Testing aim assist differences

[individual]

I figured out a method of testing aim assist differences. I'm really not bored enough to do this myself, but maybe somebody else here isn't?

1. Start a private match or split-screen match.
2. Have somebody (as an enemy) stand right next to you.
3. Spin and measure the time it takes to do a full circle.

While aiming down sights, there's a difference in turning speed, but from the hip, it seems to be the same for any gun.

If you think there could be a difference, though, you could spin with no enemy next to you as a control and make comparisons.

[mannon]

I would definitely suggest using a control with no enemy. It might also be good to make the test repeatable by having both players in precisely repeatable positions. I'm thinking perhaps wedged into two corners of the same room or something.

You could also test it with several revolutions instead of just one to improve accuracy, especially considering you're unlikely to have any given end of a revolution fall exactly on the right frame of animation to be pixel accurate to 360 degrees.

On the other hand if you record the video of it in HD you should be able to use visual landmarks and some math to determine how close the angle got by that frame and even tell how many degrees you are turning per frame at any given point. Though you would have to know what the field of view is for that weapon's ADS.

Anyway it sounds like a reasonable test in theory. On the other hand, there's several possible factors to the aim asisst and I think range is probably one of them.

This also does not measure the snap to target effect you get when you ADS near a target. For the best example of that go into SP and play the little tutorial at the beginning of the game where the targets pop up. You can snap to them when you ADS from waaayyyy off target, but once you are looking down the sights you can still track all over and make yourself miss and stuff. In fact I'm honestly not sure if the autoaim will even help you with a stationary target.

I know that it will pick up targets that move in front of your cross hairs. But other than the snap to target I always feel like I'm on my own for stationary targets. Hard to tell, (for me).

And like I said there could be other variables such as range. Some weapons could well have better autoaim at shorter vs longer ranges. You could test that too, but it's just more work.

For me what I'd really like to know more about is the snap to target. It almost certainly is range dependent, and also seems to kick in less based on how much of the target is exposed. It also feels to me like sniper and thermal scopes don't seem to snap or do it less somehow than iron, RDS, Holo, and acog. The acog however seems to actually increase the range of your snap.

I'm not sure exactly how these factors work although I suspect that the game gives you a certain radius around a target where your cross hairs will snap to it when you ADS, and that the radius can be increased or decreased based on factors like range and possibly weapon and attachments and visibility of your target, ect. But I don't know, these are only my impressions.

I suppose it's testable too, but it sounds like a PITA.

[n1gh7]

Ok I'm going to tell you why this won't work.

The joysticks for controllers are terrible. Most of the controllers' joysticks are imperfect in their manufacturing. This means that if you physically turn them exactly 90 degrees to the right using the outer shell of the controller as a landmark you are going to not be "truly" turning them to the right in the computer's eyes because of how the sticks can be seated and so on.

From what I understand aim assist (not auto aim. that is completely different) works the same amount on everything. What is changed between guns is the range at which it works.

The best way to test this (and it is still very imprecise) is for you to be far away as possible from a dummy enemy controller and simply walking side to side getting closer every single time to the dummy. When the dummy's aim moves that is the point at which the aim assist range is.

mannon: the snap to target effect is not in multiplayer. It only exists in single player. Turning it off in the menu does not change multiplayer.

[individual]

Dec 2, 2009 4:04:04 GMT -5 n1gh7 said:

Ok I'm going to tell you why this won't work.

The joysticks for controllers are terrible. Most of the controllers' joysticks are imperfect in their manufacturing. This means that if you physically turn them exactly 90 degrees to the right using the outer shell of the controller as a landmark you are going to not be "truly" turning them to the right in the computer's eyes because of how the sticks can be seated and so on.

Can you turn with the keyboard on PC? If so, you could do it on that.

[n1gh7]

Dec 2, 2009 12:36:39 GMT -5 individual said:

Dec 2, 2009 4:04:04 GMT -5 n1gh7 said:

Ok I'm going to tell you why this won't work.

The joysticks for controllers are terrible. Most of the controllers' joysticks are imperfect in their manufacturing. This means that if you physically turn them exactly 90 degrees to the right using the outer shell of the controller as a landmark you are going to not be "truly" turning them to the right in the computer's eyes because of how the sticks can be seated and so on.

Can you turn with the keyboard on PC? If so, you could do it on that.

No. looking on PC is different. you don't look based on how far you are from a "center point." You look based on how far the mouse moves IRL and that is translated into a specific amount of degrees that you turn in the game. I guess it could still work but it is still impractical to test.

[mannon]

And it would not match the console versions either. I'm not even sure the PC version uses aim assist, but it's got to be much much lower than on console.

[toysrme]

forever ago me and a buddy figured out in MW2 with your sensitivity set to 10 (360) and the target inside the auto-aim range of an acog M40, you can spin constant 360's, fire at any point in time and kill every time.
It snaps the target & let's go faster than you can see sometimes LoL

at one point in COD4 i wrote a macro that spun the acogm40 sight in a perfect, very small 360* cricle & fired. absolutely rediculous up close inside the range of it. the sight would just jump over a boatload farther than your input actually would be & shots would connect.
never been sure exactly how auto-aim works when you're moving & shooting in-play. (because alot of us XIM/mouse console guys find it HELPS you about 50-60-75% of the time & then will fight the living daylights out of you and keep you from connecting shots).


ive got another one that works on holosights in MW2 (but not RDS), but (again) thankfully the max auto-aim range isn't so far away on AR's as to even make it useful (it takes advantage of the snap affect by releasing ADS for 2-4 in game frames & then ADS'ing again.) you also learn REAL fast when shooting anything at range that the minimum HF spread size during the transition is still fuckin huge LoL!
overall accuracy is greatly reduced outside of auto-aim range & it really does some whacky things... Sometimes you can take a TAR and it'll snap every bullet on someone's body until they register dead, then the next kill no matter how you fight itll decide only the first bullet is worthy of hitting the target...






btw I know I said like a month+ ago Id make an ACOG M40 demonstration video to show just how massive auto-aim can be. Just never got around to it (encoding times are just too high to get good quality from 480i bullshit HAVA videos LoL). I DID however, find a USB flash card I had recorded like 7 matches one night & made a montage of them LoL.

There are a couple of really good auto-aim flick-shot demonstrations too. One of them is crazy. You can see the speed I flick shoot a mouse, My sight is WAAAAAY behind the guy running across a hallway (because I was late on a guy) and the INSTANT I hit the fire button (takes on average like 3-4 frames on 360 for a fire input to register visually) the rate the view is looking accellerates like mad, far beyond my movement rate, fires & the guy falls over dead.

If I can get around to it Ill find a program thatll slow-mo it for you and get it on the utube.

and there is DEFINATELY about 3 frames worth of bullet flight on console ACOGM40 when you play frame by frame. You CAN NOT have your sight 100% legit on the back half of a player moving at a reasonable range & hit them. The bullet takes too long to get there. If it were 100% hit scan and they were lagging their ass off, then they would have fallen over dead 20 feet away.

[toysrme]

The game, on a PC with a mouse, interprits movement on a linear set scale (that you can adjust) by watching how many pixels (and what direction) your mouse CURSOR has been told to move AFTER the OS/Driver is done playing with it. (some people CLAIM they watch the mouse dpi, which is utter horse-doo-doo)
Console works the same as PC with a gamepad does.


The reason COD is so badass on console, is because unlike every other console game you play. COD allows you the ability to move on a pixel by pixel basis, has the least amount of exponential accelleration applied (around 19-25%), has 100% progressive movement rate(instead of a movement rate based on several speed "steps"), has the smallest deadzone that is practical & the deadzone is PERFECTLY CIRCULAR!

Other games give you a minimum movement amount that is massive. Where you may literally be moving in 1/2* steps inside a 360* circle. COD continues to allow you to move on a per-pixel basis inside your 360* circle.
Having a perfectly circular deadzone is extremely nice. Most games are square or retangular, which combined with not having progressive look speeds REALLY fucks up diagonal movement.

[toysrme]

and n1gh7, the best visual "quick & dirty" representation I can giv eyou about auto-aim/aim-assist (same bullshit) is that it extends in a circle(or square) roughly the same diameter as the weapon's hip-fire spread.

when a body crosses into the hip-fire cone it grabs them & tracks them until they leave. my theory is that may be set so that its ability to track is also proportional to the speed & direction when you are currently looking. probably not true as to how it works, but that's definately true when it comes to the "end result" of however it does actually work.

[toysrme]

(give or take a tiny hair)
This is the maximum range of auto-aim. The closer in from this range you get, the more profound the affect it has.




While I can't fully quantify it (because while I see it in play all the time, I haven't tried too beyond repeating about 15-20 kills with RDS/Holo until it became obvious there is a difference)


With steadyaim HF size as a frame of reference...
The RDS's auto-aim would be a circle around the INSIDE of the white marks.
Holo's would be a circle around the OUTSIDE of the white marks.



Using a bash of the script I talked about above, (which ADS's and releases to shoulder fire to initiate whatever the tracking doo-doo AA does...)
Using the holo sight with the target on the outside edge of the HF spread. At a point around 50-75% of the way through the clip, the weapon will decide that it needs to move towards the target. Once this happens within the span of 5 bullets it will hit the target, hit it again then cleave it with a headshot
*edit*
(Ok Ive done it 10 more times, it cleaves it with a headshot the *majority* of the time LoL. Not sure exactly what changed, but Im finally getting them where they just fall over dead no headshot. Not sure, could have moved abit sitting here.)
Every
Single
gosh darn golly gee whiz
Time
...
Using an RDS if I line the target up on the outer edge of the white marks, the weapon WILL NOT EVER center itself on the target. It wont even begin to try.
On the inside of the mark, around 1/2-3/4 of the way through the clip; it will eventually decide it needs to snap towards the target while firing & kill it with a clean headshot.
every
single
time

So... That is basically how far the weapon is allowed to jump.


And don't just think it's recoil coaxing it into it. Itll do it right, or left LAWLZ

[dgcool]

toysrme, your final analysis is that Holo does have greater aim assist than RDS and is over all better? Also, we are speaking about ADS auto aim, not hipfire, correct? I may be confusing your screenshots with your description I believe.

[mannon]

If it doesn't happen until several shots have been fired that does sort of suggest that it's your recoil eventually nudging the aim close enough to snap in and auto aim. But it does still sound like there could be more auto aim on the holo, which makes sense enough to me.

I suppose that might be more useful on an SMG you are less likely to be outside of auto aim range with and the reticle would be less of a PITA at closer ranges too. It's practically unusable at long range though, since it's as big or bigger than your target.

I remember in SP aiming at targets and having to actually move the reticle off target to get a bead on them. I would line it up under the target, note it's elevation to a landmark, and then raise the reticle up to the target which I couldn't see at all at that point. Very annoying. ;3 I almost always traded it in asap for a an RDS or an ACOG.

[toysrme]

no, because as i stated, it will grab from any direction.
unless you want to make the case that recoil always recoils towards the target, which we both know is not true LoL!

[mannon]

No but it will recoil both to the right and the left and back again as you keep firing. Your shot grouping almost always has bullets on both sides of your aim point which means sooner or later in the clip the recoil will push you closer to the target regardless of what side it's on.

I doubt that aim assist actually gets stronger the longer you go full auto. The gun simply getting closer by random kick seems more likely. Which also means that your results could be from higher recoil, but I don't think the Holo increases recoil so I would doubt it.

I would think that the best way to test aim assist would be firing single shots to eliminate recoil completely as a factor, though it might be difficult to do that.

[ssog]

Dec 4, 2009 19:26:09 GMT -5 mannon said:

No but it will recoil both to the right and the left and back again as you keep firing. Your shot grouping almost always has bullets on both sides of your aim point which means sooner or later in the clip the recoil will push you closer to the target regardless of what side it's on.

I doubt that aim assist actually gets stronger the longer you go full auto. The gun simply getting closer by random kick seems more likely. Which also means that your results could be from higher recoil, but I don't think the Holo increases recoil so I would doubt it.

I would think that the best way to test aim assist would be firing single shots to eliminate recoil completely as a factor, though it might be difficult to do that.

From what I understand, the reticle is coming to rest on the enemy. If that's the case, then it's clearly aim assist. Recoil might move the gun randomly to the left or the right, but when you stop firing, it always centers back on where it was aiming before you started.

[mannon]

It does, but wouldn't the aim assist key off of where the sights are at any given moment rather than where they will come to a rest if you stop firing?

In other words if you get a lucky recoil on the first round that puts your sights closer to the target might that not be just enough closer for the aim assist to decide to kick in and nail the target for you?

If the aim assist is kicking in halfway through the clip rather than on the first shot then the only thing I see that has changed is that the recoil has knocked the sights around randomly for a bit.

I find it rather unlikely that aim assist gets stronger the longer you keep firing on full auto. But the recoil+aim assist could explain these observations.