Fractal Softworks Forum

Please login or register.

Login with username, password and session length
Pages: 1 [2]

Author Topic: Basic AI flux management error  (Read 6989 times)

TaLaR

  • Admiral
  • *****
  • Posts: 2794
    • View Profile
Re: Basic AI flux management error
« Reply #15 on: April 21, 2018, 10:55:21 PM »

This seems to also break scenario where faster ship wins slightly in every engagement, but needs to retreat-vent after (since taking armor/hull damage allows victim to somewhat recover position in terms of flux war and I do not want to trade armor/hull).
Which is a very common thing I do, since faster and smaller ships usually simply can't win against a larger opponent in single engagement (at acceptable cost).

100 bonus is also really a lot. I don't think I'd be able to maneuver around several enemies like that with Medusa for example. Or even most frigate vs DE scenarios.

I think it gives too much advantage to slow and heavy ships or side with numeric advantatge. Basically, every engagement against larger opponent is like getting pushed by a super-maneuverable Onslaught (unlike it's current straight and fixed burn).

Near instant nature of boost (no need to wait flux to drop) also raises difficulty bar for correctly using it (both player and AI).

And I guess, I just like the fact that even after you lose engagement round flux-wise, it's not the end as long as you can orderly disengage. To me the wide gap between barely a win and completely overwhelming an opponent doesn't seem like a bad thing.
« Last Edit: April 21, 2018, 11:58:08 PM by TaLaR »
Logged

TaLaR

  • Admiral
  • *****
  • Posts: 2794
    • View Profile
Re: Basic AI flux management error
« Reply #16 on: April 22, 2018, 06:15:38 AM »

The basic approach is it doesn't try to be perfect, but uses a sliding scale that includes its flux, enemy flux, and its soft flux, vs the shield-damage-factor, with a personality modifier thrown in for good measure.
Now this may be somewhat deeper, but can think of one more factor:

Efficiency of enemy weapons against my shields, with extreme case being Hammerhead during Accelerated Ammo Feeder activation. It's weapons are temporarily so effective, that firing almost anything at it is better than letting it spend flux on shooting (as long as it's combined soft flux from firing and incoming beams is above vent rate, which is almost always the case).
Logged

Megas

  • Admiral
  • *****
  • Posts: 12157
    • View Profile
Re: Basic AI flux management error
« Reply #17 on: April 22, 2018, 07:06:42 AM »

* half applies when shield/phase cloak is off
So, for phase ships, instead of moving forward three times faster, the phase ship moves slower due to losing top speed.  In that case, bring back the old cloak so that phase ships can actually brawl, and AI not wasting peak performance.  AI dancing ineffectively and running out of gas is great when the enemy does it, but not when my side does it.  New cloak is great for Quantum Disruptor or invulnerability-frame-kill shenanigans.  For anything else, it is annoying.

As for the rest, may hurt some high-tech.  Thanks to how short-ranged and inefficient energy weapons are, it is hard for some high-tech ships (with max vents and near max capacitors) to win the flux war and have enough flux left to hurt the enemy.  Aside from Paragon, they seem built to rush-in, dump flux, take a small bite out of armor/hull, then withdraw.

Triple-lance Odyssey might be worse, if enemies approach faster than Odyssey can back-pedal.  Thanks to shield nerf, it needs to kite and hang back to outrange enemies and snipe with triple lances.  If it is forced to brawl, it dies.  Terrible shot range (with non-lances), and terrible shield (and defenses), and mediocre firepower for its size, Odyssey cannot brawl like other capitals, including other so-called battlecruisers.
« Last Edit: April 22, 2018, 07:26:41 AM by Megas »
Logged

Thaago

  • Global Moderator
  • Admiral
  • *****
  • Posts: 7214
  • Harpoon Affectionado
    • View Profile
Re: Basic AI flux management error
« Reply #18 on: April 22, 2018, 10:01:28 AM »

Re: 0 Flux bonus potential changes

Not being tied to flux, but rather half shields half weapons, I do not see helping to stop a flux high ship from retreating with just a little damage to then rinse/repeat. Reason: ships that are high on flux and loosing drop their shields and stop firing, while ships that are winning keep theirs up. Even with a penalty to backing up, this will help overwhelmed enemies disengage. Diverting power away from weapons/engines to shields 'makes sense', but I don't think it works with this design goal.

It also widens the scenarios of "jeez, if the AI just lowered its shield and took a few hits it would get away", which is currently true only for tricking it to put up its shields in chase scenarios. I think you'd really need to work on the shield/weapon AI to take competent advantage of the speed bonus.

In regards to the bonus applying more towards forward motion than backwards: The only time I can really think of the enemy moving backwards at 0 flux is when kiting outside of weapons range against a superior enemy, as with a frigate swarm, or when a battle line of destroyers scatters away from a charging Onslaught. This I think would be an anti-kiting measure that gives initiative to the attacker - I'd have to play to see, but it might help when being outnumbered by frigates in a cruiser, which is an annoying reality given the AI fleet distributions.
Logged

Cyan Leader

  • Admiral
  • *****
  • Posts: 718
    • View Profile
Re: Basic AI flux management error
« Reply #19 on: April 22, 2018, 10:12:24 AM »

What would this break?

Retreat missions.
Logged

Alex

  • Administrator
  • Admiral
  • *****
  • Posts: 24116
    • View Profile
Re: Basic AI flux management error
« Reply #20 on: April 22, 2018, 10:18:28 AM »

Re: movement mechanic - thank you for the feedback! The AI stuff seems the most troublesome, honestly; what it can get away with as far as 0-flux it can't here since it'd be a too-common in combat consideration.


Now this may be somewhat deeper, but can think of one more factor:

Efficiency of enemy weapons against my shields, with extreme case being Hammerhead during Accelerated Ammo Feeder activation. It's weapons are temporarily so effective, that firing almost anything at it is better than letting it spend flux on shooting (as long as it's combined soft flux from firing and incoming beams is above vent rate, which is almost always the case).

Yeah, I think that's getting too deep :) Especially since stuff like light machine guns - theoretically at the extreme of that range - are nonetheless useless to use that against since their flux cost is *too* low.
Logged

SafariJohn

  • Admiral
  • *****
  • Posts: 3021
    • View Profile
Re: Basic AI flux management error
« Reply #21 on: April 22, 2018, 03:25:59 PM »

I could see the 0-flux speed boost being reworked to a 0-flux travel boost. Let it give a huge speed bonus, but also penalize acceleration/deceleration. Would that be abusable?

The most immediate trouble I see is that shield-less ships and phase ships would be unable to deactivate it by raising shields like normal ships could. That can be solved with a new control key, but that's its own problem...



Spitball idea: give ships a "tactical engine" type specified in their codex entry that describes their 0-flux or whatever boost. Some ships get the 50 speed boost, some get the above, some get some other thing.
Logged

xenoargh

  • Admiral
  • *****
  • Posts: 5078
  • naively breaking things!
    • View Profile
Re: Basic AI flux management error
« Reply #22 on: April 22, 2018, 04:23:37 PM »

Quote
The basic approach is it doesn't try to be perfect, but uses a sliding scale that includes its flux, enemy flux, and its soft flux, vs the shield-damage-factor, with a personality modifier thrown in for good measure. The sliding scale means there's a high tolerance to doing something inefficient initially but that drops quickly as the flux situation changes, meaning there's a low probability of certain weapons *never* getting fired, but it won't keep firing inefficient weapons for long.
That's a pretty cool way to go about that.  I was managing this largely by preventing high-flux weapons to fire once Soft Flux exceeded about 65%, which let the small stuff keep going a little bit, unless the enemy was getting pushed into near-Overload, and then allowing it. 

But I definitely like this idea better, since it'd prevent the AI from constantly screwing up with high-Flux weapons.  That's the biggest weak spot the AI has, and building around it is pretty fundamental to current gameplay.  I don't think there's any perfect way to handle the > 50% issue; that's... finicky, and players get it wrong, too.
Logged
Please check out my SS projects :)
Xeno's Mod Pack

Thaago

  • Global Moderator
  • Admiral
  • *****
  • Posts: 7214
  • Harpoon Affectionado
    • View Profile
Re: Basic AI flux management error
« Reply #23 on: April 22, 2018, 05:50:48 PM »

Quote
The basic approach is it doesn't try to be perfect, but uses a sliding scale that includes its flux, enemy flux, and its soft flux, vs the shield-damage-factor, with a personality modifier thrown in for good measure. The sliding scale means there's a high tolerance to doing something inefficient initially but that drops quickly as the flux situation changes, meaning there's a low probability of certain weapons *never* getting fired, but it won't keep firing inefficient weapons for long.
That's a pretty cool way to go about that.  I was managing this largely by preventing high-flux weapons to fire once Soft Flux exceeded about 65%, which let the small stuff keep going a little bit, unless the enemy was getting pushed into near-Overload, and then allowing it. 

But I definitely like this idea better, since it'd prevent the AI from constantly screwing up with high-Flux weapons.  That's the biggest weak spot the AI has, and building around it is pretty fundamental to current gameplay.  I don't think there's any perfect way to handle the > 50% issue; that's... finicky, and players get it wrong, too.

I will second players getting it wrong - I've been doing a Falcon + Phase Beam run and I've screwed up the flux plenty of times and gotten myself spanked. Thank goodness the ship actually has some armor and hull :P.
Logged

TaLaR

  • Admiral
  • *****
  • Posts: 2794
    • View Profile
Re: Basic AI flux management error
« Reply #24 on: April 22, 2018, 08:52:14 PM »

Now this may be somewhat deeper, but can think of one more factor:

Efficiency of enemy weapons against my shields, with extreme case being Hammerhead during Accelerated Ammo Feeder activation. It's weapons are temporarily so effective, that firing almost anything at it is better than letting it spend flux on shooting (as long as it's combined soft flux from firing and incoming beams is above vent rate, which is almost always the case).

Yeah, I think that's getting too deep :) Especially since stuff like light machine guns - theoretically at the extreme of that range - are nonetheless useless to use that against since their flux cost is *too* low.

Is it useless though? Once enemy is within LMG range the only options to are to run away (if you are faster), overload/disable their guns or eventually die.
Letting LMG hit shield without fighting back is likely the worst usage of flux at this point (especially if their KE shots are not mixed with enough HE). The earlier you go out with all weapons, the better (while armor lasts).
Logged

Wyvern

  • Admiral
  • *****
  • Posts: 3803
    • View Profile
Re: Basic AI flux management error
« Reply #25 on: April 23, 2018, 09:57:51 AM »

I'm really excited to see this in-game... and I'm also very curious what it does for ballistic-based ships; those tend to be much harder to make in AI-friendly flux-neutral builds.
Logged
Wyvern is 100% correct about the math.

Alex

  • Administrator
  • Admiral
  • *****
  • Posts: 24116
    • View Profile
Re: Basic AI flux management error
« Reply #26 on: April 23, 2018, 10:12:47 AM »

Is it useless though? Once enemy is within LMG range the only options to are to run away (if you are faster), overload/disable their guns or eventually die.
Letting LMG hit shield without fighting back is likely the worst usage of flux at this point (especially if their KE shots are not mixed with enough HE). The earlier you go out with all weapons, the better (while armor lasts).

Well, no-one said to keep the shields on :) It's useless in the sense that it's not going to stop the LMGs from firing. A possible best use of the flux at that point is to fire whatever is efficient-ish to take down the shields and then use heavy hitters like the HB vs armor.

This may not always be the best approach, but it will be in some cases. Point being that this is not a great metric because it's not clear-cut.


I'm really excited to see this in-game... and I'm also very curious what it does for ballistic-based ships; those tend to be much harder to make in AI-friendly flux-neutral builds.

Got any ballistic loadouts that work well in player hands but don't work for the AI because of flux management? Would love to give one a whirl.
Logged

Wyvern

  • Admiral
  • *****
  • Posts: 3803
    • View Profile
Re: Basic AI flux management error
« Reply #27 on: April 23, 2018, 11:18:06 AM »

Got any ballistic loadouts that work well in player hands but don't work for the AI because of flux management? Would love to give one a whirl.
Almost anything with an assault chaingun or storm needler?  Probably some builds with gauss cannons, too.  I'll have to do some testing this evening, come up with actual specific variants.  (Though, the stock SO-Hammerhead design is probably a good starting point - not to claim that it works 'well' in player hands, as 2x chaingun is silly - but it certainly works better for the player than the AI.  1x chaingun 1x HMG would be a much stronger variant, imo.)
Logged
Wyvern is 100% correct about the math.

Alex

  • Administrator
  • Admiral
  • *****
  • Posts: 24116
    • View Profile
Re: Basic AI flux management error
« Reply #28 on: April 23, 2018, 02:20:12 PM »

Hmm, that one doesn't seem like a great example because its flux generation is below its dissipation rate.
Logged
Pages: 1 [2]