* Second specialization is now left open to the player. I had previously assumed it would be Duelist, but this was unreasonable since Ranger and Bard are both strong.
* These calculations are accurate for the 1.02 patch (which is the same as 1.01 with the dex hotfix as far as the DPS calcs are concerned).
Warning, wall of text incoming. Skip down to sections III and IV for the tl;dr version, but I recommend checking out the gameplay mechanics stuff if you're interested. I'll keep this guide updated as we discover mistakes and gear improvements.
This is a continuation of a thread over at the now-baleeted daforums.bioware.com (used to be at http://daforums.biow...06563&forum=135). We were discussing various DW rogue builds and discovered some interesting things about gameplay mechanics along the way. I made a spreadsheet and did some DPS calculations for four different DW builds, with the results summarized below. For those who didn't follow the original thread, I'll repeat the pertinent gameplay mechanic information below in section I.
DISCLAIMERS: this is pure theorycrafting. We were just trying to come up with the highest-DPS endgame DW builds we could think of, so we assume optimal stat allocation and access to the best gear. This may not be realistic for everyone. I may run some figures with less expensive gear in the future (equipment suggestions welcome!) This is not a guide for how to build the "best" rogue. DPS is not everything to everyone. The calculations don't quantify utility or fun factor or anything like that. I just hope this is a helpful starting point for people planning their own rogues.
PLATFORM: all of my analysis assumes PC version 1.02 (or 1.01 with the dex hotfix).
NOTE FOR READERS OF THE OLD THREAD: The numbers here are slightly different from the last ones I posted in the old thread. That's because
TABLE OF CONTENTS
I: Gameplay mechanics
II: Stats and gear
III: DPS chart
IV: Conclusions
I. GAMEPLAY MECHANICS
A. DAMAGE FORMULA
I pieced together the damage formula as best I could using the toolset to view the source code. Those of you following along at home can look at core_h.nss and combat_damage_h.nss to see whether I got things right.
At an abstract level, the formula for expected damage on a backstab looks like this:
C * (AVERAGE(W, W * R) + 0.375 * X * Y ) + O - A
C is the critical hit/backstab modifier,
W is the weapon's base damage,
R is the weapon's damage range multiplier. The game randomizes your weapon's damage between W and W * R. For daggers and longswords, R is 1.5; for axes it's 1.4. For example, a dagger with a listed base damage of 6.4 will actually do between 6.4 and 1.5 * 6.4 = 9.6 damage. Thus the expected weapon damage will be AVERAGE(W, W * R).
X is the attribute modifier (e.g. 1.10 for axes, 0.425 for daggers),
Y is the number of relevant attribute points above 10 in each relevant attribute.
O is all other sources of bonus damage, and
A is the target's armor minus the attacker's armor penetration, floored at 0. Note that the game randomizes armor to between 70-100% of its "list value" in the damage calculation, but I don't account for this for technical reasons (calculating the expected armor value in Excel is beyond my puny spreadsheet skillz). This error is not large.
C defaults to 1.5 but can be raised up to 3.5 by gear. (Thanks gorboth for pointing out that the cap is 3.5, not 2!)
Why is the bonus damage from attributes multiplied by 0.375? Because, when dual-wielding, the game randomizes your attribute damage bonus to between 25-50% of its list value before throwing it in the damage calcs. This is one source of random variation in damage output. Thus your expected attribute damage bonus will be only 37.5% of its list value. (For every other weapon style besides dual-wielding, the random factor is between 50-75% of list value.)
Here is a list of all the sources of bonus damage that I accounted for (everything that goes into O):
* Exploit Weakness, which adds an expected (0.52/3) * (cunning-10) bonus damage to each backstab. Thanks Twenynge for correcting me on this (I'd prevoiusly thought it was 0.2 * (cunning-10)).
* Tainted Blood, which adds 5 + 0.2 * (cunning-10) bonus damage
* Bonus damage from gear, including runes and bonus damage on offhand weapons (offhand bonuses seem to contribute to mainhand attacks)
* Bonus damage from levelling (for rogues this is 0.2 damage per level)
Party buffs are NOT accounted for because the variations are endless.
Here is a list of everything that goes into the target's armor, A:
* Target's nominal armor value
* Armor penetration from the weapon
* Bonus AP from gear, including bonus AP from offhand weapons (offhand bonuses seem to contribute to mainhand attacks)
* Cunning's AP bonus of (cunning-10)/7
B: DUAL WIELD ATTACK TIMING
I did a bunch of video capture experiments to determine that attack timing for dual wield weapons works as follows:
Step 1: for each weapon you're wielding, start with 1.5s and substract that weapon's speed modifier (see [[http://dragonage.gul...p/items/weapons for the modifiers). Thus for daggers this value will be 1.0s and for axes it will be 1.4s. The 1.5s starting point comes from the base attack interval for the dual-wield style, as defined in core_h.nss, and verified by my in-game testing.
Step 2: Average the two numbers above and call the result T. This is your attack interval.
Whether backstabbing or attacking from the front, you will alternate hands every T seconds. Don't believe the backstab animation, which looks like it's attacking with the mainhand only. Also, when backstabbing, it appears that the mainhand runes trigger (or have a chance to trigger) on every attack, even when the offhand should be attacking. Offhand runes never trigger. I can only assume this is a bug.
Because your attack speed is the average of your two hands, the DPS-conscious rogue with a fullsize weapon in her mainhand will usually want to use a dagger in her offhand, to speed up her attacks. This also saves her attribute points that would have gone to dex for DW Mastery.
Haste and Momentum reduce T by 25% and 30%, respectively, at the END of the attack timing calculation. Thus I ignore them for purposes of the DPS calculations, since they'll affect all builds by the same proportion.
Modifié par Discobird, 13 décembre 2009 - 12:07 .
Retour en haut
In the old thread there were several people saying they only found 3 (including me). If a lot of people here say they found 4 I'll go ahead and make that fix. Thanks.
