Chance to hit (accuracy) - does not sense in 3.1 (?)

The place to discuss balance changes for future versions of the game.
(Master releases & 3.X)
Deus Siddis
Trained
Trained
Posts: 235
Joined: 18 Aug 2007, 06:58

Re: Chance to hit (accuracy) - does not sense in 3.1 (?)

Post by Deus Siddis » 22 Mar 2013, 07:51

Iluvalar wrote: Ok ! I'm listening to your suggestions. How would you fix that ?
Probably not all of the following adjustments will be necessary but I like to cast a wide net--

* Make sure the unit AI understands it's optimal attack range and the optimal attack range of it's target, under the new accuracy system. If small random evasive maneuvers makes a noticeable difference to unit survivability then consider turning this feature back on by default. You don't want the player to have to do unnecessary micromanagement of units.

* Figure out what to do with accuracy upgrades for non-homing weapons, if it turns out they make too little of a difference and can't be fixed. They could be focused more on target prediction or replaced with higher 'muzzle velocity' upgrades or could be dropped altogether.

* If average hit rates go up then you can scale down DPS on weapons and weapon upgrades so that battles last about the same amount of time, to preserve pacing.

* If some bodies or cyborgs seem to be at a disadvantage under the new system then their hitpoints or collision radius can be adjusted. If both accuracy increases and mobility/positioning is better rewarded, then these two things may partially cancel each other out.

* If some weapon categories become too accurate or inaccurate then adjust their damage rates or even mass or cost.

Per
Warzone 2100 Team Member
Warzone 2100 Team Member
Posts: 3777
Joined: 03 Aug 2006, 19:39

Re: Chance to hit (accuracy) - does not sense in 3.1 (?)

Post by Per » 22 Mar 2013, 11:00

Does the size of the opponent's droid/structure matter? If the hit roll succeeds, the projectile will aim straight at the middle of the bounding box, and for short and medium range projectiles, there is simply not enough time for the droid to move far enough to "dodge the bullet". The project will hit. If there is a miss, then size will matter and the result is unpredictable, non-intuitive and poorly balanced, as Iluvalar correctly points out. The problem arises because of the combination of two different accuracy systems.

If the target is far away, and the projectile is not homing, then size will also matter, even if hit roll succeeds. This is the case where a physics based system seems intuitively right. If the droid is moving, it will be harder to hit, and this will presumably visually "look right" to the player. This intuition is mistaken. If it were the case that projectiles would always target where the droid was when the projectile is fired, then it would create a very nice visual effect, where the missiles would always hit behind a moving droid. It would be intuitive how to deal with such projectiles, it would be easy to understand, and it would make such projectiles completely useless against approaching enemies. Hence we have target prediction code that that takes movement into account and makes such projectiles useful but non-intuitive and (ironically) very hard to predict for the droid owner. I think it is practically infeasible to fool the target prediction by micro-managing your droids with random movement -- experienced players please tell me if I am wrong about this. Even if you could predict such projectiles, the hit roll misses makes understanding the pattern very hard.

And no, you do not want to add random movement to all droids. It will just look silly and make user control harder. If you add something to one side of the balance equation that cancels out something on the other side, it had better be good looking and very intuitive, since it will have no real effect.

So we will fix the problems by simply removing either physics based or hit roll based accuracy, or both. I favour removing the hit roll, although I am fine with either of the other two alternatives as well. Perhaps we should do a straw poll.

Cyp
Evitcani
Evitcani
Posts: 778
Joined: 17 Jan 2010, 23:35

Re: Chance to hit (accuracy) - does not sense in 3.1 (?)

Post by Cyp » 22 Mar 2013, 12:50

Per wrote:… I think it is practically infeasible to fool the target prediction by micro-managing your droids with random movement -- experienced players please tell me if I am wrong about this. …
Probably true, with the notable exception of long-range non-homing artillery, where it even often misses due to droids changing direction on their own (due to pathfinding or similar).
Per wrote:So we will fix the problems by simply removing either physics based or hit roll based accuracy, or both. I favour removing the hit roll, although I am fine with either of the other two alternatives as well. Perhaps we should do a straw poll.
Think hit roll should be changed to a gaussian distribution in the firing angle (or for homing, maybe a per-shot-fixed gaussian displacement in the target location), but just removing hit roll would probably make more sense than what there is now (but would probably make mini-rockets much more powerful in early game than they already are, if not adjusting something to compensate).

User avatar
aubergine
Professional
Professional
Posts: 3459
Joined: 10 Oct 2010, 00:58
Contact:

Re: Chance to hit (accuracy) - does not sense in 3.1 (?)

Post by aubergine » 22 Mar 2013, 16:46

Per wrote:… I think it is practically infeasible to fool the target prediction by micro-managing your droids with random movement -- experienced players please tell me if I am wrong about this. …
I regularly halt units when I see indirect non-homing projectiles heading for them - the projectiles don't detect that the unit has stopped and thus blast the ground where the unit would have been if it didn't stop. In WZ 3.2, halting units is much easier (new keyboard shortcut) so I would imagine this technique will become easier to use and thus more effective.
"Dedicated to discovering Warzone artefacts, and sharing them freely for the benefit of the community."
-- https://warzone.atlassian.net/wiki/display/GO

Originway
Trained
Trained
Posts: 412
Joined: 08 Aug 2012, 06:22

Re: Chance to hit (accuracy) - does not sense in 3.1 (?)

Post by Originway » 22 Mar 2013, 19:48

Cyp wrote:
Per wrote:… I think it is practically infeasible to fool the target prediction by micro-managing your droids with random movement -- experienced players please tell me if I am wrong about this. …
Probably true, with the notable exception of long-range non-homing artillery, where it even often misses due to droids changing direction on their own (due to pathfinding or similar).
Per wrote:So we will fix the problems by simply removing either physics based or hit roll based accuracy, or both. I favour removing the hit roll, although I am fine with either of the other two alternatives as well. Perhaps we should do a straw poll.
Think hit roll should be changed to a gaussian distribution in the firing angle (or for homing, maybe a per-shot-fixed gaussian displacement in the target location), but just removing hit roll would probably make more sense than what there is now (but would probably make mini-rockets much more powerful in early game than they already are, if not adjusting something to compensate).
would it not change the whole way campaign was balanced if the current way was removed?
then you got the new targeting system here http://developer.wz2100.net/ticket/3748 to see how it works and which we are waiting for someone to make us builds to test :annoyed:

User avatar
Shadow Wolf TJC
Regular
Regular
Posts: 1047
Joined: 16 Apr 2011, 05:12
Location: Raleigh, NC

Re: Chance to hit (accuracy) - does not sense in 3.1 (?)

Post by Shadow Wolf TJC » 24 Mar 2013, 09:46

Per wrote:So we will fix the problems by simply removing either physics based or hit roll based accuracy, or both. I favour removing the hit roll, although I am fine with either of the other two alternatives as well. Perhaps we should do a straw poll.
I'm in favor of removing the old hit roll based accuracy system, and replacing it with a Supreme Commander inspired system where weapons have a "max offset angle" variable, which controls how far away (in degrees) from the target the weapons may deviate from the target when fired (a value of 0 would mean the weapon would have perfect accuracy at all ranges, but could still miss if the target's velocity changed). We could also add a "turn rate" variable to homing weapons that would determine how fast homing projectiles can turn to face their targets (so that there's a chance that even they could miss their targets, even if no terrain is present to block them).

Since we're switching from a .txt based format to a more modular .ini based format, I'd imagine that this change would be easy to implement for a future beta of 3.2 without breaking a whole lot of compatability.

By the way, do you think we could add a BEAM_DIRECT constant to the movement variable in weapons.ini that would tell the game to draw a colored line from the weapon to the target, and instantly damage the target if the beam weapon hits, instead of creating a projectile to send to the target (which may or may not hit)?
Creator of Warzone 2100: Contingency!
Founder of Wikizone 2100: http://wikizone2100.wikia.com/wiki/Wikizone_2100

raycast
Trained
Trained
Posts: 131
Joined: 12 Sep 2012, 19:16

Re: Chance to hit (accuracy) - does not sense in 3.1 (?)

Post by raycast » 24 Mar 2013, 14:17

Originway wrote: then you got the new targeting system here http://developer.wz2100.net/ticket/3748 to see how it works and which we are waiting for someone to make us builds to test :annoyed:
You really need to learn how to build it yourself.

That makes it so much easier to try out stuff yourself. It's really not hard (at least not on Linux), you just need to actually give it a try. Programming in C++ is a very valueable skill anyway.

Post Reply