Weapon Skills

[dasva]

Iirc all the kat/gkt ones with an element are hybrid. I've tested a few but forget which ones. On top of that the ranged ones mentioned are I think. Also swd has basically an elemental and a hybrid version of same ws basically. Iirc the higher lvl ones are the hybrids. Like burning/shining blade is full magic while seraph/rlb are hybrid. I think some of the other weapon types that basically get 2 weaponskills of same type in a row do that but haven't really tested them.

As far as which are hybrid easiest way to test is to find ws afected by mab and pdif

[fantasticdan]

In the process of going through the pages, I have found some more weaponskills that strike me as odd:

1) Skewer - 3 hits, 35% STR mod, chance to crit with a polearm. Why was this never as good as Raging Rush?

It couldn't always crit, IIRC they only added that sometime during toau. I remember it not performing well compared to penta thou even afterwards. 5.0 ftp 20% str/20% dex vs 3.0 ftp 35% str and probably a pretty low chance to crit; it doesn't look that great on paper to me.

[Byrthnoth]

Raging Rush is 3.0 fTP, 35% STR, and chance to crit. Penta Thrust has a ~20% attack penalty. Level 99 Gungnir was D100 and level 99 Bravura was D99. I know that Drakes averaged less at 75 merit camps (apart from Lolibri) than Raging Rush, so why was Skewer never considered? Either we were just too set in our ways when the patch hit or there's some kind of penalty associated with it.


All magical and ranged WSs (where it's easy to determined fTP) have anchor points/base fTPs that are multiples of 1/16. This led to the assumption that fTP anchor points are likely represented in /16 fractions. The Weapon Skills that do not conform to the 1/16 fTP rule on the wiki right now are:
Tachi: Ageha - 2.8 doesn't fit anything
Resolution - This is ridiculously precise and also in 32nds.
Entropy - 200TP value is in 32nds
Stardiver - 200TP value is in 32nds
Spirits Within - 300TP value is in 32nds, but this is also a weird WS

Now, we know that fTPs can be modified on the /256 scale because WS Gorgets/Belts scale that way. There are a few options that I see:
1) Most of the WSs that we're getting as /32 now have high modifiers and were released after Primeval brew, both of which makes it much easier to determine WS fTP. - It possible that SE has been doing this /32 crap all along and we just didn't know because we weren't testing with high enough resolution?
2) With the exception of Spirits Within, our fTPs are wrong for the above WSs.
3) It should be re-named the 32nds rule.

[shepardG]

I wish you could figure out how to make blade: ku not suck :-/. IMO, one of the top 5 WS animations still.

[Byrthnoth]

If those are actually its stats, nothing will save it. The only thing I can think of is that it might have a ridiculous accuracy boost at 100TP.

[Niiro]

Hybrid WSs are affected by base damage, though. It's complicated: http://www.bluegartr.com/threads/104...=1#post5052148

Also, I found the Chi mods after I posted that.

I know, I meant that would be an easy way to identify them because unlike magic-only WSs they are effected by base DMG.

Can also test if it stacks with SA or if it can miss (blinding potions), yes to either and you know it's a hybrid.

[dasva]

It's actually a lot easier for the non ranged ones. Hybrids use pdif so will vary with same setup while magical stay the same

[Casey_]

It couldn't always crit, IIRC they only added that sometime during toau. I remember it not performing well compared to penta thou even afterwards. 5.0 ftp 20% str/20% dex vs 3.0 ftp 35% str and probably a pretty low chance to crit; it doesn't look that great on paper to me.

Career lolDRG here, it used to be Accuracy varies with TP; It also rarely was used although I sort of remember it not being particularly bad, it's just that penta and drakes always did better probably because of more hits. By the time it was changed to a crit WS I forgot it existed except on ultra evasion mobs like byakko, at which point I was probably using wheeling thrust due to terrible gear.

[Prothescar]

DRG had pretty poor gear options at 75 compared to WAR too. Taking advantage of Skewer would have been difficult.

[fantasticdan]

Raging Rush is 3.0 fTP, 35% STR, and chance to crit. Penta Thrust has a ~20% attack penalty. Level 99 Gungnir was D100 and level 99 Bravura was D99. I know that Drakes averaged less at 75 merit camps (apart from Lolibri) than Raging Rush, so why was Skewer never considered? Either we were just too set in our ways when the patch hit or there's some kind of penalty associated with it.

I never knew penta had an attack penalty, was it even known at 75? Honestly it's weird that it does, I remember doing really well with it as war and sam against piercing weak stuff. Is it possible the penalty is first hit only? I suspect a lot of bonus/penalty on oldschool ws are first hit only.

But yeah, drgs gear options were incredibly bad at 75, on top of the same drawbacks it has now. Forget e body, it couldn't even wear regular haubergeon/hauberk.

[Foldypaws]

All magical and ranged WSs (where it's easy to determined fTP) have anchor points/base fTPs that are multiples of 1/16. This led to the assumption that fTP anchor points are likely represented in /16 fractions. The Weapon Skills that do not conform to the 1/16 fTP rule on the wiki right now are:
Tachi: Ageha - 2.8 doesn't fit anything

I did the Ageha testing, and it could easily be 45/16 (2.8125) rather than 2.8.

[Byrthnoth]

So you ruled out 2.75 then?

[Yugl]

May be old news, but supplying data to show physical WS do floor after applying the fTP term.

This was used to test competing WSC% equations (They didn't work, will supply that post soon). Using existing WSC%, we have . . .

LV99 RNG + Brew + Namas
Equation 1 is our known equation:
Theoretical 679+163 = 842
842*2.75 = 2315.5

2315.5*3 = 6946.5
2315*3 = 6945 (Observed damage)

Spoiler: show

[Yugl]

Eliminating competing WSC% equations (May as well add data for the wiki to show competing ones don't work):

Current: Floor(Alpha(Floor(Attribute%*Attribute + Attribute2%*Attribute2)))
Competing 1: Floor(Alpha*Attribute%*Attribute) + Floor(Alpha*Attribute2%*Attribute2)
Competing 2: Floor((Alpha*Attribute%*Attribute)+(Alpha*Attribut e2%*Attribute2))

Check 1: Brew, Vision Bow, Ruszor, Piercing Arrow
WD+fSTR = 179 (fSTR=36, WD=143)

Current WSC: 347
CPT1 WSC: 348
CPT2 WSC: 347

(347 + 179) = 526
(348+ 179) = 527

526*3 = 1578 (Observed DMG; either Current or CPT2)
527*3 = 1581

Spoiler: show

Check 2:Brew, Yoichi, Ruszor, Namas
WD+fSTR=163 (fSTR=34, WD=129)

Current WSC:679
CPT1 WSC: 679
CPT2 WSC: 678

(679+163) = 842
(678+163) = 841

842*2.75 = 2315.5
Floor(2315.5) = 2315
2315*3 = 6945 (Observed DMG; Current WSC)

841*2.75 = 2312.75
Floor(2312.75) = 2312
2312*3 = 6936

Spoiler: show

[Foldypaws]

So you ruled out 2.75 then?

I... think so? I'd have to go back and look.

[Motenten]

Tachi: Ageha testing

TP Bonus GK, should have no effect on results.

Str: 129 (14 from Hasso)
Chr: 80

dStr of +72 should cap fStr
Figured was safe to test on Bubu mobs

Assuming WSC% is accurate, WSC should be 34

Weapon: 105
fStr: 19
WSC: 34

Total base: 158

3.0 pDif damage at fTP of:
2.8000: 1326
2.8125: 1332

Observed (sorted): 1294, 1295, 1307, 1310, 1314, 1318, 1328, 1329, 1331, 1338

Off from expected. Restarted on starter zone mobs just to be absolutely sure.

1320, 1308, 1305, 1340, 1308, 1308, 1297, 1301, 1329, 1324, 1294, 1282, 1308, 1305, 1336, 1326, 1302, 1305, 1315, 1293

Numbers are still in the same range, so going to combine them.

Complete list (sorted, with more points added after):
1282, 1293, 1293, 1294, 1294, 1295, 1297, 1300, 1301, 1302, 1305, 1305, 1305, 1307, 1308, 1308, 1308, 1308, 1310, 1314, 1315, 1316, 1318, 1320, 1324, 1326, 1328, 1329, 1329, 1331, 1336, 1338, 1340, 1343, 1345

From what I remember, 2-handed weapons almost never seem to go below 3.0 pDif on crit tests like this. That means we can look at this as a full or near-full 3.0 - 3.15 range.

If it's all 3.0+, that puts 3.0 at 1281. However there's a notable gap between 1282 and 1293, which implies a 3.0 at 1293 instead. (Either that, or a completely broken distribution range, if crits can go below 3.0 with the expected rate of about 1/6)

I did get one damage result of 1354, but I had changed a couple things just before that, and am not absolutely sure they were changed back before I got that weaponskill result. If it's valid, though, it fits the max for a 1293 3.0 (max = 1357), though it's suspicious that there's nothing aside from that showing up in the upper 10 point range.

1281 == fTP base damage of 427
1293 == fTP base damage of 431

base damage == 158

fTP for 427: 2.703
fTP for 431: 2.728

If we go for /32 increments, 2 + 23/32 = 2.71875
That would put 3.0 at 1287

If we accept 1302 - 1305 as the 3.0 marker due to standard distribution split (regardless of the lack of values at the upper end of the range), that gives a 2.747 to 2.753 fTP. 1302 as 3.0 would fit exactly with 2.75.



May have to do another test with gorget/belt later to cross-reference values.

[Byrthnoth]

1293 looks like it might be the 3.0 point. If you take 1% intervals above that, you get:
[1293:1305] - 12 values
[1306:1318] - 10 values
[1319:1331] - 7 values
[1332:1344] - 4 values
[1344:1357] - 1 value

However, the reason for a 3.0 frequency spike is generally because points below 3.0 can be elevated above it. If we're at capped pDIF for a 2H weapon, we would predict few points below 3.0 even before the secondary randomizer. Why do we still see a frequency spike?

I think we might just need more data. That set has 51% coverage of the expected range.

[Ophannus]

1) Skewer - 3 hits, 35% STR mod, chance to crit with a polearm. Why was this never as good as Raging Rush?

Also don't forget that WARs have things like Berserk and Warcry while DRG mostly sub SAM and are usually running 150-200 attack less than WARs. WARs also have better base STR, better access to attack/str gear in general. Finally Great Axes tend to have slightly higher base DMG than polearms(lances) .

[Motenten]

Adding more with no gorget:

1314, 1338, 1320, 1305, 1308, 1300, 1293, 1343, 1320, 1315, 1247**, 1330, 1354**, 1299, 1313

1247 would be 2.893 pDif if 1293 is 3.0. 2.893 is within expected bounds (min should be 2.875), though obviously extremely rare.
And I got another instance of 1354, so it's valid.

[1239:1292] - 3 values
[1293:1305] - 16 values
[1306:1318] - 14 values
[1319:1331] - 10 values
[1332:1344] - 6 values
[1344:1357] - 2 values

With 51 weaponskills, I don't see the pattern magically changing anytime soon. The graph looks largely flat, fairly dense, with a couple spikes at either end.

So, doing some with gorget+belt.

1455, 1423, 1408, 1390, 1400, 1399, 1402, 1453, 1401, 1432, 1445, 1433, 1428, 1432, 1443, 1451, 1422, 1403, 1406, 1410, 1390, 1423, 1446, 1409, 1405, 1341**, 1392, 1426, 1423,

This set seems to cover a ~1.05 spread fairly quickly. 1390 * 1.05 = 1459, and max so far is 1455. That puts 3.0 at 1386 or 1389; nothing in the data favors one over the other.


Assuming baseline 3.0 was 1293, and thus base damage of 431:

If belt+gorget 3.0 is 1386, base damage is 462
200/1024 is a gain between 30.01 and 31.99
Base damage has to be between 153.65 and 163.79

If belt+gorget 3.0 is 1389, base damage is 463
200/1024 is a gain between 31.01 and 32.99
Base damage has to be between 158.77 and 168.90

Given that base damage is 158 (assuming 50% Chr mod), that means the 3.0 with belt+gorget should be 1386//462.

2.73 * 158 = 431.34 >> 431
2.9253 * 158 = 462.1974 >> 462

That gives us the final numbers, but the fTP is ugly.


However, supposing my base damage calculation is off. If fTP should be 2.75, base damage would be 157
2.75 * 157 = 431.75 >> 431
2.9453125 * 157 = 462.4140625 >> 462

That also fits the observed values, and is an acceptable value for base fTP.

157
- 105 for weapon = 52 from fStr + WSC


Determine normal melee crit range.
Assuming 19 fStr, that's 124 base damage, and expect 372 - 390.
Coincidentally, the two melee crits that I've landed during testing were 372 and 390. So we can assume that that is correct, and holds to well-established melee damage calculation.

That means 33 points are leftover for WSC, instead of the calculated 34.

80 Chr * 50% * 0.85 alpha = 34

Is alpha not exact? If it's 870/1024 (0.8496), that would drop it below 34, truncating to 33. If it were, 0.85 would still match results the vast majority of the time.

Need to find a test using a weaponskill with solidly known values, and find a stat value that can differentiate those alphas.

[Byrthnoth]

Look up at Yugl's Namas math. I think that shows it is 0.85.

870/1024 would have resulted in 678 base damage from WSC.