Currently I think the autoaim only takes velocity into account.

This makes barrel rolling easy, and boils fights down to who-can-stay-at-the-right-distance-and-time-weapons. This, in my opinion, is all wrong.

Without wanting to get into heavy physics, a basic roll can't be that hard to track. It just adds a central acceleration to the constant speed, meaning a uniformly changing velocity.

Why can't the autoaim track this?

Consider this for a minute. How do SCPs evade fire? Barrel-rolling alone, mainly. How do light fighters evade fire? Strafing.

Take out the barrel-roll as a viable dodging maneuvre, fights get quicker and more intense. Which is more fun. Backrollers get shafted. The person on the DEFENSIVE, is now at a DISADVANTAGE. People turboing away can get SHOT DOWN (if the autoaim takes into account inverse exponential acceleration, which I think is how the drag works. May have used the wrong jargon)

Turrets become USEFUL.
Fighters get SELF-AWARE.
Thrust becomes more major as a balancing issue for the heavier ships.

I can't put this very well. So I'm hoping someone will know what I'm getting at and phrase this better... but this single change might make balancing a HELL of a lot easier, and combat just... more fun.

I real fighter combat the barrel roll is a vary useful tactic. And thus I think it should remain in VO. The only ships that can do well with it are the Rev Cs and some vults. How ever if you know what you are doing you can take them out. I will agree that back-rolling takes a ton of fun outta a fight.

I think timing should play a large role in combat, it's one of those things you can't really teach but you just have to learn the hard way. I think it is what separates the newer combat guys from the older ones.

Just my thoughts, I am a heavy barrel roller so I am a wee bit bias...

While I personally don't use the backrolling method in combat it's a viable option for some and in no way is it any different then barrel rolling or strafing or diamond dodging or any other tactic. I think your being a bit too critical about the combat mechanics UncleDave, perhaps you should simply take a break for a while if your getting bored. The Auto-tracking is just fine as it is now, and making the combats last less time IMO is the WORST idea I've heard. Part of the reason why I think the auto-aim doesn't do what you suggest is that once the round leaves the cannon your no longer in control of it, and since the target is moving it does not hit. Perhaps try to get closer when you start your attack. Combat that is too quick OR too long are not enjoyable, though I'd much rather be in a long dog fight then constantly buying a new ship ever minute, or waiting for combatants every minute while they fly back.

Long dogfights are only interesting when they're exciting. Not when it's just a matter of who can hold two buttons down the longest.

True, but making auto-aim track barrel rolls is going to change that how?

John. You seem to have this uncanny knack for missing the point.

Barrel-rolling is a tactic which levels the field agility-wise.

Backrolling draws out fights to a ridiculous extent.

If autoaim can track barrel rolls or similar repetitive evasive maneuvres, then players have to rely on fast reactions and dodging skill rather than holding down a pair of buttons.

Borb: it separates the newbies from the vets, right. But its like the banana skin trick in Mario Kart. Its just a little trick. Its not based on skill, just experience. This is a part of what's causing Vendetta's elitist aura right now. And since when was Vendetta based on "real" fighter combat? Dumb analogy.

I am a heavy barrel-roller. This would nuke my current fighting style, and yet I still think it would improve gameplay.

Hell, if fights get TOO short, just double the HP on all the ships. Its not a problem.

I can't be the only one that thinks this is a good idea.

I think I agree UncleDave, but adding banana peels would require a lot of rebalancing. They might make mines much less usefull as well.

Why REDUCE the skill required to play this game?

To fight rollers, turn off autoaim and aim your own shots. It's not easy, but it's an extra skill you can learn. Try it, it eats up rollers and dodge-binds for lunch.

Though this is all a moot point, since adding this tweak will tank the game balance like many have noted.

I would love to see auto-aim handle barrel rolling with ease.
I've discussed it with Incarnate briefly, and based on his reaction, I think altering the auto-aim code to account for these kind of repetitive techniques is more complicated than we think.

Sadly, I don't see any end to the barrel rolling in the near future. I use it for botting, just to get the job done, but in PvP I really don't understand how it's fun.
I don't know about the rest of you, but I want to have fun in Vendetta, it's not about winning or losing a fight, it's about how fun the fight is, and I never have fun with a barrel roller, or barrel rolling myself.

softy2: So you think everything is balanced, ok.

Errr... Dave, how the hell would you accurately account for the roll? You take a shot during one instant in time - the autoaim can't take changing acceleration into account because at that instant accelration is for all intents and purposes constant. The reason the autoaim misses so much is because it can't possibly predict if the user is going to change what buttons he's hitting, which is more or less what you're asking it to do.

Tell you what. Do a quick google on the physics of moving objects, then write an algorithm that will calculate where a shot from one moving object to another moving object (both with changing accelerations) should go to score a hit. I think then you'll see what what you're asking is a bit over the top.

A person can hit where a constant roller is going to be because we expect patterns to arise after seeing them a certain number of times. Vendetta has no brain with which to anticipate patterns - so it would have to store data for some seconds previous to a shot to be able to make the same decision a pilot does when he tries to hit a roller with the autoaim off. You're basically asking a1k0n to go in there and program a learning autoaim...

The last thing we need is Vendetta to become SkyNet.

it is possible to get a center of rotation from an equation. thus, you can calculate acceleration. you can add that acceleration to the velocity, for each shot. this also includes calculating where to shoot for the amount of time for the shot to reach the estimated target flight path. All possible, only with constant acceleration. True that people will change motion, but when you're doing a true roll, isn't that constant acceleration in the normal (using n and t co-ord) direction? It's possible to make something to calculate and break a uniform circular motion, but that wouldn't exactly always work.

By the way, physics of moving objects is kinetics (or is it kinematics?) and generally covered in dynamics. Anyway, to end a roller, you should start off with firing a flare right into the center of the circle. from there, there are plenty of other things you can do.

They think they have it solved... and then the player starts spinning in a repetitive manner.

(All of Dave's suggestions are based on strengthening heavy ships! no suprise :P )

But yes, regarding the trouble of modifying the autoaim. I don't know how the code handles autoaiming, but lets say they broke the current direction into 3 vectors, a way that makes sense really. Basically an x,y, and z vector corresponding the horizontal velocity, vertical velocity, and backwards/fowards velocity. So, let's say you're at the top of a roll. (Z we'll ignore since it isn't important). To be accurate, your Y component is going straight up(Depending on how you're strafing) and you x component is going straight sideways. So it'll extrapolate your future position to be at a 45 degree angle outwards from where you are. HOWEVER, what this does not calculate is how the roll affects next x and y coordinate. so lets say you've rolled another 45 degrees. Suddenly your extrapolation would be outwards at the 90(I'm not sure on this one, just doing it in my head). However, in fact you're not going outwards really; you're remaining in a circle.

It basically is what you get when you draw an object tied to a string going around in circles. Velocity is always perpendicular, with force directing straight in. (Which, since there is no string, means velocity out).

That's only if they do it that way, but you can see a possible problem in programming it. If you tried to extrapolate the position of the object in a future moment in time using its current velocities, you'd be wrong because of how it acts.

Now, what some people are saying, like fnugget, is that we should work out there is a string in the middle. Now, this would be a problem since:
A)you don't circle strafe in exactly the same location every single time, you'd have a mobile center.
B)this would require a level of ai autoaim currentl doesn't have.

Possibly, what alkon would have to do is he'd need to redo how the autoaim works, which would royally suck.

Dave : Nope I think lots of things are borked, especially ships. But I do think that if you add more computer-aid into combat, then that's a step backward.

(Actually, I don't see how my comment can be construed as "everything is balanced". If anything else, making the autoaim predict acceleration will make barrel roll AGT users even more deadly, since light ships can't dodge AGT anymore. )

Fnugget : Ships motion is dynamics. Autoaim prediction is kinematics.

Pirogoeth : Exactly. In high brow language, barrel roll acceleration vector is constantly changing, so to really make autoaim able to hit rollers, you have to solve 3rd order equations, which is 2 orders up from the current linear predictor.

levels of autoaim:
0. shoot at current position.
1. shoot at current position, correct for current velocity.
2. shoot at current position, correct for current velocity and current acceleration.
3. shoot at current position, correct for current velocity, current acceleration and current change in acceleration.

Current autoaim does: 1.
Autoaim needed to hit a perfect, constant barrel roll: 3.

One 'sample' of the enemy probably contains position and velocity.
to implement 2: you take 2 samples of the enemy, calculate the acceleration by taking the difference in velocity.
to implement 3: you take 3 samples of the enemy, calculate the acceleration by taking the difference in velocity between 1 and 2 AND 2 and 3. Then you calculate the difference in acceleration from those two results.

In a real dogfight this is highly inaccurate and would probably send your aiming reticle flying all over the place...

If they made the sample points literally one process apart, it shouldn't hop around too much. Unfortunately that is still a lot of work.

...and there's no way I'm going to make the autoaim solve a third order system (as softy2 mentioned) because it would become extremely sensitive to any minor acceleration and completely throw off just about everybody's shots. I don't think it could be made to work given the imperfect information we have in a networked environment. (We don't have any way to get "current acceleration" but we can -estimate- it, and "current change in acceleration" is not a differentiable function! Acceleration changes instantaneously.)

It's already a quadratic equation to solve for simple linear dynamics.

awwwwwwwwww

pleeeeeeeeease a1k0n, it'll be fun....
(the above is sarcasm for the humor impared amoung you)

EDIT
maybe what we need is to make lateral and reverse thrusts have a seperate value from the main and turbo thrusts... then you can have more variation in ship style, (for instance the centurion could be a very fast, high accel ship, but not be as good at reverse and side to side... making it more of a hit and run fighter, but it has to be pointing in the direction it wants to run... that would also be a good way to give ships like the hornet an advantage... make them better at dodging without making them able to run as well.... now i'm rambling so i'll stop now.

Dave, the only weapon that would benefit from all this is the Gatling Turret. Others, you have to shoot close enough to the target thingymobbober that this wouldn't make a difference. As for not being able to hit a barrel roller, turn off your sacred autoaim for once! It works wonders!

Even if this was within the range of possiblity, why would you want a weapon, where you basically shoot, and it almost automatically hits the target!?! There's this thing called aiming, and it might be a lost tradition by now, but there some of us still care to use it, and it really does break the barrel roles if you're good at it. Look at Eldrad for example. The other day, I was fighting his cent in my prom, and he basically screwed even the Nuet's rudementary auto-aim, and took me down to 4% from long range.

Now *that* is narrow-minded.

Technical difficulties aside, the gatling turret would most certainly *not* be the only weapon to benefit. In fact, the gatling turret (being slow-speed) would be considerably easier to dodge with strafe than, say, a stream of N3s.

IF YOU HAVE TO STRAFE TO DODGE, THEN LIGHT SHIPS HAVE THE ADVANTAGE. Why is that so difficult to comprehend?

Oh, and empirical evidence is a no-no. We all know Eldrad is a Shaolin Grand Master by now, but this would be for us lesser mortals. ;P
By your "evidence" the centurion is balanced against the skycommand. Thanks for sharing that!