Should Might stay multiplicative or return to additive?

[Dr. Hieronymous Alloy]

Do you have any thoughts on Perception? How does the math shift if we move back to a 50% critical instead of a 25%?

[MaxQuest]

Do you have any thoughts on Perception? How does the math shift if we move back to a 50% critical instead of a 25%?

I do.

 

I find it a bit strange that:

- you deal twice less damage on graze

- and deal only a quarter more on crit

 

Although it's understandable, that critMult was nerfed because:

- there are talents that increase it

- there is x1.3 multiplicative bonus from overpenetration and 1.3 * 1.25 = 1.625. But we do not overpenetrate on every crit; nor can easily estimate how often did we get this bonus due to crit.

 

If there are no grazes, no over/under penetrations, and critMultiplier = 1.5, then going from acc = def to acc = def + 1, provides the same relative gain as going from 10 MIG to 11 MIG, i.e: 3.0%

 

 

But dunno, perhaps will have to remember how to derivate equations, or make a more detailed table for PER, before asserting anything related to it.

Edited December 18, 2017 by MaxQuest

[Dr <3]

I like everything you said in your previous post about might. Very precious work

[dunehunter]

@MaxQuest, I just turned back and look at your v4 solution again, I like it a lot and I have a question, if you make weapon quality a separate multiplier, would that make it unfair for caster because spells doesn’t benefi from weapon quality? So maybe power level can be treated as caster’s ‘weapon quality’ and be used as a multiplier too?

 

In this way, devs can design magic weapon for casters to balance meleer, like a wand that grant caster +1 power level to spells (roughly = fine weapon), +2 power level = exceptional weapon and so on.

Edited December 27, 2017 by dunehunter

[MaxQuest]

if you make weapon quality a separate multiplier, would that make it unfair for caster because spells doesn’t benefi from weapon quality? So maybe power level can be treated as caster’s ‘weapon quality’ and be used as a multiplier too?

> if weapon quality is NOT a separate multiplier (relative to MIG/STR)

- it's unfair to weapon-based damage-dealers, because the effect of MIG/STR becomes heavily diluted.

 

> if weapon quality is a separate multiplier (relative to MIG/STR)

- the fairness of weapon_dmg vs spell_dmg is not disturbed much if at all, because:

-- in case of cipher powers: cipher gets higher focus gain

-- in case of chanter: he manages to unlock more potent chants, invocations and get invocation upgrades.

-- in case of ex-vancian casters: higher ranks usually contain more powerful spells. And if something is weak now, it can be slightly tweaked-up. Not to mention that this will also indirectly buff single-class casters as they get access to higher ranks earlier.

- MIG has same impact on auto-attacking dps as DEX

 

In this way, devs can design magic weapon for casters to balance meleer, like a wand that grant caster +1 power level to spells (roughly = fine weapon), +2 power level = exceptional weapon and so on.

Many spells already benefit from power level which gradually increases with character's level.

We can view weapon upgrading to fine/exceptional/superb/legendary at level 4/8/12/16 as gradual as well.

And the devs task is to match these between them. (note: it may not necessary be 1:1 proportion like +45% dmg to spells at power level 5. It could be even half or third of it, provided that base damage / effect of spells goes up with ranks, e.g. Ray of Fire vs Hand of Weal and Woe)

Edited December 27, 2017 by MaxQuest

[dunehunter]

Thanks for the answer that makes a lot sense.

 

But if weapon quality is unique multiplier, this will make low quality weapon much less appealing to users than PoE 1. Specially the enchantment system is reworked now, if we cannot upgrade any weapon to lendary, then they will be useless in high level according to v4 solution.

[MaxQuest]

When Josh mentioned that weapons we'll find closer to the end of the game will come with higher capacity, forumers instantly pointed out that this will narrow the amount of viable weapons in the late chapters; plus it will make player feel uncertain about investing materials into the current weapon, because it may become gated by a great margin compared to a new shiny he can potentially find around the corner.

 

That said, I can't answer this question completely because I need to take a look at the actual enchanting system in Deadfire first.

 

But if it was like in PoE1, I'd say that non-legendary weapons would not necessary be useless as it would depend on what you are trading for. For example a player would not enchant superb to legendary, if:

- this will make it impossible to apply lash, because (c + 0.45) * 1.25 > (c + 0.55), and there is no DR in Deadfire

- he can have some cc effect instead (especially if he is not a dedicated damage dealer)

[draego]

Thanks for the answer that makes a lot sense.

 

But if weapon quality is unique multiplier, this will make low quality weapon much less appealing to users than PoE 1. Specially the enchantment system is reworked now, if we cannot upgrade any weapon to lendary, then they will be useless in high level according to v4 solution.

 

You wont be able to upgrade from fine to legendary anymore. At least that is the last i heard of it. The quality enchants will work like others now in POE2 and can only be upgraded one grade. This is really a sucky change but that is where its stood months ago. Hope this changes.

Edited December 27, 2017 by draego

[Dr. Hieronymous Alloy]

 

That said, I can't answer this question completely because I need to take a look at the actual enchanting system in Deadfire first.

 

 

From what Josh has said recently I get the impression that they're going a route that doesn't directly map onto the system in the first game. Like, a given weapon might have two separate upgrade "paths", which lead to discrete benefits, but both of which are still relatively unique to that weapon. One thing he mentioned specifically in the stream is that (for example) there might be only a small number of weapons in the whole game with a "burning lash" type enchantment.

 

We really don't know enough yet to make any predictions based on that system.

[MaxQuest]

From what Josh has said recently I get the impression that they're going a route that doesn't directly map onto the system in the first game. Like, a given weapon might have two separate upgrade "paths", which lead to discrete benefits, but both of which are still relatively unique to that weapon.

Yeap, I remember that.

 

We really don't know enough yet to make any predictions based on that system.

Exactly. Not to mention that his original idea could undergo a series of stealthy transformations (like current might and attack/casting speed being changed without being mentioned in transcripts) or be ditched altogether (like trinkets or reloading duration affected by armor)

[clouseau64]

hello all,

 

i did a bit of empirical testing with various weapons, damage reducing modals, and various form od additive damage bonus.

 

the 25% bonus from crit is indeed additive but negative damage modals and graze are multiplicative. i am struggling to find the exact formula but it looks like an exponential of the sum of negative coefficient as with no strength bonus nor positive bonuses a graze with parrying blade yields 33% of damage output. likewise grazing with a x% additive bonus versus 0 increase damage by a bit more than x/2 and the figure seem to increase with the size of X.

 

Does anyone have the code of damage formula?

 

I also think there might be a bug for crit when used with a negative damage modal, it looks like the 25% only applies to the damage bonus ( ie doing 1.25x30%=37.5% instead of 55%).

 

Does any of you have the same result?

 

I find it fair to have graze as multiplicative otherwise perception is a dump stat for any non CC caster / debuffer as it can be easily offset with damage bonus. On the other hand crit need to become multiplicative too.

[clouseau64]

looks like destructive channeling is also bugged on graze as i don't see any increase in damage whatsoever when it grazes

[Dr. Hieronymous Alloy]

 

 

I find it fair to have graze as multiplicative otherwise perception is a dump stat for any non CC caster / debuffer as it can be easily offset with damage bonus. On the other hand crit need to become multiplicative too.

 

 

Even with multiplicative crits, at 25% crit bonus instead of the 50% it gave last game, Perception is a dump stat right now relative to Might and Dex, giving around a 20% damage bonus if you stack it to 20, whereas Might and Dex each give 30%. To even them out, Per would have to give a 50% (multiplicative) crit bonus.

Edited December 31, 2017 by Dr. Hieronymous Alloy

[MaxQuest]

the 25% bonus from crit is indeed additive but negative damage modals and graze are multiplicative.

Yeap. Crit damage bonus is additive.

 

As for graze, it looks like a multiplicative malus: example. But if might is multiplicative (see below) something doesn't add up.

 

Does anyone have the code of damage formula?

There is no easy answer to that, because we have several sources that seem to contradict each other.

 

> According to UI tooltip steps: all damage coefficients are additive.

> According to Source Code (assuming I understood it right) all damage coefficients are additive.

> According to UI tooltip result (if we try to figure out how the end result was achieved): Might is applied multiplicatively, while all other damage coefficients are applied additively. And BMac confirms that this indeed is the case here, and tooltip steps - being just an UI problem.

 

Funny enough, none of the above would explain the graze above.

In this case looks like all damage coefficients (including Might) are additive, while Graze itself is a multiplication by x0.5

 

I find it fair to have graze as multiplicative otherwise perception is a dump stat for any non CC caster / debuffer as it can be easily offset with damage bonus. On the other hand crit need to become multiplicative too.

Agreed that both graze and crit should be applied multiplicatively. This is also reflected in suggestion v4.

 

As for perception - it was a flat stat for dps'ers in PoE1, and a bump stat for debuffers because it made cc reliable, and helped a lot in bringing defenses of high-def bosses down. And in Beta 2 there are no bosses, no Broodmothers/Battery Sirens, nor cc is worthy the cast time yet.

Edited December 31, 2017 by MaxQuest

[clouseau64]

i got something pretty similar to your example, and i can only arrive to the conclusion that formula for graze is inconsistant with formula for hit /crit. in your example i am not sure that soul whip and biting whip actually stack ( it depends if your example is prepatch or not). the only way i come close to the result is by doing 0.5x1.18x1.40x1.15x19.6=18.6 ( so multiplying all damage bonus over the penalty and assuming soulwhip doesn't apply). i got close form displayed result with other might value and high damage coefficient.

 

perception definitely needs to be buffed and i agree with your comments on crits needing to be @0.5 mult. they completely debased that stat when they changed the interrupt mechanism ( definitely not for the best unfortunately). Given that i play potd i wouldn't fully dump it but there is clearly no point in raising it over 10 dps wise. I have a high perception on unbroken/skald as might is of no use on a tank now and their CC ( paralyse, - armor and the stun) is super solid with fair perception.

 

i wish the devs were using basic heuristics when they modifiy ressources value so as not to create unnecessary imbalances

[hilfazer]

 

the 25% bonus from crit is indeed additive but negative damage modals and graze are multiplicative.

Yeap. Crit damage bonus is additive.

 

As for graze, it looks like a multiplicative malus: example. But if might is multiplicative (see below) something doesn't add up.

 

 

Graze is multiplicative while crit is additive.

Graze is - 50% while crit is +25%.

Crit increases PEN while graze does not lower it (or does it?).

 

Are they trying to make this game harder to grasp?

[MaxQuest]

i got something pretty similar to your example, and i can only arrive to the conclusion that formula for graze is inconsistant with formula for hit /crit.

Yeap

 

in your example i am not sure that soul whip and biting whip actually stack ( it depends if your example is prepatch or not).

That example is from Beta 1. I don't know if soul and bitting whips stack now, but they 100% did back then.

Example: image

16.2 * (1 + 0.5 + 0.2 + 0.4 + 0.2 + 0.15) * 1.15 = 45.64 ~= 45.7

 

the only way i come close to the result is by doing 0.5x1.18x1.40x1.15x19.6=18.6 ( so multiplying all damage bonus over the penalty and assuming soulwhip doesn't apply). i got close form displayed result with other might value and high damage coefficient.

I was thinking about additive all the way

19.6 * (1 + 0.4 + 0.2 + 0.15 + 0.18) * 0.5 = 18.9... but that's still off from 18.5

 

Graze is multiplicative while crit is additive.

Graze is - 50% while crit is +25%.

Crit increases PEN while graze does not lower it (or does it?).

That's pretty accurate. And yes, graze does not lower PEN. Edited December 31, 2017 by MaxQuest

[gGeorg]

I am fine with additive, multiplicative, v4 or whatever, BUT

for sake of sanity,

 

weapon stat(STR) formula uses multiplicative

spell stat(RES) formula uses multiplicative

 

OR

 

weapon stat(STR) formula uses additive

spell stat(RES) formula uses additive

 

Let base stats apply the same operator to get the dmg. Weapons, spells, skills, whatever, .... .

 

----------------------------------

Do Not mix things up like:

weapon stat(STR) formula uses multiplicative

spell stat(RES) formula uses additive

Creating a system where 1 point of Strength adds different volume of dmg than 1 point of Resolve is wrong by design.

Edited January 1, 2018 by gGeorg

[MaxQuest]

I've been testing how exactly is damage calculated ingame.

 

First step was to check if tooltips lie and if yes how much:

 

 

 

Result:

+ Final Damage is correct. It matches the overhead red numbers and is indeed the value that is subtracted from target's hp.

+ Pre-PEN Damage is correct. It matches the actual damage dealt when neither over or under-penetrate.

+ Rolled Damage has no reason to be incorrect, since it's the value we start from.

- Over-penetration multiplication is incorrect. No way 26.8 * 1.3 will give you 32.7. That's because over-penetration bonus... is additive, as you'll see later.

- Tooltip steps are incorrect. That's because might bonus is actually applied multiplicatively, but with a big twist, as you'll see later.

 

The next step was registering various damage tooltips and trying to find the formula that would match them all. Few examples:

 

 

 

 

 

 

 

---

And now I think I've finally figured it out. Tada:

 

FinalDamage (in beta2) = (RolledDamage * DamageMultiplier + AdditiveDamageBonus) * PostAdditiveDamageMultiplier

 

DamageMultiplier is influenced by:

- weapon quality bonus (e.g. fine/exceptional/superb)

- weapon type bonus (e.g. sharp)

- bonus damage talents (e.g. two-handed style, sneak attack, soul whip)

- crit bonus

- over-penetration bonus

 

PostAdditiveDamageMultiplier is influenced by:

- might damage coefficient

- modal malus (like -50% from daggers modal)

- graze malus

- under-penetration malus

 

As for AdditiveDamageBonus, am not completely sure but it can include flat damage bonuses; think of Novice's Suffering from PoE1.

 

Question: Now, how are these multipliers actually calculated? Additive or multiplicative?

Answer: additive with a big twist:

- all damage coefficients are broken into steps

- now, if it's value is above 1, the step will be (value - 1)

- and if the value is below 1, the step will be (1 - 1 / value)

- after that all these steps are added up, into one big coefficient

- if the value of this coefficient is above 0, the group multiplier will be (coefficient + 1)

- and if the value of this coefficient is below 0, the group multiplier will be [1 / (1 - coefficient]

 

Question: Isn't that too complicated?

Answer: Yes it is.

 

IMPORTANT DISCLAIMER:

The presented formula is for Deadfire beta2.

 

In beta4 the following changes were implemented:

- maluses from modals, graze and underpenetration were moved to DamageMultiplier group. I.e. all effects are aggregated in additive manner, while might bonus is applied multiplicatively.

 

And in release version (v1.0.1):

- all effects (including Might) do now belong to DamageMultiplier group. I.e. all steps are aggregated in additive manner.

 

---

Now a few examples:

 

For example if you graze or hit with fine sabre:

DamageMultiplier_Coefficient = (1.15 - 1) + (1.20 - 1) = 0.35

DamageMultiplier = 0.35 + 1 = 1.35

 

Or you have 20 MIG and graze:

PostAdditiveDamageMultiplier_Coefficient = (1.30 - 1) + (1 - 1 / 0.5) = 0.3 + 1 - 2 = -0.7

PostAdditiveDamageMultiplier = 1 / (1 - -0.7) = 1/1.7 = 0.588

 

Or you have 20 MIG and graze, and also under-penetrate:

PostAdditiveDamageMultiplier_Coefficient = (1.30 - 1) + (1 - 1 / 0.5) + (1 - 1 / 0.25) = -0.7 + 1 - 4 = -3.7

PostAdditiveDamageMultiplier = 1 / (1 - -3.7) = 1/4.7 = 0.212

 

And a concrete example: this tooltip

 

RolledDamage = 17.2

17.2 * 1.35 * 0.588 = 13.65 ~= 13.6 -> pre-penetration damage

17.2 * 1.35 * 0.212 = 4.92 ~= 4.8 -> final damage

 

Rounding errors are kinda big, most likely because of double inversion of damage maluses.

Edited June 12, 2018 by MaxQuest

[dunehunter]

Overpenetration is an additive ?! So devs lies about this?

[MaxQuest]

Overpenetration is an additive ?! So devs lies about this?

It's... more complicated than that.

But yes, overpenetration bonus is additive with the bonuses from: crit, weapon quality/type, and damage talents like sneak attack.

 

P.S. I'd be aware of how graze diminishes the effect of might... At 20 MIG and graze, the PostAdditiveDamageMultiplier is only 0.588.

Edited January 13, 2018 by MaxQuest

[dunehunter]

FinalDamage = (RolledDamage * DamageMultiplier + AdditiveDamageBonus) * PostAdditiveDamageMultiplier

 

DamageMultiplier is influenced by:

- weapon quality bonus (e.g. fine/exceptional/superb)

- weapon type bonus (e.g. sharp)

- bonus damage talents (e.g. two-handed style, sneak attack, soul whip)

- crit bonus

- over-penetration bonus

 

PostAdditiveDamageMultiplier is influenced by:

- might damage coefficient

- modal malus (like -50% from daggers modal)

- graze malus

- under-penetration malus

 

 

Why I have the feeling that they wanna change every bonus into an additive and every malus into a multiplier? So 1. you cannot to do super high spike damage; 2. any malus will be very punishing.

[IndiraLightfoot]

The obligatory stupid question of the day: How can rolled damage be in decimals, like 15.2 or 17.2?

 

And then the hardest question, how can we clean up this system, preferably using integers all the way, so it gets simpler and easier to follow and learn by heart (at least approximate)? 

[MaxQuest]

Why I have the feeling that they wanna change every bonus into an additive and every malus into a multiplier? So 1. you cannot to do super high spike damage; 2. any malus will be very punishing.

Current system seem to have... "feel" attached to it. Like under-pen graze deals too low damage, let's increase it. But the damage can skyrocket, let's move over-penetration bonus to the preAdditive category. Either so, or it's a side effect of multiplier double-inversion shenanigans.

 

The obligatory stupid question of the day: How can rolled damage be in decimals, like 15.2 or 17.2?

It's float. Another question would be: why?

I could be short-sighted on this, but I don't see any reason why rolled damage shouldn't have integer value.

 

And then the hardest question, how can we clean up this system, preferably using integers all the way, so it gets simpler and easier to follow and learn by heart (at least approximate)?

Well, the current system could get rid of double inversion in the first place.

And have the damage to be easier to compute in mind. Or at least let us estimate in mind how bigger/less the damage is with current modifiers than without. E.g. we graze and under-pen: 0.5 * 0.25 = 1/8

 

The easiest solution with minimum changes, would be: to move all bonuses (except penetration) to the first group (i.e. DamageMultiplier) and let them be aggregated in additive manner (without inversion). While penetration multiplier (for both under and over-penetration) to be attached to PostAdditiveDamageMultiplier and be applied in multiplicative manner (also without inversion).

Example:

> a under-pen hit with 20 MIG, sneak attack, two-handed style:

> finalDmg = {rolledDmg * [(1.3 - 1) + (1.5 - 1) + (1.2 - 1)] + AdditiveDamageBonus} * 0.25

 

^ That's basically the same as in PoE1: all additive. Just the end result is being multiplied by penetration multiplier (x0.25, x0.5, x0.75, x1.0, x1.3)

(and a variation of this could have MIG belong to PostAdditiveDamageMultiplier too; as consequence MIG would be more important and in-line with DEX's +3% bonus; but builds with Sneak Attack and Bitting Whip that used flat MIG value will suffer)

 

Let's call this suggestion v5.

I have updated the related post with system comparisons: link to img chart.

 

P.S. I would still advocate for v4. But if there is a tight time schedule, v5 will somewhat do to,.. as it requires the least effort to migrate to.

Edited January 14, 2018 by MaxQuest

[IndiraLightfoot]

I love your calculations, MaxQuest!

 

And your quick and dirty fix (v5) is definitely a big improvement. Also, a bonus that it follows PoE1.

 

As for damage rolls being float, well, it's a true conundrum. It really doesn't fit a CRPG or any kind of game system user-end-wise.