Enmity testing - Still Going!

[Kaeko]

Ok, so, correcting earlier formulas:

96/86 = (x + 25/256) / x
96x = 86x + (86*25)/256
10x=(86+25)/256
x=(86*25)/(256*10)
x=(43*5)/256 = 215/256 (~.8398 ) @ 100tp

At 200 TP, we're getting 105 damage, 105/86*215=262.5, so either 262 or 263?
At 300 TP, we're getting 192 damage, 192/86*215=480/256.

You're not accounting for the chance that the 96 and 86 damages are being floored. If you consider this, then "96" really means anything from 96 to 97 (exclusive) and "86" means anything from 86 to 87 (exclusive). If you consider this, than the possible range of "X" is quite large and not limited to only 215.

EDIT:

Based on these ranges, X can be anything from 196/256 to 242/256

[FailureMidgard]

You're not accounting for the chance that the 96 and 86 damages are being floored. If you consider this, then "96" really means anything from 96 to 97 (exclusive) and "86" means anything from 86 to 87 (exclusive). If you consider this, than the possible range of "X" is quite large and not limited to only 215.

EDIT:

Based on these ranges, X can be anything from 196/256 to 242/256

I agree there is a range, your range is too large, or there is something wrong with my math, however.

Repeating the math for my last post, using just one of your assumed numbers.

Given: 100tp>86 dmg.
Assumed: x=196
86 = floor(D*x/256)
86 = floor(D * 196/256)
If 86=D*196/256, then D=112.32
If D=112, then 112*196/256=85.75, floors to 85. Wrong.
If D=113, then 113*196/256=86.51, floors to 86. Possible.
If D=114, then 114*196/256=87.28, floors to 87. Wrong.
D=113

Given: 100tp+gorget>96 dmg.
Assumed: Gorget makes new x value (call it x') equal to old x + 25. x' = x + 25, or x' = 221.
Math:
96 = floor(D*221/256) = floor(113*221/256) = floor(97.55) = 97
196 is definitely not correct.

If you see a flaw with this methodology, let me know. I don't really feel like repeating it for your entire range. I'm going to try to test with a different # of dispels relatively prime to 3 later, to try to narrow it down.

Anyway, I do agree 215 may not be exactly correct, but I think the range is lower than you are saying.

[Kaeko]

OK this is my personal idea on how this works. This is contrary to the numbers Midgard posted so I'll try to explain my own reasoning and how the raw data fits and supports it.

My personal belief is that the modifier for CE/VE will be a "nice" number. The one I am working with is 0.1 (10%). Given this modifier, I came up with these 3 fTP values:

CE Modifier = 0.1
F100 = 231/256
F200 = 282/256
F300 = 512/256

(Excludes VE from formula, and does not include cap situations)

Damage = floor(floor(CE*0.1)*fTP)

************************

All Raw Data compared to this formula to see if it fits. Ones that do not fit are in red.

320CE;100TP: 28 Damage >> Calculated = 28.9
960CE;100TP: 86 Damage >> Calculated = 86.6
1920CE;100TP: 172 Damage >> Calculated = 173.3
960CE;100TP(GORGET): 96 Damage >> Calculated = 96.0
320CE;150TP: 32 Damage >> Calculated = 32.1
960CE;150TP: 96 Damage >> Calculated = 96.2
1920CE;150TP: 192 Damage >> Calculated = 192.4
320CE;200TP: 35 Damage >> Calculated = 35.3
332CE;200TP: 36 Damage >> Calculated = 36.4
960CE;200TP: 105 Damage >> Calculated = 105.8
1920CE;200TP: 212 Damage >> Calculated = 211.5
960CE;250TP: 148 Damage >> Calculated = 148.9
1920CE;250TP: 297 Damage >> Calculated = 297.8
960CE;300TP: 192 Damage >> Calculated = 192.0
1920CE;300TP: 384 Damage >> Calculated = 384.0

Only a couple of the 1920 CE (6 dispels) don't fit. Not sure what to make of that atm.

Another interesting piece I found if you assume those fTP and CE mods is how the fTP affects the CE cap.

We know the CE caps for 100, 150, 200, and 250 TP.

100TP: 360
150TP: 400
200TP: 440
250TP: 620

If we actually just say that the original "CE Cap" is an even 400, these caps make complete sense in terms of their fTP modifiers:

100TP: 400 * 231/256 = 360.9
150TP: 400 * ((231+282)/2)/256 = 400.8
200TP: 400 * 282/256 = 440.6
250TP: 400 * ((282+512)/2)/256 = 620.3

These caps fit perfectly with current raw data, too. In addition, it sort of agrees with Suiram's guess that if the exact amount of CE to cap your CE component is exactly 4,000 CE.

I personally like "nice numbers" but I don't want to force things to be nice when they may not be. However, this does seem to fit nearly all the data thus far.

************************

[Kaeko]

I agree there is a range, your range is too large, or there is something wrong with my math, however.

Repeating the math for my last post, using just one of your assumed numbers.

Given: 100tp>86 dmg.
Assumed: x=196
86 = floor(D*x/256)
86 = floor(D * 196/256)
If 86=D*196/256, then D=112.32
If D=112, then 112*196/256=85.75, floors to 85. Wrong.
If D=113, then 113*196/256=86.51, floors to 86. Possible.
If D=114, then 114*196/256=87.28, floors to 87. Wrong.
D=113

Given: 100tp+gorget>96 dmg.
Assumed: Gorget makes new x value (call it x') equal to old x + 25. x' = x + 25, or x' = 221.
Math:
96 = floor(D*221/256) = floor(113*221/256) = floor(97.55) = 97
196 is definitely not correct.

If you see a flaw with this methodology, let me know. I don't really feel like repeating it for your entire range. I'm going to try to test with a different # of dispels relatively prime to 3 later, to try to narrow it down.

Anyway, I do agree 215 may not be exactly correct, but I think the range is lower than you are saying.

I got the max range simply by generating the maximum difference when dividing 96/86 with the floors. These 2 maximum differences are generated by 97/86 and 96/87. That's all I did to generate it.

I am not saying 196/256 makes sense with the data. I am only saying that given that single calculation in which you based x=215 on, you could get a a huge range with a lower bound of 196. You could obviously eliminate many of these with other data.

[FailureMidgard]

OK this is my personal idea on how this works. This is contrary to the numbers Midgard posted so I'll try to explain my own reasoning and how the raw data fits and supports it.

My personal belief is that the modifier for CE/VE will be a "nice" number. The one I am working with is 0.1 (10%). Given this modifier, I came up with these 3 fTP values:

CE Modifier = 0.1
F100 = 231/256
F200 = 282/256
F300 = 512/256

(Excludes VE from formula, and does not include cap situations)

Damage = floor(floor(CE*0.1)*fTP)

************************

All Raw Data compared to this formula to see if it fits. Ones that do not fit are in red.

320CE;100TP: 28 Damage >> Calculated = 28.9
960CE;100TP: 86 Damage >> Calculated = 86.6
1920CE;100TP: 172 Damage >> Calculated = 173.3
960CE;100TP(GORGET): 96 Damage >> Calculated = 96.0
320CE;150TP: 32 Damage >> Calculated = 32.1
960CE;150TP: 96 Damage >> Calculated = 96.2
1920CE;150TP: 192 Damage >> Calculated = 192.4
320CE;200TP: 35 Damage >> Calculated = 35.3
332CE;200TP: 36 Damage >> Calculated = 36.4
960CE;200TP: 105 Damage >> Calculated = 105.8
1920CE;200TP: 212 Damage >> Calculated = 211.5
960CE;250TP: 148 Damage >> Calculated = 148.9
1920CE;250TP: 297 Damage >> Calculated = 297.8
960CE;300TP: 192 Damage >> Calculated = 192.0
1920CE;300TP: 384 Damage >> Calculated = 384.0

Only a couple of the 1920 CE (6 dispels) don't fit. Not sure what to make of that atm.

Another interesting piece I found if you assume those fTP and CE mods is how the fTP affects the CE cap.

We know the CE caps for 100, 150, 200, and 250 TP.

100TP: 360
150TP: 400
200TP: 440
250TP: 620

If we actually just say that the original "CE Cap" is an even 400, these caps make complete sense in terms of their fTP modifiers:

100TP: 400 * 231/256 = 360.9
150TP: 400 * ((231+282)/2)/256 = 400.8
200TP: 400 * 282/256 = 440.6
250TP: 400 * ((282+512)/2)/256 = 620.3

These caps fit perfectly with current raw data, too. In addition, it sort of agrees with Suiram's guess that if the exact amount of CE to cap your CE component is exactly 4,000 CE.

I personally like "nice numbers" but I don't want to force things to be nice when they may not be. However, this does seem to fit nearly all the data thus far.

************************

960CE:

103.9TP = 86 dmg

906*.1=96
96*floor(231+3.9/100*51)/256=96*floor(232.989)/256=96*232/256=87

[Fisban]

Sorry to interrupt the Atonement talk for a second.

I was wondering if anyone had done testing on the difference in enmity formulas for Cure I,II,III,V (mainly Cure III) and Cure IV. I don't recall seeing anything on Kanican LJ regarding that.

Sorry if this is posted someplace obvious, I just haven't seen it yet...

Thanks, and keep up the good work!

[Siniroth]

Kanican - Enmity Table (Curing)

Kanican - Enmity Testing (Part VIII)

[Fisban]

I have seen those posts...

I guess I am asking, are there different CE and VE modifiers for Cure IV opposed to Cure I, II, III, and V?

Pretty sure SE has stated that Cure IV generates more enmity than Cure V?

Therefore having 1 set of cure modifiers doesn't work for calculating enmity generated for all cures? Perhaps its just Cure V that has a different modifier?

[Siniroth]

Kanican - Enmity Table (WHM)

Spell	CE	VE	Level	MP	Casting
Cure V	400	700	61	135	2.50

Cure V is a static enmity amount, cure IV is based on the number cured

[Fisban]

Gotcha...

[Kaeko]

If you see on the table that it says -- C -- that means it follows the formula. That means Cure 1-4, all curagas, all waltzs, most blu heals etc. all follow this same formula.

EDIT:

It's easier just to list exceptions:

Item heals like Potions
Exuviation
Cura
Cure V

As far as the idea that Cure V could simply have a different modifier as opposed to a static amount of enmity, this was easily tested out by curing someone for 0 HP with Cure V and receiving the same enmity as curing someone for 900+. As far as different cures have different modifiers, I probably should have put it in the testing - I didn't bother with it since at that point (over a year since testing started) I already knew it absolutely did not. The actual test for that is probably buried somewhere in this thread.

I included an extra line in the table:

"All abilities and spells designated with at "-- C --" follow this formula unless indicated otherwise."

just to clear up any confusion.

[Fisban]

Thanks, I guess I just never ventured over to the WHM table to see the Cure V numbers...

[Kaeko]

Update on testing with High Jump.

I redid a similar test to Niv's high jump test using a 72RDM/36DRG. High Jump produced -30% for him, I got exactly -50%, similar to the KI poster. This suggests another variable, perhaps level or the act of subbing /DRG (i.e. the difference is whether or not DRG is main or sub)?

EDIT: -60% with legs, i'm leaning towards sub being the differentiating factor.

-30% when subbed
-50% when main
-60% when main /w AF2 legs

Something like that.

Suiram/Niv, can you try repeating this test on 36DRG main? If I am correct, you will get -50% reduction.

[Suiram]

I went with Fival, a PLD in my LS who has atonement, and tested 100% TP, 1920 CE gorget damage. It was 192, which (still assuming it's done in 256ths), narrows down the range of possible values to 208/256 through 222/256 for fTP without gorget at 100% TP. Would have done it with higher CE but I didn't realize the range would still be so big at 1920.


Can maybe try the High Jump 37DRG (would have to be 37, the only reason it was 36 in that test was because Niv's RDM was 72) over the weekend.

[Spekkio]

Update on testing with High Jump.

I redid a similar test to Niv's high jump test using a 72RDM/36DRG. High Jump produced -30% for him, I got exactly -50%, similar to the KI poster. This suggests another variable, perhaps level or the act of subbing /DRG (i.e. the difference is whether or not DRG is main or sub)?

EDIT: -60% with legs, i'm leaning towards sub being the differentiating factor.

-30% when subbed
-50% when main
-60% when main /w AF2 legs

Something like that.

Suiram/Niv, can you try repeating this test on 36DRG main? If I am correct, you will get -50% reduction.

i seem to remember something about the player using HJ had to be at the top of the hate list for it to be effective. does your testing show this to be true?

[RyaWHM]

I have what's probably a silly question...

In the testing of HP determining enmity lost when hit:

1500 hp -> 1350 hp, should be same enmity as

1000 hp -> 900 hp.

But my question is ... does the source of HP matter? Has anyone noticed an exception with either "ConvertsHPtoMP" or certain slots?

Examples of exceptions like this:
Sorcerer's Ring
Parade Gorget
Sublimation
+% not being doubled by drinks
(...I'm sure there's more, that's all I can think of atm)

[Thalandor46]

Might be a bit of a pain to test, but can High Jump be tested with a miss, to see if the enmity down still procs? I think its been debated before, but I know I've seen hate move away from me after a missed High Jump on a few occasions before. Very curious about the results.

You guys rock! :D

[Therin]

I just finished my Light Gorget so I'm good to do some additional testing on Atonement if necessary. Just post here what you're looking for and I'll see it done.

[Kaeko]

I have what's probably a silly question...

In the testing of HP determining enmity lost when hit:

1500 hp -> 1350 hp, should be same enmity as

1000 hp -> 900 hp.

But my question is ... does the source of HP matter? Has anyone noticed an exception with either "ConvertsHPtoMP" or certain slots?

Examples of exceptions like this:
Sorcerer's Ring
Parade Gorget
Sublimation
+% not being doubled by drinks
(...I'm sure there's more, that's all I can think of atm)

It doesn't. I did a lot of my HP changes using Zenith gear when starting out the damage taken tests so I would have noticed this pretty sure.

[Kaeko]

Regarding the HJ tests, I can confirm that the HJ enmity down works even when the mob is not looking at you; however, I don't know about missing/hitting affecting it. It always hit when we tested.

Do blinding potions bring the jump accuracy to 0%?