Author Topic: Collision Model Question  (Read 1039 times)

Offline crims

  • Nickel Member
  • ***
  • Posts: 411
Collision Model Question
« Reply #15 on: August 14, 2004, 05:36:54 PM »
Quote
Originally posted by Karnak
You are not punished for a better connection.



Well then can you explain why only 1 Plane takes dammage >?
If we are in the same place at the same time and only 1 gets damage. This is the Question that I have asked ?



Crims
479th Raiders
Life's to Short Don't Forget To Laugh 479Th Raiders FG

Offline Arlo

  • Radioactive Member
  • *******
  • Posts: 24760
Collision Model Question
« Reply #16 on: August 14, 2004, 05:48:08 PM »
Quote
Originally posted by crims
Well then can you explain why only 1 Plane takes dammage >?
If we are in the same place at the same time and only 1 gets damage. This is the Question that I have asked ?
 


Because no matter how good or bad your connection is, the internet inherently has lag and a server has x amount of delay between your signal and your opponent's and sometimes has to anticipate where either of you will be in relationship to the other for x number of milliseconds and corrects that constantly via HT's excellent smoothing code?

When you fire x% of bullets in a situation where you're on a wildly evasive opponent's six o'clock, part of your hit percentage loss is probably due to such lag, but not enough to make it impossible to kill the guy. (Question: What's the highest individual hit percentage currently in AHII?)

When you're closing (or even flying circles with) an opponent and he actually moved to avoid and you didn't (or visa versa) and that little teeny tiny inherent bit of lagtime told the server that your frontend saw a collision and his didn't (or visa versa) - why penalize the guy who actually successfully avoided the collision just because of the inherent internet lag?

Offline ZZ3

  • Zinc Member
  • *
  • Posts: 71
Collision Model Question
« Reply #17 on: August 14, 2004, 07:53:42 PM »
That really doesnt answer the question.

Offline Ghosth

  • AH Training Corps (retired)
  • Plutonium Member
  • *******
  • Posts: 8497
      • http://332nd.org
Collision Model Question
« Reply #18 on: August 14, 2004, 08:01:48 PM »
With gunnery lag doesn't come into play.

You shot at something you saw at on your FE. You either hit it or not. If you did it sends a packet to the other guy saying you got hit here.  

His FE says ok this many hits here tears off a wing & he dies.

With collisions lag is very much in play.

Get a squadmate or freind to follow you.

Somewhere between 1k & 500 feet back.

Now compare notes on the distance he sees vs the distance you see.

It can be as little as 10 - 50 feet. Which still makes a HUGE difference in a collision.

OR, as much as 4 to 500 feet.

So now you pass nose to nose, one guy sees that he misses by 300 yards. The other guy sees a collision.

Your going to kill the guy that saw that he missed it?

Or not kill the guy who didn't avoid it?

Either alternative is totally porked. IT CAN ONLY BE THE WAY IT IS!

The problem is there was a problem with collsions. Namely there were hardly any.  So people forgot to DUCK!

Start flying to not hit the other guy and the collision problem is gone.

Anyone have a link handy for hoofs article on netlag?  Should be required reading before your allowed to log on.

Offline crims

  • Nickel Member
  • ***
  • Posts: 411
Collision Model Question
« Reply #19 on: August 14, 2004, 11:32:53 PM »
So KNife fighting is out of the question now. And also good Loop fighting, And turn fighting. Because If I fly to Close to the Enemy I will hit him. And if There is a Post of Internet Lag that I should read before I log in.......Maybe Hitech should account for it also in the collision model. All of this Info Is Great But if the Program < or Internet lag > sees one plane hit and the other one not hit. That is My Question. How can one plane get damige and one not if they collide? Who did I collide with if there is only 1 plane damaged?




Crims
479th Raiders
Life's to Short Don't Forget To Laugh 479Th Raiders FG

Offline Arlo

  • Radioactive Member
  • *******
  • Posts: 24760
Collision Model Question
« Reply #20 on: August 15, 2004, 03:18:27 AM »
Quote
Originally posted by crims
So KNife fighting is out of the question now. And also good Loop fighting, And turn fighting.  


No they're not. You just have to learn to accept that there WILL be times that either you or your opponent will come out of a collision unscathed. There will be times neither of you do.

Offline ramzey

  • Gold Member
  • *****
  • Posts: 3223
Collision Model Question
« Reply #21 on: August 15, 2004, 03:31:38 AM »
i dont understand one thing
HTC "cut" effective range of shooting and same time rise collision radius?

thats make not sence for me

Offline Pongo

  • Platinum Member
  • ******
  • Posts: 6701
Collision Model Question
« Reply #22 on: August 15, 2004, 03:37:53 AM »
If there is a collision, it should always affect both or niether..no matter what.
If one guy avoids it..you both avoided it. Period.
HT has it backwards.

Offline zorstorer

  • Silver Member
  • ****
  • Posts: 950
Collision Model Question
« Reply #23 on: August 15, 2004, 04:15:25 AM »
Didn't the Russians use ramming as a combat technique?

Offline Zwerg

  • Copper Member
  • **
  • Posts: 125
Collision Model Question
« Reply #24 on: August 15, 2004, 05:27:05 AM »
Quote
Originally posted by Pongo
If there is a collision, it should always affect both or niether..no matter what.
If one guy avoids it..you both avoided it. Period.
HT has it backwards.


Would mean: the AH server has to track the collision messages...
Don't think they could do it at 14.95$

Offline phookat

  • Nickel Member
  • ***
  • Posts: 629
Collision Model Question
« Reply #25 on: August 15, 2004, 10:31:27 AM »
Quote
Originally posted by Pongo
If there is a collision, it should always affect both or niether..no matter what.
If one guy avoids it..you both avoided it. Period.
HT has it backwards.


This is a distributed environment.  There is never one completely correct "master" or synchronized database.  There is always error.  If we implement your policy and err on the side of more collisions (only one collides --> both collide), then you encourage ramming as a method of killing.  If you err the other way (only one avoids --> both avoid), then you encourage staying on target without consideration of collision.

IOW, your policy sets up a prisoner's dilemma.

The best solution in light of the distributed database is to have each FE responsible for collisions on his own plane.  That way there's no prisoner's dilemma.  Ramming can still happen, but you can't know for sure that your ram will kill the other guy, so ramming is discouraged.  Disregarding collisions is also discouraged, for the same reason.  


Now if they could just fix the fact that planes are "colliding" when they are 50 feet away...we'd be set.  There is error here too, of course--the collision bubble can be a bit too big or a bit too small.  I say it should err on the small side.  If the bubble is too big, you can't really tell when you are about to collide to avoid it.  And a slightly too small collision bubble will work to detect "real" collisions (how often does your wing hit by 5" but not 10").

Offline ZZ3

  • Zinc Member
  • *
  • Posts: 71
Collision Model Question
« Reply #26 on: August 15, 2004, 12:20:04 PM »
If it can not be done with any effectiveness, then turn it off.
As it is now, it's flawed, and very unfair.

Offline Arlo

  • Radioactive Member
  • *******
  • Posts: 24760
Collision Model Question
« Reply #27 on: August 15, 2004, 01:02:55 PM »
Quote
Originally posted by ZZ3
If it can not be done with any effectiveness, then turn it off.
As it is now, it's flawed, and very unfair.


Actually .... as it is right now is just fine. And it certainly isn't unfair. The server reacts to what you get on your FE. If you don't avoid ... you collide. If you avoid, you don't collide. Don't get much more fair than that. Work on collision avoidance. If the other guy is intact, he obviously did.

Offline hitech

  • Administrator
  • Administrator
  • *****
  • Posts: 12425
      • http://www.hitechcreations.com
Collision Model Question
« Reply #28 on: August 15, 2004, 01:09:59 PM »
ramzey: The radius shouldn't have increased but your planes are more sollid then in AH1, there were possible gapes in the detection in ah1. Ill do some more testing to make sure it is working as designed.


HiTech

Offline Flayed1

  • Silver Member
  • ****
  • Posts: 1091
Collision Model Question
« Reply #29 on: August 15, 2004, 02:05:32 PM »
I haven't had 1 collision yet and I usually only fly the A6M2 and FM2 (turn fighters) killed about 5 or 6 guys so far when I can catch them and I've been at 200 or closer to kill them. :)  had 1 close call as a P38's entire tail section flew by my right wing.
From the ashes of the old we rise to fly again. Behold The Phoenix Wing!