Spirit Generator Attack Speeds Redux

[Scrangos]

This was useful but it didnt answer some questions for me... So I ran my own test using a similar methodology (frame counting at 30fps capture). Im posting my results for anyone whos interested, some might differ a slight amount with the OP. Could be my human error. Im willing to upload the video I used to test these.

Notes, this was done with a 2H weapon, Staff at 1.00 attack speed. No attack speed modifiers at all. Captured at 30 FPS.
My findings when cycling attacks and doing one cycle to rest differed. FoT, Reach and Crippling came out worse, Hundred fists cut off a few frames.
Tests were done unruned, sample dps asumptions are pretty much asumptions till I test the runes... At least Flurry is accurate though.

FoT  
Frames - Atacks per Second - Best Dps% (Tclap)

61  1.475  213.875%/s

18  1.666  241.666%/s
19  1.579  228.955%/s
24  1.250  181.250%/s

162.25%/s  Average w/o rune
8.85 spirit / second


Reach  Individual frames here seem a little iffy but thats what I got.
F    APS     Dps Best (Scattered  ||  Foresight on)
74  1.216    157.300%/s     157.230%/s

17  1.764    194.040%/s     228.967%/s
16  1.875    206.250%/s     243.375%/s
41  0.732    124.440%/s     95.0136%/s

133.25%/s average without runes
7.26 spirit/second

Crippling
F    APS     Dps Best (Mangle)
77  1.16     164.880%/s

25  1.2      171.600%/s
22  1.363    195.000%/s
30  1        143.000%/s

127.6%avg/sec unruned
6.96 spirit/sec

Hundred Fists
F    APS     Dps Best (Flurry)
81  1.111    196.10%/s

25   1.2     168.00%/s
27   1.111   155.54%/s   (222.2%/s HoLightning)
29   1.034   258.50%/s

155.4%avg/s unruned
177.6%avg/s HoLightning
178.71%avg/s Stacked Blazing

6.66 spirit/sec

Feel free to draw conclusions or point out mistakes.

To me, the best 2 spirit gen combination seems to be reach foresight (get foresight on, then only do the first two) with the last hit of hundred fists flurry.

It gives less spirit than using FoT though, and you dont get a teleport.

[yovargas]

So Hundred Fists is the slowest of the four? That seems odd...I kinda don't get WothF. I keep playing around with it but feel like I don't understand the point. Unruned it feels extremely similar to FoT. If it's just a slower FoT , why does it exist??

[Scrangos]

So Hundred Fists is the slowest of the four? That seems odd...I kinda don't get WothF. I keep playing around with it but feel like I don't understand the point. Unruned it feels extremely similar to FoT. If it's just a slower FoT , why does it exist??

Seems slightly poorly thought out. Hundred is a cone though, FoT is singletarget cept the last hit, barring the tclap passive. Seems to be more aoe oriented, but crippling just outdoes it if you go mangle and you get the slow ontop. Without mangle existing however, Hundred would fill a good niche of having good aoe and pretty good single target too.

Like reach you can shift use it to strike at range-ish.

[Athenau]

Scrangos:

I didn't attempt to identify the frame count of each individual strike--I wasn't sure where each attack ended or began so I only measured the length of a full combo. The rest position was unambiguous so measuring using that as a start/end point should be relatively accurate.

One caveat is that the animation for a single strike, including the frames to return to rest is different from the animation within a combo since you don't return to rest until you stop attacking. So attempting to measure the frame count of a single strike using that method may be misleading--that's another reason why I only attempted to measure full combos.

Your FoT numbers do seem consistent with mine though.

As for the best combo, reach/foresight + windforce flurry is a tiny bit better damage wise than FoT/Thunderclap + Flurry (~185% average damage per strike vs 180%, both at the same speed) IF that's the only damage bonus you're running. Damage boosts stack additively, so if (for example) you're running blazing wrath (+15%) and combination strike (+24% in a three generator build) you get:

Reach/Foresight + Flurry: (110% + 110% + 250%)/3 * 1.57 = 246% damage per strike
FoT/Thunderclap + Flurry: (145% + 145% + 250%)/3 * 1.39 = 250% damage per strike

TBH the damage difference is noise, and thunderclap offers so much more utility. The teleport is amazing. It's instant, ignores projectiles, and most importantly, free. Plus you can stunlock regular mobs really easily with it. I do like the range on reach though.

WoTHF has one reason to exist, and that is so you can scream IMMA FIRING MAH LAZOR while you pew pew everything with windforce flurry. That rune is godlike.

[Such Violent Storms]

If you can get a good critical hit chance, you can maintain 3 stacks of the Blazing Fist rune for 15% ias and movement speed. That seems useful like it would increase damage a good amount.

[Scrangos]

Scrangos:

I didn't attempt to identify the frame count of each individual strike--I wasn't sure where each attack ended or began so I only measured the length of a full combo. The rest position was unambiguous so measuring using that as a start/end point should be relatively accurate.

One caveat is that the animation for a single strike, including the frames to return to rest is different from the animation within a combo since you don't return to rest until you stop attacking. So attempting to measure the frame count of a single strike using that method may be misleading--that's another reason why I only attempted to measure full combos.

My first few tests were using the rest position, but I did another method using a stopwatch by doing 30 full cycles back to back of each and there were some discrepancies. For most combos (Cept hundred fists) theres a frame where the character changes the pose abruptly signaling the start of the next attack (feet and hand are very telling).

My results showed that FoT, Crippling and reach have more of a lag at the end before another attack can start than you can see when going into rest. And that hundred fists actually cuts before it would get into rest and goes into the next attack faster.

I also tried measuring every single strike separately, and was getting a very different set of results.


As a generic/singletarget/utility/more spirit attack FoT does completely overwhelm reach however it might work well in certain builds, like running the snare on conviction and having foes line up, constantly hitting 3-4 with all your attacks. Also in a 2 gen attack you will probably have two more sources of damage (Ally and sweeping wind most likely) which benefit from the extra 18% from reach. It does however give up utility and spirit power to do so which can be dangerous (Specially if you play hardcore).



Blazing fist is kinda subpar, just using flurry would give more dps as a single spirit. If you want to combine reach with blazing fist to get both the IAS and the damage, it would boost considerably other spells that rely on dps rather than weapon damage (like sweeping wind and ally) but keeping up flurry is pretty much a fulltime thing. Even a simple cc can cause it to drop.

That said... asuming you can keep up both, and ignoring when you use reach to do the buff.. and you somehow can keep the stacks up only using the third hundred fists attack with no other damage and aspeed buffs which would cause other combos to gain more than this combo would... itd be something like...

263.235%/s
279.795%/s
196.439%/s
246%/s average.
Its still a bigger boost to the other moves that arent the spirit builders but... I dont think its possible to sustain it.

[Athenau]

For most combos (Cept hundred fists) theres a frame where the character changes the pose abruptly signaling the start of the next attack (feet and hand are very telling).

By "next attack" do you mean next attack within a cycle or next combo? If the latter you could use any identifiable frame and just measure the period between re-occurrences, it doesn't have a to be a startup frame.

Edit: I stepped through one of my vids and I see the abrupt feet transition you're talking about with Thunderclap x2 + Windforce Flurry. Using that a single combo takes 42 frames, so the calculated attack speed multiplier is 1.33x, or a tiny bit faster than the results in my OP. Good find! This makes it easier to time things without doing long tests.

My results showed that FoT, Crippling and reach have more of a lag at the end before another attack can start than you can see when going into rest. And that hundred fists actually cuts before it would get into rest and goes into the next attack faster.

Yes, so if you measure a single cycle you'll introduce some error that way. I measured 8 cycles back to back so a margin of a couple frames won't affect the results too much. I also tried some longer trials (20 cycles ~ 60 attacks) and my results didn't change. Mostly the limitation on longer tests was human error since it's hard to do mixed combos are higher attack speeds for long without screwing up--the timing is surprisingly tight for an ARPG.

[Scrangos]

Next attack within a cycle and next combo both... just the transition from one hit to the next.

From some other thread I read that while the engine works on an internal framecap, if you end up having to approximate frames, on some cycles itll use a certain amount of frames, and on others itll use a different one. Approaching the correct amount of attack speed over 100 attacks.