Finding pet skill cast time

Brael - Dreamweaver
Brael - Dreamweaver Posts: 1,430 Arc User
edited October 2009 in Venomancer
Ok, I just ran a couple of tests doing this with a magmite. It's level 65, like all other magmites has a .6 attack/sec speed, and when relevant was using bash 4 (8 sec cooldown).

For my test mob I was using Chintien in 1-1 solo mode. The method used for testing was to run fraps and record the fights at 25 fps, then play them back and count the number of attacks (red lines above the mob) during the test. When using skills, I did the same thing to count the total, then went through again and counted the times the skill went into its cooldown timer, to separate melee from skills, I subtracted the number of skill uses from the total.

I saved the videos incase anyone wants to view them and count for themselves, but I would need a way to shrink the file sizes. They're both right around 1.3 gigs each after importing to after effects, reducing the resolution to 1/3 (and I had already reduced my resolution prior to making them too), and decreasing the quality, so I'm not sure how to make it smaller while still allowing someone to count the hits, other than maybe rendering with only the area where attacks show up and cooldown timers in view. If there's any interest from people in checking the work I'll do that.

Test 1 - No bash.
Duration - 209 seconds
First attack - 0:02:19 (2.76 seconds)
Last attack - 3:28:09 (208.36 seconds)
Fight time - 205.6 seconds
Melee attacks - 121
Attacks expected - 123 (205.6*.6)

Test 2 - Bash
Duration - 244 seconds
First attack - 0:07:07 (7.28 seconds)
Last attack - 4:00:17 (240.68 seconds)
Fight time - 233.4 seconds
Total attacks - 143
Bash attacks - 25
Melee attacks - 118
Total expected - 169 ((233.4*.6)+(233.4/8))
Bash expected - 29 (233.4/8)
Melee expected - 140 (233.4*.6)

Going off of this data, it looks like the cast time on skills hinders attack rates quite a bit. In the first test, two attacks appear to be missing. I can rule out bad measurements on my end as I was measuring frame by frame which means I either counted wrong or the server doesn't actually deal damage from the pet in 1.67 second increments. Going by the way two consecutive spells will occasionally land at shorter intervals (when I cast a heal or venomous while holding the key down it's generally cast, cast, 1-2 sec pause, cast) I think the server is actually staggering the damage a little. This can be seen with pet melee too by watching the numbers float up, there's not an equal distribution between attacks so they're not occurring at the same time. Sometimes two attacks will overlap, or be closer than normal.

Unfortunately, my expired license copy of after effects (so it could only be opened the one time) decided to crash right when I realized I can count the number of frames between skill uses and subtract the cooldown time to get the cast time, but I think there's enough data to figure it out anyways. Going by timestamps, there's somewhere between 1.01 and 2 seconds inbetween each bash use. Looking at the above expected attacks of 140 and seeing that there's 118, while also knowing that it's .6 attacks per second, it can be said there's 36.67 seconds of melee not accounted for. Since there's 25 bashes, 36.67/25=1.4668 seconds. That fits into the expected range, and when adjusting that to numbers the game tends to use, it looks like a fairly safe bet to say it's a 1.5 second cast time on skills, and that melee cooldown timers don't go down while the pet is mid cast.

What this means to you, using bash for an example:
.5 attack/sec - Pet gives up 75% damage to gain 140%-200% damage. Actual gain of 65%-125% damage.
.6 attack/sec - Pet gives up 89.82% damage to gain 140%-200% damage. Actual gain of 50.18%-110.18% damage.
.8 attack/sec - Pet gives up 120% damage to gain 140%-200% damage. Actual gain of 20%-80% damage.
1 attack/sec - Pet gives up 150% damage to gain 140%-200% damage. Actual gain of -10%-50% damage (I guess this explains why a nix starts with FR5, FR1 would actually be detrimental).

Applying this same idea to damage skills, comparing bash 5 and FR 5 on a .8/sec pet
DD skill=9.5 sec cycle, 80% damage gain (200-120), 8.42%/sec gain
Dot skill=16.5 sec cycle, 180% damage gain (300-120), 10.91%/sec gain

Since FR is reduced by level differences in PvE though (but only 2/3 of the damage is reduced) the break even point would be a 44.36% damage reduction which with the way the tables work, means FR is more damage than Bash provided it goes full duration on anything up to 19 levels above the pet.

Anyways, maybe this information will help someone or maybe it won't. If nothing else it will help my pet skill calculator be a bit more exact since I can take skill cast timers into account now.
Post edited by Brael - Dreamweaver on

Comments

  • Taranta - Dreamweaver
    Taranta - Dreamweaver Posts: 387 Arc User
    edited October 2009
    Woah...Wall o' text. But, there is a nice little factoid there, for those willing to read. Thanks for the info!
    I shall dominate the Dreamweaver server through the use of Baked goods...and only baked goods.^^[SIGPIC][/SIGPIC]

    Venomancers are Female. Barbarians are male. It will forever be this way. So suck it up and deal with it already.
  • Brael - Dreamweaver
    Brael - Dreamweaver Posts: 1,430 Arc User
    edited October 2009
    Text walls are necessary to explain methods :/