Player Features and Ideas Discussion

Right now, turrets always aim by tracking their target along its entire path.

This is only one of two major kinds of real world aiming strategies. The other is ambush aiming: instead of following your target, you use your brain to figure out a point up ahead where it will probably pass by, and you aim carefully at that point, and then shoot it when it arrives.


If Eve were real life, pilots would use both kinds of aiming frequently for different situations. This would be a major part of actual pilot tactics, and a major element separating good from bad pilots. Eve is missing this, and thus missing an easy opportunity for complexity and depth. So let's add ambush aiming:

TRACKING AIM:
Would work the same way eve aiming does currently
Good for: Complex movement patterns that are hard to predict (like when multiple ships are orbiting one another, or one is orbiting a fast moving ship, creating spirals, etc.). Since the paths are hard to predict, tracking is best, because you only have to predict a moment ahead of time.

Bad for: Any situation where the target is moving too quickly to track (e.g. frigate orbiting a battleship).

AMBUSH AIM:
Would work by recognizing "valid" predictable movement patterns of your target, and then "predicting" much further ahead in space than tracking aim does. The turrets would then aim at some point far enough ahead that they can reach that point before the target gets there (taking into account their own tracking speed). Then they shoot the target as it passes by.

This is good and bad in the opposite situations as tracking aim. It's bad when movement is complex, because you are unable to predict movement far enough ahead to ambush. But it's good for when things are moving quickly but predictably (like an orbit), because no matter how slow your guns are, you can still hit them after a little while observing.

How it would work in computer code:
The game already knows if your target is orbiting something or flying in a straight line, or whatnot. It uses this knowledge to very efficiently determine whether the target would APPEAR to be moving in a simple pattern to the shooter's perspective, without having to actually do any complicated calculations at all.

Situations that the game will consider to be valid patterns of movement for ambushing:
A) Target is orbiting a stationary or very slow object (including you!)
B) Target is approaching an orbit around a slow object.
C) Target is not moving.
D) Target is moving in a straight line (for example, the pilot double clicked in space, or aligned to a celestial)
E) Target is approaching a slow object.
F) Target is doing any of the above, but orbiting/approaching a fast object, IF that object is very far away (the further it is, the faster it can be moving and still qualify. This is just a precalculated lookup table, not calculated in the moment)

In any of these situations, the game just extrapolates where the object will be at a few different ticks in the future, and the turret tracks toward the first one it can reach in time. Thus essentially giving up a little bit of rate of fire in exchange for transveral velocity becoming irrelevant.

In any other situations that do not qualify, the turret will determine the instantaneous tangent of current movement, and use that for extrapolation (which will almost always miss).



WHY?
Because perfect, invulnerable speed tanking is dumb, and incredibly unrealistic, and makes it seem like space age civilizations have caveman technology. It also robs us of natural strategic decisions we could be making. If I took any one of you players, and put you in a first person cockpit flight simulator, and gave you slowly responding guns, you would instinctively switch to ambush aiming almost immediately for fast moving targets... Why not have the option to do so in eve's third person mode?

And this wouldn't make speed tanking obsolete by any means. It would just make speed tankers not 100% invulnerable. You could still avoid shots by adjusting your orbit every so often, and even if you didn't, ambush aiming inherently has fewer opportunities to hit (it has to wait for you to pass the ambush point, and thus the guns will be firing more slowly than their normal rate of fire as they wait)

It adds more player skill to space combat, since you have to strategically decide when to use tracking versus ambush aiming yourself (button toggles it on the HUD). Indeed, you may even learn to put yourself into positions that allow you to use one or the other methods to maximum effect (like slowing your ship down when fast small enemies are buzzing around you, to make their movements more predictable)


If we are going to have a 3rd person slow-response, strategic shooter (which is what EVE is), then we need lots of options for strategies. Currently we don't have very many at all, and space combat is dangerously close to a simple grind, that is determined much more by your fitting skills in the station than by your skills as an actual pilot.

I see what you mean, but I do not think that anything more complex than now would be required.

I tried to explain above, but maybe didn't do a good enough job. The computer could "cheat" in ambush-mode just like it "cheats" in tracking mode, by using all kinds of shortcuts available due to the game having omnisicient knowledge of all the ships' movements and instructions.

For instance, a frigate is orbiting you in your battleship, which is moving slowly relative to the distance of orbit (30m/s and 5km radius of orbit):
1) The game recognizes this as a "valid" ambush situation, using a simple formula like:
d = distance from shooter to target
s = speed of thing being orbited or approached by the target (=0 if target is just moving in a straight line by clicking in space)
a = ambush situation = 1 if this is a situation on the list in my OP. = 0 if not.
c = some constant used for tweaking during testing until mechanic is balanced.

formula: a * (d / s) * c

If the result is greater than some number, then ambush tracking can commence! if not, it defaults to attempting to do tracking-aim, basically. or just gives up, I don't know. Something less effective.

2) Once ambush aim is cleared to commence, the game does another simple arithmetic formula, something like:
t = tracking speed of turret
v = current transversal velocity of target
k = some other constant.

Time until guns fire is then something like:

time = (v / t) * k

So it doesn't have to actually figure out all the geometry. It could just approximately "guess" about how long it would take. If transversal is high, that would usually mean that a gun would have to look further ahead to pick an ambush spot. If the tracking speed of the turret is high, then it would have to look less far ahead. And a constant lets the results be tweaked to balance.

3) Then, every tick after that until the guns fire, recalculate the first formula (a * (d / s) * c) to make sure that the target is still within valid parameters and hasn't taken some evasive action.

4) Then when the time arrives, if nothing has changed and the ship is still on grid, etc., fire and hit (with any falloff calculations etc. as normally calculated already)!


Really, it could be as simple as that. Look up the situation in a table of valid situations, then do one simple arithmetic formula to determine whether the situation is within quantitative thresholds. Then calculate time until firing, and checkup with the first formula in the meantime.

Very similar if not identical to the math requirements of existing tracking aim

A reasonable shortcut, but falls short on another count. Imagine a typical spiraling approach, where you move by doubleclicking every time as you approach, to keep angular speed above a certain threshold. (Angular is what's actually used in the tracking formula, but apparently, back when most lore on the subject was collected, you could only see transversal.) By all rights, the user doing that is not moving in a 'straight line' -- doubleclicks will never be so perfectly on the same level as would be required to produce an actual straight line.

But how would the code know that? Your formula does not provide for how long should anyone keep moving in a straight line before it's 'a straight line'. Suppose it does require that no movement commands have been issued in the last server tick. Now, I program my mouse button to send a doubleclick every second while it's held down -- a perfectly EULA-legal application of mouse capabilities. Poof, I'm immune to ambush mode because every server tick I'm moving slightly differently.

While the idea of adding more tactical options available to everyone, whether they fit modules or not, is interesting, I'm afraid implementing it from that particular angle either requires way too much work on deep fundamentals to consider, or is far too easily defeated, or is grossly unfair to people actually doing something clever when it isn't easy to defeat.

...You kinda killed the discussion right there. The disadvantage of Titans, I suppose, is that they're too expensive to actually use, right?

You must always predict the future location of a ship when firing weapons: even lasers have a travel time. It's safe to assume that whatever virtual smarts are powering the targeting systems of our virtual spaceships are already doing this flight path prediction, otherwise we'd never be able to hit moving spaceships.