Fractal Softworks Forum

Starsector => General Discussion => Topic started by: angrytigerp on March 25, 2019, 12:11:39 PM

Title: Can someone explain the (invisible-ish) "autoaim effectiveness" stat?
Post by: angrytigerp on March 25, 2019, 12:11:39 PM
I'm really, really unclear on how the Gunnery Control Implants skill with "+% to autoaim effectiveness" (or, if you want, the mod-added Gunnery Control AI hullmod that basically gives you a minor boost in every category GCI covers, including the nominal autoaim effectiveness stat) affects gameplay.

Whenever I use the weapon arcs mod to draw autoaimed groups (or use the vanilla method of selecting a group, then selecting an empty group, still drawing the last group's arcs), the aim line appears to track with the same mechanical computer-controlled precision as it ever does.

Perhaps I'm not paying close enough attention, but does the aim point "lag behind" and take longer to reacquire/retrack a target's changes in velocity or something more with a lower autoaim stat? That's was I'm guessing it means but... Right now, the perk just sort of exists as an awkward in-the-way part of the GCI skill, which I otherwise want for projectile speed and weapon range.
Title: Re: Can someone explain the (invisible-ish) "autoaim effectiveness" stat?
Post by: Kittah Khan on March 25, 2019, 12:22:40 PM
It's to have the weapons that fire automatically(non-selected weapon groups with autofire) correct for player and target movement.

By default, weapons autoaim at the target's current position. Causing shots at moving targets to miss.
Improved autoaim accuracy shifts the position the weapon is aiming for towards the interception point.
Title: Re: Can someone explain the (invisible-ish) "autoaim effectiveness" stat?
Post by: Alex on March 25, 2019, 12:25:36 PM
The way it works is when shooting at a target, an auto-firing weapon will  initially lag behind, but - with more time spent firing at the same target - it will gradually catch up.

A better "target leading accuracy" both reduces the amount of initial "lag" and the time it takes to catch up.

Note that CR also factors in - at 100% CR, the initial aim is at the best possible quality, and likewise if the ship's captain has Gunnery Implants. So, having both is overkill, but GI helps once CR gets below 100%. Integrated Point Defense AI also makes PD weapons have the best possible aim under all circumstances.

For AI captains, the "target leading accuracy" also affects their aim as far as pointing the ship itself at a target, with a separate value tracking how long they've been aiming at the same target, and with the aim gradually improving (if possible, i.e. if it didn't already start out the best possible due to GI etc).

If you're curious, the way "lagging behind" works is by underestimating how long it'll take for shots to reach the target, and by gradually improving this estimate. As a consequence, beam weapons are not affected by aim quality, being essentially hitscan. The worst possible aim quality, iirc, assumes shots are twice as fast as they actually are.

(Edit: oops, partially ninja'ed...)
Title: Re: Can someone explain the (invisible-ish) "autoaim effectiveness" stat?
Post by: angrytigerp on March 25, 2019, 01:06:20 PM
Thanks for the replies, I think I get it. Also, sorry I used like all the wrong terms and values :(

So, to read it back in my own words to make sure I'm understanding correctly:

If a ship has perfect "target leading accuracy" (between CR, skills, and hullmods, etc.), then it will fire when and only when the weapon is aligned in such a way that a hit would be guaranteed, assuming the target maintains course and speed and the weapon travels straight down the centerline (usually not because recoil, but that's an aside).

When the value is less than 100% or 1.00 or whatever, the ship will start firing while the turret or weapon is still traversing to the aim point, with the lower the value leading to an earlier start time for shooting. This can waste flux at best (high RoF weapons) or ruin a salvo at worst (low RoF weapons).

When a target changes velocity, the weapon(s) must redo the same math they did when first acquiring the target, with the lower value leading to the same premature-to-aimed firing. However, if a target maintained identical course and speed (or even very similar, with ship size/hitboxes allowing for some differentiation), the weapon(s) will after a time determined by gunnery skill eventually settle into a perfect firing arc as described above, and assuming a situation where the target and shooter stay at the same relative velocities to one another, will have "perfect" aim.

Sorry for the wall of text, but am I understanding this right?
Title: Re: Can someone explain the (invisible-ish) "autoaim effectiveness" stat?
Post by: Alex on March 25, 2019, 01:15:59 PM
Close, but not quiiiiiite :)

If a ship has perfect "target leading accuracy" (between CR, skills, and hullmods, etc.), then it will fire when and only when the weapon is aligned in such a way that a hit would be guaranteed, assuming the target maintains course and speed and the weapon travels straight down the centerline (usually not because recoil, but that's an aside).

A hit is not going to be 100% guaranteed regardless, as the calculation isn't perfect, but aside from that, yes. (So, for example, a zero-recoil point defense weapon is likely to end up being less effective than the same weapon with some recoil, as the randomness will help it avoid some situations where it'll say, reliably miss a small missile. But that's getting pretty deep.)

When the value is less than 100% or 1.00 or whatever, the ship will start firing while the turret or weapon is still traversing to the aim point, with the lower the value leading to an earlier start time for shooting. This can waste flux at best (high RoF weapons) or ruin a salvo at worst (low RoF weapons).

Not necessarily an earlier start time to the shooting - if it's traversing from another direction, it may lead to a later start time (i.e. after it swings past where it ought to have fired ideally). Or, if the target isn't moving laterally relative to the weapon, it may be the same start time. Hopefully that makes sense.

When a target changes velocity, the weapon(s) must redo the same math they did when first acquiring the target, with the lower value leading to the same premature-to-aimed firing. However, if a target maintained identical course and speed (or even very similar, with ship size/hitboxes allowing for some differentiation), the weapon(s) will after a time determined by gunnery skill eventually settle into a perfect firing arc as described above, and assuming a situation where the target and shooter stay at the same relative velocities to one another, will have "perfect" aim.

No - the math is redone every frame (or close to it) anyway, since the relative velocities are pretty much always changing. The improvements to aim quality come with time spent firing at the same target, regardless of its movements. The aim quality resets to its base (CR + GI etc) value when a weapon switches targets or stops firing for a couple of seconds.
Title: Re: Can someone explain the (invisible-ish) "autoaim effectiveness" stat?
Post by: angrytigerp on March 25, 2019, 01:31:28 PM
Hmm... So, the turret may over OR undercompensate, either way, as a result of lower TLA? And I'm curious then, does that mean that the weapon won't fire until/unless it has that "ideal firing solution"? Because you seem to nip my "prefiring" thing in the bud, but maybe I'm misunderstanding the hell out of this.

It's clear to me though that you guys (indie devs the world over) making your own engine and everything are frickin' smart, though, and for what it's worth I really am grateful you're clever enough to make this stuff work!
Title: Re: Can someone explain the (invisible-ish) "autoaim effectiveness" stat?
Post by: Alex on March 25, 2019, 02:15:56 PM
Case 1:

<-----------T

       I
       X

Case 2:

<-T----------

       I
       X


X is the gun, I is its barrel, so in both cases it's pointing up. T is the target, moving from right to left. Assuming imperfect "aim quality" on the gun.

In case 1, it'll turn to the right, and fire after it needed to to score hits, since it's aiming at an older position of the target than it needs to, essentially.

In case 2, it'll turn to the left, and fire before it needed to to score hits, since it's aiming at an older position of the target than it needs to.

In both cases, it's going to be hitting somewhere behind the target's vector, it's just that firing either "earlier" or "later" than optimal may be needed to do that.


It's clear to me though that you guys (indie devs the world over) making your own engine and everything are frickin' smart, though, and for what it's worth I really am grateful you're clever enough to make this stuff work!

I'm probably just making this sound too complicated :) All it's doing, really, is "leading the target, just not by quite enough". And then gradually adjusting.
Title: Re: Can someone explain the (invisible-ish) "autoaim effectiveness" stat?
Post by: angrytigerp on March 25, 2019, 02:39:47 PM
Case 1:

<-----------T

       I
       X

Case 2:

<-T----------

       I
       X


X is the gun, I is its barrel, so in both cases it's pointing up. T is the target, moving from right to left. Assuming imperfect "aim quality" on the gun.

In case 1, it'll turn to the right, and fire after it needed to to score hits, since it's aiming at an older position of the target than it needs to, essentially.

In case 2, it'll turn to the left, and fire before it needed to to score hits, since it's aiming at an older position of the target than it needs to.

In both cases, it's going to be hitting somewhere behind the target's vector, it's just that firing either "earlier" or "later" than optimal may be needed to do that.


It's clear to me though that you guys (indie devs the world over) making your own engine and everything are frickin' smart, though, and for what it's worth I really am grateful you're clever enough to make this stuff work!

I'm probably just making this sound too complicated :) All it's doing, really, is "leading the target, just not by quite enough". And then gradually adjusting.

Ooooohhhhh

So it can be before OR after firing solution is acquired?
Title: Re: Can someone explain the (invisible-ish) "autoaim effectiveness" stat?
Post by: Alex on March 25, 2019, 03:28:34 PM
I'm not sure what you mean by "a firing solution is acquired".

Let me put it another way: it's going to aim so that its shots hit (or, rather, miss) behind the target. Whether that necessitates firing before or after it would fire if it was aiming to hit the target depends on their relative velocities, the facing of the gun, etc. All I'm trying to say is that "firing behind the target" does *not* always equate to "firing earlier". Sometimes it does, other times it doesn't.
Title: Re: Can someone explain the (invisible-ish) "autoaim effectiveness" stat?
Post by: angrytigerp on March 25, 2019, 09:49:43 PM
I'm not sure what you mean by "a firing solution is acquired".

Let me put it another way: it's going to aim so that its shots hit (or, rather, miss) behind the target. Whether that necessitates firing before or after it would fire if it was aiming to hit the target depends on their relative velocities, the facing of the gun, etc. All I'm trying to say is that "firing behind the target" does *not* always equate to "firing earlier". Sometimes it does, other times it doesn't.

Sorry, too much military (Navy, space and wet) fiction. Firing solution is roughly equivalent to "weapons lock" for non-guided weapons.

Okay, gonna see if I'm picking it up one more time, and if not... Well, I'll just trust it's a good thing:

So a weapon is always trying to aim at "center mass" (or rough equivalent) of a target, and TLA dictates how quickly the weapon is brought to bear on that aimpoint?