Final Form Magian Weapons

**ccl** · 2010-09-22 17:46

I can't test now cause i'm in abyssea, but tp def matter, when I was trying early I did over 3k at 300 tp, tp seems to make a difference just from eyeballing on abyssea NMs too.

**FailureMidgard** · 2010-09-22 18:45

Originally Posted by Byrthnoth

Other interesting tests:

1) Its description does not imply it has any TP mods. Do 300TP Tachi: Fudos do the same damage as 100TP Tachi: Fudos? It should be obvious and take about one WS to test.

Huh? Description: Deals double damage. Damage varies with TP. Masamune: Aftermath.

**Motenten** · 2010-09-22 18:52

While I'd considered addressing the other points, that'd just add noise at this point. My primary objection to your estimate is this: that you *were* including the (obviously) <3.0 value as part of an average in order to measure the difference between sets, when doing so clearly skews the average downward, increasing the difference between sets, and pushing fTP towards the low end (ie: 1.8-1.9).

And I'll agree that the actual value can't be determined without additional data points.

**FailureMidgard** · 2010-09-22 19:08

Originally Posted by Motenten

While I'd considered addressing the other points, that'd just add noise at this point. My primary objection to your estimate is this: that you *were* including the (obviously) <3.0 value as part of an average in order to measure the difference between sets, when doing so clearly skews the average downward, increasing the difference between sets, and pushing fTP towards the low end (ie: 1.8-1.9).

And I'll agree that the actual value can't be determined without additional data points.

Well, I dunno, from my own testing back when the relic augments first came out, and from the data I've seen here, I've seen absolutely nothing to support, and everything to disagree with the idea that "it's very rare to see numbers under 3.0" yet Byrth quotes it, as do other people I generally trust. I guess my point is that as far as ftp is concerned, I'm ok with using that average. I agree entirely my calculations for the STR mod were nowhere near accurate, and just a guess, but I don't see any legitimate proof or reasoning to assume that the numbers showing up are "most likely 3.0+" and that the high numbers there aren't 3.0 with the outlier being even lower. Use the average or call it inconclusive and drop it. Dropping an outlier out of 3 data points (that's off by less than 3%) and then including points that could easily be off by 3+% isn't a very good way to go about it.

Anyway, should just sit and wait for better data, we all agree, I was just annoyed because it felt way too much like people trying to bend the data to fit the # they wanted, which is a pretty bad thing to do in my mind, even if it does end up correct in the end.

**Byrthnoth** · 2010-09-22 19:21

Wow, somehow I glazed over when I was reading that and didn't see the "Damage Varies with TP." I seriously think they may make Relics OAT now. As it stands, total Amano killer. The only way that Amano could ever maybe come out on top is if you could 4-hit Amano and not 4-hit this weapon... or if they made Amano OAT.

Also, I know I shouldn't have written "2.0" but I meant "2", as in, one significant figure and approximately 2. I was just too used to seeing fTPs listed to at least one decimal place.

**FailureMidgard** · 2010-09-22 19:38

Originally Posted by Byrthnoth

Wow, somehow I glazed over when I was reading that and didn't see the "Damage Varies with TP." I seriously think they may make Relics OAT now. As it stands, total Amano killer. The only way that Amano could ever maybe come out on top is if you could 4-hit Amano and not 4-hit this weapon... or if they made Amano OAT.

Also, I know I shouldn't have written "2.0" but I meant "2", as in, one significant figure and approximately 2. I was just too used to seeing fTPs listed to at least one decimal place.

2 as opposed to 2.0 makes me a lot more comfortable with it, but at the same time is saying 1.5-2.49, so not really a tight range or anything.

And ya, as I said, the emp-1 according to some really fast and not detailed math is around equal to a full upgraded amano in an easily accuracy capped situation. And really, a few months ago I would have been ok with Amano being replaced by Emp. weapons, but I'm really pissed off they put me through killing 8000 mobs, only to make it inferior afterwards.

**Motenten** · 2010-09-22 22:05

I've seen absolutely nothing to support, and everything to disagree with the idea that "it's very rare to see numbers under 3.0" yet Byrth quotes it, as do other people I generally trust.

Lemme see.. Yes, I still have them on imageshack. Here's a couple graphs of crit frequencies from a test I did a while back. The sudden vertical increase is when it crosses 3.0. First one is with a Behemoth Knife, the second one is from a BK+1. Both on savannah dhalmels, capped fstr, no piercing bonus, etc.

http://img51.imageshack.us/i/bkcrits.png/
http://img193.imageshack.us/i/bk1crits.png/

While "very rare" is an overstatement, I would still say the majority of (capped) crits should fall in the 3.0-3.15 range.

**ccl** · 2010-09-22 22:46

I did some more test naked and with snow gorget, didn't had time to do with gear, will probaly be able to do some more this weekend if needed.

Naked

Spoiler: show

with snow gorget.

Spoiler: show

Also fought Karkinos early, I only did 3-4 gekko and 3-4 fudo, no song or roll or atma, gekko seemed to do more, so it seems it either doesn't have an attack bonus, or just a weaker one than gekko, but just did a few ws so not sure on that yet.

**Byrthnoth** · 2010-09-23 00:48

Okay, we should probably wait for more data points. Either something is weird, it isn't adding up the way we think, or they aren't "doubling damage" the way we assumed. Right now, if you calculate it from that you get a base fTP of 3.0. Now! 100TP Gekko is 1.5. Maybe when they say "double damage" they mean 1.5*2 = 3.0 pDIF and there's no extra term. Then adding the gorget bumps it up to 3.1 and everything makes sense.

This would give the weapon ~191 base damage during WS, or ~69 from mods. Considering he has 83 STR naked and assuming it's purely STR mods, that would mean it's approximately a 100% STR mod (with Alpha at .85). This seems unlikely. When 65 STR were added (55 damage) you would predict 2402 damage, but he got under 2200. So it's possible that it isn't purely a STR mod. It could be STR50-60% and something-else ~40-50% or so though.

**Drystyn** · 2010-09-23 01:28

Sorry to distract from the math, but...

Roark - Quetz has a Maschu+1 (the WoE GA) if anyone on the server wants to bug him with Ukko's Fury questions. (And he'll probably be annoyed that I directed people to him, but whatever.)

**Motenten** · 2010-09-23 01:56

Without gorget:

1790 max at 100 TP, 1803 max at 104 TP. Increase of 0.7%. A growth of 0.25 fTP over 100>200 range would give a theoretical 0.5% to 1.0% gain (assuming starting between 1.0 and 2.0) over 4 TP. Large enough to treat them separately here.

Min value at 100 TP is 1711. Ratio from 1711 to 1790 is 1.046, so very good chances of getting full 3.0>3.15 range coverage, and supporting 1790 as max or near max. 1711 supports up to 1796 as max.

@104 TP: 1662, 1665 and 1729 as the lower results, 1803 as highest. Ratio of 1803 to 1729 is 1.043. Ratio against 1662 is 1.085, so the 1660's should be considered outliers. 1803 a very high probability of max damage, but 1729 could support up to 1815 as a max.

With gorget:

Max at:
100 TP: 1837
104 TP: 1863

Min at:
100 TP: 1701, 1745, 1746, 1747
104 TP: 1747, 1772, 1775, 1779x2

For 104 TP: 1747 is well over the max ratio, making it an outlier. 1772 is slightly over the 1.05 ratio (1.051). That indicates that 1772 is below 3.0, though probably very close. There are also two instances of 1779 and one of 1775 for 104 TP; if 1775 was 3.0, that would put the max value at 1863 (matches observed max). We can probably consider 1863 to be the max.

For 100 TP: Even if 1747 were 3.0, the max would only reach 1835. 1745 can only support perhaps 1833. We must therefore consider these outliers. The lowest value other than those is 1763, which supports a max of 1851.

@100 TP: Max of 1837-1851 with gorget, 1790-1796 without. 41-61 differential.
@104 TP: Max of 1863 with gorget, 1803-1815 without. 48-60 differential.

The implied base fTP would be:

@100 TP: 2.9 - 4.3
@104 TP: 3.0 - 3.7

One more comparison, using the average values of weaponskills above the apparent 3.0 mark:

@100 TP: 1755 - 1797 = +42 differential, 4.1 fTP
@104 TP: 1767 - 1816 = +49 differential, 3.6 fTP
[0.7% increase on average non-gorget, 1.1% increase in average with gorget]

Combining all ranges, 3.6-3.7 fTP seems to be most probable.

Max fStr for a d103 weapon would be 19. 83 base str should cap fStr on lvl 0 mobs. Total base damage: 122

@100 TP without gorget, max crit was 1790-1796 (the narrowest range that I'm fairly confident of). Divide by 3.15 puts it at 568.25 to 570.15.

If fTP is indeed ~3.6, divide that out to get 157.8 - 158.4.

158 - 122 = 36 base damage from WSC. Factoring out alpha, that puts the WSC multiplier at about 50%.

However all that leaves the "doubling damage" unaccounted for.

Given the scale of things, most likely the "does double damage" gets applied at the fTP stage, doubling the fTP modifier (before gorget; if 0.2 was used in the above checks, fTP would be upwards near 8.0). That would mean 'base' fTP would be around 1.8-1.9.

If that's the case, then we could try a forward estimate with the +str build.

103+19 = 122 base
83+65 = 148 * 50% * 0.85 = 62 WSC
Total base = 184

184 * (1.8 * 2 + 0.1) * 3.15 =
184 * 3.7 * 3.15 =
2144
(slightly low)

or

184 * (1.9 * 2 + 0.1) * 3.15 =
184 * 3.9 * 3.15 =
2260
(a bit high)

or

184 * (1.85 * 2 + 0.1) * 3.15 =
184 * 3.8 * 3.15 =
2202
(reasonably close)

So ~1.85 fTP & 50% str would be my new guess, rather closer to Failure's estimate than the rough 2.0 starting estimate.

**FailureMidgard** · 2010-09-23 03:45

Originally Posted by Motenten

Lemme see.. Yes, I still have them on imageshack. Here's a couple graphs of crit frequencies from a test I did a while back. The sudden vertical increase is when it crosses 3.0. First one is with a Behemoth Knife, the second one is from a BK+1. Both on savannah dhalmels, capped fstr, no piercing bonus, etc.

http://img51.imageshack.us/i/bkcrits.png/
http://img193.imageshack.us/i/bk1crits.png/

While "very rare" is an overstatement, I would still say the majority of (capped) crits should fall in the 3.0-3.15 range.

Looking at the first graph only, 268 over crit cap, 173 under crit cap (could be off by 1 or 2, close enough, to lazy to go through 2nd graph). Using that as a guideline, 60% chance for any given value to be over 3.0? Really not convincing me that with 4 datapoints that the 3 high ones must be over 3.0. Especially considering that the difference between the low and 2.9 pdif there is a larger difference than 2.9 to 3.15. So ya, you've just convinced me more that my observations were correct, and while the majority may be over 3.0 and the frequency of any given point over 3.0 is significantly higher, the actual distribution is nowhere near as skewed as Byrth was trying to imply (aka - if there is a huge frequency spike, you can assume it's over 3.0 but the actual chance of a number being over 3.0 given a very small # of datapoints is barely higher than 50/50). And you need a much larger margin than what you were originally looking for to safely conclude it's over 3.0.

Originally Posted by ccl

I did some more test naked and with snow gorget, didn't had time to do with gear, will probaly be able to do some more this weekend if needed.

Naked

Spoiler: show

with snow gorget.

Spoiler: show

Also fought Karkinos early, I only did 3-4 gekko and 3-4 fudo, no song or roll or atma, gekko seemed to do more, so it seems it either doesn't have an attack bonus, or just a weaker one than gekko, but just did a few ws so not sure on that yet.

No Gorget, 100TP: 1755 average.
No Gorget, 104TP, dropping outliers: 1768

Gorget, 100TP, dropping outlier: 1784
Gorget, 104TP: 1812

1755/n = 1784/(n+.1) ~6.0 ftp

1768/n = 1812/(n+.1) ~4.0 ftp

The averages aren't really making any sense, but assuming Motenten is close on the maximums, the numbers there do seem to fit all the data quite nicely (just must not have gotten a good sample to work with averages).

In any case, Motenten's estimate this time looks correct to me, so I'd go with it for now. If you get more time and want to test Ccl, it'd be cool if we could get some data points @ 200 and 300 TP also, to get an idea of how that scales (and ideally 10-15 at each, not 3-4, heh).

But, I'd like to point out that stuff like "The max possible value for 3.15 is 1854." as opposed to "Max at: 104 TP: 1863" is exactly why before I was saying you couldn't just toss data points you didn't like with 3 to work with. Data points you were assuming were at 3.0 were actually closer to 2.85-2.9, and while ftp2.0 was a point that would fit the data, it wasn't the best fit and was found via skewing the data. Anyway, I think your calculations now are much more conclusive and appear to be mostly accurate.

Seems to destroy Kaiten. Wonder what the AM procrate is...

**Myrrh_Quetz** · 2010-09-23 05:51

http://i29.photobucket.com/albums/c2...inus6staff.png

:D

At 85 my Garuda costs me 0 mp and that's without using Avatar's Favor. I'm not sure I want to do another Avatar staff though. Avatar killshot trials are so gay.

**Fereydoon** · 2010-09-23 10:56

At least you can team up now!

**TehKerm** · 2010-09-23 13:07

Anyone got round to testing out the Rudra's storm on THF at all?

**Byrthnoth** · 2010-09-23 13:25

Haven't talked to him, not on Quetz, etc, but:
http://www.ffxiah.com/screenshots/49148

Properties look like Light/Frag (unfortunate that we can't self-light, but convenient for smart DRGs and SAMs). He claims that the WS has a 300-400 damage on KJ and randomly deals double damage.

............. . . . . . . . . . . . lol

**Nameless** · 2010-09-23 13:33

Doesn't look like anything special to be honest as I expected, just from looking at the damage compare to Kasha. G/K/Y were never the high end damage WS, they were just extremely consistent regardless of the mobs. It looks in par with RR and KJ, I don't know what he was talking about 300-400 damage on KJ, I doublt his KJ is only 1100 on a mob like that when Kasha is already doing 1100.

**Massimo** · 2010-09-23 13:37

Originally Posted by Byrthnoth

Haven't talked to him, not on Quetz, etc, but:
http://www.ffxiah.com/screenshots/49148

Properties look like Light/Frag (unfortunate that we can't self-light, but convenient for smart DRGs and SAMs). He claims that the WS has a 300-400 damage on KJ and randomly deals double damage.

............. . . . . . . . . . . . lol

Friend was doing Nyzul with him the other day, and reported his weaponskills doing pretty shit damage. Maybe he's just gimp, I don't know, but it disappointed me quite a bit.

**Gadritan** · 2010-09-23 13:45

Well, as he says, he's "missing stuff." http://www.ffxiah.com/player/Quetzal...oark#item-sets

Full STR/DA/WC/GA merits. Nothing on zerk at all. He has some decent gear with some meh choices for a lot of stuff, so I wouldn't put too much stock into his performance with it.

**Byrthnoth** · 2010-09-23 13:49

We need to know his buffs before we can guess how good it is, but even assuming it's only as good as Raging Rush and the Aftermath has a shitty (10%) proc rate the Empyreal GA is very competitive with Bravura. Bravura has the advantage of being much easier to 6-hit than the Empyreal, but that could change at any point if they add more STP gear. Regardless, I consider anything that involves 50 Glavoid or Chloris items unobtainable.

If the WS is actually the ~20% better than KJ that he's claiming, Maschu+1 is probably the best non-relic/magian/empyreal.

Edit:
If his WSs were actually doing shit, then those 2 we see in the screenshot were probably Mighty Strikes'd and aren't very impressive.

Looking at his sets, he is missing some stuff (Pole grip, Forager's Mantle are by far the easiest improvements), but his performance with that WS should let us know about how good it is. The last 10 STR he's missing isn't going to overturn the WS hierarchy.

Thread: Final Form Magian Weapons

Thread Tools

Similar Threads

Final Magian/Empyreon Armor.

Good Camps for Magian Weapons

Final weapon skill quests

relic weapons 1-4 form help

Anyone trying for their Relic Weapon?

Latest Threads

Up & Coming Threads

Hottest Threads