If person A turbo's towards person B and shoots energy with auto aim on the shots go widely off target. This isn't much of a problem for rockets since you can turn off the auto aim and even if you shot isn't perfect the prox radius will make it work. But since energy only hit if their spot on this basically makes turboing at a target never work without their cooperation. Is it possible to fix the auto aim so that it doesn't do this? If so are there reasons it shouldn't be done? (Note this isn't very time sensitive, but it would be nice to know if it's on, or put on, a to-do list)

Maybe energy weapons should get a small proxy too?

A few energy weapons look like they should have them.

Charge cannon, gauss cannon, tachyon blaster, gravitons, plasma cannon to name a few... they should at least have a blast radius that matches the animation.

Gauss cannon does damage if any part of the projectile encounters a ship. Since the projectils is very large, that makes damaging things very easy.

I have never liked the idea of energy with proxy. Something does need fixing though. Not really sure what it would be however.

Giving some energy weapons prox is a completely separate idea, it is not a fix to the problem that turboing kills auto aim. If you removed all non-prox/non-homing weapons then the point would be such a small issue that it wouldn't be worth paying attention to but I don't think that's a very wise solution.

i think fully charged charged cannons, gausses, plasma cannons and, rail guns (cause it uses ammo) should have splash

*Thinks of fixing energy targeting*
*Thinks of Gauss rushers*
*shudders*

Rails musn't have splash, they must be faster:P

Think of Jango Fett's dogfight with Obi Wan in Episode 1 in the asteroid field. Perhaps some weapons like gravitrons and phasers can have splash, while other weapons like gauss and tachyon do not.

Exploding energy weapons are a completely separate issue. The real problem is that the targeting software ('autoaim') is pretty bad at the job (it doesn't take into account any sort of constant acceleration, rotation, or what not) and its miserable targeting something without it - hense the reason everyone uses gauss.

SC: episode 2...
and i am completely against giving the gauss or plasma cannon any splash damage.
they are powerful enough as is (even too powerful).
i think that the autoaim is not the highest priority atm (since one can play with the problem in existence).

Oops, Ep2. I agree with Eldrad, though, the bot is easily fooled, giving more experienced players a powerful edge. What if something like a Ragnarok with plasmas could pin a valk coming at him? All of a sudden, people would be a lot more careful.

The plasma's do deserve an upgrade, but that is a completely different issue. If you like, we can start a thread on what needs to happen to balance all the weapons, or a different thread on 'unfilled niche weapons'.

The autoaim calculation was surprisingly more difficult than I had thought, and we actually had to do a sort of gradient descent to solve it, and we only do two iterations. In other words, the autoaim calculation isn't exact, and it could possibly be wrong in the case of high relative velocities.

I want to look at it and see if I can solve that problem, and also smooth it out so it doesn't jump around so much. It also has no idea whether or not the problem is solvable: if you're flying backwards at 400m/s with an Unreal engine, it's impossible to hit anything in front of you unless it also has an Unreal engine, but the autoaim still points uselessly towards your target.

But we've a million other things to do. This probably isn't too hard, but who knows when we'll get to it.

a1k0n: I'm aware that its a very difficult problem, otherwise human agility would be very easy to mimic. The solution is probably as difficult to obtain as the solution to the equation for moving your arm to catch a softball thrown at you. I can understand why you've been putting it off, I'm just commenting that it is infact an underlying problem.

Although I haven't thought hard about the math, I'm not surprised that the only solution that you have involves an approximation. How much time does it kill to go to three or four iterations? Most numerical methods converge fairly well by four iterations or so.

Also, maybe there is a way to precache approximations to make the method converge faster?

The 'bug' you describe with the autoaim not realizing it can't hit the target with a negative velocity weapon isn't so much of a bug as it is the algorithm catching an error condition. You wouldn't want it to handle the situation in any other way (though you'd probably would want it never to switch to 'red' or 'yellow' in the reticule whether pointing at the target or not).

a1, cool. It's not that big of an issue, I'm just happy knowing it'll be dealt with at some point. Oh the lack of smoothness is due to the auto aim... I always thought the minimum change in angles seemed ridiculously large. Thanks for answering my concern.

Cele, it's worse than you think. Not only does it incorrectly deal with constant acceleration, but it doesn't deal with high velocity differences well either even when no one is accelerating.