Test Server Findings

[VZX01]

because it seems Externator -> Evisceration is darkness

This seems false.

[noodles355]

Looks like my proposed set for Thief, replacing hands with Skadi, feet with Thf AF3 and body with NotToci'sQQ

[VZX01]

Denali +5 wristbands, Kecks, and Gamashes are very good for Exenterator. Now we need to know how obtainable are they.
Use Ocelomeh for head and Loki's for body.

[Banalaty]

Unless you are force critting extenerator, bring back the old Denali jacket over lokis. 10 agi, 9atk>11 agi without the crit bonus being a factor.

[InsaniacTHF]

Haha, check it out!

So if DAII is 2% it's possible they did add TAII and we just didn't notice cause it's only 1% lol.

[Byrthnoth]

So if DAII is 2% it's possible they did add TAII and we just didn't notice cause it's only 1% lol.

I think I have a fairly large parse at 5.XX% TA from two nights ago. I do not remember where/if I posted it, but I can continue it until I exclude 6% if it doesn't already.

[Yugl]

With 1015 attack and mob's 327 defense I'm at ~3.1 cRatio and I did WS with pDIF around ~2.218

2.218*(20/21)/1.2=~1.76
1.76/3.1=0.567

Its only a proof that penalty cant be higher than 44% but thats still a lot.

@Raelia That is exactly how I understand it too and thats why I was confused by what Yugl is writing. I expected more from a math person like you Yugl!

I didn't write the page; I just read it. As it's written, the page says 2.25/2.7XX/Multiply by 1-1.05. There is no and cannot be any link to Masamune's data, so I cannot look at that for clarification purposes.

[Motenten]

Re: Resolution

pDif calculations are complicated. I worked with Masa on his data, but we never did reach a final formula (motivation to finish it kind of petered out). I have a fairly solid, though incomplete, rundown of how it's calculated, though. (I can post my algorithmic writeup if people are interested.)

Max pDif, for example, is not 1.2*cRatio, it's a value of ~0.225 * cRatio (0.225 approximates 230/1024) with a cap of approx. 378/1024 (0.369) that gets added to cRatio. Cap cRatio on 2-handed weapons is 2.25, so that becomes 2.25 + min(0.369, 2.25*0.225) == 2.25 + min(0.369, 0.506) == 2.25 + 0.369 = 2.619. Then with the random multiplier it can reach 2.619 * 1.05 = ~2.75. 2.75 * 32 = 88 (almost exactly), which you will almost never see, so max melee hit you'll likely see is 87 (same as observed).

Using 2.2 * 1.2 = 2.64 * 1.05 = 2.772 * 32 = 88.704, and 88 would have likely been the high value seen for melee hits.


Atoreis: No sample size was given. How many weaponskills were performed for each? Did you parse them so that you could differentiate between the lvl 81 and lvl 82 birds?

Going to make a guess: you mentioned 300 hits for melee to get your max melee hit. With a 7-hit build, that would be around 50 weaponskills. With the third test cut short, I'm going to guess 20 each for the first two sets and 10 for the last.

With 20 weaponskills, there's a reasonable chance that the highest damage weaponskill had all 5 hits landed in the top 20% of the pDif range. The average over those 5 hits was then probably at about the 90th percentile of the full pDif range, if spread evenly across the top 20%. With margin of error, it's most likely between 0.8 and 0.9.

First set:

Observed high pDif: 1.5

With 1.5 as a high-end pDif, cRatio would have been below 1.5, which means the potential for a high amount of clustering at 1.0. Did you see several weaponskill damage values fairly close to each other at somewhere around 665 damage? Actually, that's probably about the minimum damage you saw, if I'm guessing correctly. (Please include all data points when doing these sorts of tests; they help in the calculations.)

Very rough estimate: 1.3 cRatio. Max pDif of 1.6. Min pDif of 1.0. Approx 1/4 of all hits fall below 1.0 and are reset to 1.0.

Dropping from 2.25 to 1.3 implies about a 40% reduction in attack.


Second set:

Observed high pDif: 1.9

Predicted cRatio: 1.65 (range of 1.17 to 2.02).

Nominal cRatio: 901/327 ~= 2.75

Reduction: 40%


Third set:

Observed high pDif: 2.21

Predicted cRatio: 1.93 (range: 1.5 - 2.3)

Nominal cRatio: 1015/327 ~= 3.10

Reduction: ~38%


Overall conclusion: Most likely a 40% reduction in attack.

[Mojo]

Seems like all of those attack penalized weaponskills are worthless

[Prothescar]

Ugh, seemed pretty high, but did not expect it to actually be that high. Definitely standard work by SE though. Going to get some more possibly accurate data later, no real reason to believe it's less or more than 40 though.

[Nightfyre]

SE:

[Taint]

It is the test server, it could all be ironed out before the real launch.....or not

[Prothescar]

like heavy strike? if anything they'll make it worse : D_

[fantasticdan]

So is the new katana WS worse than jin outside of abyssea than? And does sword have the same attack penalty?

[Prothescar]

If sword has an attack penalty while it gains an attack boost from TP I'd lol so hard. It would very well explain why the attack boost seems incredibly small or nonexistent.

[Dtroyy]

If sword has an attack penalty while it gains an attack boost from TP I'd lol so hard. It would very well explain why the attack boost seems incredibly small or nonexistent.

Wouldn't suprise me lol

[fantasticdan]

Well katana has the same description and that apparently has the attack penalty...

[Prothescar]

http://i384.photobucket.com/albums/o...uo/BALANCE.png

[Neisan]

If sword has an attack penalty while it gains an attack boost from TP I'd lol so hard. It would very well explain why the attack boost seems incredibly small or nonexistent.

Piercing Arrow, ignore 0 defense at 100 TP etc.

[Foldypaws]

Sidewinder, Accuracy varies with TP, -60 ACC @ 100 TP.