Critical Damage Balancing/ mitigation....

[curryinahurry]

The problem with using the word Damage + a second word that causes similar but different defense benefits should be pretty obvious.  The simple solution would have been to rename one of the two values so that they are distinct in the players' minds.  A term like Armor Rating or Armor Value helps to give the impression of difference from Damage Reduction.  Then the designers just have to come up with some narrative to explain the differences.   It's pretty much how 99.9% of advertising works.

[GordonHalfman]

I wonder if they made a single armour rating or class value whether both DR and Res (or whatever they want to call it) could have a fixed relationship with it? i.e DR = AC and Res(%) = 2*AC or something like that. I can't remember what the typical values were before. Doesn't feel like it should be an issue either way.

[Sensuki]

Yeah Armor Rating or something would have been more sensible, absolutely.

[archangel979]

I have been playing a lot of cRPG and aRPGs. In any of them, if the game allowed players to stack critical chance and/or critical damage, those builds were always dominant and optimal and it always killed the fun of the building different characters.

IE games were cool because you could only minimally control critical chance and you could not control critical damage. One of the games where you could make different characters without one option being obviously dominant (well except for Kensai/Mages :D but even they were not about critical hits ).

 

In Both NWN games the most effective characters were critical hits based.

Edited January 10, 2015 by archangel979

[morhilane]

I wonder if they made a single armour rating or class value whether both DR and Res (or whatever they want to call it) could have a fixed relationship with it? i.e DR = AC and Res(%) = 2*AC or something like that. I can't remember what the typical values were before. Doesn't feel like it should be an issue either way.

 

You don't want to call it Armor Class, people will think it gets used in the hit resolution equation and wonder why there is more than one value too...

 

Anyway, they didn't remove the second value just to put it back in, the system of two damage resistance stats was redundant and added a layer of complexity to the damage calculation. I personally think they should remove all the percentage based stuff and go with fixed values everywhere. Screw the pretty curves on the graphs I say.

[IndiraLightfoot]

Perhaps I'm just imagining things, but isn't the obvious solution to make critical hits rare enough in occurrence as well as not having an insane effect on damage?

[Shevek]

Crits arent all that crazy in PoE and they werent a real big issue just one patch ago.

Edited January 10, 2015 by Shevek

[Luckmann]

 

I wonder if they made a single armour rating or class value whether both DR and Res (or whatever they want to call it) could have a fixed relationship with it? i.e DR = AC and Res(%) = 2*AC or something like that. I can't remember what the typical values were before. Doesn't feel like it should be an issue either way.

 

You don't want to call it Armor Class, people will think it gets used in the hit resolution equation and wonder why there is more than one value too...

 

Anyway, they didn't remove the second value just to put it back in, the system of two damage resistance stats was redundant and added a layer of complexity to the damage calculation. I personally think they should remove all the percentage based stuff and go with fixed values everywhere. Screw the pretty curves on the graphs I say.

 

 

Difficulty to balance notwithstanding, I'd personally prefer not one or two aspects to armour, but three; Damage Resistance (Percentage decrease to physical dmg), Damage Threshold (Flat decrease to dmg) and Dodge Modifier(s) (Penalty or even bonus to Dodge).

 

Gives you the possibility to really differentiate between various forms of armour, magic or otherwise, and specialize based on that.

 

The armour system in PoE is currently actually very flat, scaling more or less linearly between DT and Action Speed. It's not nearly as varied as the weapons, where I actually feel they've done a competent job of differentiating between all the various weapons (although some are clearly worse or better than other, atm, the bits and pieces are there).

 

It's just crazy that currently, you're usually best off just going in naked and try to kill everything before it kills you, and even if they revampt the values, it'll most likely end up just being a choice between really heavy armour, the lightest possible armour, or no armour, with no niching or specialization within the various forms of armour (ring mails vs. scale mails, full plate vs. other heavy armour, etc).

[Sensuki]

While I think three armor types would probably be over the top, I do agree that they have done a better job differentiating how weapons work (which could be done even better, but it's passable).

 

I don't really like how the weapon styles work though. Using a shield is more of a 'because I have to' in this game rather than something I actually want to do, because it really really hurts your character's DPS and there's no cool shield-based abilities or anything that other games have.
 

Weapons still aren't balanced vs DT, so I'm usually using two handers, ranged weapons or dual-wielding (for fast attacks on rogues with damage multipliers).

 

I'm glad pollaxes are finally fixed. They're actually pretty nasty.

Edited January 10, 2015 by Sensuki

[Hormalakh]

 

I am also surprised by the conclusion that the old DR (now removed) was a more complex notion than the new DR (old DT) system.

 

To me it just seems that saying if you're wearing light armor all damage is reduced by 10%, medium 15%, etc would be easier to understand than saying that a fixed amount would be reduced. This is especially true when all other systems continue to use a percentage based adjustment.

 

I, too, think the above proposed modifications might be worthwhile to look at - they seem like interesting changes. I especially like the might attribute rebalancing. I think as it stands, while might is no longer the most obvious choice, it truly does seem a bit OP compared to other attributes.

 

I think the confusing part was having both the static (DT) and % (DR) based damage reduction at the same time. They decided to stick to static and renamed it DR because people who play PoE don't understand what DT means. Josh explained that DT came easily to the team because it is the name they used in Fallout New Vegas, but it would appear that people who play PoE didn't play FNV...

 

 

I didn't know playing FNV was a prerequisite to playing this game. I personally don't mind that armor is less complex. I want to be able to make changes based on information and complex calculations, while doable, are not fun and make decision making tough to do because you have to be able to filter the right sorts of information.

 

they should aim for simplicity without triviality. Intuitive designs are always the best.

Edited January 10, 2015 by Hormalakh

[Luckmann]

While I think three armor types would probably be over the top, I do agree that they have done a better job differentiating how weapons work (which could be done even better, but it's passable).

 

I don't really like how the weapon styles work though. Using a shield is more of a 'because I have to' in this game rather than something I actually want to do, because it really really hurts your character's DPS and there's no cool shield-based abilities or anything that other games have.

 

Weapons still aren't balanced vs DT, so I'm usually using two handers, ranged weapons or dual-wielding (for fast attacks on rogues with damage multipliers).

 

I'm glad pollaxes are finally fixed. They're actually pretty nasty.

Oh, yes, Shields, which brings me to another possible armour modifier; Block Rate, which could essentially be a Hit-To-Graze and Graze-to-Miss modifier.

 

And Shield-based Abilities should really have been a given, such as Knockdown (which now Fighters can take either way, sadly) or Shield Push/Slam for Stunning, or Shieldwall for radically increasing Block Rate for a limited time.

 

...makes me think that there really should be Abilities that can be taken by anyone with the prerequisite Talents, such as unlocking Knockback, Shield Slam and Shieldwall Abilities when you take the "Weapon and Shield Style" Talent. This would almost necessitate an increase in the number of Talents per level (again, I still think it should be 1 Talent/lvl) and Abilities per level (Maybe more set choices for some that currently have free choices (Fighter, Paladin, etc), and add free choices for those that doesn't have any (Priests, etc).

 

But now I'm really just musing.

[Doppelschwert]

One good point raised in this thread was what i mentioned as well a week ago:
 

The problem is not only in the crit chance but also how I assume crits are computed:
I guess its
 
 

Total_Damage = [(Damage+Bonus_Damage)*Critmultiplier - DR],

 

where Damage is the base damage of an action modified by might and Bonus_Damage comes from buffs, although it would be nice if someone could confirm that. A better way would be either

Total_Damage = [Damage*Critmultiplier+Bonus_Damage- DR]

or

Total_Damage = [ Damage+Bonus_Damage - DR]*Critmultiplier,

 

the difference between the two being (Critmultiplier-1)*[bonus_Damage-DR].

 

 

I can't look into the game so my proposed formula is only a guess, but I think it holds. Maybe someone who can look into the code can confirm this, as it would explains a lot of the game behaviour and swinginess. Using the formula

 

Total_Damage = [ Damage+Bonus_Damage - DR]*Critmultiplier
 

as proposed in this thread as well is a good balanced way to do this from a mathematical point of view, as it turns critical hits into a linear increase in damage, compared to some nonlinear extreme value when damage reduction starts to grow. In particular, this smoothes the usefulness of critical hits over all weapon types:

A fast low damage weapon will not benefit any more from critical hits than a high damage slow weapon. Assuming that this spectrum of speed/damage is properly balanced against the armor system at some point, this is IMO preferable:

Given my formula holds true, that means that fast weapons with low damage are relatively more useful against high DR values when they crit, so increasing your crit chance basically undermindes the question of which weapon to use against which which range of DR values (which was the design goal of that system in the first place).

It's a bland solution because crits basically become only free attacks, but its mathematical soundest in order to prevent gaps big gaps between crits and normal hits.

 

If you insist on having some nonlinearity, which is probably more fun, then go for the other formula

 

Total_Damage = [Damage*Multipliers+Bonus_Damage- DR]

 

When I played the game it seemed to me that when I critical sneak attacked someone, the damage was calculated as

 

Total_Damage= (Weapon_Damage+Sneak_Damage+Chanter_Buff)*(Might_Multiplier (+/* not sure)  Critical_Multiplier) - DR

 

Just by having all percentile calculations only apply to the base damage would help the damage range a lot at making the game way more mathematicaly robust towards big gaps between hits and crits. I wouldn't even apply the multipliers on the weapon modifications like the 'fine' damage bonus.  Still, some actual ingame code would be nice to see how these formulas work.

Edited January 10, 2015 by Doppelschwert

[Mr. Magniloquent]

Problem is there are too many things to factor in.  It isn't just crit bonuses, it is your might bonus, your damage roll, bonuses from talents, bonuses on the weapon itself, temp spell buffs, blah blah blah.  The math for one single swing of your sword in this game is in short way too damn complicated.  The numbers in general need to be reduced, while the graze penalty needs to be reduced as well.  Changing just one modifier won't work.

 

It really comes down to the change in how character no longer miss on a large percentages of attacks. When everything hits, you need a very sophisticated way to mitigate damage without resorting to the banalities of HP bloat that plagues every Japanese game, boss fight, and MMO. Damage Threshold is an intuitive way to screen out frequent grazing, and Damage Resistance is the logical conclusion beyond this for having armor mitigate damage rather than evasion.

 

I personally have no problems at all with any of this though. Things can get out of hand very quickly, but that gives combat more visceral and threatening appeal that I appreciate. With their knockout system behind it, I don't feel that the wide range of weapon damage is terrible as those attacks are not resource based. I feel this system would be much easier to handle if weapons had a single damage value rather than range. The ultimate damage would still be determined by the wielder due to the accuracy resolution and armor design. Doing this would eliminate the worst of the "swinginess" offenses that people dislike.

[Sensuki]

Think I found the problem guys, I need to ask the devs some questions and then you'll probably hear something about this.

[Karkarov]

It really comes down to the change in how character no longer miss on a large percentages of attacks.

Fair enough but attacks always hitting has been part of Eternity's design since day one.  It is well past time we ironed things out.  Personally I think removing the extra armor mechanics was a good idea.  The math (like I said earlier) is already pointlessly complex.  Taking out one layer of it isn't going to hurt.  Hell take out more than one.  They just need to tune the numbers more.

[Sensuki]

I have to ask Josh how damage calculation is supposed to work (which I just did on SA). If I'm right, how he thinks it works and how it's actually coded are two different things.

[Sensuki]

Okay so, I got an answer, and here is my bug report

 

http://forums.obsidian.net/topic/70016-possible-bugs-with-damage-calculation-weapon-damage/

 

If some other people could use my testing methodology and try and produce the same results, that would be appreciated.

[Doppelschwert]

Thanks for digging this up!

I think pointing out the code is already enough to show that there are errors in the way the game handles the formulas, but I guess some further testing can't hurt as well. It also explains a great part of the 400 damage rogue.

[Void3dge]

I find the whole argument about damage math pointless unless the design goal is clearly defined in the first place. Before discussing further which dps formula is the best one, we should answer questions like :

 

- Shall weapons/armour belongs in well-defined categories ?

- What parameters describes weapons/armour ?

- What parameters describes characters ?

- Do we want weapons / weapon styles / armours to be equally effective in all situations ?

- ....

 

For instance, if the design goal is like :

- Fast weapons should do more damage to naked folk

- Slow weapons have more DR bypass

- Heavy weapons should benefits more from might

- Two-handed weapons crits should cut

- Damage multipliers (Critical strikes / grazes / sneaks) sums up and affects the net damage output

- Critical strikes / grazes / sneaks affects the net damage output

 

We could use a formula like

Total_Damage= (Weapon_Damage*Might_Multiplier - min(0,DR-DRbypass) )*(1+critical_multiplier+sneak_multiplier)

 

Another design could be like :

- DR is applied on the final damage

- Might give a net damage increase before DR

- Armours have a stat which reduce critical multiplier

- Sneaks give a flat extra damage

- Critical strikes / grazes / affects base weapon damage

 

With the damage formula :

Total_Damage= (Weapon_Damage*(1+critical_multiplier-armour_crit_reduction) + sneak_damage)*Might_Multiplier - DR

 

Neither of these formulas is right or wrong. The question is : what gameplay do we want ?

Edited January 12, 2015 by Void3dge

[Sensuki]

This is how it's supposed to work though

 

 

and I don't think it does work like that at the moment (but I can't be sure until the devs confirm it)

[Doppelschwert]

I find the whole argument about damage math pointless unless the design goal is clearly defined in the first place. Before discussing further which dps formula is the best one, we should answer questions like :

 

- Shall weapons/armour belongs in well-defined categories ?

- What parameters describes weapons/armour ?

- What parameters describes characters ?

- Do we want weapons / weapon styles / armours to be equally effective in all situations ?

- ....

 

 

 

- snip -

 

 

 

Neither of these formulas is right or wrong. The question is : what gameplay do we want ?

 

I actually think it's more important to have a dps formula that is mathematically robust against scaling. It's nicer if you can have all values small and still contribute meaningful compared to some JRPG scaling where you start with 83 HP at lvl 1 and end up with 12507 HP at lvl 50. As long as every factor is concerned in your formula, you can still tweak the individual values to give an incentive (although of course less than tuning the formula).

If, however, the formula scales badly toward extreme allocation of points because the mathematical growth of the function is hardly contained, then not only keeping values small but also balancing becomes rather difficult or not possible at all. When you set up your system in a way that the highest DR you can get from armor is 12 and the enemies have DR in way higher ranges just to make them feasible encounters in the first place, you screwed up in my book. Besides, as sensuki pointed out, we know what the design intention is anyway.

 

It's mathematical clear that the system is more robust if there are less multiplications and more summations, especially when you can limit the input of any factor.

Edited January 12, 2015 by Doppelschwert

[Sensuki]

Okay I got a dev response on the issue after a PM convo with NCarver

 

Yes,

 

Damage Calculations are multiplicative in game currently, so any damage that you see will be calculated in this way.

 

As to whether or not this will be changing or staying the same, I'm not certain.

 

Though, for the examples you listed it seems as though all of the data is calculating correctly.

 

For example:

 

If base damage is 10 and the user has 20 Might (30%), Sneak Attack (50%), Reckless Assault (20%), and the fine Quality Mod (15%) then the result on a hit would be 26.91 before DR. 10 x 1.3 x 1.5 x 1.2 x 1.15 = 26.91 Damage.

 

If this were additive then the result would be 21.5 Damage. 0.3 + 0.5 + 0.2 + 0.15 = 1.15 + 1 (for base effect) = 2.15 x 10 - 21.5 Damage.

 

I hope this helps to clarify.

 

-N

 

So contrary to initial beliefs, damage bonuses are multiplicative rather than additive. Not sure what's going to happen there but there's no harm in discussing what we'd like to see happen there.

Edited January 13, 2015 by Sensuki

[illathid]

Well I know I'd prefer for damage bonuses to be additive rather than multiplicative. And I've thought so for quite some time (remember saying so way back when rogue balance was being discussed).

[Sensuki]

I would be inclined to agree.

[archangel979]

multiplicative bonuses are bad. Burn them!