67 réponses à ce sujet

[RangerSG]

Hmm, the ogre in the beginning (Tower of Ishal) always grabs my tanks. And unless my PC is a mage, they die horribly as a result because I've got no one to heal them, the grab-punching bag thing always kills them because the circle mage you get can't heal. Now, I'm sure I'm missing something on that. But to say 'all that grabs' is dragons. Or 'you'll survive the grab,' erm....not in the early game. Ogres grab, and if it's early game, they kill you when they do.

[Chains-Gore]

Stun the Ogre, it'll drop your tank.

[Fleapants]

The next ogres after the Tower should be in, hum, Deep Roads or the Brecillian Forest?

Most goes Redcliffe/Haven or Circle Tower first, but even then, a Cone of Cold or Dirty Fighting (can't be resisted) will release your tank easily.

[Warkow]

Ugh, I don't agree with much of this post. Threat being broken stands out also, i'm glad it is not as easy mode as wow, I am pretty sure that some of the threat in here is intentional, ie sometimes you just HATE the magem and no stinkin Shield is going to tell me otherwise. I enjoy that you can't 'puppet' every mob. GJ BW.

[Theramond]

Here is a replication of the previous experiment. Instead of 3 trials each, I did 15 trials each. For the control group, strength was only pumped to 38. Lastly, the highwaymen at Lothering were attacked instead. As said previously, this is designed to measure the real time effects of attribute allocation alone. Because the experiment was done in a highly controlled environment, the experiment does not reflect actual game play circumstances. In addition, the opposing NPC's used in the experiment are not representative of all opposing NPC's within the game.

[Update] a Str_Con group was added post-hoc. Str_Con is not a possible configuration without using the developer console. I included this group as it acts as a secondary control group from which the experimental groups can be compared.

[Update] another control group "Control_Arm" was added. I took the control group and added massive dragon bone armor which can be compared to the original control to measure the effects of adding high quality armor.

The details of the attribute allocation are as below:
Control group:

Str_Dex group:

Str_Dex_Con group:

Str_Con group:

Control_Arm:


Demo of a trial:
www.youtube.com/watch

Here are the results:
Plots:



A side-note: from further analysis it appears that the variability across trials mostly has to do with the variability of the A.I.'s behavior (that is the talent/spells being used, the time they were activated, and whether or not it connected with the target). For example in trial 4 for the Control_Arm group, the PC was inflicted by "sunder armor" very early on in the trial. That appears to be why the PC lasted for only 4 seconds, despite being armed with massive dragon bone armor.

Raw Data:

Modifié par Theramond, 30 décembre 2009 - 09:16 .

[2late2die]

@Theramond: Wow, that's some serious testing. I'm sure I'm not speaking just for myself when I say that we appreciate your work on this. And, to me at least, the results really make sense.



It seems that a more balanced approach will yield better results. I'm guessing that, as some have pointed out, benefits of defense are either capped at some fixed %, or decline the more points you put in there beyond a certain limit. So adding constitution at that point would be better than defense.



There does seem to be a consensus that the best build for sword & shield warrior is 38str, 26dex and rest in con. Obviously depending on play style variations are possible such as 42str (I personally like having more stamina as I find it allows me to be more flexible during combat, and so I don't use massive armor), or maybe few more dex to get less hits.



I can definitely see how a tank with 53dex and 13con wouldn't be that great. Sure he's not gonna get hit often but when he does get hit it's gonna hurt something fierce.



At the end of the day, while I like having good builds, hardcore min/maxing is just not fun. I would even put points in willpower, because if I'm creating a party protector that will be a leader and an inspiration to the rest of the characters, it just doesn't make sense that he'll have paltry 10 willpower.

[RobotXYZ]

One thing I wondered was why the physical resistance is 37 for each build?  Doesn't Con increase that?


I also wonder if your character is attacking while all of this is going on.  A character with more dex will hit more often.  A character with more strength will also hit more often and for more damage.  Con in no way improves hit and damage.

By killing enemies who are dealing damage to you (or knocking them down) your character will also reduce the damage he is taking.

Not sure how important that facet is..


Offense is predicted to be the most important with fights against few enemies.  To the limit of few enemies killing your lone opponent reduces the threat to zero.  While against 1000 enemies killing 1 of them does not help.

Offense is predicted to be more important the longer you last.  If you have a healer making your effective HP pool larger then that is more time to kill enemies.

Modifié par RobotXYZ, 30 décembre 2009 - 03:05 .

[Lord Phoebus]

I think Theramond didn't get his dex high enough to really see the benefits (you still get hit at 101), once your defense hits around 120 (gear+modes) you can just stand in a group of trash enemies and none of them can hit you. Still I'm wondering why he was losing that encounter, the last three characters should have been able to handle it solo.

Modifié par Lord Phoebus, 30 décembre 2009 - 03:46 .

[Fleapants]

It would be very interesting to see what 30 points of armor would add to the test results ^^

[Theramond]

Fleapants wrote...

It would be very interesting to see what 30 points of armor would add to the test results ^^

Done.

[Theramond]

Lord Phoebus wrote...

I think Theramond didn't get his dex high enough to really see the benefits (you still get hit at 101), once your defense hits around 120 (gear+modes) you can just stand in a group of trash enemies and none of them can hit you. Still I'm wondering why he was losing that encounter, the last three characters should have been able to handle it solo.

I can test that too, but maybe some other time. I've been doing this all day. I did do a couple of trials were I brought defense up to 150, and the PC lasted 12-20 seconds. Constitution did not seem as great. So perhaps higher dex outperformes constitution in the long run. But then again, I only did a couple trials.

Modifié par Theramond, 30 décembre 2009 - 04:49 .

[DragoonKain3]

Definitely, higher DEX outperforms CON by a large margin in the long run. This is because DEX runs on increasing returns the more you have of it, and there is no cap in evade %. It is very practical to achieve 100% evade rate against trash mobs, and 100% evade rate even against reds assuming you run Heroic Defense 100% of the time.

Perform a test with...

26 base STR
all the rest in DEX (+50)
Fade (+4)
Warrior bonus (+4)
Human (+1)
Level 20 (+20)
Helm of Honnleath (+2)
Dalish gloves/boots (+5)
Felon's Coat (+15)
Andruil's Blessing (+2)
Magister's Shield (+6)
Key to the City (+2)
Lifegiver
Rose of Thorns (+2)
Champion's Shield (+15)
Shield Defense on (+15)
Rally on (+10)

That's +153 DEF modifiers. Add on to that the Warrior's 45 base DEF, and he has a whopping 198 base DEF. And I can guarantee you that he will last infinite seconds against lothering chumps just standing there.

I mean, if my DEX tank Alistair can solo Revenants without ever getting hit even without being optimized through stats, nor wearing optimized gear, how much more for the warrior above?

Modifié par DragoonKain3, 30 décembre 2009 - 05:58 .

[Haplose]

Dexterity adds the same amount to Physical Resistance as Con.



The test isn't very practical as a source of information. Like DragoonKain wrote it ignores equipment but also mostly ignores party synergy.



At 101 Defence you might still get hit almost every time.

But you can easily buff that by 50 or even close to 100 points!

Meanwhile the Armour rating can get only minor buffs for melee classes (Shield Wall +5?).



Where's Rally for +10 (maybe +20, if two chars have it)?

Stone Aura (ok, maybe not).

Where's Heroic Defence, Glyph of Warding (I'm not sure... 20 or 30 Defence each)?

You can pass the 150 Defence benchamrk very early (soon after you recruit Wynne infact) and then you start being near-unhittable. Before buffs, at Defence lower then 100 it's hardly worth it, agreed, but afterwards it soon becomes uber.

I always have Wynne autocast Glyph of Warding and Heroic Defence on Alistair. Sure it costs mana.. but then I hardly have to heal him!

[Theramond]

Haplose wrote...

Dexterity adds the same amount to Physical Resistance as Con.

The test isn't very practical as a source of information. Like DragoonKain wrote it ignores equipment but also mostly ignores party synergy.

At 101 Defence you might still get hit almost every time.
But you can easily buff that by 50 or even close to 100 points!
Meanwhile the Armour rating can get only minor buffs for melee classes (Shield Wall +5?).

Where's Rally for +10 (maybe +20, if two chars have it)?
Stone Aura (ok, maybe not).
Where's Heroic Defence, Glyph of Warding (I'm not sure... 20 or 30 Defence each)?
You can pass the 150 Defence benchamrk very early (soon after you recruit Wynne infact) and then you start being near-unhittable. Before buffs, at Defence lower then 100 it's hardly worth it, agreed, but afterwards it soon becomes uber.
I always have Wynne autocast Glyph of Warding and Heroic Defence on Alistair. Sure it costs mana.. but then I hardly have to heal him!

It ignores a lot of things, that is how it was designed (it is a reductionist approach). So, yet, there is some impracticality to it, there is an overall  purpose for ignoring many factors that come into play. If you took all factors into account in one get-go, the test results would be extremely variable. Also, I can certainly simulate the effects of having multiple defense buffs, and see if there is an accumulative effect across the different groups. If so, then the information should certainly reflect what you and others have been saying. Ultimately, what is going to make Dex better than Con (IMO), is whether or not if there is some sort of accumulative effect, which is something I can test for.

Modifié par Theramond, 30 décembre 2009 - 10:19 .

[shree420]

Glyph of Warding is particularly uber...30 seconds, I believe, of +30 defense, the biggest defense buff in the game. My Lvl 15 Templar main was unhittable when fighting a Revenant with this on. This is with 30 Dex, I believe - have a lot of defense boosting items on him. Can't wait to boost the base Dex past 40.



This is also why I think Shield Wall is a relative waste; Shield Defense is better, although it's one of the lower-level talents. With Shield Mastery and Expertise, SW gives you +5 armor over SD, but SD gives you +5 defense over SW.



SW does have the advantage of no direct knockdowns, but IMHO that's a narrow category. Plus it decreases your damage by 20%, which for a tank already dealing lesser damage than his DW or 2H counterparts is aggravating. SD has no such penalty( penalty to attack at the start, but Shield Balance removes it).



Any thoughts on this?

[Haplose]

Theramond wrote...

Haplose wrote...

Dexterity adds the same amount to Physical Resistance as Con.

The test isn't very practical as a source of information. Like DragoonKain wrote it ignores equipment but also mostly ignores party synergy.

At 101 Defence you might still get hit almost every time.
But you can easily buff that by 50 or even close to 100 points!
Meanwhile the Armour rating can get only minor buffs for melee classes (Shield Wall +5?).

Where's Rally for +10 (maybe +20, if two chars have it)?
Stone Aura (ok, maybe not).
Where's Heroic Defence, Glyph of Warding (I'm not sure... 20 or 30 Defence each)?
You can pass the 150 Defence benchamrk very early (soon after you recruit Wynne infact) and then you start being near-unhittable. Before buffs, at Defence lower then 100 it's hardly worth it, agreed, but afterwards it soon becomes uber.
I always have Wynne autocast Glyph of Warding and Heroic Defence on Alistair. Sure it costs mana.. but then I hardly have to heal him!

It ignores a lot of things, that is how it was designed (it is a reductionist approach). So, yet, there is some impracticality to it, there is an overall  purpose for ignoring many factors that come into play. If you took all factors into account in one get-go, the test results would be extremely variable. Also, I can certainly simulate the effects of having multiple defense buffs, and see if there is an accumulative effect across the different groups. If so, then the information should certainly reflect what you and others have been saying. Ultimately, what is going to make Dex better than Con (IMO), is whether or not if there is some sort of accumulative effect, which is something I can test for.

Yes, I'm sure you could test it. I'm not sure what the result will be. But I think it's very possible that the result will show that when you stack all +Def sources both the Dex tanks and Con tanks become nearly unhittable (again proving the superiority of Con tanks).
And that may very well be true for the endgame, if you focus on the Defence aspect somewhat.
Still the results won't really be relevant for like 90% of the game inbetween...

The test that would interest me more would be the test somewhere in the 15 level range, with maybe 1/4 or 1/2 Defence boosts available. Realistically you will be spending most of the game closer to such conditions.

[shree420]

I think you've done a great job of testing it.



I'd suggest testing these tank builds across a wide variety of saves across a typical playthrough. The bandits at Lothering might or might not be a representative sample of enemies you find throughout the game. It'd also be interesting to see how the survivability evolves.



Off the top of my head, a Con tank might be superior at the start when Dex bonuses to defense don't really matter much, but then lose ground as the increase in HP fails to keep up with the increase in enemy damage potential. How it fares in boss fights is also a thing to look out for.

[Haplose]

shree420 wrote...

Glyph of Warding is particularly uber...30 seconds, I believe, of +30 defense, the biggest defense buff in the game. My Lvl 15 Templar main was unhittable when fighting a Revenant with this on. This is with 30 Dex, I believe - have a lot of defense boosting items on him. Can't wait to boost the base Dex past 40.

This is also why I think Shield Wall is a relative waste; Shield Defense is better, although it's one of the lower-level talents. With Shield Mastery and Expertise, SW gives you +5 armor over SD, but SD gives you +5 defense over SW.

SW does have the advantage of no direct knockdowns, but IMHO that's a narrow category. Plus it decreases your damage by 20%, which for a tank already dealing lesser damage than his DW or 2H counterparts is aggravating. SD has no such penalty( penalty to attack at the start, but Shield Balance removes it).

Any thoughts on this?

Shield Wall vs Shield Defence is a tough call. Being knocked down is pretty bad. I would think the extra defence from Shield Defence can be better initially for a Dex tank, when 10 points can make a serious difference.
Later on I imagine it looses it's effectiveness. These 10 points become rather easy to compensate with buffs and equipment. Something you can't do with knockdown immunity. And the +5 Armour would also be nice once you're rarely getting hit - mostly with autohit attacks.

Shield Wall would always be superior when you face knockdown-spamming enemies. Also always for Con tanks which are getting hit almost everytime.

So I would say there is likely a time in the game where Shield Defence is better for a Dex tank. It's difficult to say how long it lasts however.
I would say Shield Wall still wins overall (at some point for Dex tanks, always for Con tanks and always when facing knockdown spamming enemies).

[Zaragorne]

I just put all of Shale the golems points into health and use him to tank

[battleship potemkin village]

Fleapants wrote...

battleship potemkin village wrote...

Most games of this type (I'm thinking of Diablo II) will cap the percentage chance to avoid being hit (just like the resistances seem to be capped at 75%). If we assume that this is so, then the best "tank" build would run DEX up to the limit and put the rest in CON (which I'm told helps with physical resistance).

P.S.: Nice chart.

Str, Dex and Con increases physical resistance by the same amount, but PH isn't really that important as it's used to avoid knockdowns and grabs. Your tank will be immune to targeted KDs with enough talent points into shield spec, and you should be able to kill Ogres well before they get to grab, or at least well before they inflict any real damage.
That leaves the two dragons, which will grab you no matter what.

Well then, nevermind.

[Creature 1]

Theramond wrote...

I've decided to stop fangirling Zevran and fangirl you instead. 

[LynxAQ]

I wouldnt stop so quickly. I have no doubt that the graph making, research etc took some time but the data on the graphs are meaningless and leave out so many variables that they can only be taken with a grain of salt.



You have gone the wrong way completely with your tests. Which stat is the best to increase. Your tests say nothing and answer nothing of that question.



You needed to do the tests (keeping simple numbers to make my point) startinh with 10 con vs 10 dex for example. Run each numerous times against the same opponent (by numerous I mean 50+). Then increase the questioned stats by 2 each or something and rerun the tests again numerous times, increase each by 2 again, rinse and repeat. That way you would get an accurate picture of which stat was increasing a tank survivability the most.



BTW you can become unhittable in this game. There is no hard cap, which inherently means dexterity will be the best at the end of the day.



Also some people seem to have physical resistance wrong. It reduces damage from grabs, not reduces the chance of happening, same as normal elemental resistances.

[Theramond]

A +50 defense buff was added to the PC across all groups and trials to measure the possibility of an accumulative effect. This was measured against the previous data, which was used as a baseline.

Here are pictures detailing the stats:

Control:

Str_Dex:

Str_Dex_Con:

Str_Con:

Control_Arm:


Here are the results:

Plots:

Baseline:

+50 Defense Buff:

Raw Data:

Modifié par Theramond, 31 décembre 2009 - 09:11 .

[Shidi]

What I have learned from this thread:

1. This game is just like World of Warcraft

2. Excel spreadsheets are more fun than Dragon Age

3. LynxAQ is a helpful and pleasant person.

[shree420]

Haplose wrote...

shree420 wrote...

Glyph of Warding is particularly uber...30 seconds, I believe, of +30 defense, the biggest defense buff in the game. My Lvl 15 Templar main was unhittable when fighting a Revenant with this on. This is with 30 Dex, I believe - have a lot of defense boosting items on him. Can't wait to boost the base Dex past 40.

This is also why I think Shield Wall is a relative waste; Shield Defense is better, although it's one of the lower-level talents. With Shield Mastery and Expertise, SW gives you +5 armor over SD, but SD gives you +5 defense over SW.

SW does have the advantage of no direct knockdowns, but IMHO that's a narrow category. Plus it decreases your damage by 20%, which for a tank already dealing lesser damage than his DW or 2H counterparts is aggravating. SD has no such penalty( penalty to attack at the start, but Shield Balance removes it).

Any thoughts on this?

Shield Wall vs Shield Defence is a tough call. Being knocked down is pretty bad. I would think the extra defence from Shield Defence can be better initially for a Dex tank, when 10 points can make a serious difference.
Later on I imagine it looses it's effectiveness. These 10 points become rather easy to compensate with buffs and equipment. Something you can't do with knockdown immunity. And the +5 Armour would also be nice once you're rarely getting hit - mostly with autohit attacks.

Shield Wall would always be superior when you face knockdown-spamming enemies. Also always for Con tanks which are getting hit almost everytime.

So I would say there is likely a time in the game where Shield Defence is better for a Dex tank. It's difficult to say how long it lasts however.
I would say Shield Wall still wins overall (at some point for Dex tanks, always for Con tanks and always when facing knockdown spamming enemies).

I'd think that direct knockdown immunity is not that frequently required, and even when it happens, it's not that serious. And you do have the option of choosing Shield Wall then.

My main thrust was not about the defensive advantages - though I'd take better DEF than better Armor, given that I don't like CON tanking. +5 Def takes you into the unhittable zone vs. whites earlier.

The problem is the nerfed damage - 20% - which for a Tier 3 talent sounds like too much punishment. Like I said earlier, there's too much of a damage gap between S&S and 2H and DW, and this just widens it.

Against knockdown spamming enemies, yes, it's a benefit, but for most of the game Shield Defense should be your choice IMO.