Weapon Range

[DogFish]

I was playing around with weapon range in the arena this morning and got some strange results. Was hoping someone could replicate the results and explain these.

Purpose: Attempt to determine the range of 6 things, Feint(with hammer), Forceful Blow, Melee(hammer), Feint(Spear), Gutting, Melee(Spear)

Experiment: Two players in the arena, call them A(ttacker), D(efender). D remains in the same spot at all times. Put A into walk mode. Move A close enough to D to cast the power/attack. Record position. Tap 's' to move back a bit. See if A can still attack. Continue until A cannot attack. Repeat around the maximum distance to verify distance.

Results: Attacker, Defender, Distance
Feint(Hammer): 1837.04,1108.56, 1840.26,1107.58, 3.37
ForcefulBlow: 1837.75,1108.23, 1840.26,1107.58, 2.59
Melee(Hammer):1837.73,1108.25, 1840.26,1107.58, 2.62

Feint(Spear): 1836.06,1108.69, 1840.26,1107.58, 4.39
Gutting: 1836.06,1108.69, 1840.26,1107.58, 4.39
Melee(Spear): 1836.96,1108.45, 1840.26,1107.58, 3.41

These results are notably higher than the Range 2 on a Spear.


Questions:
1) Can anyone replicate these results?
2) Feint and Gutting are Range 0 spells. The spear claims to have Range 2 in the item description.
2a) Melee range isn't even the range specified in the item description, why?/how does melee range relate to item Range?
2b) Range 0 spells have range 0.8-1.0 m further than the melee range, why?/What really is Range 0?

[Kitsuni]

Well to put this simply, players are not actually where you see them, due to ping times and positioning errors (position bug).

Also, positioning is less accurate the further away someone gets.

Find an ally and both mount a horse and ride to a location. You will see your ally riding behind you, but he will see you riding behind him.

[DogFish]

The horse example can be explained by realizing there is a delay between Client B->Server and Server->Client A. Client A sees Client B in the past.

Well to put this simply, players are not actually where you see them, due to ping times and positioning errors (position bug).

Could you explain this with more details?

Ping Time:
Both A and D are stationary at the time of attack. Both A and D know their own absolute coordinates. A should know D's absolute coordinates, since D does not move. Network latency should not affect Client A's calculations for distance. If the calculation is done server side, interpolation error should not exist since both players are not moving at the time of attack.

By positioning error, are you claiming that the position of player D on client A is not the same as the absolute coordinate of player D? Or position errors due to delays in position updates?

A spear "claims" to have a range of 2. Players can Feint from 4.4 m, that's an error of 2.4 m.

Both of us have pings around 150, which I don't think is terrible. Both players are stationary, which should remove most, if not all, errors from interpolation. I would think that 2.4 m is a lot for network latency to explain all of it.

[SilverHaven]

Originally Posted by DogFish

If the calculation is done server side, interpolation error should not exist since both players are not moving at the time of attack.

If this was true, I would never be kicked or feinter from a player from 10m away. My pings are around 80-100 (ADSL), i was in arena with a friend whose pings are 300-2k (GPRS), because my pings are lower, my client would have more accurate information on 'real' positions whilest his client would have wrong ones. Right know he has my position wrong, he see me next to him and can kick me, while my real position is way too far for a kick, but it gets casted. Players with worse ping, connection, just have kind of advantage over those with good connection. Not only positional, but perhaps even resist rate. And the worst players are those teleporting from place to place, you get to them and they are 10m away in a nanosecond, impossible for you to kill them, while they can kill you. Fortunately I don't see many of those as I did in the past.

[DogFish]

You can be kicked by your friend because NGD implements Lag Compensation Time.
Your friend receives your position later because of his latency. In his client, you are in range, so his client sends a "Cast Kick" to the server. The server takes into account his latency, goes backwards in time from the last positional update that he received, and sees that according to your friend, you were in range, and thus allows the Kick to be casted.

This shouldn't apply to the experiment I described. Your friend can attack from a seemingly further range because he has incorrect information about your position. Even if Client A had enormous latency, Client A would still have the correct position for Player D, because Player D doesn't move.

Perhaps I waited too short of a time, maybe a position update hadn't been published by the server. I can try again with longer wait times before attempting to attack if the consensus so far is that the error is explained by network latency.

I agree that latency can make it appear that player's are further away than they truly are, however if both players stop, then both clients and the server should converge to the correct positions of both players. Thus this should not affect the distance calculation for weapon range.

[Kitsuni]

Originally Posted by DogFish

By positioning error, are you claiming that the position of player D on client A is not the same as the absolute coordinate of player D? Or position errors due to delays in position updates?

A spear "claims" to have a range of 2. Players can Feint from 4.4 m, that's an error of 2.4 m.

Both of us have pings around 150, which I don't think is terrible. Both players are stationary, which should remove most, if not all, errors from interpolation. I would think that 2.4 m is a lot for network latency to explain all of it.

Let me explain this better a little better. There is a reason we call this "position bug" and not how positioning is meant to be, because you can often try to hit an opponent as a warrior and get "not facing" or "out of range" even on someone who is knocked or immobilized. There is even a series of screenshots somewhere on this forum of people all in the same room around a table, and every single persons's client saw everyone else in different positions, while sitting down.

So I summarize again, the network code is relatively poor and it is buggy and inaccurate. There is no logic to it.

By the way the horse example with one player seeing the past would not occur in other games due to network prediction. Many games like Real Time Strategy (RTS) and FIrst Person Shooter (FPS), would completely fall apart if that was the case, but they don't due to well-implemented prediction models that NGD lacks.

[SilverHaven]

Originally Posted by DogFish

You can be kicked by your friend because NGD implements Lag Compensation Time.
Your friend receives your position later because of his latency. In his client, you are in range, so his client sends a "Cast Kick" to the server. The server takes into account his latency, goes backwards in time from the last positional update that he received, and sees that according to your friend, you were in range, and thus allows the Kick to be casted.

Good enough explanation for me. Though the fact is that ppl with worse connection have advantage over those with better while it would be more logical to have it the oposite way. But well, I don't even know if I care anymore

[_Enio_]

Originally Posted by Jaroslav

Good enough explanation for me. Though the fact is that ppl with worse connection have advantage over those with better while it would be more logical to have it the oposite way. But well, I don't even know if I care anymore

Its annoying, but you should know about these as you can - well id not call it abuse, but.. get less wtf moments on yourself and give more wtf moments on your opponents.

Also theres a huge difference in wether it is a direct attack/spell or an area spell.

Areas use the positions of people as seen on the server, spells use positions as seen on the client up do some margin (thats why you can get knocked from a higher then usual distance under certain movement constellations.

Extreme case:
Lets say you stand still and a barb rushs on you with normal speed, when he is 10m away you start running away back direction. Now after 2s of you already running away you get knocked - wtf - on your screen you saw him still 10m away running after you.

Now on the Barbs screen: The moment you saw him on range 10, he saw himself already on 5m near you, now when you moved on your screen he didnt see you move yet on his screen due to delays.

So he has like 2 more seconds to cross 5m before you would be moving away on his screen and of course is in range to kick you down because its a single target spell. (The server might have seen you two in 5+m distance to eachother, but the casters screen counts for these spells.)

If the barb would try to area you in such scenario he would likely missed you because he sees you standing in one spot on his screen while on the server you might be already moving away.

Note: These numbers can vary due to different ping delays and server load and movement speeds etcetc.


Key is, when you know about these mechanics you can adapt, adjust your timings and movements to get the best out of it, be it offensive or defensive.

(I.E. When someone runs towards a barb the barb can cast his area stun earlier then youd think from the visual distance assuming the target running towards him is already 5m closer from server perspective. OR a barb who is chased by another barb in a certain distance can quickly turn towards the barb and get a kick in, even when the chasing barb is hammering kick too.. stuff like this.)


Wall of text.. anyways, gnight.