophies. It's not even actually about burst damage per se, but that my ships can spread out in a wider line than the enemy fleet, but without spreading themselves so thin that enemy ships can get through. As long as that happens, then as my ships start killing the initial enemy fleet, then it thins them out enough where my entire fleet can concentrate its fire on a smaller number of enemy ships, since the rest of their DP are tied up in reinforcements (still advancing to the front lines).

You seem to be operating under an assumption that since you figured out the best possible fleet composition/build in the game(which I believe to be true) everyone should logically follow suit and use it - or at least something similar - but that's just not how it works.

No, I don't know if this is the best possible fleet composition/build. All I can say is that it was one of the best ones I tested in 0.96a, and that it performs even better now in 0.97a. It's entirely possible that other fleet compositions have improved even more in 0.97a. It's also entirely possible that there are other fleet compositions which are better that I never tried.

Putting aside that most players are, well, bad at the game and never get into the nitty gritty minutae of mechanics and stats and will never be even remotely as good as you, even the players who ARE good at the game would rather figure out their own thing rather than copy someone else's solution.

No, that's assuming that 1) I'm good at playing the game and 2) good strategies are somehow uniquely and wholly different from each other.

First, I do a lot of analysis, i.e. looking through the numbers, figuring out correlations, understanding and modeling how combat works, etc., which is a very different skillset than playing the game, i.e. controlling the flagship, intuiting/predicting what is about to occur, understanding commands, giving the correct commands at the correct times to the other ships, etc. It's the difference between an airplane designer and a pilot. You can see that I make plenty of mistakes in all of my videos (except the AI-controlled one obviously), not the least of which is that at the beginning of my 5 Ordos fight, I took multiple hits to my armor within a minute of the initial clash and started taking hull damage from there. Thus I had to do the rest of the fight without most of my armor, and thus I had to play a lot more defensively. That's a bad use of the Onslaught, but I didn't want to bother to redo it. The fight could easily have been several minutes faster if I had my armor, since the [REDACTED] are much more aggressive toward vulnerable ships, and less aggressive (i.e. less likely to commit to an assault and more likely to just sit back and take damage until they die) against pristine ships. So a lot of my fleet compositions are built around being relatively fault-tolerant because the player himself makes plenty of mistakes. The completion times that I post are in spite of all these mistakes; a more competent player could easily post better times.

Second, there is nothing particularly unique about military concepts like outflanking the enemy, local concentration of force, etc. These are simply good principles to follow in any combat game, or military combat in general, and have been around since long before you or I were born. The specific implementation may vary from game to game and different players may use different ways to go about them (in this case, different ships, different weapons, etc.), but those principles are pretty time-tested. Players who are good at the game would figure out how to use them to the maximum, rather than avoid them to not "copy someone else's solution". That makes little sense; if someone else already figured out a good solution, why would you bother using a solution that is inferior? If someone says "maxing out your flux is a good idea", do you then say "actually now that you said it, I'm going to do something completely different, because I don't want to copy someone else's solution"? Rejecting good advice for the sake of being different is a bad way to go through life.

Finally:

I don't know how did you manage to misread that as

you're probably just getting too much noise from your flagship to even notice.

You've been doing the same one thing for years and apparently can't comprehend that most other people don't.

You seem to be operating under an assumption that

You keep making up stuff about my state of mind which 1) I have never said, 2) are untrue, and 3) have little relationship with any of the points I'm making, while avoiding addressing any of the points that I do make. This is not only ad hominem but is classic straw man. Please address the points actually being made instead of making things up about other posters.

Conquest's main issue is being broadside. AI doesn't "crabwalk" nearly enough to use broadside efficiently.

Yeah, that's one of the limitations of the Conquest. IIRC Alex has said broadside ships don't move directly toward their target, they move diagonally toward their target instead, even though this means they take longer to get within weapon range of their target. A better AI would go straight for their target, turning as needed so that the weapons face the target, since the ships can turn in any direction independent of their direction of travel. As a result I order them a bit "further in" toward the middle than where I want them to go. For example, for a "left-facing" Conquest (broadside layout where the weapons are on the left), I'll order them to go more to the left than where I actually want them to go, knowing that they'll drift to the right.

For this reason I'm also experimenting with sticking a shorter range weapon on them, say a Pulse Laser, purely to encourage them to move closer more. But obviously if they get too close then they'll take more damage to shields which is bad, so it's a bit of a balancing act.

You know what, it would be interesting for people to post their capital builds so we could actually compare their damage outputs.

Yeah, that's pretty much what I've been doing, comparing the best possible builds against a static test fleet (double Ordos). I in the player-controlled flagship Onslaught to serve as the main lightning rod running into the middle of the enemy fleet (i.e. I'm in the center controlling the flow of battle, tanking when needed, picking the best target to kill, etc.), with a Gryphon on either side taking care of any flankers (so that whichever ship I'm testing does not get their DPS reduced by having to chase down strays, that DPS hit goes to the Gryphons instead). So this way the ship I'm testing is basically spending most their time shooting at the enemy, with some amount of tanking. That's the basis for my opinion on the damage capabilities of the different ships.

2
General Discussion / Re: Damage Bonuses
« on: March 22, 2024, 05:04:30 PM »
Sorry, don't quite understand the question. Could you rephrase it?

Oh, I just meant if the code is such that there's a specific function that says "if X1 effect then modify stat by Y1, if X2 effect then modify stat by Y2", etc., or if you're having to look through each effect function to see how it modifies a stat. It looks like it's the latter, so you'd have to look through them one by one.

EDIT: Just to give an example of how counter-intuitive these get. Void explosions from Cascade Emitter are improved by bonuses that improve energy weapons. Void explosions from Rift Beam and Rift Lance are improved by bonuses that improve missile weapons. They look pretty much the same in-game.

Hmm wonder if that's a bug then. I mean there's no reason why they should be missile-based, unless it says so somewhere in the description. Or if it's one of those "hidden" effects.

3
General Discussion / Re: Damage Bonuses
« on: March 20, 2024, 11:07:18 AM »
Here's a list based on code digging, as opposed to testing:

Ahh very cool! That's exactly what I was looking for!

I started looking into this based on the observation that Target Analysis seemed to do "more" than what its numbers would suggest, as I mentioned here, and testing showed that it was behaving multiplicatively as opposed to additively with other damage modifiers. Having a complete list is going to be very helpful in figuring out how to build a fleet. Thanks!

It seems like then the only remaining thing to worry about are the "special cases" weapons, i.e. Squall, Breach, certain [REDACTED] weapons, etc., which are much more case-by-case rather than general. Likely a lot more straightforward to test and/or look into code since those are basically known exceptions.

Is there a specific function in the code that lists this out, or is the list based on searching through the code for all instances of a particular modifier function/variable?

4
General Discussion / Re: Damage Bonuses
« on: March 20, 2024, 10:18:05 AM »
Graviton is a damage taken modifier, applied to the target ship. getDamageToTargetShieldsMult() is a damage dealt modifier, applied to the firing ship. Imagine a hypothetical "deal 10% more damage to shields" officer skill, that would be it. I think nothing in vanilla currently uses it, though.

As already explained by float, damage taken modifiers are never additive with damage dealt modifiers, and have several "partitions" among them as well.

Ahh that's a good point, since every other ship that's hitting the same target ship also benefits from it, so it's affecting the target ship rather than the firing ship. So it's not just a different category but a different effect altogether. Hmm. Wonder if I have to test all the different damage taken modifiers as well (for example, how Graviton Beam stacks with Entropy Amplifier, etc.).

But at least this examination shows that some effects are more "powerful" than others. For example the fact that Graviton Beam's effect will multiply with everything on the firing ship (and on other firing ships) means that its 5% is better than the 5% from Tactical Drills.

EDIT: On top of that, there are also combat listeners that can modify damage.

Yeah I'd like to list out all the possible damage modifiers in vanilla and figure out how they interact with each other (add or multiply). I wonder if those will end up being special cases. Will take some effort to figure out, I guess.

5
General Discussion / Re: should we just nerf the Onslaught?
« on: March 20, 2024, 12:07:27 AM »
Yes, that's exactly my point. You've been doing the same one thing for years and apparently can't comprehend that most other people don't.

Your point is that...you don't understand combat strategies well enough and that therefore other people's sense of balance should be based on that? I was pointing out that "I didn't know about this concept and therefore your sense of balance shouldn't be based on it" is a bad argument and you're actually doubling down on this position? You're basing the validity of other people's opinions of ship balance based on what you yourself don't understand about combat strategies?

No, actually, putting ships in best possible scenario is the worst way to rate them because it will obfuscate their shortcomings.

The whole point of figuring out good ship loadouts, fleet compositions, combat strategies, etc. is to maximize your strengths and minimize your weaknesses. Nobody cares about how bad a shield-shunted Paragon is (except for meme purposes) or what happens if you don't put any missiles on a Gryphon (or a Pegasus). Ship balance isn't based around how bad your fleet is, it's based on how good you can make it. It's not about putting ships in the best possible scenario (which would be more like sic them against a Kite), it's about how to use them the most effectively. This is the most effective way I know of to kill enemy fleets, and that's why my testing is done in this way. Until someone posts a better way to handle combat, I don't see why you would choose a different, less-effective approach toward killing enemy fleets in evaluating different ships.

6
General Discussion / Re: Damage Bonuses
« on: March 19, 2024, 11:01:28 PM »
This is just a side effect of damage bonuses being partitioned two ways: by weapon type and by hull size.

Oh huh, I didn't notice that. Thanks for pointing that out, it makes a lot of sense. I haven't looked at the code but it makes sense that the code would do something like "sum up all the weapon type bonuses and then apply them as a multiplicative modifier" followed by "sum up all the hull size type bonuses and then apply them as a multiplicative modifier" etc. for other modifier types.

Targeting Feed uses the first partition, so it should actually be a "regular" bonus under your classification, unless there's something else going on that I'm missing.

Well it's based on testing in the sim, using a Heron with Dagger against the mid-tech practice target. The Dagger's damage against the shields (shield efficiency of 1.4) was:

700 without any damage bonuses
770 with 100% CR
1050 with Targeting Feed
1155 with 100% CR and Targeting Feed

That only works out if the CR bonus multiplies rather than adds with the Targeting Feed bonus. So it's not in the "regular" category.

I should note: It's not necessarily the case that Targeting Feed is part of the "special" category, since there were no bonuses in that category that I could use it with to confirm that they add rather than multiply. So it's possible that it's part of a different category altogether. It stacks multiplicatively with Graviton Beam so it's not in that category either. So all I can really say is that it's not part of the "regular" category (which is where the CR bonus resides) and it's not in the same category as Graviton Beam. For right now there's nothing in the "special" category that works at the same time as Targeting Feed so functionally it doesn't matter if it's lumped in together with them or in its own category.

I would expect there to be a third type as well - getDamageToTargetHullMult(), getDamageToTargetShieldsMult(), getDamageToTargetWeaponsMult(), etc.
So the "100% to engines and weapons" from elite TA would likely be (1 + Z). Willing to test for that?
Based on the code, Beam Scatterer would be in the (1 + X) group, while PD skills and terminator core in the (1 + Y) group.

Sure, I don't know how to directly see damage numbers for weapons or engines (or fighters or missiles to test the PD bonus) though. I think there was a mod somewhere along the line that will show weapon/engine health, but I can't find it offhand.

Yes, High Scatter Amplifier is part of the same group as HEF, tested using Sunder on mid-tech practice target.

All these categories stack multiplicatively, so yeah, there are a lot more than just 2 partitions, but I think the hull size and weapon type ones make up the majority of damage bonuses.

Yeah would be nice to list out all the possible damage bonuses and sort them by category, to know just which ones stack additively (within same category) or multiplicatively (across different categories). Not sure what other damage bonuses are out there though, hmm.

One thing that I just checked is that the damage bonus to shields from Graviton Beam does multiply with both the "regular" and "special" categories. Tested using Executor with Graviton Beam and Gigacannon against the mid-tech practice target (keeping the beam location on the shield separated enough to have two different shield damage numbers). So I assume that would be an example of getDamageToTargetShieldsMult(). So it's actually quite powerful then. Tested that 1 Graviton gives +5% overall damage bonus, while 3 Gravitons give +10% overall damage bonus, just like it says "on the tin".

7
General Discussion / Re: should we just nerf the Onslaught?
« on: March 19, 2024, 10:12:18 PM »
FYI, I have never seen anyone else's fleet do it. Even when Draba posts his own minmaxed comps which take 5 Ordo at a time this doesn't happen. So again let me reiterate the point that your specific fleet composition is completely bonkers and you should be extremely careful when trying to use it as basis for balance changes.

Uh, so you're saying that because you weren't aware of a combat strategy, therefore someone else should be "extremely careful" about considering it in balance discussions? I don't see why everyone else's opinions about balance should revolve around your own knowledge or ignorance of the different possible combat strategies in Starsector.

I've been posting about this way of defeating [REDACTED] fleets for literally years. All of my ship comparison testing is done this way, because it's the most effective way I know of currently. (In other words, it puts all the ships being tested in the best possible light.)  This is before I started using my current flagship Onslaught + 2 Gryphons + "other" paradigm. So it can be done with all sorts of different ships. My own 5-Ordos video was based on this exact strategy, done on the first try, and was the fastest completion posted by a significant margin. Whatever strategies other players use for their multi-Ordos videos is up to them. They're more about being able to say "yes I did it" than serving a practical purpose anyway, since there's no real reason to go beyond about 3 Ordos depending on your fleet, and the posters have to do some sort of FP shenanigans (inflate their fleet in terms of FP) in order to get that many Ordos fleets to be willing to fight them in the first place.

Different posters have posted about doing this concept of surrounding the enemy fleet so that you can focus all your fire on a few enemy ships in relative safety, since they're basically streaming in a few at a time; in fact there's a military term for this that someone else uses on the forums (that I only know because I didn't know what it was and had to Google it, but I forgot the term now). If you're not aware of this approach to combat in Starsector or other games, then it makes your own opinions about combat balance highly suspect. This is a pretty basic concept in a lot of these types of games.

8
Suggestions / Re: Hull restoration giving a 3rd s-mod instead of BOTB
« on: March 19, 2024, 07:39:39 PM »
I think that if we're going to have a fleetwide CR bonus besides crew training, then it really shouldn't be in the leadership tree because it would feel like it has too much of a monopoly on the mechanic. We have combat endurance in red, crew training in green, and then one of the other two trees should get the last bonus. Cybernetic augmentation for example could donate it's current +/- damage bonus to BOTB (thematically fitting but a bit weird in a tree with no combat skills, although that could be spun as a positive) and instead it could give 15% cr to ships with officers.

Yeah all good points. Seems like there are quite a few players who are on board with putting the 3rd s-mod onto Hull Restoration (or some other Industry capstone, or the Industry tree in general), which makes sense to me. Then it's a matter of what to put for the leadership capstone (alongside the extra deployment) and then for the tech capstone, to adjust things to be "about as strong".


9
General Discussion / Re: How is battle difficulty exp determined?
« on: March 18, 2024, 11:18:56 PM »
Does it matter what skills you have chosen, or is it just plain lvl 15?

For the player, it just counts the level, regardless of whether or not you've taken any skills. Whether skills are elite or not do not affect the XP bonus. For officers, they can stay at level 1 (or any other level) and build up extra XP, since they don't actually get a level until they take another skill, whereas the player character automatically gains a level as soon as the XP is gained.

10
General Discussion / Re: should we just nerf the Onslaught?
« on: March 18, 2024, 11:14:30 PM »
It has definitely felt subpar ever since 0.95, so I would like to know what is the secret ingredient to making it good again.

It's been good since at least 0.96a (and probably before) when I started testing player fleets against double Ordos using the paradigm of player-controlled flagship Onslaught as center anchor + 2 Gryphons (or however many is needed for the player fleet to add up to ~200 DP) as side flankers + whatever ship is being tested. The best option, in terms of fastest battle completion time, was + 6 Gryphons (i.e flagship Onslaught + 8 Gryphons), but + 3 Conquests (i.e. flagship Onslaught + 2 Gryphons + 3 Conquests) was the second fastest, faster than + 3 Onslaughts, + 3 Legion XIVs, or pretty much anything else other than more Gryphons. No particular secret, other than maybe sticking a PD in the same weapon group to force autofire on; the weapons were Squall/Locust/dual Mjolnir/dual HVD/dual Harpoons. In 0.96a I used IR Autolance + s-mod Expanded Magazines, but now in 0.97a I'm experimenting with other weapons there as well.

Half of Conquest is two large missile slots which have been nerfed about 4-7 times in last two patches(What was it - three Squall nerfs, EMR nerf, two Missile Specialization nerfs and a buff to Remnant PD?) It's weakest it's ever been, you're probably just getting too much noise from your flagship to even notice.

No, Squalls are important but they only form a small part of the Conquest's damage, like roughly 15%. Mjolnir is the biggest damage dealer, and I use two of them. In fact the shield part of the two Mjolnirs' overall damage is higher than the Squall's shield damage, but the Mjolnir also provides a lot of armor/hull damage while the Squall provides little else. Each Mjolnir provides around 1.5x of the Squall's overall damage.

My flagship Onslaught actually took a heavy nerf, largely due to the PCL nerf. It used to fire at once per second and elite Missile Spec giving +50% fire rate, so the Onslaught with 4 of them used to fire them at the equivalent of 6 per second. It now fires at once every 2 seconds with eMS only giving +25% fire rate, so now the Onslaught fires at the equivalent of 2.5 per second. So I can no longer haphazardly charge in, nor assume that the PCL will kill all incoming Reapers. My overall battle DPS has dropped from around 1600-1800 DPS in 0.96a to around 1200-1500 DPS now, roughly 20% less overall battle DPS. And this is using Cybernetic Augmentation for 12% or 14% extra damage on the flagship and less damage taken (depending on if I'm putting the 4th point into elite EWM or Neural Link).

In 0.96a, the Gryphon was at roughly 430 DPS for 20 DP, the highest ratio I found for ships under AI control against double Ordos, while the Conquest was at around 800 DPS for 40 DP. Since 0.97a, the Conquest's DPS has actually gone up. In fact the Conquest is having so little trouble that I'm experimenting with putting Pulse Lasers or Heavy Blasters in the medium energy slot alongside the above weapons just to encourage the Conquests to move in closer so that they don't idle at long range so much. I don't know if it's the new Cybernetic Augmentation (because the DPS increase is a lot more than what's expected from just +6% damage increase) or just that the new enemy fleet deploy order actually makes things easier for the Conquest and other player ships in general. At any rate, nowadays this fleet relies more on the Conquests relative to the flagship Onslaught to steamroll through double Ordos than in 0.96a. Not sure if double Ordos has gotten easier for other ships as well, but it's easier at least for the Conquest.

It’s very average in a lot of areas but I don’t think it’s “weak.” I just don’t think the endgame can be won by sustained firepower alone, which is what it excels at. You have to have some burst and/or decent defense to weather being swarmed.

Well the firepower is so high that my fleet basically advances toward the enemy spawn point at the top of the map, meaning that even against Ordos it's killing enemy ships faster than they can march in. The Conquests use Harpoon Pods linked to the Squall/Locust, meaning they'll fire at the beginning of the battle, which is the most important time to burn through the initial enemy fleet as quickly as possible. That's the only time when my fleet (which is only 200 DP) is facing the entire 240 DP of the enemy fleet. After that, my fleet is only facing around 1/3 of the enemy fleet (so, around 80 DP's worth of enemy ships) at any given time, with the reinforcements still on their way to the front lines. So it's basically, burst firepower initially, then sustained firepower for the rest of the fight. I posted an example of this in 0.96a here; in 0.97a the Conquests are doing even better.

11
General Discussion / Re: should we just nerf the Onslaught?
« on: March 18, 2024, 12:44:45 AM »
We never saw 5 Conquests beat 5 ordos. Tragic.

Clearly the Conquest needs to be buffed.

I'm laying low about the Conquest's abilities until I finish my playthrough and hope Alex doesn't notice how powerful it is.

12
General Discussion / Re: Is the Hephaestus at a good spot?
« on: March 17, 2024, 10:29:00 PM »
There are minor inaccuracies, but never noticed continuous beams being overreported.

Well the bug is not necessarily that frequent, so it may not necessarily show up in a couple of tests in the sim. But it shows up pretty often in full battles. For example, attached is a screenshot of DCR reporting an Eagle with a single Phase Lance doing over 124k hull damage and two IR Autolances doing nearly 53k hull damage in a double Ordos fight. This is a fight where, if there are no player ships using any beams, the total hull damage for the entire double Ordos fleet is usually around 320k or so, but this fight registered 588k of hull damage. Obviously a Phase Lance and two IR Autolances did not do 177k of hull damage during the fight. This is in version 0.97a-RC10. So yeah you can't count on the beam damage as reported by DCR.

I *might* do some testing of IR Autolance damage when I have time, to see if there's any way to get an accurate picture of what its actual in-battle damage output is. But as I mentioned the only way I have for estimating beam damage is statistical; basically, having done the same double Ordos fight multiple times with different fleet compositions (no beams), I know the average armor damage and hull damage that is usually done to the test double Ordos fleet. They appear independent of battle time, whereas shield damage is correlated with battle time (somewhere around 1k more shield damage for every extra second of battle time, when total fleet DPS is around 3000-5000 DPS; basically, the first 1000 DPS just goes toward double Ordos shield regen while the remaining DPS is what actually goes to overcoming their shields and then armor and hull). So in principle I can do a bunch of runs with say mass Eagles using mass IR Autolances, subtract out the damage coming from projectile weapons (which are pretty accurate), and then the remainder is what the IR Autolances did. But every run inherently returns different results so I have to do enough runs to get a good average to draw any conclusions.

I've thought about making a weapon with the same stats as IR Autolance but make it projectile-based so that DCR gives accurate results. However, projectile-based also means hard flux and I'm not sure if that would skew the results.

I think the effect is on shield damage. High range beams will often fire before the enemies they are hitting are able to fire themselves. As a result this will tick up shield damage but that shield damage will be swiftly dissipated away. So there are going to be a lot of instances where a Sunder will be ticking up 450+ shield dmg/second but like… actually doing nothing*.

That's actually a different issue, i.e. how much is a point of hard flux worth compared to a point of soft flux, when the enemy ship can dissipate soft flux easily but hard flux requires lowering shields (unless they have Field Modulation or a couple of other exceptions). The bug I'm talking about affects all damage types though as far as I know.

It’s not dissimilar to a ship that has only kinetics shooting an enemy with good armor that vents in their face and puts the shields up again. The shield damage number will be catastrophically high but the actual effect on the fight will be low.

No, because it's pretty inefficient DPS-wise to wait for the enemy ship to back off, vent, come back again, etc. (even if it's different enemy ships taking turns). It's actually faster to just go ahead and kill them. Sure, you get more total damage done, but your overall damage-per-second rate actually decreases. That's because the rate of enemy ship flux regen, considering that they need to back off to a safe distance (behind other ships if need be), vent, then reenter the front lines, is very low. Your other non-anti-shield weapons can easily do more damage to armor and hull in the same time that it takes for enemy ships to regen their flux.

I never trust damage statistics. Ever since I played Overwatch, realising people who would just gun a tank instead of sniping key targets would be always claiming high damage values as a justification for their skill. In the case of Starsector numbers, weapon's value.

Sure. If you have a better metric for comparing weapons, including a way for someone else to measure it (i.e. not just "I feel weapon A is better than weapon B just because", but a way where you define the "goodness" of the weapon, and someone else can conduct the same testing for that "goodness" and come to the same or similar conclusions), feel free to share it. Until then, not only is it pointless navel-gazing as Draba said, but it's also just an intellectually lazy way of rejecting someone else's evidence without taking the time to put forth a coherent argument.

In some sense, the only number that Detailed Combat Results gives that's actually "important" is the overall battle completion time. Everything else is there to help diagnose the reasons for that time. Sure, there are some situations where the numbers may be skewed in one way or another; for example, if there's a (player-controlled) Doom spewing mines everywhere, it makes the enemy ships point their shields in the wrong direction, leading to less total shield damage (since more of your fleet's damage will go directly toward armor and hull instead of shields).

But that's not at play here. We're just simply looking at comparing different weapons under the same (or similar) circumstances, i.e. just the bog-standard firing at targets. The AI isn't smart enough to snipe targets or whatever to "game" the results. It's not going give the complete picture, because no single metric does. But it reduces weapons down to an easily-understood metric that people can use as a basis for comparison and discussion about different weapons. Nobody claims that a car's miles-per-gallon is all you need to know about the car, nor that it's completely accurate, but it gives an easy basis for people to compare different cars.

Or put it another way: If you don't trust damage statistics of weapons, then on what basis do you form an opinion as to whether weapon A is better than weapon B? How do you determine which weapon to put in a given weapon slot without making any reference to how much damage it does? In Starsector, the only way to win a battle is to do enough damage to destroy opposing ships (or get them to retreat). Measuring the damage your fleet does seems like a natural thing to do in this context.

Oh, you mean in-game. Uhhhhhhhhhh, well it depends. Efficiency. Passive effects like Graviton beam shield increase. Forcing the operating of enemy PD can increase flux spending, reducing their ability to deter enemies with weapon fire.

I don't see how any of these examples have any coherent connection to why you don't trust damage statistics. All of it looks like random stuff that's vaguely battle-related but nothing specific to "this is why damage statistics are misleading", or are things which measuring weapon damage already accounts for. For example, what does projectiles forcing repositioning of the enemy have to do with you using a tactical laser? Poor strategy implementation would come out in -- surprise surprise -- the damage numbers: if your strategy is poor, then your damage numbers will be low. All this looks like random examples of stuff that happens in combat without any connection to your position.

13
Suggestions / Re: Hull restoration giving a 3rd s-mod instead of BOTB
« on: March 17, 2024, 07:47:50 PM »
Yeah I mentioned this last year about skill changes, and the reasoning is pretty much the same:

I would say, have BotB give the deployment bonus and the +5% CR per s-mod, while Hull Restoration gives the 3rd s-mod instead of +5% CR per s-mod. That conceptually fits better with what each skill tree is supposed to be about, and splits up the "overly convenient" benefits of deploying more ships and an additional s-mod into different capstones.

BotB is already very useful for the beginning deployment bonus, although I think it by itself is somewhat too light for a capstone. So the "other" effect could be more minor. I figured 5% CR per s-mod would make it interesting since min-maxers would still have to take both BotB and HR to actually get 100% CR without getting Combat Endurance and Crew Training. Thematically leadership giving more crew performance fits with what CR represents, while a third s-mod is more of a production thing that would fit the Industry tree better.

14
General Discussion / Re: Is the Hephaestus at a good spot?
« on: March 17, 2024, 12:03:35 AM »
I am not sure I would say that. It’s not exactly a “fast” armor stripper for very high armor compared to the best options in the game. But even if it’s doing minimum armor damage of 10%(skilled) the entire time it’s still doing 96 armor damage per second.

Oh I'm not saying Heph's not good at armor, I just mean that I think of it as anti-hull first, and anti-armor second. So it's more like "great anti-hull with pretty good anti-armor" as opposed to something that you get for the armor-breaking ability.

I am not sure why I cannot write this calculus (line intergrals was a long time ago) to get a theoretical armor kill time but like. Killing 2000 armor with one of these will take less than 20 seconds clearly. 10 seconds seems like a high estimate.

Well I did a derivation of it in the past (see here), assuming that each shot hits the same armor cell, and taking account the inner/outer cell dynamic. Attached is a graph of the time to fully strip armor for 1) Mjolnir, Heph, and Hellbore, using the approximate assumption of beam-like continuous DPS (as opposed to discrete shots), 2) Mjolnir and Heph plotting the number of discrete shots it takes for each armor rating in increments of 50 armor rating, and 3) Hellbore plotting the number of discrete shots it takes for each armor rating in increments of 1 (easy to find since it's just the breakpoints between 1 and 2, 2 and 3, etc.). You can see that continuous damage vs discrete shots doesn't really affect the solution that much, except in extreme cases like the Hellbore, so approximating it as continuous damage is usually a pretty good approximation (and makes it much easier to work with).

For certain values of armor, Heph actually beats Hellbore in time to fully strip armor. Of course, Heph also uses up more flux, but at this point, raw damage is going to be more important. You can also see that even at higher values of target armor, Heph takes only around 25% longer than Hellbore to fully strip armor. Plus this doesn't account for Target Analysis, (elite) Ballistic Mastery, etc.

The fatal flaw in this analysis is the assumption that all shots will hit the same armor cell. That's never going to happen, and different weapons will have a different spread. Mjolnir has a very high hit rate due to its faster projectile speed, with Heph not far behind, while the Hellbore's slower projectile speed and its wider spread means that a lot of shots will hit all over the target ship or simply miss. CapnHector's "probability wave" approach (in the threads here and here) accounts for different projectile hit distributions, and I think represents the most sophisticated analytical model of Starsector combat to date.

However, I've been going more toward analyzing the results from the Detailed Combat Results mod, since it represents actual, experimental data on how each weapon performs. That's where it comes out that the Hellbore's higher hit strength relative to the Heph is pretty much canceled out by its lower hit rate, so in practice they end up doing about the same damage per point of flux spent (this was in 0.96a, Heph is probably actually better now). And Heph's much higher DPS is what kills enemy ships faster, which is what you're looking for (the weapon set which maximizes the ship's kill rate of other ships, or alternately, minimizes the time-to-kill). Hence why I say, in practice, it's always been Mjolnir or Heph that ends up being the best in the large ballistic slot, at least thus far for the ships I've tested. Their high DPS is simply better than anything else you can put into that slot. Maybe the new Storm Needler is good, but then you'll have to find some good anti-armor and anti-hull weapons to complement it, and I'm not sure what you'd put in the other slots to make that work.

My bad, HB means Heavy Blaster to me. The comparison certainly seemed a bit random, but looked on brand after thumper and ACG :)

Oh, fair enough, however in the post you were replying to, "HB" was referring to Hellbore, heh. Another example where OoA (overuse of acronyms) leads to more confusion, not less.

Regarding the inner/outer cells, Vanshilar is 100% correct. However, it’s hard to eyeball a weapon’s performance when you have residual outer cells contributing fractions of the total armor well past the failure of the inner cells. All that to say, even though armor doesn’t work in way that I outlined in the guide or how it’s being presented above, it’s useful to estimate a weapon’s damage over time using the simplification and comparing it against other simplified examples.

Oh, I think the way you outlined it is correct, it's just that it didn't account for the inner/outer cell in some of the analysis. For 100% to 20% of the base armor, the results end up being the same (if multiplying the DPS by 80%). It's going from 20% to 0% armor (when the damage to inner cells are passing through to hull, while the damage to outer cells are continuing to hit armor) where it's different.

All that to say, the HAG is still doing the Lord’s work and doing it well, despite other options being equally good or better. I don’t compare the Hellbore and HAG against each other but I do make comparisons against the Mjolnir. They share some of the same space. Likewise, I don’t think the HIL or Plasma Cannons are direct competitors.

Yeah I feel like it depends on what the ship needs. Mjolnir is better than Heph at anti-shield and anti-hull, but Heph is better at anti-armor. Against [REDACTED], Mjolnir does something like 20% more DPS to hull, but Heph is something like 25% more flux efficient spreadsheet-wise. But in practice, high DPS leads to flux efficiency on its own, since the enemy ship has less time to generate flux and thus does less damage to you, plus has less of an opportunity to retreat and regen. So it's always a bit of a toss-up. On my flagship Onslaught I use center Mjolnir with Heph on the sides to get the best of both worlds in a sense; if I'm focusing on finishing off ships and flux is an issue, I can turn off my anti-shield weapons and the Mjolnir usually does enough shield damage to keep their shields down, while Heph is more flux efficient. But either one is usually pretty good.

Eh based on the discussion above I'll also include a graph using the continuous DPS assumption for time-to-kill, hitting same armor cell, for the Mjolnir (533 DPS, 400 energy hit strength), Heph (480 DPS, 120 HE hit strength), Plasma Cannon (750 DPS, 500 energy hit strength), Hellbore (250 DPS, 750 HE hit strength), and HIL (500 DPS, 250 HE hit strength). No bonuses of any kind. It's noticeable that up to around 1000 armor or so, the Heph, Plasma Cannon, and Hellbore all have similar times. HIL is best but it doesn't do much to shields since it's a beam (soft flux). In theory, the ballistics will actually do a bit better by comparison since they can have (elite) Ballistic Mastery.

Tachyon is more difficult to analyze due to the scripted damage. For what it's worth though, I took an Executor, used a practice target that's 1750 armor, 100k hull, stuck a Tach and a HIL on it, so there were 3 distinct damage areas: the Tach hit area, the HIL hit area, and the Tach's scripted damage which hit at the center of the practice target. None of them overlapped so they were totally separate. No bonuses of any kind. Firing both weapons simultaneously and continuously, when the practice target blew up the HIL had done about 52k hull damage. So the Tachyon's main beam (hitting its armor area then hull underneath) and the Tachyon's scripted damage (hitting its own armor area in the center and then hull underneath) combined did around 48k hull damage. So pretty close to the HIL in overall damage output. However, on an actual target, that scripted damage hits all over the place so it's hard to account for it analytically (it'll waste a lot of its damage on hitting armor all over the place, but it'll also disable weapons and engines which help quite a bit), and unfortunately, Detailed Combat Results does not report beam damage accurately, so it's hard to know if it's better or not.

Since this has turned kind of into a weapon balance thread, I'm going to throw out a hot take: I think the IR autolance is actually slightly overtuned right now! Gravitons are in a good place for what they are, but that hull melting, intelligent AI, instant hit blamo is just really good.

Even hotter take: The IR Autolance + smodded Expanded Magazines is in a pretty decent spot, but Detailed Combat Results' buggy reporting of beam damage (usually inflating beam damage) makes people think it's better than it actually is. Oh I use it all the time, I think usually I use it more than Graviton, but I sort of ballpark it as being roughly 1-2 times the armor/hull damage of an HVD (since I use HVD pretty frequently, this lets me ballpark what the actual damage likely was from DCR), and pretty-close-to-zero shield damage since it's soft flux. It's hard to tell though since DCR results are inaccurate for beams; I haven't found a good way to test it. In theory I can gather it statistically (compare battle results of no-beam runs with battle results of many IRAL) and infer the hull damage dealt, but that takes a lot of work.

In 0.96a my Conquests used s-modded Expanded Magazines just for the one IRAL on their medium energy slot, to help kill off smaller targets more quickly. That's a good complement to their projectile weapons.

15
Blog Posts / Re: Simulator Enhancements
« on: March 13, 2024, 02:32:55 PM »
Nope, and also not the Guardian. I think those are the only vanilla ships that are excluded. :)

...what about the mothership?

Edit: wanted to mention, Java upgrade sounds good! Yes it's good for 32-bit compatibility, in case people are still using it; I myself play on a desktop that is over a decade old (though it's 64-bit), so there are probably quite a few people who are managing to squeak along on older computers.

Pages: [1] 2 3 ... 39