Prime WS Information, Testing, Discussion

Imo you can make due with less. If I was going to add samples here, I would add them to your 1k No Fotia base case. What kind of kills you here is that it's your least precise sample (only 86% coverage).

Confirming that With Fotia is gorget only?

Correct. This was my weaponskill set for the first line of testing to determine the tp scaling. I did all the tests as Monk/Warrior. So no WSD trait of any sort from any source.

ItemSet 402029

I did a series with this, and a series with fotia gorget added for each tp value. Fotia gorget was the only change from the first and second sets at each tp value. I had 296 strength and 412 dex in this set.

Then for the strength testing I added accessories that raised my strength but didn't raise my dex. JSE cape with 30 strength and DA rate, gere ring, cornelia's belt, schere earring, and odnowa earring +1, and republican platinum medal (I'm currently a citizen of windurst so the regain is inactive and won't affect data).

I have fewer dex accessories but still enough to get the job done. I used a JSE cape for 30 dex and no other mods, and threw in ilabrat and odr earring to raise my dex for that test.

I can continue gathering data for the sets with lower variance. I'll try to get everything to at least 1.047 spread. There are a few that are higher. So some of the data points should be fine. I'll update with the new values when I have something for you guys to review.

4.95+ would be ideal.
Ofc that doesnt need to be done for every threshold or wsc change, but at least for 1000TP with and without Fotia and maybe for 1000TP no fotia with some DEX or STR change.

That being said it is something around those values (extremely close)

WSC 40% STR/DEX
1000 3.136
2000 7.385
3000 11.67

EDIT: I've removed the chart I had here a few days ago to post a more updated revised version on page 81 of this thread. I'm leaving the rest of this post here along with the following discussion for reference.

That correlates with the data from a japaneese test done here https://x.com/za_ibd/status/1946560930952904805

That test reports

STR/DEX 44%
fTP Values:
1000 TP-778/256 - 3.04
2000 TP-1812/256 - 7.08
3000 TP-2827/256 - 11.04

I'll try to get a bigger spread on the values. The 1k is being far more stubborn than some of the others. I swear everything wants to land in the 7900 to 8200 range, which isn't a big enough spread. I'll keep playing around until I get something better.

My biggest takeaway from all this is that we completely overshot the WSC stat mods on the wiki. Whether the str/dex mod is 40 or 45, that's a pretty big difference from the wiki's reported 60% str/60% dex. The tp scaling is close on all counts between the wiki, the japaneese test I linked, and my own data so far. There's some room for variance but we're close. The secondary stat mods though are quite a ways off from what the wiki has.

I'll keep working with it. No ETA on when I'll report back. I'll try to get as good of a sample as I can. This could take quite a while though.

If it's 44% it's for sure not fTP he listed

fTP that fits for 44% and numbers gathered by you are:

2.992
7.05
11.155

I would lean more toward what you initially suggested Simon. I think it's far more likely the strength and dex mods are 40%. I think there's a very good chance that the conclusion the japaneese test came to is flawed, just as whatever tests were done that led to the wiki stating a 60% str and 60% dex mod. I'll keep working at it. But that part of our previous understanding is clearly wrong.

I did 50 more weaponskills at the 1000 tp mark with no fotia last night and managed to come up with 8244 as a high and 7868 as a low, which is a lot closer to what we want but still only 0.0477 difference. I'll keep playing with that one for a while and then work on some of the others too. Hopefully I can get samples at 4.9% at least for everything. I'll make a side project out of it for a while.

I suspect the values may end up being

40% Str/40% dex
1000 3.0
2000 7.5
3000 11.5

That's the direction I think we're heading toward. Hopefully a better sample will be able to confirm that.

It cant be that far from the values I posted earlier for 40%WSC. Combined spread from all of you numbers was enough to pin point it pretty accurately for specific WSC%.

I can't do math anymore. I'm forgetting some trait / factor (perhaps native MNK WSD or something?)

3 + FLOOR(0.11*(487+277)) + 213 + FLOOR( (3+213)/9 + 8) = 331 base damage
FLOOR(296*.4) + FLOOR(412*.4) = 282 WSC with 40% STR/DEX
613 damage * 3.8 pDIF cap for 99 MNK with H2H * fTP * randomizer = weapon skill damage

For 1k TP and 1.0 randomizer we know that WS damage must be between 7852 (CEIL 8244/1.05) and 7868, but that would imply an fTP of ~3.375 with 40% mods.

What +12.5% factor am I forgetting?

Monk has no native WSD, and all testing was done Monk/War. Did you account for Hand to Hand skill affecting base damage? I posted that in my screenshot. My H2H skill is 482. I know I forgot to account for it in a post I made last week so maybe that's where you're off?

Edit: Oh I bet you forgot damage limit II. Monk doesn't have WSD trait, but it DOES have damage limit 2 trait. Monk natively has some PDL.

https://www.bg-wiki.com/ffxi/Damage_Limit%2B

Hmm.. I was using bgwiki's 3.8 approximation instead of 3.5 + 77/256, but that's basically the same thing. It must be something else.

You are forgetting 8% pdl from AM1 on prime 5 (and 10% at 2000 and 12% at 3000)

That'd do it. The prime aftermath kicks in before the weaponskill goes off. It's impossible to not have it present in this testing. I unequip varga and re equip it after every weaponskill to zero out my TP for the next ask sash/boost cycle so every 1k, 2k, and 3k will have the appropriate corresponding AM effect.

https://www.ffxiah.com/forum/topic/57318/prime-ws-information-testing-discussion/35/#3680658

^ This has it at +6%~12% based on TP for stage 5, but it's also weird that we treat job trait as additive with the cap and PDL buff as multiplicative. Is that true? If so, does PDL buff affect the job trait as well? If it was additive, +6% would instead be +0.2

With *1.06 stuck onto to the pDIF cap, now I get 3.180 ~ 3.186 with those mods. If it was actually *1.08, it would be 3.032~3.038.

AFAIK, yes. Sam for example gets Dmg limit+ II at +51/256 (~0.2) per wiki. If it were like Aria or the prime weapons aftermath, that's 20% PDL, which is frigging huge.

But you don't get the kind of dmg increase subbing Sam or drk on a job without a native dmg limit+ trait that you get from prime aftermath or Aria.

PDL trait and gear is completely different thing. It was discussed and repeated countless times. PDL in gear and Aria is multiplicative % effect. Trait is just +0.1 per level. Under pdif page on bgwiki, you have all the caps listed for each type of the weapon for each job. This is based on pdl trait on the jobs and individual weapon types.

OK, interesting. I didn't see that on the pdif page anywhere, so I have added it.

Technically the data supports a pretty wide range of possibilities still. The WSC could be between 219 and 330 based on the gorget tests, which is a wide mod range. Two things would help precision:
1. Use both gorget and belt, not just gorget. If you're swapping belts anyway to avoid Ask's 5% WSD, may as well put on Fotia!
2. Get one TP waypoint of with/without Fotias that is really precise, because it will allow the calculation of your base damage.

The simplest version I've found that fits most of the data is 3/7/11 with 45%/45%. It doesn't seem to be worth going more granular than that with my (low) level of certainty about the order the flooring steps are applied in + uncertainty about the magnitude of modifier, because you can kind of make anything from 25% to 45% fit at this point.


Predicted should be between Low and MinLow, where MinLow is High/1.05

If you try to fit into just 1000TP data then yes. If you try to fit into ALL the data he provided. It's much more narrow.

The fotia items with wsd+10 are actually ftp+0.1.

Fotia is + 25/256 fTP not pdif :)

The fotia gorget should be all we need to determine the fTP scaling values if I get proper range spreads. I was able to get the 1k with no fotia last night to slightly over 1.049% which is looking a lot better, but I must've done a few hundred weaponskills to get there. I didn't get to start the 1k tp with fotia yet, and there are quite a few other values to broaden, and the 2 and 3k will require twice as many boost cycles to farm up than the 1k, so I'll be at it for a while. I can work on it in my leisure while I'm doing housework or w/e so it's not really a big deal to do, but it's going to take a lot longer than I initially thought to get more precise values across the whole spectrum.

With that said, is the 5% multiplier randomized where each value has an equal chance to show up, or is it on a bell curve where the outer values show up with much lower frequency than the ones in the center? I saw the 8244 show up twice now, but the lowest I've seen is 7858. Almost everything is in the 7900-8200 range and whenever I see one of those I just ignore it and repeat the process. I do occasionally see numbers more toward the higher and lower ends, but nothing beyond a 1.0491% spread so far so there have to be a few values I haven't seen yet on either end. I think for starters I'll try to get everything to 1.049 or better and go from there, so I'll move on to the 1k with fotia tonight.

Personal note by the way; while the scaling is obviously lower than the wiki I'm really starting to enjoy maru kala in limbus. When I have full buffs even when berserk is down my attack is still high enough for maru kala to be competitive. I initially thought the best way to play it was to swap between varga when impetus is up and spam smites then swap to godhands for the impetus down phase, but I'm finding that maru kala keeps up plenty well even when impetus is down. It's proving to be a very satisfying weapon to use. The self gravitation to darkness skillchain basically offsets any dps loss I would have had otherwise. It's quite a formidable weapon.

Fotia is ftp +0.1 not pdif

You can use solver to find (x+0.1)/x = minfotia/min
This gives you an ftp value that ignores stats, pdif, fstr, wpn dmg, wsd, pdl only dependent on keeping those things constant and randomizer (which you are eliminating with sample spread)

This is technically true, but ultimately the base damage is something like:
DHigh = (MinLow with - Low without)/(25/256)
DLow = (MinLow without - Low with)/(25/256)

So, if you are using 50/256 instead of 25/256, you have twice the effect size (for the delta) and denominator. however, because you multipled the numerator and denominator by 2, rounding becomes less significant.

Do you think it would be more efficient to scrap the numbers I collected and start over with 2 fotia pieces?

Surely not! However, if you're going to do a hundred 1k TP WSs with Fotia, I'd use both so we get maximum resolution. You're likely to exceed or match both your extrema for the 1k set with that sample size anyway, so you may as well have it with better resolution.

Good point. I haven't done a whole lot with the 1k set and that'll be the easiest to get 2 fotia piece values for. I can do the 1k with fotia spread with 2 fotia pieces

I mispoke.

Do stage 5 have an occasional increase in damage during the weaponskill? I’ll average 74k zesho meppo and sometimes it’ll spike up to 81k+ around every 10 weaponskills. Is anyone else noticing that or can explain it.

I updated my post with more data. I'll keep going, just had a break from collecting for a bit

da or ta proc maybe?

ma proc, different TP values when firing it off, PDL AM not being up if its being utilized, change in buffs/debuffs etc