Print

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

[Alex]

Hm, just found out that Devastator is worthless against some enemies. Proximity fuse is triggered by enemy collision radius, so if enemy long and thin ship faces you with broadside, it will be immune to Devastator (because collision radius is far enough from it's hull). Even somewhat fat Conquest qualifies, and mod-land is full of needle-like ships.

Quick note on this - here's what the proximity fuse trigger area for ships looks like:



Full size: https://i.imgur.com/xJDje29.png


The circle is the collision radius, the sort-of-oval (actually 4 90-degree oval sections) is the actual trigger area. Ignore the rectangles; those are supposed to be the area of the sprite but the code rendering those only works properly when the ship is pointing up, so it looks off.

Given that the range of the Devastator's explosion is 50, and its trigger radius is 30, anywhere there's more than 20 pixels between the oval and the collision bounds of the hull, it'll detonate but not hit. There's indeed a dead zone on the sides, but it's not too extreme - if, say, you've got a pair of Devastators on the Dominator *trying* to fire at that dead zone, a lot of the shells will hit anyway.

The AI also uses the same oval approximation for rangefinding and such.

As long as the sprite size isn't excessive compared to the ship, the oval approximation should work decently well for long/thin ships. But if, say, the sprite has 100 pixels of padding on each side, then that could cause it to be way off.

(Thinking about it now, it should probably figure out a width/height based on the collision bounds, not the sprite, but, well, it doesn't. There are other reasons not to have over-large sprites, anyway.)

[TaLaR]

@Alex
Interesting, well, I didn't get all the details right.
Still, last thing you'd expect is missing a Capital point-blank. Definitely going to use Devastator way less than before.

EDIT: I suppose, it's not that much a problem, when you know you can't hit Conquest center and actively avoid firing such areas.

[Alex]

I guess it depends on what you're going up against. I will say that missing shots like this by proximity weapons with a narrow-enough explosion/trigger band is more or less "priced in".

[Wyvern]

Suggestion: When a proximity device with a blast radius greater than or equal to its fuse radius detonates, the ship that triggered that detonation should always get hit by it, without any further bounds-checking.

[Alex]

Hmm, yeah. Let me make a note to look into it. It gets a bit tricky since it'd still need to go through the proper collision-checking pipeline to see whether it hits shields, where it hits on the armor grid, and so on.

[xenoargh]

I haven't gotten to playtest much, or read the current list of replies.  I think it's fair to say that this still needs some tweaking.  After viewing in the context of Rebal last night, it went some weird places until I tuned things a bit, and it's still not right yet, but the systems are generally working as expected.  I think I'll have it producing sensible results sometime this weekend.

[intrinsic_parity]

1. I disagree. Midtech ships have both energy and ballistic slots so these weapons are competing within the weapon load out. If energy were much worse than ballistics, it might even be optimal to leave energy slots empty. This is sometimes true even now. I would rather leave medium energy slots empty on an eagle than put pulse lasers in them because the range is so bad compared to the ballistics and the ship doesn't have the dissipation to use them effectively anyway. Additionally, ships with energy weapons fight ships with ballistics so they must be balanced with respect to one another to some degree, even if the ships that use them are not on equal footing wrt flux stats.

Mid tech ships almost never have hybrid or universal slots on them and you can easily compare between if you want to without adding weights.

You should definitely not leave the medium slots on your Eagle empty. Pulse lasers may be inefficient at fleet range but if you do not want a high damage fallback weapon you can always put Gravitons in

I think you missed the point?
The OP was suggesting that weapons should be balanced within their respective damage types only and I was pointing out a situation where inter-damage type balance is relevant since on mid-tech, weapons compete for the same ordinance points if not the same mounts. Idk what you are talking about with weights?

When I was talking about the pulse laser, I was giving an example of where energy and ballistic weapons compete for the same ordinance points. The eagle doesn't have the flux stats for 6 medium hard flux weapons (assuming the ship is under AI control so all weapons are firing simultaneously) so I must choose what weapons to use. This was not saying I leave medium mounts empty, just that I would rather leave them empty than use a pulse laser, which could indicate balance issues.

[xenoargh]

1) there is a direct tradable equivalence (like say, OP for Flux dissipation)

Well, that's how it works, actually.  Everything gets compared on an OP basis:

(((AV7/$T$3)+(F7/$T$2)))/L7)  in the final series of operations is largely dominant of the results.  

Translated:

((kill_power / damage_div) + (range / range_div)) / OPs

Why in that order of operations?  Because kill_power's a pretty complex synthesis that weighs overall efficiency of damage vs. Flux vs. overall TTK vs. Flux, after miss rates are taken into consideration, bursts are factored in, etc., etc. etc.- I'm afraid that's still a little more "black box" than I'd like, but I've tried my best to expose everything so that the logic's visible, along with the weighting.

Range in the last equation because it is the most important stat of all, and pricing it correctly in relation to kill_power is very tricky, and it cannot be included in earlier parts of the synthesis in a meaningful way, other than as a factor in calculating the hit_percent (i.e., smaller angular miss rates matter a lot more at long ranges).

So, the coin of this land is OPs.  OPs spent on any given weapon can't be spent on other weapons, Flux or Hull Mods.  It's the only token that's actually worth using as a comparator.


@intrinsic_parity:  Agreed on all points.  Universals and Hybrids are a thing.  We need to quit trying to create "balance" based on game conditions that haven't really pertained since SS 0.5... uh... 5 years ago?

[Blothorn]

I think you missed the point?
The OP was suggesting that weapons should be balanced within their respective damage types only and I was pointing out a situation where inter-damage type balance is relevant since on mid-tech, weapons compete for the same ordinance points if not the same mounts. Idk what you are talking about with weights?

When I was talking about the pulse laser, I was giving an example of where energy and ballistic weapons compete for the same ordinance points. The eagle doesn't have the flux stats for 6 medium hard flux weapons (assuming the ship is under AI control so all weapons are firing simultaneously) so I must choose what weapons to use. This was not saying I leave medium mounts empty, just that I would rather leave them empty than use a pulse laser, which could indicate balance issues.

I would rather leave slots empty than put three Plasma Cannons on an Odyssey--I do not think that points to balance problems but merely the fact that some weapons are situational. The pulse laser is a bit of a flux-dump weapon (albeit not as much of one as the Heavy Blaster); the Graviton Beam is an efficient support weapon. I would almost never put a PL on an eagle--but I would also almost never put a GB on an Aurora.

[Blothorn]

The circle is the collision radius, the sort-of-oval (actually 4 90-degree oval sections) is the actual trigger area. Ignore the rectangles; those are supposed to be the area of the sprite but the code rendering those only works properly when the ship is pointing up, so it looks off.

Given that the range of the Devastator's explosion is 50, and its trigger radius is 30, anywhere there's more than 20 pixels between the oval and the collision bounds of the hull, it'll detonate but not hit. There's indeed a dead zone on the sides, but it's not too extreme - if, say, you've got a pair of Devastators on the Dominator *trying* to fire at that dead zone, a lot of the shells will hit anyway.

The AI also uses the same oval approximation for rangefinding and such.

As long as the sprite size isn't excessive compared to the ship, the oval approximation should work decently well for long/thin ships. But if, say, the sprite has 100 pixels of padding on each side, then that could cause it to be way off.

(Thinking about it now, it should probably figure out a width/height based on the collision bounds, not the sprite, but, well, it doesn't. There are other reasons not to have over-large sprites, anyway.)

Somewhat off-topic, but is the oval approximation exposed in the API? It looks like it could be useful for modded system AIs and such that are doing rangefinding.

[xenoargh]

is the oval approximation exposed in the API?

Yes.  

It's a little more expensive to access than the circle, so it needs to generally used a little more circumspectly; usually it's best-tested after the circle, but before testing against the collision segments.

OK, gotta go again.  I'll be back when I've gotten a chance to plug away at the guts of the thing s'more.  It's not perfect, but I don't think I'm tilting at windmills.  Thanks to everybody for keeping this (relatively) drama-free; I think this is a solvable problem, it's just Hard.

[intrinsic_parity]

That's why I said 'could'. Although I can't actually think of a late game ship load out I would intentionally use with a pulse laser. I wouldn't mind it's current state except that there's not really another option for the same role. Heavy blaster is more of an armor cracker and costs way to much flux to use on anything smaller than a cruiser (under AI control without an officer). Graviton deals soft flux. There are no other brawl/assault type energy weapons. I would rather see a new energy weapon added that is a high tier assault energy weapon. Pulse laser is like an ablest, it's usually serviceable but highly mediocre, however there are multiple other medium kinetic weapons to use instead of arblest that fill the same role. Pulse laser is fine as a cheap easily available assault weapon, I just want other better options for late game.

[Philder]

@xenoargh

I should note that while I read your OP and checked out your spreadsheet I didn't read any other comments so I apologize if I'm repeating anything. I've been making efforts in the same direction as you but I'm slightly further along so I'm just commenting to give you advice and constructive criticism. Only generally, though, because a whollistic argument would be far beyond how much work I want to put into this. Sorry. (also, the numbered points do not correspond to your own points)

1. Okay, first off! Big thing: You MUST account for the dynamics of armor and hull. There are no 'ifs' 'ands' or 'buts' about this. You MUST. If you don't, then all your work is wasted. Armor is highly dynamic and can make a difference of over 600% in the actual damage caused to armor hp. Even hull, at only 5% of max armor, makes a huge difference in outcomes. In addition, there is also an interesting hidden dynamic that I haven't yet heard anyone ever mention, and that's the hidden positive efficiency of increased cascading effect of lower shot damage against armor; as an example, lets just say weapon A does 1000 dmg per shot and weapon B does 500. Weapon A will do a single shot at, lets just say, 1000 armor, but weapon B will do two shots; one at 100 armor and a second against the now lowered armor, doing more damage to armor hp than the first one. In addition to that, you also have to take into account armor overkill, when the armor is finally penetrated and the leftover damage is applied to hull. That will affect your armor-only calculations.

2. Although on the surface TTK is highly useful, it's too distilled and isolated compared to the complete picture of all that's going on. It's more of a very rough rating of armor dps, but IMO it's not all that useful because it doesn't take into account other significant stats. I'm still working on a whollistic evaluation of these things so I don't really have anything else to add that won't take more work explaining than I'm up to.

3. OP is not a good metric to measure everything against. Different mount sizes as well as types change the meaning of a given amount of OP. There are too many weapons that have no equivalents across the mount sizes, some of which are so unique as to be completely unequivocal. Rather than OP costs, it's simply a matter of whether you need it or not. Same for mount types.

In addition to that, in Vanilla, which ship these weapons will be going on is also a huge factor on their effects. As a simple example, energy weapons usually go on hightech ships with much higher base flux stats and move speeds.

4. Weapon spread isn't always a bad thing. The AI's predictive target leading is based on the targets current vector, so a ship can evade it by simply changing its vector...which they do if they have the mobility to gets its hitbox out of the way. Weapon spread, however, can 'accidentally' aim a shot at the targets evaded position, make weapons with spread more effective against more evasive targets; rather than a 0% hit chance from a perfectly accurate weapon, a very inaccurate weapon may have 15% hit chance.

5. In addition to what I've already mentioned, you are missing a fair few other important considerations. Some are rather more abstract and meta. Sorry I can't detail them all for you.

6. Hint Hint: You don't actually require a solid baseline. You can create a comparison for a single member within a set against every other member within that set, for every member within that set, and then compare those comparisons with each other. Poly-order meta


Finally, I just want to say that I think it's great you're working on this stuff. You've still some way to go, but you're getting there. Keep up the good work.

[Goumindong]

1) there is a direct tradable equivalence (like say, OP for Flux dissipation)

Well, that's how it works, actually.  Everything gets compared on an OP basis:

(((AV7/$T$3)+(F7/$T$2)))/L7)  in the final series of operations is largely dominant of the results.  

Translated:

((kill_power / damage_div) + (range / range_div)) / OPs

Why in that order of operations?  Because kill_power's a pretty complex synthesis that weighs overall efficiency of damage vs. Flux vs. overall TTK vs. Flux, after miss rates are taken into consideration, bursts are factored in, etc., etc. etc.- I'm afraid that's still a little more "black box" than I'd like, but I've tried my best to expose everything so that the logic's visible, along with the weighting.

Range in the last equation because it is the most important stat of all, and pricing it correctly in relation to kill_power is very tricky, and it cannot be included in earlier parts of the synthesis in a meaningful way, other than as a factor in calculating the hit_percent (i.e., smaller angular miss rates matter a lot more at long ranges).

So, the coin of this land is OPs.  OPs spent on any given weapon can't be spent on other weapons, Flux or Hull Mods.  It's the only token that's actually worth using as a comparator.


@intrinsic_parity:  Agreed on all points.  Universals and Hybrids are a thing.  We need to quit trying to create "balance" based on game conditions that haven't really pertained since SS 0.5... uh... 5 years ago?

If kill power is a “complex synthesis” it either needs to generated by measurement (point 2) because it’s not a direct tradable equivalence

Kill power only make sense if you can directly trade TTK for DPS or Flux at an established rate for all weapons

This is why adding OP works for flux efficiency in the other thread. You can trade 1 OP for 10 flux/second so long as you have non-maxed vents.. So a weapon that costs 9 OP vs a weapon that costs 10 OP has an effective reduction in firing cost by 10 flux/second.  This is a direct trade off that applies as we fit the weapon and is consistent across all ships.

Kill power does not do this.

If you are pricing weapon qualities by hand (and adjusting them!) then you done f’d up. You will be ordering your list rather than measuring an order or value. The final product isn’t an evaluation of what is best based on the numbers it’s numbers based on your evaluation of what is best.

[xenoargh]

OK, got the latest version done.  I'm much happier with this one; I ended up with a good way to reconcile kill_power with range and it finally came together well.  

I've included, on a separate sheet, the stat-lines for the Light Kinetics from Rebal after this was applied; they differ subtly from Vanilla in various ways, but aren't startlingly different after getting everything really near par.

A couple of highlights, just demonstrating how the systems work:

1.  The Railgun paid for 100 more range and more efficiency with a little inaccuracy at maximum range.

2.  The Dual AC paid 1 OP to gain a little more kill_power and a little more accuracy than the Light AC.   Which one's better?  Situational; the Light AC's better for Flux-starved builds, but the Dual's probably better for SO themes.

3.  The Light Needler's significantly nerfed; it both misses a bit more and it costs more Flux to fire.  It has superior kill_power to the Railgun and pays for it mainly by spending 2 more OPs.

4.  The LMG / Dual LMG are actually significantly different, and are good case studies for what happened to the outlier cases. 

Essentially, at 3 OPs, the LMG has to pay for its relatively-good firepower in multiple ways.  It's less-accurate, has a lower kill_power value (timed bursts vs. constant fire) and it costs 1 more Flux to fire.  It's simply not as good.  I thought about cutting its damage and thus kill_power quite a lot, but better to leave it sometimes luckily useful as PD rather than universally useless (15 damage is pretty key; a burst of 6 shots from a LMG that actually connects will kill the lighter rockets, but not a Harpoon).

So, if you want a Kinetic PD weapon that's reliable... DLMG.  If you want a less-reliable one that can be used as an efficient belly-gun offensively, the LMG's still OK.