Slow, Hojo, and Slow II

[nekura]

The Elegy instrument comes one level after Battlefield Elegy so probably that, since Battlefield is 25%. :3

[UnknownGamer]

I went through Rheya's data in addition to Releena's and found the sources for previously mentioned info about Slow II:

1) Merit levels in Slow II affect the potency cap
There are 3 cases in Rheya's data where at presumptively capped MND (83, 93-96, and 108), changing from level 1 Slow II to level 2 resulted in a definitive change in slow potency. A 60 second Utsusemi recast turns into 61 in each case. There was no extra info from Releena simply because all of Releena's data exceeding +75 MND are of either Slow I or Slow II level 3, and the slow % ranges for level 3 still overlap with level 2.

2) Either +75 MND is not the slow cap or the amount of slow % added by Slow II differs at capped MND
Releena's data contains 3 high-MND entries that peg Slow II level 3 at 35.56% to 38.88%. Rheya's data shows Slow II level 1 at 33.33% to 35.55%. Releena has the lone entry for capped MND Slow I (45s -> 58s at +86 MND) which pegs that at 28.89% to 31.10%.

The difference between Slow I and Slow II level 1 at cap calculates out to 2.23% to 6.66%, which is inconsistent with the difference between these two at 3-4 MND, which is 6.68% to 11.10% according to Rheya's data (keep in mind the max may be underreported because Slow I was cast with 1 more MND).

A similar disparity arises when comparing Slow I and Slow II level 3. Releena's data: at +86 MND, Slow II level 3 is 4.45% to 8.88% more potent than Slow I. But at 0 MND, Slow II level 3 is 11.12% to 14.80% more potent.

[Tsuko_Asura]

The Elegy instrument comes one level after Battlefield Elegy so probably that, since Battlefield is 25%. :3

Thanks, I suppose that was painfully obvious.

I tend to forget what life was like before Carnage

[Kirschy]

It was a mistake for me to rush and post my findings without completing my tests and neglect to post my data or testing proceedures. Also after examining my testing methods, the margin of error is higher than I made it out to be. I also did some further testing and organised my data so I could share it /w all. This post will be long and boring, so I made pretti graphs(or at least tried to). Cliffnotes at bottom.

First off, this is how I conducted majority of my experiments:
-I found a target mob, and aggro'd it.
-I used fraps at 30 fps, and recorded 11 attacks.
-I opened the video in VirtualDub, and checked for the first frame in the first attack animation. I recorded this number.
-I then checked for the first frame in the last(11th) attack animation, and recorded this number.
-Subtract the two numbers, and divide by 10 (This gives you the average frames between the start of two attacks.)
-(Optional)Repeat this test 5-10 times, and get the average frames per delay.

Some notes about this testing proceedure:
I used this test a lot when I was calculating haste%. I used a 999 Delay Weapon majority of the time, which made the testing extremely accurate. Each frame I was off by only resulted in a .2% error. Calculating slow% on enemies, the largest delay I worked with was around 360. This is .56% error per frame. After doing over 250+ tests, I feel safe to say I am being tough when I give myself a margin of error at +/- 2 frames.

I think it's important to note that there are inherent inaccuracys in this test. There is slight amount of lag. The game can only display animations at 30 frames per second making precision down to 1 delay impossible. We don't know how the game calculates delay, if there can be decimals, where truncation occurs, etc...

There are usualli 3 attack animations per mob. I had a friend who editted DATs and made each of the three attack animations the same for some mobs. Even with this, the amount of frames that occur between attacks is not always the same. Furthermore, each attack animation can vary in actual frames in the DAT. (For example a 15 frame attack animation, can appear in game to last 30-33 frames, and is not always the same.)

After calculating the average frames of an attack, you can divide by 30 to determine how many seconds per attack. And then multiply this number by 60 to get the FFXI Delay. (Or just Multiply by 2.)

To make one last point about this testing procedure. Despite all of what I wrote above, the tests do come out to be fairly precise. I don't think there is a more accurate way to determine delay changes.

Testing(The following are in FRAMES!):

Huge Hornet:
181.3
181.4
181.3
181.0
181.7
182.0
181.5
181.6
------
Avg~181.5

Slow /w 120+ MND:
234.0
234.3
233.4
233.7
234.8
------
Avg~233.84

233.84/181.5 = 1.288 (28.8% Slow) (Expected: 29.3%)

Carnage Elegy:
271.8
272.2
271.0
272.2
271.9
-----
Avg~271.82

271.82/181.5 = 1.498 (49.8% Slow)

In a previous thread, I said that Carnage Elegy was closer to 51%. This was because I used the base delay at 360(180 frames). (271.82/180 = 1.51%) I now believe this test to be more accurate, so I think Elegy is capped at 512/1024(50%).

Battlefield Elegy /w +0 Instrument:
226.7

226.7/181.5 = 1.249 (24.9% Slow)

Battlefield Elegy /w +2 Instrument:
237.6
237.8
-----
Avg~237.7

237.7/181.5 = 1.3096 (30.96% Slow)

It appears +2 Instrument gives +6% slow. Probably +3% per instrument.

Slow II /w 1 Merit:
244.8
244.7
-----
Avg~244.75

244.75/181.5 = 1.348 (34.8% Slow)

I originalli had Slow II cap at 36.1%. This was again a mistake on my part as I used 180 frames base instead of 181.5. 244.75/180 = 1.3597 (35.97% Slow)

Slow II /w 2 Merits:
245.4
245.2
245.2
244.9
-----
Avg~245.17

245.17/181.5 = 1.3508 (35.08% Slow)

As above, I incorrectly used 180 frames as the base delay, but this was wrong. Slow II /w 1 Merit and Slow II /w 2 Merits cap extremely close to one another. I mistook this for no change in cap.

Slow II /w 3 Merits:
249.0
248.5
248.4
248.3
248.5
-----
Avg~ 248.54

248.54/181.5 = 1.369 (36.9% Slow)

This result is almost 2% higher than Slow II /w 2 merits. I cannot explain why I can't get the difference between 1 and 2, but I can see a difference /w 3. It might just be the inaccuracy in my testing. In either sense, extra Slow II merits do appear to raise the cap.

Robber Crab:
Banish:
64 MND - 13 Damage
65 MND - 13 Damage
66 MND - 14 Damage <<<
67 MND - 15 Damage
68 MND - 16 Damage

Robber Crab Base:
121.2
121.9
121.5
121.8
-----
Avg~121.6

Robber Crab /w Slow II 3 Merits @ 67 MND(dMND=+1):
152.6
151.7
152.5
152.8
152.4
152.6
------
152.43

152.43/121.6 = 1.2536 (25.36% Slow)

Robber Crab /w Slow II 2 Merits @ 66 MND(dMND=0):
151.2
151.7
151.5
152.1
-----
Avg~151.625

151.625/121.6 = 1.2469 (24.69% Slow)

To double check that slow stacks the way I expected it to, I casted Carnage Elegy + Slow 1 on a bee and got 325 frames. 325/181.5 = 1.79 (79% Slow. 50%[Elegy] + 29%[Slow]).

The rest of testing was done to map out Slow 1 and Slow 2 at different potencies. Below are the graphs that I came up with.

Slow I
(The expected portion of the graph follows the equation I posted on the first page. It also has margin of error lines for up to 3 frames since I often only tested once at each MND value.)
Slow II /w 2 Merits

It seems like most of my data supports Releena's tests. Her's are more precise as using recasts for testing is going to be much more accurate than using delay. Since the testing appears to follow the same, it would seem appropriate to conclude that pvp testing /w Slow is OK.

TY to Hesasa for letting me abuse his RDM, and Tinkil for letting me get some Slow II /w 3 Merit tests done.

=============
***Cliffnotes***
=============
tl;dr

-My experiment is not precise enough. I also made some errors.
-Slow 2 /w 3 merits has a higher cap than Slow 2 /w 2 merits. Future increases will probably increase the cap further.
-Slow I caps at 29%ish
-Slow II /w 1 merit caps at 35%ish
-Slow II /w 2 merits caps at 35%ish
-I don't know why Slow II /w 1 and 2 appear to be the same
-Slow II /w 3 merits caps at 37%ish
-Carnage Elegy caps at 50%ish
-Battlefield Elegy is 25%ish /w +0 Instrument
-Battlefield Elegy is 31%ish /w +2 Instrument
-Slows stacks as expected.
-Slow I Graph
-Slow II /w 2 Merits

[Kirschy]

Retested Slow II /w 1 Merit:
145.5
245.2
245.6
244.5
-----
Avg~245.2

245.2/181.5=1.351(35.1% Slow)

I don't understand why it almost matches my Slow II data :/

[Virtuous Saint]

Very nice testing, good job.

[UnknownGamer]

Very nice work, seems consistent with all of the recast data and reinforces the conclusions hiding in Releena's/Rheya's data sets.

One thing they tested and which you haven't, though, is what happens when you cast Slow I at dMND = 0 or lower? Kind of a last thing to confirm before we start to rework one of the Slow equations. It also has a practical implication, meaning if you don't have high MND, it may be even more important to use Slow II over Slow.

Recast data is exact, but the numbers all come out in ranges, and it's the job of the experimenter to collect data from a variety of recasts to narrow that down mathematically. If we lack the spells to do this, then direct measurement would be more precise.

[majriti]

can someone please do better testing with ninja magic, including increased magic skill rather then base 269 skill

[Aux]

=============
***Cliffnotes***
=============

-My experiment is not precise enough. I also made some errors.
-Slow 2 /w 3 merits has a higher cap than Slow 2 /w 2 merits. Future increases will probably increase the cap further.
-Slow I caps at 29%ish
-Slow II /w 1 merit caps at 35%ish
-Slow II /w 2 merits caps at 35%ish
-I don't know why Slow II /w 1 and 2 appear to be the same
-Slow II /w 3 merits caps at 37%ish
-Carnage Elegy caps at 50%ish
-Battlefield Elegy is 25%ish /w +0 Instrument
-Battlefield Elegy is 31%ish /w +2 Instrument
-Slows stacks as expected.
-Slow I Graph
-Slow II /w 2 Merits

Kirschy, thanks much for your initial testing with Slow II. Now post-update wise and the merits boost, how would a:
-Slow II w/ 4 merits
-Slow II w/ 5 merits
test out in conjunction with the other 3 forms of Slow II? I realize that to obtain these it'll require some vast meriting and not that you or anyone else supplying us with information would have all the time in the world to get immediate merits, so I humbly ask if you have any rough guess or figures where those would fit in?

[Kirschy]

Unfortunately my character has been locked for over two weeks now. While I have an alt character I can log on, everyone seems to be busy with the new stuffs.

I was kinda hoping someone would run some pvp tests /w 4/5 merits in Slow II to see how recasts are affected. Using Slow II on recast like Dread Spikes would be accurate to 1%.

[AzazaelFenrir]

So as someone who has some extra RDM merits now, is there any benefit to going Slow II level 2 over Slow II level 1? If I read the tests on this page right then they seem to be identical, is this accurate?

[Zoot]

Unfortunately my character has been locked for over two weeks now. While I have an alt character I can log on, everyone seems to be busy with the new stuffs.

I was kinda hoping someone would run some pvp tests /w 4/5 merits in Slow II to see how recasts are affected. Using Slow II on recast like Dread Spikes would be accurate to 1%.

We did this with feather barrier and reraise, but decided that the results were too innaccurate to post results (yet) and have been too busy to go back and retest. It was also 10AM and we hadn't slept because some crazy ballista people decided they wanted to PvP it up for a few hours.

I'm a little hesitant to post what the results did show because I'm afraid people will take it as fact. Also, rel is the math whiz, not me, and he's been busy.

Basically slow II lv3->slow II lv4 showed no direct increase in potency where mnd=mnd, but for the same strange reason slow II lv4->slowII lv5 showed an increase. There was also an increase where dMND = 76 in both slow II lv4 over slow II lv3 and slow II lv 5 over slow II lv4, but dMND = 93 did not show an increase over slow II lv4 at level 5.

[Kirschy]

Feather Barrier seems like a very accurate way to test. If youre not getting expected results, there is probably a lot more testing to do to rework a model. If you ever get around to running more tests and posting your results, myself and I'm sure others would be interested in seeing what you've found.

Feather Barrier is 120 second recast. Assuming 10/1024 is about 1% for the most part, and 8-9/1024 is needed to change the recast by a whole second, it seems very accurate for testing.

As for Slow II /w 1,2, and 3 merits. In capped conditions on bees, there appears to be no change in thier attack delay with 1 and 2 merits. But with recast testing in pvp, there does appear a small change. With the 3rd merit there's a clear decrease in delay. It's possible attack animations only register at certian slow %s or delays. When it comes down it, it appears that the same number of attacks occur during a set period /w Slow II 1 and 2 merits.

[Zoot]

Feather Barrier seems like a very accurate way to test. If youre not getting expected results, there is probably a lot more testing to do to rework a model. If you ever get around to running more tests and posting your results, myself and I'm sure others would be interested in seeing what you've found.

This is what releena implied, but we made a mistake and went ahead with a shorter recast spell to start (my fault, really). Thus, we ended up with a comparison for Lv4 to Lv5 but innaccurate comparisons for Lv3 to Lv4. In order to be 100% accurate, we are going to test with more than one longer recast spell, and find some bloke with level 3 to correct the mistake of leaving out feather barrier.

[Aux]

According to a post by Maiev of Fenrir, it seems that Maiev has capped out Slow II at 5 Merits, and is also looking to do some testing. If anyone is on Fenrir willing to assist Maiev of some sort, this would be a tremendous opportunity to look in-depth to the real cap of Slow II.

[Kirschy]

I know Maiev, he's a cute brownhair taru =p. I'll go bother him later and see if he'll do tests /w me.

[Kirschy]

Played around a bit /w Maiev tonite. Thought I would share what we found~

The first thing we did is test using my fraps method, and Slow II /w 5 merits came out to be ~39% Slow. This is noticably larger than Merit 1/2, and Merit 3 with this testing method. The cap definetly changes.

Luckily, Maiev had some ballista points left, so I logged on a friends DRK account and we headed to Jeuno to get one of those Whitepage pvp things. The test setup was as follows:

The DRK I used was DRK/SCH (Lv1 SCH) /w a Spike Necklace only. 45 Base MND - 6 MND = 39 Total MND. (I also controlled a BRD.)
Maiev controlled himself as RDM and had a variety of MND gear. Maiev also controlled a BLM.

Maiev would hit a desired MND and then Slow II'd the DRK. I casted Dread Spikes, recorded the recast. Then Maiev had the BLM nuke/kill the little taru DRK. (This was to avoid waiting 4+ min on recast...)

MND : Recast
146 : 4:10
114 : 4:10
109 : 4:08
104 : 4:07
99 : 4:06
94 : 4:04
89 : 4:03
84 : 4:02
77 : 4:00

Using etudes + eq I bumped the DRK up to 77 MND to match the RDM.
@ dMND = 0 : 3:49
Using Eq I had the BRD/WHM at 77 MND to match the RDM.
@ dMND = 0 : 1:16 (Reraise)


So what does all of this mean?

-Slow II caps at dMND = 75.
-Slow II /w 5 Merits caps between 399/1024 - 404/1024. (Probably 400/1024 - 39%).
-Slow II /w 5 Merits @ dMND = 0 is between 279/1024 - 284/1024 (Probably 280/1024 - 27.3%).
-Slow II has a lower MNDotency ratio than Slow.

With this data, and combining the results from my previous tests and Releena's, I would go one step further and propose this revised model of Slow II which should encompass most tests:

y=Merits in Slow II
dMND = (Casters MND - Target's MND)

While dMND >= 0:
Slow II% = ([230] + [y * 10] + [floor(dMND * 1.6)])/1024
Cap @ (350+[y*10])/1024

{Note: While I think this formula predicts the correct slow% with high accuracy, it probably needs a tweak or two.}

In regards to why my fraps method is showing Slow II /w 1 or 2 merits being equal potency... I believe the game engine uses some of appromixated/truncated/rounded info. What we see on the screen is an approximation for the serverside calculations.

Thank You to Maiev for staying up past 3AM to do tests /w me .

[shaddix]

I believe the game engine uses some of appromixated/truncated/rounded info. What we see on the screen is an approximation for the serverside calculations.

Off-topic, but I wonder if this is why we see no improvement in short cure animations with clogs alone. But a dark arts cure3-5 with clogs alone does show improvement.

[Virtuous Saint]

Great job with the testing Kirschy!

[Umbra]

So with my extremely slow thinking today due to getting no sleep:

-Slow II has a lower MNDotency ratio than Slow.

I take this to mean it takes less mnd to hit this supposed 39% cap according to this latest round of testing? 5/5 slow 2 is as good as it sounds then?

Also, the theory about additional merits (past 3) only affecting the total mnd needed to hit the theoretical 37% cap has basically been debunked and the cap is actually raised then?