Utsusemi / Dualwield -- Recast+Cap

[genome]

Ok i just did 2 more tests with 25% haste vs 30% haste and attack speed.

I felt very uncomfortable with my accuracy in timing them, but still repeating it twice with each setup (the cap for recast 25%, and 30%), I basically picked an EP colibri, then started swinging staff at it, pressed /clock as 1 hit landed and counted that as #1, then pressed again clock right as the 10th hit landed (without double attack gear or sub and not counting counters but obviously counting misses if any).
For all cases at 25% and 30% I got 42 seconds for 10 hits...

It's a pain to count and measure this way, so I might have messed up, the difference would be just 2.8 seconds. Still so far I am inclined to think that haste from gear also caps at 25% for attack speed too..

Anybody did more accurate tests for this or has some suggestions on how to test it more accurately (something that is not incredibly tedious plz).

Divisortheory if your parser will be able to do that, please hurry up and finish it =P

[gemini~cusp]

As a non~nin, actually have an idiotic question about dual~wield Does it matter what the delay and the damage is on the main vs. sub (i.e. is it better to have shorter delay on main or is there no difference and is it better to have the higher damage wpn on main).

Guess I wanted to throw that in since we're on the dual~wield question

[Ryko]

Dual wield calculation is based off of the delay combined so it doesnt matter which is mainhanded, due to the way weapon skills are calculated though it's typically a smart idea to keep the higher damage weapon in your main hand.

[Nny]

Very very easy way to taste things regarding haste:

Record a video via fraps and record at 30 FPS.

Open it up in something like Adobe Preimere where you can easily go from frame to frame to pinpoint when you swing next and get it down to an exact amount of seconds (Number of frames to next swing/30)

The only problem with this is that the game doesn't actually have a set delay that shown by swining animation nor logs. There are different swining animations, and each one takes a different amount of time. So you have to makes sure you have a long enough source to be able to get a good average between all the swings.

I don't suggest using a parser to test haste, would be a lot more accurate to do it via video.

[weeee]

graphically its too inaccurate

[Nny]

You mean because of the different graphic animations? The same goes for reading the logs.

[genome]

Very very easy way to taste things regarding haste:

Record a video via fraps and record at 30 FPS.

Open it up in something like Adobe Preimere where you can easily go from frame to frame to pinpoint when you swing next and get it down to an exact amount of seconds (Number of frames to next swing/30)

The only problem with this is that the game doesn't actually have a set delay that shown by swining animation nor logs. There are different swining animations, and each one takes a different amount of time. So you have to makes sure you have a long enough source to be able to get a good average between all the swings.

I don't suggest using a parser to test haste, would be a lot more accurate to do it via video.

Thanks that's a very good idea, and especially i'll be able to go back and look at my test again if i have doubts or think of something that could have been wrong. I'll try it again soon, if i find there isn't a cap on attack speed and i just counted wrong i'll post here again

[Dryhus]

Take the net delay of 36000 and divide by the total hits in each case to get the delay per swing.

Why 36,000? -- because in your example scenario you are testing for 10 min intervals?

60 delay/second * 60 seconds/minute * 10 minutes = 36000

[draxyl]

9x4x1,000 = 36,000 too.

[Rag]

So the cap is 25%haste (Equip) + 25%haste(Spells)

What about fast cast trait? i know rdm/nin lv75 Got better utsi recast than my nin with +17Haste%

[Vail]

So the cap is 25%haste (Equip) + 25%haste(Spells)

What about fast cast trait? i know rdm/nin lv75 Got better utsi recast than my nin with +17Haste%

Recast Rates cap at 50%, w/ or w/o fast cast
though technicly sense you'll cast faster, your recast timers will start sooner (if that makes sense)

[divisortheory]

Since there was talk of parser earlier in this thread, I decided to bump this one rather than start a new one. I decided to do the ultima test of the parser and use it in Beseiged with all damage filters OFF. Needless to say it was laggy as FUCK, but you can definitely see that it's coming along.

First screenshot I took


A few minutes later


A few minutes later


If you wanna know how much HP the typical Qutrub Wastrel has, add up all the TTL numbers in the second screenshot, since that was a Qutrub Wasterl from about 95% to 0% with no other mobs around.

[Russta]

That looks really nice. Rough idea how far off from some degree of completion it is?

[divisortheory]

20% for completion, and maybe 40% towards where I actually put out a Technology Preview (e.g. buggy, not feature complete) release that people can use. If you look carefully at the screenshot you can see all kinds of things that don't even make sense. Like why is everyone's accuracy 100%, for example. On top of that, the columns in the display plus about 5 more which are not shown are currently the only things it can calculate. I need to add a few more. Also, something weird but POL seems to crash with Access Violations a lot more often since I've started using/testing this parser. I'm wondering if that program I'm using to read the game memory doesn't have some sort of bug in it.

A likely potential holdup to me releasing anything at all is that I'm going to have to figure out how/where the chat log is store in memory on my own, because I'm not going to release the parser with a dependency on yet another program. So either I need to figure it out myself or remove the real time capability, which was one of my design goals in the first place. If anybody who read this forum are good at figuring stuff like that out, it would save me an unknown amount of time if someone PMed me with a description of how/where the chat log is stored in memory, as well as how to automatically detect where it is in memory so that parser doesn't break every time there's an update. Naturally, if someone did this they would receive appropriate credit.

[Nny]

Getting back to equipment haste cap on delay, was bored so decided to test it.

I tested 25%, 35%, and 48%. 25% because that's the spell casting cap, 35% because 25% is half of 50% (The spell total haste cap, whereas for delay the total haste cap is 70), and 48% cuz that's the most I could get via equipment. Used a 420 delay great katana

Average attack time was 5.2445 (or 314.67 delay).

420 * .75 = 315

So yeah, equipment haste is capped at 25%. Excuse me while I go cry in a corner.

I'm not going to post the video unless somebody really wants to see them, since it's really basic stuff.

[Nny]

Something else interesting, equipment with "Enchantment : Haste" does not count towards the equipment haste cap. The only piece I know of with it is Haste Belt, but it just got a hell of a lot better for all WAR/DRK/PLDs.

[Heian]

Kalia the work you are doing on that parser is awesome man, keep up the good work.

Oh and to the OP also look at this thread.

http://www.theorderls.com/phpBB2/viewtopic.php?t=8867

There is alot of data on how the nerf affected Dual Wield Trait increases and Delay effects on TP.

[Kaylia]

Something else interesting, equipment with "Enchantment : Haste" does not count towards the equipment haste cap. The only piece I know of with it is Haste Belt, but it just got a hell of a lot better for all WAR/DRK/PLDs.

It was already good in the first place But how do you hit the cap on war/drk anyway?


5% head, 3% feet, 5% legs. Another 4% from dusk hand I guess. If you use speed belt with that, you're still far from the cap.

[Nny]

It was already good in the first place But how do you hit the cap on war/drk anyway?

5% head, 3% feet, 5% legs. Another 4% from dusk hand I guess. If you use speed belt with that, you're still far from the cap.

Rune Chopper. 8) Fuck /NIN and Ridill.

Oh and you can't remove Haste Belt to add on Speed Belt, the Haste effect will come off.

[genome]

Getting back to equipment haste cap on delay, was bored so decided to test it.

I tested 25%, 35%, and 48%. 25% because that's the spell casting cap, 35% because 25% is half of 50% (The spell total haste cap, whereas for delay the total haste cap is 70), and 48% cuz that's the most I could get via equipment. Used a 420 delay great katana

Average attack time was 5.2445 (or 314.67 delay).

420 * .75 = 315

So yeah, equipment haste is capped at 25%. Excuse me while I go cry in a corner.

I'm not going to post the video unless somebody really wants to see them, since it's really basic stuff.

Yeah I noticed that too (haste delay also capped at 25% from equip), thanks for verifying this for me, I wasn't sure if it was just me