Jump to content

I Think I Isolated The Probable *causes* Of Hit Detection Issues


12 replies to this topic

#1 JimboFBX

    Member

  • PipPipPipPipPipPip
  • 345 posts

Posted 26 July 2013 - 10:59 PM

These are theories of course. If I owned the code I would do some research before making such declarations of the cause. Any one of these could be true or at least a theory to investigate.

1. Servers have too much strain and are de-prioritizing ballistics hit calculation, resulting in reduced accuracy. This results in projectiles "passing through" things they shouldn't. I notice that hit accuracy seems to vary from game to game and when I played at a time when less people did I noticed great hit detection.

2. For SRMs there is a very small (maybe 1 tick?) delay from when the missile explodes to when the damage is done. If a mech's hitbox moves away from the missile in that 1 tick it won't count because you set the explosion radius to dinky. This, combined with #1, creates issues because when the server load is higher the time between ticks is larger.

3. HSR doesn't appear to take into account where the USER thinks they are. Basically it's like this:

If I'm "playing" tick number 10000 then by the time my position gets to the server it'll be tick 10020. I think what's happening is I shoot at tick 10000, but my position is actually where I'm at at tick 10020. What I don't get is that I would think if this was the case we would see drastic issues with hit detection when we move, but instead we only see minor issues. Perhaps there is something that's reducing this effect, if this is happening? Another thing to consider is processing order - are we processing HSR prior to processing a player's position in a tick?

4. HSR might not be taking into account the 100 ms interpolation that's typically on by default for cryengine 3 unless you guys disabled it.

5. Player positions might be extrapolated. This is common in games without lag compensation - you extrapolate positions based on pings so that by the time your packet gets there it's correct. The unreal engine does this. If positions are extrapolated then I'm actually seeing mechs ahead of where they really are.

#2 FREDtheDEAD

    Member

  • PipPipPipPipPipPip
  • The 1 Percent
  • 406 posts
  • LocationSouth Autstralia

Posted 27 July 2013 - 10:40 PM

With players reporting being hit but not taking damage (and the 'being hit on the opposite side' issues that have been reported over the months) there's a chance there are other hit detection bugs apart from the apparently flawed HSR and interpolation code.

A developer said they found a problem with client lag jitter. He heavily implied that they were exploring increasing frequency of UDP 'pings' to solve it. That was a month or so ago.

#3 Fire and Salt

    Member

  • PipPipPipPipPipPipPip
  • 526 posts
  • LocationFlorida

Posted 28 July 2013 - 01:36 AM

if a weapon hits you on you end, does that necessarily mean that you should take damage?

Lets say a enemy shoots at a rock. You are running towards that rock, but are not yet there. The enemy sees himself shoot the rock, just before you get there.

Now on your end, it takes a second for the server to say -"hey - this guy shot that rock"

By the time your computer gets the message, you are already in front of that rock. So on your end, it looks like he hit you - but really he hit the rock and you were just in the way.





Hmmmm......................

#4 Roland

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • 8,260 posts

Posted 28 July 2013 - 01:35 PM

I've noticed some things today that I think definitely link the current hit detection problems to the shooters OWN movement, rather than the target.

However, I think it's something different from what the OP describes.

In cases where I am moving east, firing at a target north, my shots will seemingly miss to the right of him (that is, to the east of him, in the direction of my movement), even if they appear to hit perfectly on my screen. He will see them mising to the east.

I'm not sure what is causing this, but I think it may be a root cause of a lot of the hit detection issues.

#5 Asmosis

    Member

  • PipPipPipPipPipPipPipPipPip
  • Ace Of Spades
  • 2,118 posts

Posted 28 July 2013 - 05:21 PM

View PostFire and Salt, on 28 July 2013 - 01:36 AM, said:

if a weapon hits you on you end, does that necessarily mean that you should take damage?

Lets say a enemy shoots at a rock. You are running towards that rock, but are not yet there. The enemy sees himself shoot the rock, just before you get there.

Now on your end, it takes a second for the server to say -"hey - this guy shot that rock"

By the time your computer gets the message, you are already in front of that rock. So on your end, it looks like he hit you - but really he hit the rock and you were just in the way.





Hmmmm......................


This.

with the way HSR is setup, taking damage when you didnt get "hit" or not taking damage when you did get "hit" should be expected anytime someone with 100+ ping is on the other team, so pretty much every game.

It does *NOT* mean that person has an unfair advantage. they still take the time to aim and shoot at you after all, your just a lil late to hear about it. by the same token, they experience an almost identical thing since you shoot them, and due to their lag they are just as late to hear/see/feel your shots.

Theres all sorts of room for bugs and miscalculations though.

#6 mack sabbath

    Member

  • PipPipPipPipPipPipPipPip
  • The God
  • The God
  • 1,073 posts
  • Twitch: Link
  • LocationUSA

Posted 28 July 2013 - 05:58 PM

Why do all the weapons work flawlessly in Testing Grounds, then turn to poop the next actual match?

#7 Theronlas

    Member

  • PipPip
  • 25 posts

Posted 28 July 2013 - 09:50 PM

View PostDie Primate Die, on 28 July 2013 - 05:58 PM, said:

Why do all the weapons work flawlessly in Testing Grounds, then turn to poop the next actual match?


Because the target mechs in testing ground never moved. Thus there is no difference between the server's and your client's ideas of where the weapons hit. thus HSR is largely irrelevant and any issues will be impossible to detect.

That is assuming the testing ground is actually server side. Given that none of the mech tree upgrades appear to function there at all, it is possible that the testing ground is actually client side.

Edited by Theronlas, 28 July 2013 - 09:54 PM.


#8 Deathlike

    Member

  • PipPipPipPipPipPipPipPipPipPipPipPipPipPip
  • Littlest Helper
  • Littlest Helper
  • 29,240 posts
  • Location#NOToTaterBalance #BadBalanceOverlordIsBad

Posted 28 July 2013 - 09:59 PM

There are times where my shots, whether projectile (PPC/Gauss) or hitscan (LL, meds) seem to successfully deal a "hit" according to the cursor (when it turns red, it's a registered hit, according to the game/server) when I'm occasionally "shooting through a hill". I'm getting the strong feeling that something is seriously out of sync between the client and the server when it comes to the physics that is running on the server.

Edited by Deathlike, 28 July 2013 - 09:59 PM.


#9 JimboFBX

    Member

  • PipPipPipPipPipPip
  • 345 posts

Posted 28 July 2013 - 11:46 PM

testing grounds is just singleplayer. notice how it loads much faster

#10 Elizander

    Member

  • PipPipPipPipPipPipPipPipPipPip
  • Legendary Founder
  • Legendary Founder
  • 7,540 posts
  • LocationPhilippines

Posted 29 July 2013 - 03:07 AM

Testing Grounds is 100% on your computer. There is no server to talk to. :)

#11 Nebelfeuer

    Member

  • PipPipPipPipPipPip
  • 302 posts

Posted 29 July 2013 - 03:18 AM

As far as conection issues (ping) are concerned we will probably have to live with that and adjust our aim accordingly per target.
Server strain seems to be a mayor part though that should be adressed if possible.

#12 ManDaisy

    Member

  • PipPipPipPipPipPipPipPipPip
  • Elite Founder
  • Elite Founder
  • 3,272 posts
  • LocationKing Of Flower Beds

Posted 29 July 2013 - 09:34 AM

I am concerned distance is being calculated incorrectly when carrying over the snapshot when the rewind occurs. If that isn't preservered then you will have decreased damage on smaller fast targets between the calculation time and no damage do to range max outs due to targets moving significantly since the time of fire.

#13 B0oN

    Member

  • PipPipPipPipPipPipPipPipPip
  • 2,870 posts

Posted 29 July 2013 - 09:55 AM

View PostDie Primate Die, on 28 July 2013 - 05:58 PM, said:

Why do all the weapons work flawlessly in Testing Grounds, then turn to poop the next actual match?


Because testing ground runs "local" which means it is being computed solely by your PC, so 0 latency .





3 user(s) are reading this topic

0 members, 3 guests, 0 anonymous users