#Deadfire Q&A Twitch Stream with Josh Sawyer and Lead Designer Bobby Null

[Heijoushin]

1. Will there be a place to buy cosmetic items early in game? (cloaks of different colours, pirate hats etc.) PoE1 had all those assets in game, but it was hard to get them early without killing a civilian.

2. Any idea how many more portraits we can look forward to?

[blotter]

Q&A's over. AFAIK neither of those questions were addressed.

[Doppelschwert]

Q&A's over. AFAIK neither of those questions were addressed.

 

 

Social media sucks. This forum should get much more attention than some random guys on twitter...

[Heijoushin]

 

Q&A's over. AFAIK neither of those questions were addressed.

 

 

Social media sucks. This forum should get much more attention than some random guys on twitter...

 

 

Now, now, while I agree social media is a scourge on the world, it was my fault for posting the questions late.

[blotter]

It seemed like they answered more forum questions this time, but maybe that's just because they addressed 2/10 of mine instead of 0/8 like before.

[PK htiw klaw eriF]

They should do a Q&A here on the forums sometime.

[Xaratas]

Thanks Aarik for sneaking my questions on the paper

[ushas]

Pillars of Eternity II: Deadfire Q&A #4 Transcript by Fereed:

https://www.reddit.com/r/projecteternity/comments/5vsoiy/pillars_of_eternity_ii_deadfire_qa_4_transcript/

[Madscientist]

Thanks for answering one of my questions.

[MaxQuest]

Aaand backed)

 

We are getting lot's of new content, customizable AI, ships+upgrades plus Josh is finally answering mech questions.

Through the addressed questions weren't understood in the exact way I meant, I am sold anyway)

I know Obsidian will deliver a great story. There is no doubt in that. But now we are sure that the team really looks at multi-classing emerging issues and is serious about mech balance; so there is a solid chance to play a balanced version of PoE2 before 2019. And what more can we ask?

 

---

 

 

 

Just a few notes about stream questions, which I'd like to add later. Maybe these remarks will be of use. Maybe not.

Right now I have time to address the 2nd answered question, i.e.: 

Q: If Empower adds a flat +3 to power level, won't its relative effect become much weaker in the late game?

JS: It's always +3 levels; it's always whatever you're adding on top of anything else.

BN: And our levels can scale past whatever max level is.

JS: Yeah, they can scale past the end. So just because you have maxed out your power levels, that can go above that as well.

I didn't mean a power_level cap. It is obvious that there would be no such thing, as that would kill the idea of empower at max power_level completely

 

What I meant is: we know that spells and abilities are going to scale with power level. And this creates the following discrepancy:

- if you are level 2 wizard with power_level 1, and you empower a spell; your power_level goes from 1 to 4

- if you are level 20 wizard with power_level 10, and you empower a spell; your power_level goes from 10 to 13

 

I.e. late game characters will benefit significantly less from empower. While in early game, characters will viceversa, have a much higher relative gain.

 

If that damage coefficient is applied linearly (to power_level) and is applied multiplicativelly:

- the first wizard will see an increase of +300% (1->4)

- the second wizard will see an increase of +30% (10->13)

 

But if it is applied additivelly, there is still a sensible difference, moreover, it would quite buff 10/10 equal splits, as it would be easier to compensate for the loss in dps, via other damage coefficients.

 

 

 

---

 

Ushas already posted link to transcript #4. I will add a short recap:

 

- Transcript #1

- Transcript #2

- Transcript #3

- Transcript #4

Edited February 24, 2017 by MaxQuest

[Boeroer]

But MayQuest - that's the case for all things that get raised by a fixed amount every level. Like deflection, endurance, accuracy and so on. The relative gain of +10 accuracy is lower on higher levels than it is on lower. If you have 30 ACC and gain +10 ACC it's a bigger increase (percentage wise) than getting raised from 90 to 100.

[JerekKruger]

But MayQuest - that's the case for all things that get raised by a fixed amount every level. Like deflection, endurance, accuracy and so on. The relative gain of +10 accuracy is lower on higher levels than it is on lower. If you have 30 ACC and gain +10 ACC it's a bigger increase (percentage wise) than getting raised from 90 to 100.

 

Not true for Deflection and Accuracy since what actually matters for them is not their absolute value, but rather their value relative to whatever you're fighting. To take your example, +10 ACC when I currently have 30 is exactly the same as +10 when you currently have 90, assuming you're fighting Deflection 30 and 90 enemies respectively.

[Boeroer]

Sure, but that wasn't the point. The point was that the percentage of the increase is lower at higher levels if you stick to a flat number that is added every level. That's true for Empower but also for all the things I listed - also for deflection and accuracy. Because you always get the same bonus on levelup no matter how large your pool already is. 

Edited February 25, 2017 by Boeroer

[JerekKruger]

Sure, but that wasn't the point. The point was that the percentage of the increase is lower at higher levels if you stick to a flat number that is added every level. That's true for Empower but also for all the things I listed - also for deflection and accuracy. Because you always get the same bonus on levelup no matter how large your pool already is. 

 

Okay that's true, but surely Max's point was that in the case of Empower it's a problem.

[Boeroer]

Yeah - but even then my thinking is: Why is this a problem of Empower while it's not a problem with all the other stats which get absolute bonuses on level up (and not percentage based ones)?

 

I mean you could say it is a problem that the fixed endurance bonus has less impact at higher levels - but nobody complained about that, right?

 

It's always like this: if you level up from lvl 1 to 2 it feels like you took a big step - and you did. If you level up from 19 to 20 it's hardly noticable (except you get a very potent new ability).

That's also the reason why in later levels the new abilites have to be more powerful than the older ones - else a level up wouldn't do much for you anymore.

Edited February 25, 2017 by Boeroer

[Dr. Hieronymous Alloy]

It's also just fine for each progressive level up to give you "less", subjectively speaking, as long as the rate of change is proportionally constant. You play less and less of the game at or above each successively higher level.

[Boeroer]

Yeah - also, a percentage bonus based on actual stats would lead to absurd and exponential power growth.

[MaxQuest]

Okay that's true, but surely Max's point was that in the case of Empower it's a problem.

Exactly)

But MaxQuest - that's the case for all things that get raised by a fixed amount every level. Like deflection, endurance, accuracy and so on. The relative gain of +10 accuracy is lower on higher levels than it is on lower. If you have 30 ACC and gain +10 ACC it's a bigger increase (percentage wise) than getting raised from 90 to 100.

Intrinsic diminishing returns by themselves are not a problem. (imho well-balanced intrinsic returns are actually a god send)

The problem is if there is too big discrepancy in relative gain that we get from +m to a stat in:

- situation when that stat is low, versus

- situation when that stat is high

Yeah - but even then my thinking is: Why is this a problem of Empower while it's not a problem with all the other stats which get absolute bonuses on level up (and not percentage based ones)?

Let me explain, why I see the problem in Empower, but not in the aforementioned: accuracy, deflection, endurance.

 

1. Accuracy (in PoE1)

 

 

- Let's imagine that you have 30 ACC, and your enemy has 30 in respective defence.

- Let's imagine that you have gained +10 ACC. This would shift your attack resolution to the right, 

 .- basically resulting in 10% of total swings resulting in a crit instead of miss.

 .- or we could say that 66% of misses were converted to crits. same thing.

 

Sure going from 30 to 40 ACC, versus enemy with 30 DEF, would net you a higher relative gain than going from 90 to 100. Because in first case part of your misses were upgraded to crits; while in second part of your hits were upgraded to crits.

 

The more ACC you get, the less useful (in general) it gets.

But, there is a counter-weight. Even few of them:

- onCrit effects

- enemies defences also go up, as the game progresses

- existence of enemies with VERY high defences

 

If your ACC goes up from 90 to 100, against enemy with 100 DEF; it's completely the same thing as:

If your ACC goes up from 30 to  40, against enemy with  30 DEF.

 

 

2. Deflection (in PoE1)

 

Deflection is affected by increasing-returns from the start.

If your enemy has 30 ACC, and your DEF is increased from 30 to 40, this would shift attack resolution by 10 to the left,

 .- basically resulting in 10% of total swings resulting in a miss instead of hit.

 .- or we could say that 20% of incoming hits were converted to misses.

 

Going from 90 to 100 DEF, versus enemy with 30 ACC, would net us a higher relative gain than going from 30 to 40. Because the closer we are to (100 + enemyAcc), the closer we are to taking no damage at all.

 

The more DEF you get, the more useful (in general) it gets.

But, there is a counter-weight. Even few of them:

- enemies accuracy also goes up, as the game progresses

- there are many stuff that targets other defences as well; and player has to balance those

- existence of crowd-control abilities; that to some extent allow us to just cc the enemy, while significantly dumping def in order to free stats/talents for other facets.

 

3. Endurance (Constitution) (in PoE1)

 

Indeed, the higher your current maximum endurance is, the less relative gain you get on each level-up.

 

It is not a stat that we won't to be as high as possible. But rather high enough. 

Because if your character is dead, it doesn't matter how high his other stats are.

And if he is alive, and has some endurance/health left at the moment you rest, it means that points in constitution were partially wasted.

 

The thing is:

- it also has a sweet spot which is not static (i.e. it varies depending on party composition and play style)

- it can be of similar use in early and late game

- even if it wasn't perfectly balanced, nobody would complain about that, as we would either gladly put more points in con (if it was more useful), or gladly have free points to put in other stats (if con was less useful).

 

So as we have seen, the above stats although do have intistic diminishing returns, do not result in a problem. Because their gradual decline in relative gain from +flat value; is counter-weighted.

 

---

Now we have arrived to Empower.

 

 

First, what do we know about it?:

- it's something that makes our spells hit harder. As was already said, Empowered Fireball will deal more damage. Empowered missiles will fire more of.. missiles.

- empower adds +3 to power_level; this means that power_level is part of damage formula

- power_levels range from 0 to 10 (at 20th character level)

 

Now, let's take for example Minnoleta's Missiles. We know that:

- this spell releases x missiles

- number of missiles scales with power_level. Must be some: count = c + a * power_level

- enemies have no flat DR in PoE2

- this spell will ignore target's AR

- total_damage = damage_of_1_missile * count

- damage_of_1_missile = base_damage * (1 + might_coef + hit_quality_coef + etc)

 

So what do we have?

-  1/0  wizard casts regular missiles: total_damage = damage_of_1_missile * (c + a * 1)

-  1/0  wizard casts empower missiles: total_damage = damage_of_1_missile * (c + a * 4)

- 10/10 wizard casts regular missiles: total_damage = damage_of_1_missile * (c + a * 7)

- 10/10 wizard casts empower missiles: total_damage = damage_of_1_missile * (c + a * 10)

- 20/0  wizard casts regular missiles: total_damage = damage_of_1_missile * (c + a * 10)

- 20/0  wizard casts empower missiles: total_damage = damage_of_1_missile * (c + a * 13)

 

where:

- a is how many missiles are added per 1 extra power_level; and is quite likely to be an integer as that would be the most intuitive and easier to tooltip option. And perhaps it will be equal to 1, because firing 26 missiles (especially if they would bounce) would be like fireworks. 

- c is how many missiles are fired by default 

 

---

Tbh, When I had it all in my mind, and before starting to lay it down here, I didn't consider 'c'. 

Imagine as it was zero. In that case the problem I was talking about would become apparent, even with a = 1:

-  1/0  wizard casts regular missiles: total_damage = damage_of_1_missile * 1

-  1/0  wizard casts empower missiles: total_damage = damage_of_1_missile * 4 (i.e. a whooping +300% relative gain from Empower)

- 10/10 wizard casts regular missiles: total_damage = damage_of_1_missile * 7

- 10/10 wizard casts empower missiles: total_damage = damage_of_1_missile * 10 (i.e. a +42% relative gain from Empower)

- 20/0  wizard casts regular missiles: total_damage = damage_of_1_missile * 10

- 20/0  wizard casts empower missiles: total_damage = damage_of_1_missile * 13 (i.e. a +30% relative gain from Empower)

 

The first problem would be: at low levels Empowers gives one hell of a boost.

Second problem would be: multi-class characters would have a higher gain than pure-class ones.

 

But now I actually see how this can be mitigated, or rather smoothed.

 

---

- First of all it is clear that c should be greater than zero. This will greatly reduce that whooping relative gain from Empower on low levels.

 .- with something like c =  5, and a = 1; That +300% gain will be reduced to +50% (9/6 instead of 4/1). While +30% would get reduced to +20% (18/15 instead of 13/10).

 .- with something like c = 10, and a = 1; That +300% gain will be reduced to +27% (14/11 instead of 4/1). While +30% would get reduced to +15% (23/20 instead of 13/10).

- Second, c cannot be too high, as that would heavily reduce the important of power_level and will especially hit pure classes.

- Third, as we have seen with missiles, power_level coefficient is applied multiplicatively. For the sake of consistency it should do the same for usual spells too. For example firebal damage = base_damage * sum_of_dmg_coefficients * [(c + a * power_level) / c]. 

 .- with c =  5, a = 2; Firebal at power_level 10, will deal x3.57 more damage than at power_level 1. (5.0 / 1.4); or x1.31 more than at pl 7 (5.0 / 3.8.

 .- with c =  5, a = 1; Firebal at power_level 10, will deal x2.50 more damage than at power_level 1. (3.0 / 1.2); or x1.25 more than at pl 7 (3.0 / 2.4).

 .- with c = 10, a = 1; Firebal at power_level 10, will deal x1.81 more damage than at power_level 1. (2.0 / 1.1); or x1.17 more than at pl 7 (2.0 / 1.7).

- Fourth, we could potentially apply same scalling to class-specific base abilities. E.g. Sneak Attack = base_percentage * [(c + a * power_level) / c]

 .- with base_percentage = 10%, c =  5, a = 2; Sneak Attack would take values 10% (base), 12% (pl = 1), 18% (pl = 4),.., 38% (pl = 7), 50% (pl = 10) 

 .- with base_percentage = 15%, c =  5, a = 2; Sneak Attack would take values 15% (base), 12% (pl = 1), 18% (pl = 4),.., 57% (pl = 7), 75% (pl = 10) 

 .- with base_percentage = 15%, c =  5, a = 1; Sneak Attack would take values 15% (base), 18% (pl = 1), 27% (pl = 4),.., 36% (pl = 7), 45% (pl = 10) 

 .- with base_percentage = 20%, c =  5, a = 1; Sneak Attack would take values 20% (base), 24% (pl = 1), 36% (pl = 4),.., 48% (pl = 7), 60% (pl = 10) 

 .- with base_percentage = 25%, c =  5, a = 1; Sneak Attack would take values 25% (base), 30% (pl = 1), 45% (pl = 4),.., 60% (pl = 7), 75% (pl = 10) 

 .- with base_percentage = 20%, c = 10, a = 1; Sneak Attack would take values 20% (base), 22% (pl = 1), 28% (pl = 4),.., 34% (pl = 7), 40% (pl = 10) 

 

Taking it all together it looks, like we can have one global scaling formula [(c + a * power_level) / c], which is linear, very easy to explain and tooltip at values c = 5, a = 1. Which doesn't cause a high discrepancy in relative gain from empower (+50% at pl 1 versus +20% at pl 10), and thus can much easier counter-weighted, for example by:

- more Empower usages becoming available in the late game

- higher crit rate in the late game, plus access to more damage-increasing modifiers. 

- high indirect effect on DoTs. Because when calculating total damage of a (usual) DoT it is also multiplied by Int duration modifier. (and late game int is always higher, due to equipment, high-end buffs, food and resting bonuses).

 

 

---

TL.DR. I've started from asserting that Empower providing +3 flat bonus to current power level can lead to problems. And ended up with... the conclusion that it can actually be balanced) But it heavily depends on scaling formula.

 

P.S. Yeah, I know, I am overthinking it

Edited February 27, 2017 by MaxQuest

[Boeroer]

That's what I thought: I all depends what you do with (high level) Empower.

 

For example I had the idea that they could just add other, additional effects (to the missiles in this case) if your power level is just high enough. That way you could always balance those things out if you feel a mere +3 bonus is not sufficient.

 

It may indeed be that during the beta phase it turns out that the linear rise of Empower feels too weak at higher levels - as you feared. But there are a lot of things you can do then (like you pointed out).

 

Still good that somebody points out those little hooks. So thanks!

I mean look at bashing. If I had attended the PoE1 beta I would have insisted that this stuff gets fixed. Or the monk's Lightning Strikes doing only +10% lash which was always eaten up by enemies' DR and thus was useless. If somebody would have guessed that right from the start they wouldn't have had to patch it.

[JerekKruger]

Still good that somebody points out those little hooks. So thanks!

 

Definitely. It hadn't occurred to me even though it's quite an obvious (potential) flaw in the system.

 

I mean look at bashing. If I had attended the PoE1 beta I would have insisted that this stuff gets fixed. Or the monk's Lightning Strikes doing only +10% lash which was always eaten up by enemies' DR and thus was useless. If somebody would have guessed that right from the start they wouldn't have had to patch it. 

 

Have you pledged enough for beta access for Deadfire? If so, we can blame you for every problem that remains on release

 

More seriously, I will try to spend some of my beta time actually testing things rather than simply messing around for a few hours then waiting for release.

Edited February 27, 2017 by JerekKruger

[MaxQuest]

That's what I thought: I all depends what you do with (high level) Empower.

 

For example I had the idea that they could just add other, additional effects (to the missiles in this case) if your power level is just high enough. That way you could always balance those things out if you feel a mere +3 bonus is not sufficient.

Yeap. Adding additional/secondary effects could both adjust the balance, and be actually fun. Missiles interrupting / pushing. Fireball knockbacking or adding a small dot / short panic effect / whatever)

 

I also had an idea of Empower slightly increasing penetration of the affected spell. But after some thought, decided that is should go to Might stat instead. As Might no longer has a counter-weight in form of flat DR.

 

Still good that somebody points out those little hooks. So thanks!

I mean look at bashing. If I had attended the PoE1 beta I would have insisted that this stuff gets fixed. Or the monk's Lightning Strikes doing only +10% lash which was always eaten up by enemies' DR and thus was useless. If somebody would have guessed that right from the start they wouldn't have had to patch it.

Exactly)

v3.0+ is quite well balanced, but looking at old threads... *shakes his head*. 

I mean it is great that Obsidian is open towards keeping iterating. But some of potential problems we indeed can theorycraft in advance, provided we have enough info. So there would be no need in fixing it. Also it would allow forum mech posts and wiki of being relevant and up-to-date for longer.

Edited February 27, 2017 by MaxQuest

[Boeroer]

Yes. I finally plegded for the beta access even though I don't have any windows OS. Maybe I will get one or play the game via wine (although I cringe at that thought). I just like to think that I could discover some game breaking (Carnage + Jolting Touch or Retaliation) or lame (Lightning Strikes: +10% lash, Bashing) stuff and prevent that from finding its way into the final game - thus preventing endless patching to a certain extend. Don't know if this will happen, but one can dream.

Edited February 27, 2017 by Boeroer

[ushas]

Yeah, those additional/secondary effects would be also interesting. Maybe even something unique to empower, which isn't gainable simply by getting corresponding levels or stats?

Just in case it wasn't already mentioned, Josh Sawyer in the reddit Q&A said that

You can use per-rest Empower resources to make a power extra-strong or to replenish your per-encounter resources in long fights.

​​
Not sure how it will look like. Perhaps, when already reached spell level quota, you can decide that you want to use a spell from that level again, and are able to do that in exchange for empower stuff...

Regarding what you have written, @MaxQuest, it's well thought out and interesting. Most of the time I'm just lurking, but I always like to read what you guys have to say about systems. It seems to me that we lack the needed info. Not only about the underlying by developer's words significantly changed class system. But also as you have yourself already touched at the end of the post, one would need take into account other progressions going on in the game -- the equipment, accumulation of abilities, enemies, even party composition, etc.

I agree that the proposed formula looks useful for tuning and "c" ought to be > 0, but actually also because I think nonlinear progression of classes may not be bad thing, especially when one takes into account all the other external effects (eg. the equipment progression). As such using anecdotal example, in PoE1 I haven't minded at all diminishing returns you have mentioned, on the contrary, I've eventually wished we don't get any level ups in those things (because how powerful relative to the opposition we can get). Here they also plan to make multi-classed characters within 75%-85% power of a single-classed character. A single-classed character have at hand more class based abilities to choose from and all of them use the same high power source, so even when the relative gain in isolation wouldn't look so good compare to lower levels, he might be quiet effective character overall. But it's impossible for me to evaluate those things without bigger picture, so I apologize for not giving any numbers.

On a side note, I don't dig why it has to be a system where we are required to use together class level, power level and power points for each class. I sympathize with the hard work of UI designers. From the distance it seems to me that one can simply continue the transformation into using just power points and distributing them at will between two classes. You can still gate access to abilities or spell levels by required class power points instead of class levels. Even when it's slightly different, one doesn't reach high level spells unless literally progressing corresponding source points. It may be also possible to use power points instead power levels in your calculations, and have the sneak attack progressed more gradually.

[ushas]

​

Yes. I finally plegded for the beta access even though I don't have any windows OS. Maybe I will get one or play the game via wine (although I cringe at that thought). I just like to think that I could discover some game breaking (Carnage + Jolting Touch or Retaliation) or lame (Lightning Strikes: +10% lash, Bashing) stuff and prevent that from finding its way into the final game - thus preventing endless patching to a certain extend. Don't know if this will happen, but one can dream.

I hope you will not become mad at me with wine ​

​

​Btw. I recall maintaining one thread about trap issues at the end of beta, I think it got finally fixed in 3.0+ (at least most of the problems I think). So don't worry.

But on more serious note, how do you guys like the effectiveness of forums environment? I remember shmerl tried to establish community bug tracker at github but without the official support it's not really viable.

Edited March 1, 2017 by ushas

[Madscientist]

Lots of people (including myself) have posted lots of bugs in the forum. Most of these issues have been fixed, although they managed to add some new bugs along the way.

So I think the forum effectiveness is OK.

 

The game was very buggy at release and I had to quit my first playthrough because of this.

I started again and finished the game a few patches later, it was much better than.

With version 3.0+ the game is well balanced (compared to most other single player RPG) and usually you can play it without facing game breaking problems. There is no game that is completely bug free.

 

I was not in the beta for PoE1, so I cannot say which problems have been solved before release and what could have been solved if you did the beta testing.

The PoE1 system has been created from scratch and there where lots of changes what each stat does and some other things. Now we have a system that works quite well most of the time. Tyranny (at release) was much better than PoE1 (at release). I could finish the first version without big problems. There were some bugs but most of them have been fixed. Because of the classless skill system the balance has been lost and it is easy to make very powerful (or very useless) chars. Some new things from tyranny will be in PoE2 (but fortuanatly not the same classless skill based system).

 

Now we have a very solid basic system for PoE2. The main issue of beta testing will be to test if some game mechanics and class abilities lead to totally underwhelming or overpowered results.