RNG update at Amun!

[ShadowForce]

Originally Posted by Iheartpancakes

Has anyone had any luck testing Warlock on amun?

Originally Posted by ShadowForce

I've spent a good few hours testing on Amun now, mainly on my warlock. In my honest opinion there are still too many resists present. More than 1 resist in a single pvp is too many Imo. Like it has been mentioned above; remove random resists completely and add or rework some spells in each class to provide spell resistance. There is still too much of a luck factor as things are. A dev mentioned somewhere that the update is supposed to bring in more of a skill factor. As it stands, you can master your class in certain scenarios and make every right move but still end up dying, that's not right I'm afraid. NGD, balancing sub classes to the best of your ability and then throwing in the roll of a dice contradicts and negates that said balance. People will argue that if there are no resists it is too easy to chain spells. What is failed to be mentioned though is factoring in team play. In a game predominantly designed around RvR, why should you be able to roam around the warzone alone, stumble across a lock or a hunter, and resist your way out of trouble randomly. No consequence for the risk?

In one pvp vs a marksman 3 of my spells were resisted (not casted on SOTW). Warlock is therefore still not a viable class in the warzone imo. I have a Warmaster Lock and have spent enough time playing the class to know that even with the new RNG on Amun, the resist level is still not low enough for this class to not feel frustrating.

[-Aniara-]

I just toasted a few barbs at haven, even with some resist a barb has as much chance as a ice cube in hell against a lock in wz. With no resists its not even worth trying against lock/hunter as a melee class.

And i am sure a rusty lock.

/A

[LawZ]

Alo.

Found some time to mess a little bit with new patch on Amun.

1. RNG seems working ok. There are some fancy situations from time to time, but it is quite good and robust in overall.

2. I get some weird results considering critical damage. In the photo below, i got a new (less fancy) setup this time, and got some (really few) hits on Dee. (Thank you Dee for being the dummy!)

As you can see, the "Ratio : Critical/Normal" can exceed even the "Worst Ratio : Max_Crit/Min_Normal" derived by character sheet. This could generally happen, but implies (at least) that the damage armor reduction is applied after the critical damage (?). If this is the case, it is weird in my opinion.



Anyway, just wanted to post it for further discussion. I repeat, RNG system seems better than the previous one.

Best,

[schachteana]

can i just ask one thing

what was wrong with regnums RNG? It used to produce real random numbers, right? and now, it was changed so the chance for resist/evade/whatever-chains is smaller? so, ngd basically manipulated the probability?
as far as i know, people weren't really having a problem with how probability works but rather with the amount of resists in general, which used to be (and still is on liveserver) about 30% for non-damaging stun spells. Lowering the resist rate in general would have been enough, I'd say, but definitely necessary nevertheless.
since this is probably solved now, I shouldn't complain anyway. there is so much to do for ngd, I hope they'll get around to everything one day.

[Ivramartono]

Can any1 give me an explanation how the RNG/resistrate works...? I cant find one anywhere. i dislike to estimate things, i need accurate results

[LawZ]

Originally Posted by schachteana

can i just ask one thing

what was wrong with regnums RNG? It used to produce real random numbers, right? and now, it was changed so the chance for resist/evade/whatever-chains is smaller? so, ngd basically manipulated the probability?
as far as i know, people weren't really having a problem with how probability works but rather with the amount of resists in general, which used to be (and still is on liveserver) about 30% for non-damaging stun spells. Lowering the resist rate in general would have been enough, I'd say, but definitely necessary nevertheless.
since this is probably solved now, I shouldn't complain anyway. there is so much to do for ngd, I hope they'll get around to everything one day.

As far as i understood, the previous RNG was working well in the long-term, but not in the short-term. So i guess they changed their algorithms to reach the expected probability in less hits, i.e. in short-term (check this).
Not really sure if they manipulate the probability, using any soft of conditional probabilities, but the new RNG systems seems to be more robust in the short-term than the previous one.

Originally Posted by Ivramartono

Can any1 give me an explanation how the RNG/resistrate works...? I cant find one anywhere. i dislike to estimate things, i need accurate results

I don't think NGD will ever present the exact formulas, so as to avoid any game exploit by the user-base.

As Adrian said in a previous post, there is much work to be done yet. I doubt that this patch will come soon on live servers.

[pieceofmeat]

Current RNG favours the once with bad ping or something, so id rather call it RNGD.

It so obvious that players like candy, sylaedgon, bongee and zord etc (no offence) often have much higher resist rates than others.

[Candyx]

It seemed to me that the old RNG was failing because the server couldn't keep with with the number of calls and so there were repeats in the RNG. Either that or NGD precalculates a nice long list of probabilities before hand and these are drawn from in order, but the issue being the server was too slow to allow a sync lock around the increment of which number is being drawn from and so what happens is multiple calls are made in quick succession and all receive the same random number result, since the increment from the first call hadn't been processed yet.

Either way it's fixable. I'm not 100% sure why the evade mechanics have been changed, if NGD agrees the underlying problem was the RNG then why bother changing the mechanics? Don't get me wrong, the way it used to work (still does on haven) wasn't ideal, HC had almost no value since the rate of evades is low and most of the time people aren't hitting normals.

I'd like HC brought back and for it to effect where the mean damage is (while keeping the same range). Higher hit chance increases the mean damage; moving it away from the median and towards the maximum.this would have to be based on a skewed probability distribution *but* this can be done with very few server cycles.

[LawZ]

Originally Posted by Candyx

It seemed to me that the old RNG was failing because the server couldn't keep with with the number of calls and so there were repeats in the RNG. Either that or NGD precalculates a nice long list of probabilities before hand and these are drawn from in order, but the issue being the server was too slow to allow a sync lock around the increment of which number is being drawn from and so what happens is multiple calls are made in quick succession and all receive the same random number result, since the increment from the first call hadn't been processed yet.

Either way it's fixable. I'm not 100% sure why the evade mechanics have been changed, if NGD agrees the underlying problem was the RNG then why bother changing the mechanics? Don't get me wrong, the way it used to work (still does on haven) wasn't ideal, HC had almost no value since the rate of evades is low and most of the time people aren't hitting normals.

I'd like HC brought back and for it to effect where the mean damage is (while keeping the same range). Higher hit chance increases the mean damage; moving it away from the median and towards the maximum.this would have to be based on a skewed probability distribution *but* this can be done with very few server cycles.

I was waiting for the expert to comment...!

Well i guess they assumed HC = 100% (absolute) for all players in order (maybe) to avoid the calculation "HC - Evasion" and alleviate the server computational burden? That could be a rational explanation, couldnt it?
And since they didnt know what to do with HC stat, they replaced it with Critical Damage.

[Candyx]

Originally Posted by LawZ

I was waiting for the expert to comment...!

Well i guess they assumed HC = 100% (absolute) for all players in order (maybe) to avoid the calculation "HC - Evasion" and alleviate the server computational burden? That could be a rational explanation, couldnt it?
And since they didnt know what to do with HC stat, they replaced it with Critical Damage.

If the only rational was reducing computational burden then we'll be back to playing pong at the next update. :P

Looking for values in a list/array is typically alot faster than calculating values. A million value long list of numbers can be precalculated and searched through in fractions of a second, typically much faster than doing any serious math operations. All of this seems to come to very little difference though since the RNG is changing.

My question: Should hit chance be changed to critical damage (as on test server) or be used as a method to shift the mean damage from normal hits? Whichever is best suited to the game mechanics should be chosen and code can be written and optimised to work within the constraints of the server.