Blue Mage: Best thread ever

[Durden]

I just got the orc helm for my blu, what can I do to help with testing?

[Delekii]

There's nothing to test, the numbers needed already exist, it's just a matter of math.
As repairs said, there might be tiny differences but nothing that should make you feel bad for using any one of those pieces in lieu of any other. The only point of contention is pdif - namely how much attack you get from a point of defense, and that isnt really testable to any degree of certainty.

[el_tio]

Tests should be done with AF body + hat since the unknown (unless I missed something) is what the skill tiers are after 307. It can be predicted, sure, but it's better to actually know. Unfortunately I don't have this piece yet, but if anyone on Hades hooks me up with one I'll give you 100 relic hat vs. 100 non-relic hat @ 308, 310 and 312. :D

(still working on those merits >.>)

[Repairs]

Yugl: Would that change depending on how much +STR and +VIT you had to begin with?

The .5 multiplier for CB changes based on your total str and vit. So, the more or less of those stats could impact that function of the formula. (FLOOR .83*(FLOOR(str*.5)+FLOOR(vit*.5)). You have to plug in your own stats to see how much total str/vit would effect the formula.

Also, your total str plays an additional part in the forumal (mob vit to your str). The closer you are to the mobs vit (mob vit 100, your str 100 = 0 bonus), the more benefits/returns you will see from adding str:
*mob vit 100, your str 100 = 0 bonus
*mob vit 100, your str 106 = 6 bonus (6 bonus)

*mob vit 100, your str 120 = 12 bonus
*mob vit 100, your str 126 = 15 bonus (3 bonus)

Str is used in 2 parts of the known formula, hence why str equipment shows better results for higher level mobs.

----------------------

Trying to test consistency of AF2 hat isn't easy and may not be reliable...

Testing 100 is great but from my experience with parsing imps, even 200 attempts doesn't give you very accurate numbers (when we are talking about these very small differences). For example: one day of testing (200 attempts) would show fren rip has an average of 534, while the next day of testing would show an average of 493.

You also have to test against the exact same mob over and over (level correlation, vit/str, def/blu skill) - none of which are easily done.

[Repairs]

Further to CB testing thoughts:

Do we know where the defense works into the formula? I somewhat remember hearing that it was believed to be a multiplier to the existing spell multiplier of .50, and not effecting the pdif portion? For example from my previous post, af2+1 helm had a WSC of 85; then the defense function would multiply that (mob defense 500, my defense 575 = .15 difference -> 85*1.15 = 97 (floored) WSC.

We should be able to test this theory on level1 bunnies (without CA)? and just reverse engineer it?

Edit: Sneak attack simply adds 1 to pdif? So you would know that pdif had to be somewhere between 1-3.15 (if defense has no impact)?

Theorised test: SA cannonball a bunny without coccon, then reverse engineer it to ensure it matches known formula. Then another SA CB with coccon, then reverse engineer it to see which portion of the formula is out of synch? Hmmm, guess it would have to max damage comparisons, not just any CB (i.e. 3.15 pdif)?

[Yugl]

Doesn't VIT also have a dual role since Cannonball is based on DEF? In addition, based on the formula, would it be better to balance the STR and VIT (50/50 or some other ratio) or pack more STR?

Currently it seems the best Cannonball build is:
Wing+1 (Or Jag+1)
Arch+1
Bibiki Shell/Chakram+2 *
Gnad/Relic Helm/Wivre+1/Dusk(NQ/+1) *
Fortitude/Justice Torque *
Triumph (+1)/Cassie/Heims *
AF+1 Body
Alky Bracelets
Flame Ring (2x)
Smilodon Mantle
Warwolf Belt
Enkidu Legs
Rutter Boots

*Trying to decide which of those equips to use

[Repairs]

Doesn't VIT also have a dual role since Cannonball is based on DEF? In addition, based on the formula, would it be better to balance the STR and VIT (50/50 or some other ratio) or pack more STR?

Currently it seems the best Cannonball build is:
Wing+1 (Or Jag+1)
Arch+1
Bibiki Shell/Chakram+2 *
Gnad/Relic Helm/Wivre+1/Dusk(NQ/+1) *
Fortitude/Justice Torque *
Triumph (+1)/Cassie/Heims *
AF+1 Body
Alky Bracelets
Flame Ring (2x)
Smilodon Mantle
Warwolf Belt
Enkidu Legs
Rutter Boots

*Trying to decide which of those equips to use

I don't know how/where the defense function works, I just know it makes it stronger so it's hard to make decisions. The other part I don't understand is how blu skill = attack/pdif. Until those are known, it's a toss up. The current formula has no balancing, so you are not looking for 100 str and 100 vit for max performance - in theory 150 str and 80 vit or 80 str 150 vit would be better.

Personally I use all str for HNM, just because we know the benefits (Gnad, triumphs). Also double defense songs is standard so I'm regularly 800+ defense. Adding 10 defense to 800+ is adding like 1%, so it may or maynot have more benefits then more str.

[Repairs]

Further to CB testing thoughts:

Do we know where the defense works into the formula? I somewhat remember hearing that it was believed to be a multiplier to the existing spell multiplier of .50, and not effecting the pdif portion? For example from my previous post, af2+1 helm had a WSC of 85; then the defense function would multiply that (mob defense 500, my defense 575 = .15 difference -> 85*1.15 = 97 (floored) WSC.

We should be able to test this theory on level1 bunnies (without CA)? and just reverse engineer it?

Edit: Sneak attack simply adds 1 to pdif? So you would know that pdif had to be somewhere between 1-3.15 (if defense has no impact)?

Theorised test: SA cannonball a bunny without coccon, then reverse engineer it to ensure it matches known formula. Then another SA CB with coccon, then reverse engineer it to see which portion of the formula is out of synch? Hmmm, guess it would have to max damage comparisons, not just any CB (i.e. 3.15 pdif)?

Tests:
1) Bunny has 10def? Then try to evenly match my def to 210.
2) SA CB 100+ times to find max damage (i.e. 3.15 pdif)
3) Reverse engineer to find how much WSC was increased (base)
4) SA CB with an additional 100 defense (ex. 310), for 100+ times to find max damage (i.e. 3.15 pdif)
5) Reverse engineer to find how much WSC was increased (base +100)
6) SA CB with an additional 100 defense (ex. 410), for 100+ times to find max damage (i.e. 3.15 pdif)
7) Reverse engineer to find how much WSC was increased (base +200)
8 Compare/analyze the three tests

This should work?

[Valisk]

Fortitude/Justice Torque *

Look into an Auditory Torque (DEF: 5 Latent: STR +4 VIT +4) also. Latent is in effect w/ a song or roll active. Drops from Boroka. Even if someone is monopolizing Boroka on your server, the torque probably gets trashed and they might let you have free of charge.

[Yugl]

Generally I don't have BRD songs on me when using SATACA Cannonball, but you are right, Auditory Torque is the best piece for Cannonball when the latent is active.

Repairs, it might be difficult to do that since you would have to make sure that only your DEF (Not STR or VIT included) is changing. I'm not sure how much DEF each Minne gives, but it seems you need a BRD to conduct that test.

[kenshyn]

You also have to test against the exact same mob over and over (level correlation, vit/str, def/blu skill) - none of which are easily done.

Three ways:

Test vs a paladin in Diorama - Ghelsba Outpost easily can determine def and changes.

Test with a 75 Bst vs a ep/dc crab use charm and leave to keep same mob heck bring pet food for healing they are cheap. Just keep the mob in its spawn area and you can use the same subject over and over.

If 35 Bst just sub bst and find a mob that you will not one shot and see above just slower since no pet food.

[Force-Attuned Krogoth]

A couple weeks ago I got curious about cannonball, myself. It's always seemed to me that without proper def, it's got way less atk than other spells. So I went out and performed a little test. I'm lazy, so it was qualitative and no screenshots, but I got a sense of how it works. Bunnies didn't work because I had 150 def naked, so I found the average CB number (450) and moved on. In Yuhtunga, I poked Goblin Furriers (lv 32-37). 400 atk should be near or above pDIF cap, unless I'm wildly off the mark. With 150 def and 284 blu skill, unbuffed CB was doing 250. Switching in gear with just def, and adding cocoon, I found that CB approached the 400s when my def was about 400-450. From this, it seems to me that blu skill does not work for Cannonball's attack like it does other spells. My hunch is it only affects D, and def alone determines atk.

[Repairs]

A couple weeks ago I got curious about cannonball, myself. It's always seemed to me that without proper def, it's got way less atk than other spells. So I went out and performed a little test. I'm lazy, so it was qualitative and no screenshots, but I got a sense of how it works. Bunnies didn't work because I had 150 def naked, so I found the average CB number (450) and moved on. In Yuhtunga, I poked Goblin Furriers (lv 32-37). 400 atk should be near or above pDIF cap, unless I'm wildly off the mark. With 150 def and 284 blu skill, unbuffed CB was doing 250. Switching in gear with just def, and adding cocoon, I found that CB approached the 400s when my def was about 400-450. From this, it seems to me that blu skill does not work for Cannonball's attack like it does other spells. My hunch is it only affects D, and def alone determines atk.

How many tests on the bunnies did you do? Ruff estimates? There was a previous poster who mentioned they did see a difference. The current theory of defense effecting the multipliers means that increasing your defense on level 1 bunnies would make a difference.

Anyone know how much def bunnies have?

[Alistaire]

Generally I don't have BRD songs on me when using SATACA Cannonball, but you are right, Auditory Torque is the best piece for Cannonball when the latent is active.

Repairs, it might be difficult to do that since you would have to make sure that only your DEF (Not STR or VIT included) is changing. I'm not sure how much DEF each Minne gives, but it seems you need a BRD to conduct that test.

COR rolls work too btw.

[Force-Attuned Krogoth]

Well, I can tell you that if defense affects multipliers, the difference between 150 def and 600 should be noticeable. If it doesn't show up in five trials, it's got to be small, especially compared to the effect of def on uncapped pDIF. I admit I didn't perform a rigorous test, but I'm convinced that, if skill affects atk at all, it's far less than on other spells.

In summary: Def is the vast majority (if not entirety) of the hidden "attack" stat on cannonball, and does not have any other discernable effect.

[Repairs]

SA Cannonball bunnies

Skill-292, str 66, vit 63

D = 67
Fstr = 22?
WSC = (floor(str*.5)+floor(vit*.5)) = 64; *83 = 53
Spell Multi 1.75
Total 248 – (67+22+53)*1.75

Total * PDIF (3.15) = 781 expected result

Actual result after 235 tests:
743 - max

I was thinking that my actual numbers would be higher then expected results, since the formula didn’t include any defense mods.

Is my math correct? I’m somewhat thinking fstr might be wrong? Maybe fstr of 15 (since it works with my numbers)?

[Belgaer]

Does anyone know what range Pinecone Bomb is most potent at? Thinking back, I remember doing just under 1k DMG on too weak mobs with it before at range, without using /equip swaps. But now, testing on EP mobs, melee range seems to be most potent (with STR gear)(but not by any dramatic numbers, such as going from 400 to 900 DMG as I remember...).

[Yugl]

I don't remember that spell ever doing that much DMG. I think a thread would have been dedicated to it at some point if it had.

[Karshe]

Haven't used the spell in a long time but it definitely used to do an abnormal amount of damage on too weak mobs (~900 sounds about right). Haven't tested it in a long time though, maybe it was fixed some how.

[LD]

fSTR cap is 44 on ranged spell. Entering the usual crap into the formula, there's nothing abnormal about ~1k on something with practically no def.