Jump to content

Welcome to Obsidian Forum Community
Register now to gain access to all of our features. Once registered and logged in, you will be able to create topics, post replies to existing threads, give reputation to your fellow members, get your own private messenger, post status updates, manage your profile and so much more. If you already have an account, login here - otherwise create an account for free today!
Photo

[MECHANICS] Attack Speed, Recovery Time, Reload Time

mechanics attack speed attack time recovery duration recovery time reload time

  • Please log in to reply
110 replies to this topic

#101
giftmefood

giftmefood

    (4) Theurgist

  • Members
  • 225 posts

 

 

 

Am I correct that there is no benefit to dual wielding two guns vs gun / one handed melee weapon if no abilities are in play and you just autoattack at range?

 
depends on if the guns have any special passive bonuses that you can combine, but yes, in general you are correct. in fact, if you have a particular gun or one-handed ranged-weapon that you like to use a lot, it is optimal to wield it with a one-handed melee weapon instead of a lesser second gun, if no full attacks are in play.
 
Unrelated question but I just want to get the bottom of this. Is stacking -% recovery time items and skills a good idea or does it have diminishing returns? Because based on what I read, you and MaxQuest haven't agreed on whether it has diminishing or linear returns.

IIRC we actually do agree, MaxQuest has a specific definition for "intrinsic diminishing returns." This is the idea that mathematically, with linear returns you still get less relative benefit from something the more you have of it. I agree with the math, I just more disagree with the naming.

Basically under a scenario with linear returns (or intrinsinc diminishing returns) you should be concerned with tradeoffs; it essentially means that if you're trading off between two sources of damage boost (e.g. might or dexterity) you should balance them out (as I've mentioned elsewhere it's for the same reason why maximizing the area of something with a fixed perimeter means creating a square instead of a rectangle). "True" diminishing returns means eventually it's just not worth investing in on its own, regardless of merit, which is not the case for something like recovery time.

 

 

So does that mean stacking 4 sources of -%10 recovery will be exactly twice as strong as 2 sources of -%10 recovery? So returns are linear? 


Edited by giftmefood, 10 February 2019 - 02:04 PM.


#102
Boeroer

Boeroer

    Arch-Mage

  • Members
  • 14436 posts
  • Location:Bucharest, Romania
  • Lords of the Eastern Reach Backer
  • Deadfire Backer
  • Fig Backer
  • Black Isle Bastard!
Maybe we can do a simplified example:

Let's say you have 10 sec recovery time.

Now you add a 10% recovery bonus. You are at 9 sec. Now you add an additional 10% recovery bonus. You are at 8.1.

And so on.

Not twice as strong.

But you become 10% faster than before with every step.

But you can't reach 0 that way.

Edited by Boeroer, 10 February 2019 - 03:02 PM.

  • giftmefood likes this

#103
thelee

thelee

    (9) Sorcerer

  • Members
  • 1379 posts
  • Pillars of Eternity Silver Backer
  • Kickstarter Backer
  • Deadfire Silver Backer
  • Fig Backer

 

 

 

 

Am I correct that there is no benefit to dual wielding two guns vs gun / one handed melee weapon if no abilities are in play and you just autoattack at range?

 
depends on if the guns have any special passive bonuses that you can combine, but yes, in general you are correct. in fact, if you have a particular gun or one-handed ranged-weapon that you like to use a lot, it is optimal to wield it with a one-handed melee weapon instead of a lesser second gun, if no full attacks are in play.
 
Unrelated question but I just want to get the bottom of this. Is stacking -% recovery time items and skills a good idea or does it have diminishing returns? Because based on what I read, you and MaxQuest haven't agreed on whether it has diminishing or linear returns.

IIRC we actually do agree, MaxQuest has a specific definition for "intrinsic diminishing returns." This is the idea that mathematically, with linear returns you still get less relative benefit from something the more you have of it. I agree with the math, I just more disagree with the naming.

Basically under a scenario with linear returns (or intrinsinc diminishing returns) you should be concerned with tradeoffs; it essentially means that if you're trading off between two sources of damage boost (e.g. might or dexterity) you should balance them out (as I've mentioned elsewhere it's for the same reason why maximizing the area of something with a fixed perimeter means creating a square instead of a rectangle). "True" diminishing returns means eventually it's just not worth investing in on its own, regardless of merit, which is not the case for something like recovery time.

 

 

So does that mean stacking 4 sources of -%10 recovery will be exactly twice as strong as 2 sources of -%10 recovery? So returns are linear? 

 

 

for a simplified example, "linear returns" means that a +10% damage bonus will always give you the same absolute benefit, no matter how many other damage sources you had before. so if you did 10 damage, then whether or not you had +200% damage bonuses or +0% damage bonuses, an additional +10% would always give you 1 more damage.

 

what you're talking about is actually "increasing returns." so no, 4 sources of -10% recovery will never be exactly twice as strong as 2 sources of -10% recovery, but any given -10% recovery time bonus will give you the same absolute benefit to net damage (or whatever) regardless of whether you had no prior bonus, or you already had a lot.

 

for anyone who's ever taken calculus, linear returns is essentially where the second derivative is 0, and increasing and diminishing are when the second derivative is positive or negative, respectively.


  • FleetFeet, Dr <3 and giftmefood like this

#104
MaxQuest

MaxQuest

    Arch-Mage

  • Members
  • 2279 posts
  • Deadfire Backer
  • Fig Backer

Unrelated question but I just want to get the bottom of this. Is stacking -% recovery time items and skills a good idea or does it have diminishing returns? Because based on what I read, you and MaxQuest haven't agreed on whether it has diminishing or linear returns.

It's a somewhat confusing topic.
Because on your first question there is no simple yes/no answer, due to opportunity costs.
As for the second question, as Thelee already mentioned, we actually did agree. The problem was that we were using different definitions/perspectives for the said returns. But after all it doesn't matter if the cow is darker than light-gray, or lighter than dark-gray; as we only care for how much milk she gives^^

But if you want to get to the bottom of this, then here are the two perspectives:

1. One of the common definitions for diminishing returns is this (shamelessly taken from wikipedia):

In economics, diminishing returns is the decrease in the marginal (incremental) output of a production process as the amount of a single factor of production is incrementally increased, while the amounts of all other factors of production stay constant.

Since this takes into account only absolute gains, Thelee is correct by stating that we have "linear returns" from action_speed on dps.

For example, we have 100 dps, no other modifiers, and we incrementally apply a stackable +25% action speed buff.
Now we can compare the following scenarios:

Scenario 1:
> buff1: +25% action speed

> step_sum: 0.25
> speed_coef = 1.25
> dps: 125 dps

Scenario 2:
> buff1: +25% action speed
> buff2: +25% action speed

> step_sum: 0.25 + 0.25
> speed_coef = 1.50
> dps: 150 dps

Scenario 3:
> buff1: +25% action speed
> buff2: +25% action speed
> buff3: +25% action speed
> dps: 175 dps

...and so on

So for every +x% action speed you get the same absolute gain. And this is linear returns.
Btw, Thelee also elaborates on this in his Umezawa build (check the LINEAR RETURNS appendix).

 
2.a). Now the thing is, as a player you are not necessary interested in absolute gain. Personally I am more interested in the relative one.


2.a)i) First, let's take a look at the first scenario from point 1.
The buff1, increased your dps from 100->125. That's a x1.25 relative increase.
The buff2, when applied after buff1, although being the same, increased the dps from 125->150. That's a x1.20 relative increase. So in a way we could call this "relative diminishing returns", or also "intrinsic diminishing returns".

Why "intrinsic"? Because majority of stuff in nature is subject to this.
You have one drop, and add another one to it: now you have twice more.
You have an ocean, and add a drop it. Basically nothing changed.

Although this is a somewhat deliberate use of the term. So be aware.
But since I don't have another, I use what I have)
Now to answer why am I interested in this "relative whatever it is" in the first place? Because:

2.a)ii) Opportunity cost.

Let's assume you have 100 dps.
And can select one out of two buffs:
Buff1: +25% action speed
Buff2: +25% damage

Whichever you choose, your dps will become 125. So in given context, the buffs are pretty equal.
Now let's say you have chosen Buff1, and get an opportunity to choose again.

If you choose Buff1, you will get x1.20 relative increase in dps: 125 -> 150.
If you choose Buff2, you will get x1.25 relative increase in dps: 125 -> 156.25

So again (although I am kinda twisting it), the Buff1 became less efficient than it was before.
Or let's do a less abstract example.
Sure-Handed Illa provides the awesome -20% recovery_time and -20% reload_time buffs, and the character can benefit from both if he attacks with a firearm. So the player can get quite fond of it. But if he builds a streetfighter, the relative gain from Sure-Handed Illa will be lessened; and perhaps it could be more efficient dps-wise to start looking for some damage-increase options over reload-decreasing ones.

2.a)iii) Also we could speak about limited resource pools.
Let's say you have 100dps, and can choose between +100% action speed or +100% damage bonus. But have only 2 fireball usages, after which you will have to auto-attack. It's kinda understandable that unless the combat ends too soon or is infinite, the second option will result in higher dps.

Although must also remind about the overkill damage being wasted. And that a longer cast is easier to interrupt.

2.b). And now would like to elaborate on the "-% recovery time" that you have mentioned.
Let's say you have 1.0s attack time, 4.0s recovery time, 100dps, and keep applying a stackable -20% recovery time buff.

- 0 buffs: 1.0s + 4.0000s = 5.0000s (100.0 dps)
- 1 buffs: 1.0s + 3.2000s = 4.2000s (119.0 dps) (+19.0 dps)
- 2 buffs: 1.0s + 2.6666s = 3.6666s (136.3 dps) (+17.3 dps)
- 3 buffs: 1.0s + 2.2857s = 3.2857s (152.1 dps) (+15.8 dps)
- 4 buffs: 1.0s + 2.0000s = 3.0000s (166.6 dps) (+14.5 dps)

Here not only relative dps gains are diminishing, but even the absolute ones. Because of recovery_time buffs effect on total_action_time being diluted by the unchanged attack_time.


And one more thing about returns.
Besides the relative vs absolute perspective (although absolute is more technically correct), it's also important to specify what entity do we examine.
Let's take the previous example, and say that attack time is 0s:

- 0 buffs: 0s + 4.0000s = 4.0000s (100 dps)
- 1 buffs: 0s + 3.2000s = 3.2000s (125 dps) (+25 dps)
- 2 buffs: 0s + 2.6666s = 2.6666s (150 dps) (+25 dps)
- 3 buffs: 0s + 2.2857s = 2.2857s (175 dps) (+25 dps)
- 4 buffs: 0s + 2.0000s = 2.0000s (200 dps) (+25 dps)

So my understanding is that we have here:
- (absolute) linear returns on dps
- (relative/intrinsic) diminishing returns on dps
- (absolute) diminishing returns on recovery_time
- (relative/intrinsic) diminishing returns on recovery_time
- (absolute) linear returns on recovery_speed
- (relative/intrinsic) diminishing returns on recovery_speed

And that's why it is hard to answer your question:

Is stacking -% recovery time items and skills a good idea or does it have diminishing returns?

So general answer would be: it depends)


Edited by MaxQuest, 13 February 2019 - 05:25 AM.

  • grasida, FleetFeet, Madscientist and 2 others like this

#105
Dr <3

Dr <3

    (8) Warlock

  • Members
  • 1043 posts
  • Deadfire Silver Backer
  • Fig Backer
The Bible
  • grasida likes this

#106
Madscientist

Madscientist

    (10) Necromancer

  • Members
  • 1472 posts
  • Deadfire Silver Backer
  • Fig Backer
  • Black Isle Bastard!

Thanks Max Quest

 

I do now understand the concept of absolute/relative diminishing/linear returns. (At least I think so. After some more posts I will realize I know nothing.)

But if I understand it correctly, the absolute linear ( or relative diminishing ) returns concept is only true when you have only bonusses.

 

Do I understand this correctly:

- Thanks to the double inversion, several negative modifiers will always have diminishing penalties ( both in absolute and relative numbers).

That means the combination of penalty 1 and 2 will always be less than the sum of 1+2.

 

- Things go to hell when you combine bonusses and penalties. ( I do not play computer games with a calculator on.)

Even the devs did not understand it. ( See initial version of blunted criticals.)

I do not know how much bonusses you need to compensate a penalty and I am not sure if the above concept ( absolute/relative linear/diminishing returns) changes if you combine bonusses and penalties.

 

PS:

Sometimes I have the desire to punch the guy who introduced double inversion.

Attack speed will never be easy to understand ( especially when they measure duration for some abilities and speed for others ),

but they should not have changed the damage formula from PoE1. All they needed was to define a minimum damage to avoid negative numbers.



#107
MaxQuest

MaxQuest

    Arch-Mage

  • Members
  • 2279 posts
  • Deadfire Backer
  • Fig Backer

Do I understand this correctly:

- Thanks to the double inversion, several negative modifiers will always have diminishing penalties ( both in absolute and relative numbers).

That means the combination of penalty 1 and 2 will always be less than the sum of 1+2.

I am not sure how to understand this)
The "less" is confusing me ^^

For example there are two hypothetical scenarios unrelated to Deadfire. In one you get [-3] penalty, in another [-5] penalty.

> -5 is less than -3
> but -5 penalty is bigger than -3 penalty

Thus I am unsure if to answer "less" or not.
Let me just provide a few examples instead:

 
1). Let's say you have 1.0s attack time, 4.0s recovery time, 100dps, and the enemy keeps applying a stackable -20% [action_speed] debuff.

- 0 debuffs: 1.00s + 4.00s = 5.00s (100.00 dps)
- 1 debuffs: 1.25s + 5.00s = 6.25s ( 80.00 dps) (-20.00 dps) (-20.0%) (x0.800)
- 2 debuffs: 1.50s + 6.00s = 7.50s ( 66.66 dps) (-13.33 dps) (-16.6%) (x0.833)
- 3 debuffs: 1.75s + 7.00s = 8.75s ( 57.14 dps) ( -9.52 dps) (-14.2%) (x0.857)
- 4 debuffs: 2.00s + 8.00s =10.00s ( 50.00 dps) ( -7.14 dps) (-12.5%) (x0.875)


2). And in second example, you will have the same 1.0s attack time, 4.0s recovery time, 100dps. But the enemy keeps applying a stackable -40% [action_speed] debuff.

- 0 debuffs: 1.000s + 4.000s = 5.000s (100.00 dps)
- 1 debuffs: 1.666s + 6.666s = 8.333s ( 60.00 dps) (-40.00 dps) (-40.0%) (x0.600)
- 2 debuffs: 2.333s + 9.333s =11.666s ( 42.85 dps) (-17.14 dps) (-28.5%) (x0.714)

 
Now let's try again, but with [+x% action_time] debuff instead of [-x% action_speed] debuff.

1). You have 1.0s attack time, 4.0s recovery time, 100dps, and the enemy keeps applying a stackable +25% [action_time] debuff.

- 0 debuffs: 1.00s + 4.00s = 5.00s (100.00 dps)
- 1 debuffs: 1.25s + 5.00s = 6.25s ( 80.00 dps) (-20.00 dps) (-20.0%) (x0.800)
- 2 debuffs: 1.50s + 6.00s = 7.50s ( 66.66 dps) (-13.33 dps) (-16.6%) (x0.833)
- 3 debuffs: 1.75s + 7.00s = 8.75s ( 57.14 dps) ( -9.52 dps) (-14.2%) (x0.857)
- 4 debuffs: 2.00s + 8.00s =10.00s ( 50.00 dps) ( -7.14 dps) (-12.5%) (x0.875)


2). You have 1.0s attack time, 4.0s recovery time, 100dps. But the enemy keeps applying a stackable +50% [action_time] debuff.

- 0 debuffs: 1.00s + 4.00s = 5.00s (100.00 dps)
- 1 debuffs: 1.50s + 6.00s = 7.50s ( 66.66 dps) (-33.33 dps) (-33.3%) (x0.666)
- 2 debuffs: 2.00s + 8.00s =10.00s ( 50.00 dps) (-16.66 dps) (-25.0%) (x0.750)

 
So, As you can see:
- the penalty from two [-20% action_speed] debuffs is less than from one [-40% action_speed] debuff.
- the penalty from n [-20% action_speed] debuffs is the same from n [+25% action_time] debuffs.
- the penalty from two [+25% action_time] debuffs is the same as from [+50% action_time] debuff.

Also, the more and stronger speed_decreasing/time_increasing debuffs you are already afflicted with, the lesser effect a new debuff will have.

 

- Things go to hell when you combine bonusses and penalties. ( I do not play computer games with a calculator on.)
Even the devs did not understand it. ( See initial version of blunted criticals.)

Kinda yeah)
 

I do not know how much bonusses you need to compensate a penalty and I am not sure if the above concept ( absolute/relative linear/diminishing returns) changes if you combine bonusses and penalties.

You can use the following mini "lifehack":
- take a look at your maluses
- let's say there is -25% action_speed. This means you can "compensate" for it with +33% action_speed.
- let's say there is +25% action_time. This means you can "compensate" for it with +20% action_speed.
- remove this way all maluses (and bonuses that were used for "compensation")
- and if hopefully only bonuses remained, use them in a simple, additive manner. Just convert them to x_speed first, if there are any x_time via 1/value.
- and if you remained with maluses - perhaps this is not your dps character ^^
 

Attack speed will never be easy to understand ( especially when they measure duration for some abilities and speed for others )

Because UI shows the attack/recovery duration in seconds, there could be just "attack_time" and "recovery_time" effects.
But:
- "Attack time" sounds like some army term, e.g: "attack time is 11:45. At that time we will launch THE NUKE".
- We are too used with dexterity increasing speed.

Maybe going for "action_speed" and "recovery_speed" would indeed be a better choice. Thus we would get rid of all these "_time" altogether, not to mention that internally it is converted to speed anyway. But this would also require proper UI/hover tooltips that would explain how did +30% action speed transform into -23%. Although you know what? we need this even now))
 

All they needed was to define a minimum damage to avoid negative numbers.

Seems that they were afraid of those -75% underpen and -50% graze occuring at the same time.
But yeah, letting a certain % of damage go through no matter what is one way to get around that.

Personally though, I'd prefer over/underpen coefficient being applied multiplicatively.
...And same for crit/graze,... but speaking of crits, there is bonus_PEN_onCrit and also overpen bonus, so it's a bit dangerous to touch it now.

Edited by MaxQuest, 13 February 2019 - 07:22 AM.

  • Madscientist likes this

#108
Madscientist

Madscientist

    (10) Necromancer

  • Members
  • 1472 posts
  • Deadfire Silver Backer
  • Fig Backer
  • Black Isle Bastard!

Thank you again

 

So we could say:

- penalties for attack speed have diminishing returns: two 20% penalties is a smaller DPS loss than one 40% penalty

- penalties for attack time have linear returns: two 20% penalties is the same as one 50% penalties

 

Are the results for damage equal to the results for speed or for time?

 

The thing that makes it so complicated is that most chars wear some kind of armor which slows you down.

There is also armored grace and the pet that reduces armor penalty, but you do not see the exact numbers in game.

 

to the devs:

WHY did you do this???

In PoE1 you invented a rule set that was relatively easy to understand. ( At least compared to most computer games based on PnP rules. )

In PoE2 you took the system from PoE1 and made everything more complicated. (e.g. double inversion, penetration system, . . .)

I see no benefit in this, both for players and devs.



#109
thelee

thelee

    (9) Sorcerer

  • Members
  • 1379 posts
  • Pillars of Eternity Silver Backer
  • Kickstarter Backer
  • Deadfire Silver Backer
  • Fig Backer

Thank you again
 
So we could say:
- penalties for attack speed have diminishing returns: two 20% penalties is a smaller DPS loss than one 40% penalty
- penalties for attack time have linear returns: two 20% penalties is the same as one 50% penalties

 
let's try to be deliberate about what terms we use, to avoid confusion, and I think you have a typo or two.
 
By "attack time" you really mean "recovery time" because attack time penalties do not exist. But yes, recovery time penalties are trivially linear returns.
 
Attack speed: not quite. The actual attack speed numbers you see are not the returns you get (the returns you get is your DPS essentially). A penalty to attack speed is exactly equivalent to it being inverted and applied as an "attack time" penalty (again such a thing does not actually exist in game just as an illustration), which has trivially verifiable linear returns. A -20% attack speed is exactly equivalent to +25% attack time penalty (like a +25% recovery time penalty that is also applied to the actual attack itself), and any number of -20% attack speed penalties will additively and linearly combine. The fact that two -20% attack speed penalties give you less DPS loss than a single -40% attack speed penalty has very little to do with what type returns you get. It's just a weird unit of measurement, essentially. (Because a -40% attack speed penalty is actually equivalent to a +66.6% attack time penalty, which is much larger than two +25% attack time penalties)
 
That being said, a single attack speed penalty that grows progressively larger has increasing losses, and a single recovery time bonus that grows progressively larger has increasing returns. (https://gamefaqs.gam...reasing-returns)  But I do not know of  any such attack speed penalties in the game (edit: correciton, dexterity penalties below 10 would actually be a common case of this), and I can only think of one type of recovery time bonus that does that (Mob Stance).
 

In PoE1 you invented a rule set that was relatively easy to understand. ( At least compared to most computer games based on PnP rules. )

In PoE2 you took the system from PoE1 and made everything more complicated. (e.g. double inversion, penetration system, . . .)


PoE1 was relatively easy to understand (but in truth there were actually lots of exceptions to the point that exceptions were more common than the rule), but had flaws by being a largely purely additive system.

 

PoE2 sacrifices some of the intuitiveness so that you can have meaningful penalties in areas that don't get easily trivialized because of being a simple additive system.

 

My take on this follows below: (from https://gamefaqs.gam...6599/inversions)

 

"If I could pinpoint the single-most confusing aspect of Deadfire, it's how negative modifiers are handled. TL;DR: deadfire is an additive system for positive modifiers but gets weird with negative modifiers.

 

Let's say you have a buff and a debuff. The buff gives you +20% damage. The debuff gives you -20% damage. What's the net effect on your damage?

 

Most games do one of two things: an additive combination or a multiplicative combination. Under an additive system, we just combine the two modifiers by addition: +20 + -20 = 0, so no net effect on damage. Under a multiplicative system, we convert the modifiers into multipliers, and multiply them together: so +20% becomes 1.2x and -20% becomes .8x, and then we do 1.2 * .8 = .96, so a -4% net effect on damage.

 
Deadfire does something very different, that we'll call "inversion". It's basically at its heart an additive system, but it treats negative modifiers specially.

 

Sidebar: why inversions?

A weakness with additive systems is that negative modifiers are extremely easy to trivialize. This was a problem with PoE1. Grazes in PoE1 were hits made with a -50% modifier to damage. This was appropriately weak early on, but as you got stronger and got better weapons, you would get more and more damage buffs. Eventually, this might altogether erase the impact of a graze, making it relatively less painful than before. Indeed, in PoE1, much of a rogue's power came from their sneak attack essentially making grazes "as good" as a normal hit from anyone else. Another weakness with additive systems is that if you're not careful with designing it, it becomes easy to pile on enough negative modifiers that you end up with 0% net effect on stuff. A plus, however, is that once you know a system is additive it's pretty easy to reason about. You just take every modifier you have and combine them through addition. Really easy for a player to evaluate. Plus, because everything is added together, it's real hard to end up with surprisingly degenerate situations where you can combine buffs and blow out game balance.
 
By contrast, multiplicative systems mean that negative modifiers are always impactful. If Graze was instead .5x damage, then no matter how much base damage you would do, a Graze would always do half damage. Similarly, unless you have a negative modifier that is explicitly -100% damage, it is impossible to combine too many negative modifiers and end up with 0% net effect. However, multiplicative systems suffer from its own weakness: because all buffs multiply with each other, if you're not extremely careful you can have runaway buff-stacking and end up with huge numbers that blows away game balance. It also is in some cases unintuitive to the player; in our earlier example, a player might intuitively think that a +20% buff and a -20% buff will cancel out, but such is not the case in a multiplicative system.
 
You can try to design a combined additive/multiplicative system, but you have to be very careful; probably such a system would mostly have additive effects and only a handful of very special multiplicative effects. Diablo 3, for example, uses both additive and multiplicative modifiers but favors multiplicative multiplier; the end result is that people chase multiplicative modifiers really hard since there are so many that they blow out most additive effects you can find. So it's hard to get right.
 
I believe Deadfire's designers were trying real hard to come up with a system that combined the general intuitive and ease-of-balancing benefits of an additive system but still let negative modifiers be significant without the risk of 0% effects as in a multiplicative system. I think in terms of intuitivity they failed horribly, but if you learn and understand the system, I think there's a lot to be said in favor of it. It's just rather a significant "if.""
 
 
 
P.S. Early on I would have suggested to the designers that they should have made it a completely additive system and just had a few multiplicative exceptions for graze and underpenetration, but after lots of playthroughs I can appreciate the game design impact of making any malus more impactful via inversions. It's just a huge mathematical hurdle to get over for literally everyone (including the designers themselves as apparent in the broken way that paladin and priest disposition bonuses scale) and you can't reason about it as easily (though at this point I've done so many inversion calculations that I know a lot of common case inversions off the top of my head, like -15% penalty cancels out almost +18% bonuses).

Edited by thelee, 13 February 2019 - 09:10 AM.

  • Madscientist, Dr <3 and MaxQuest like this

#110
MaxQuest

MaxQuest

    Arch-Mage

  • Members
  • 2279 posts
  • Deadfire Backer
  • Fig Backer

Thank you again
 
So we could say:
- penalties for attack action speed have diminishing returns: two 20% penalties is a smaller DPS loss than one 40% penalty
- penalties for attack action time have linear returns: two 20% 25% penalties is the same as one 50% penalties

Well, yeap we can say that.

Just few minor notes:
- there is 25% not 20% on the second line
- it's action not attack. Because we were examining relation between action related maluses and dps. And if we would examine recovery or attack phase only, than it's results would get diluted by the second phase.
- as Thelee already mentioned there are no penalties with "Attack Time" in description. Also there are no penalties with "Action Time" in description; but I used it for the sake of example.
 

Are the results for damage equal to the results for speed or for time?

Good question :)

The damage related bonuses/maluses are going internally through the same system as speed related ones.
But there is a slight difference at the end:
- final_damage = base_damage * final_damage_coefficient
- final_attack_time = base_attack_time / final_speed_coefficient

In case of final damage, we have multiplication by the damage coefficient.
In case of final attack/recovery/reload time, we have division by the speed coefficient.
 

The thing that makes it so complicated is that most chars wear some kind of armor which slows you down.

That's true)
 

There is also armored grace and the pet that reduces armor penalty, but you do not see the exact numbers in game.

Yeah, the game just shows you the resultant armor recovery penalty.

But if you are interested, the game does store armor recovery time penalties as speed penalties:
- for heavy armors it is: 0.645 (= 1/1.55)
- for medium armors it is: 0.741 (= 1/1.35)
- for light armors it is: 0.833 (= 1/1.20)

And:
- armored grace just adds +0.10 to that value (so 0.645 becomes 0.745 which becomes 1.33, i.e. 55%->33% recovery time penalty)
- abraham and cutthroat cosmo also provide +0.10
- and nalvi provides +0.08 (but as an aura)

 

to the devs:
WHY did you do this???
In PoE1 you invented a rule set that was relatively easy to understand. ( At least compared to most computer games based on PnP rules. )
In PoE2 you took the system from PoE1 and made everything more complicated. (e.g. double inversion, penetration system, . . .)
I see no benefit in this, both for players and devs.

Well... Action Speed system in PoE1 was even more complicated than this one :)
I proposed a bit of polishing to it here. But it didn't matter because Obsidian has ditched the PoE1 idea of "reduce action time at the cost of recovery time first" altogether. And I am perfectly fine with that, as it was hard to just explain how it works.

I am not fond of Action Speed system in Deadfire neither. But along with cons, it also has it's pros:
- Thelee already mentioned it above. Plus here's a related post by Ensign. I'll just add that current system smoothed the scenarios when you have a lot of maluses, and when you have a lot of bonuses.
- also current system is easy on the programmer. You just have an adjustedValue object, and throw to it bonuses, maluses, doesn't matter; and when you are done you just request the final value. I mean there is no specific order that you need to maintain. And it's the same for both speed and damage.

@Thelee, props for the Sidebars. It perfectly explains the situation. And I like to believe that it is the lack of time - being what prevented Obsidian from implementing system like in 3rd paragraph (as it required careful evaluation of each modifier).

Edit: Speaking of damage formula... I am more inclined towards something like this:

Spoiler

Edited by MaxQuest, Yesterday, 02:20 AM.

  • thelee and Madscientist like this

#111
Madscientist

Madscientist

    (10) Necromancer

  • Members
  • 1472 posts
  • Deadfire Silver Backer
  • Fig Backer
  • Black Isle Bastard!

Thanks again Thelee and MaxQuest

 

The link to the guide alone was worth it.

 

At this point I give up and leave the math to experts like you.

I have finished the game on normal and beat everything except the mega bosses.

I will not play on higher difficulties because the armor/penetration mechanic alone is annoying for some tough enemies.

 

I completely agree with this: The double inversion may be a good mathematical solution, but it is very unintuitive and most players (and some devs) will have problems to understand it.

There were lots of bug reports where players complain the damage is wrong because +20% and -20% does not result in zero.







Also tagged with one or more of these keywords: mechanics, attack speed, attack time, recovery duration, recovery time, reload time

0 user(s) are reading this topic

0 members, 0 guests, 0 anonymous users