Delay Reduction Cap Test

[kittycatrag]

Summary:
Dual wield and haste combined multiplicatively cap around 80%, just like failure said. High delay weapons and video capture were used to eyeball the effective weapon delay.
video

Setup:
thf90/nin45, dw+25
main: relic dagger, delay 999
sub: auric dagger, delay 201 dw+5
range: raider's boomerang, dw+3
head: af3, h+6
neck: tiercel, h+1
ear1: suppanomimi, dw+5
body: af3, h+5
hands: ocelot, h+3
waist: nusku, dw+5
legs: af3, h+5
feet: ballerines, h+4

4/5 merited haste samba from a dnc in pt, h+9
soul voice double march+3 with 600+ skill and whm haste, h+43.75

Expectations:
Delay if 80% cap is in place: 1200 * .2 = 240 or 4 seconds
Delay if not: 1200 * 0.57 * 0.2325 = 159.03 or 2.6505 seconds

Method:
I attacked a zvahl fortalice and recorded video at 30 FPS. The uncompressed video was viewed frame by frame, with timestamps recorded when TP values changed.

Data:
1:26.400
1:30.467
1:34.567
1:38.633
1:42.767
1:46.833
1:50.900
1:55.033
1:58.967
2:03.067
2:07.133
2:11.467
2:15.567
2:19.633
2:23.633
2:27.700

Sources of error:
Network lag
Aliasing from video capture rate
Imprecise haste values
Incorrect dual wield values
Rounding
Incorrect denominator for delay calculation

Conclusion
Just by eyeballing the data/video you can see 4ish seconds per round. I wouldn't tear down your gear to get exactly 80% delay reduction, but I'd get rid of supanomimi and charis necklace on a 5/5 haste samba dnc with a brd in the same party.

[Raen Ryong]

Pancakebunny is the best name ever. And thanks for confirming >:3

[Kirschy]

Nice to see additional confirmation, good stuffs~

[kamugi]

TEAMcat 4tw.
Good job KC

[Chun Liroy]

Why were they banned?

[Byrthnoth]

There's slightly more than 4 seconds per round (61.3 seconds for 15 4-second swing delays) and it's a pretty systematic variation. All deviations except one are positive or 0.

I think I remember some people expressing suspicions that 60 delay was actually more than 1 second in the past. This would work out to about 59 delay per second.

[Theraven]

Why were they banned?

He wasn't, that's just his normal avatar (notice it isn't animated like the banhammer one).

[Masamune]

There's slightly more than 4 seconds per round (61.3 seconds for 15 4-second swing delays) and it's a pretty systematic variation. All deviations except one are positive or 0.

I think I remember some people expressing suspicions that 60 delay was actually more than 1 second in the past. This would work out to about 59 delay per second.

I remember VZX (or StudioGobli) quantified that delay to be exactly 58.85.

Also eyeballing in a testing involving low differences is far from ideal :S.