Wow. Disintegration is equal to Lightning strike in damage per hit but it procs more often....
BOW TO YOUR NEW GOD DDO!!!
Yeah Corrosive Salt is similar damage per swing too (believed to be a 2% chance for 6 pulses of 5d10+50 per proc). Only problem is, those two effects do not make good Shroud weapons, as a fair bit of a Shroud weapon's DPS is tied up in the 'Holy' mutation that is not compatible with C-Salt or Disint.
I don't have a zerging problem.
I'm zerging. That's YOUR problem.
This post is for Slicing Winds. It can be found on the new version of Breeze from Abbot, and on the Elemental Rapier of Air from the cannith challenges.
There were a total of 112117 samples. Out of those, there were 2236 procs, for a proc rate of 1.9943%, so most likely 2%. Each proc will deal 3 hits over several seconds, although very occasionally (<1% of the time) it'll give 4 hits. Slicing Winds can proc on glancing blows, subject to the magical proc rate.
The observed damages, taking into account the helpless state of the dummy, were:
The last two entries are not a typo. For some reason, I saw a hit of 361 damage (241 damage after the dummy's helpless state is factored in) and two hits of 375 damage (250 damage after the dummy's helpless state is factored in). No one was around so it wasn't due to some buff or anything. Ignoring those 3 numbers, the average of the rest of the numbers was 124.05, and the standard deviation was 8.17. This fits pretty well with a dice of 8d10+80 (which is average 124 and standard deviation 8.12), so that is the most likely dice. Remember that there are in general 3 hits of that damage on a proc, so I'm writing it as 3*(8d10+80).Code:Damage Count 95 0 96 0 97 1 98 0 99 2 100 1 101 6 102 5 103 10 104 20 105 20 106 22 107 35 108 53 109 76 110 71 111 91 112 135 113 146 114 153 115 162 116 212 117 225 118 246 119 256 120 298 121 318 122 314 123 324 124 326 125 293 126 307 127 319 128 273 129 247 130 247 131 238 132 208 133 184 134 181 135 127 136 121 137 101 138 65 139 74 140 53 141 42 142 26 143 22 144 12 145 8 146 8 147 7 148 6 149 2 150 1 151 1 152 1 153 0 154 1 155 1 156 0 157 0 158 0 159 0 160 0 241 1 250 2
I've also updated the disintegration entry with videos, just in case people feel like looking through it and counting up the samples themselves. Have fun.
Last edited by Vanshilar; 02-02-2012 at 01:18 AM.
Good work as always.
Were you having any lag for the big hits? I ask because they are suspiciously close to 2x the average, it's conceivable (to me anyway) the game interpreted two separate hits as one big hit.
Deception can be found on: Ring of Lies, Treason, Shadow Star, Mirage, Tenebrous Stiletto, Cannith Crafting Shards
Improved Deception can be found on: Epic Midnight Greetings, Epic Phialan Spy Dagger, Upgraded Ring of Lies, Upgraded Treason.
(Yes, I am quite interested in the proc rates of these.)
Hey Vanshilar, keep up the good work here. Also, maybe you would be interested in a thorough testing of double strike?
Vanshilar,
Is this still getting updated by you?
People indicate that the proc rate on the cormyrian weapons differ in some respects. Slicing Wind is an example.
I tested the "Greater Sunburst" effect on the level 24 version of the Mabar Wraps on the training dummy.
The total sample size was 143515 hit attempts, which resulted in 136385 hits (i.e. filtering out the misses). Of those, there were 2728 Greater Sunburst procs, for a proc rate of 2.00%. The 95% confidence interval for this is (1.93%, 2.08%) so it likely really is 2%.
The damage effect on Greater Sunburst occurs slightly after it procs, so occasionally, the dummy will die after the effect procs but before its damage procs. So there are a few less procs that did damage. Anyway, the damage that I got (after factoring in the helpless damage) was:
The training dummy saved on Greater Sunburst 129 out of 2634 times, or 4.9% of the time, so yes there's a save on it. The dummy has very bad saves so naturally it basically only saves when it rolls a 20. I don't really know what the save is on the effect, since I don't think it shows up in the combat log (unless maybe in the pvp pit?).Code:Damage Count 116 1 117 3 118 0 119 3 120 4 121 6 122 6 123 5 124 9 125 13 126 12 127 12 128 9 129 6 130 9 131 4 132 6 133 9 134 5 135 2 136 4 137 0 138 1 139 0 223 0 224 1 225 0 226 0 227 2 228 1 229 3 230 3 231 3 232 4 233 4 234 9 235 13 236 8 237 15 238 18 239 19 240 30 241 32 242 29 243 50 244 45 245 68 246 63 247 56 248 92 249 83 250 98 251 109 252 92 253 92 254 108 255 121 256 102 257 111 258 118 259 106 260 85 261 100 262 88 263 90 264 65 265 61 266 51 267 44 268 41 269 28 270 36 271 28 272 15 273 16 274 13 275 12 276 8 277 9 278 1 279 0 280 3 281 3 282 0
For the hits that weren't saved, the average was 255.21 with a standard deviation of 8.96. This fits a variety of different formulae, but with an observed low value of 224, the most obvious candidate would be 40d5 + 135, although I realize that's sort of non-standard dice. 10d10 + 200 is also a good possibility. [edited due to error]
The damage works on the Abbot, at least on normal. Unlike the Radiant Blast that's on the level 20 version of the Mabar wraps, it does not do any additional damage vs undead compared with non-undead.
Actually, disintegration does less damage per proc, but procs more often. So it ends up having the same average damage output as lightning strike. Given the choice though, I think I'd rather take disintegration because there's lower chance of overkill (i.e. having the damage wasted because you do more damage than needed to kill a monster) and I think there's less things that are immune to disintegration compared with lightning strike.
That's a possibility. I just pointed it out because it's what was in the raw data.
Last edited by Vanshilar; 11-02-2012 at 12:43 PM.
Last edited by Deathdefy; 11-02-2012 at 06:38 AM.
Khyber: Aggrim (Completionist!)
In Von 3 the breakables in the Troll Ambassador optional room are slow to get to and unnecessary for ransack.
Blind insta-kills floating eye balls.
Sorry yeah I meant 40d5 + 135. Obviously 40d5 + 215 wouldn't work because the average for that would be 335. I'll edit that post to clear up any confusion.
What I'm actually doing is, given that the statistics are a mean of 255.21 and a standard deviation of 8.96, I can figure out, for each given dice face amount (d2, d3, d4, etc.), how many of that dice would be needed to match the experimentally determined standard deviation. So for example, for d2 it'd be 320.84d2, for d3 it'd be 120.31d3, etc. The formula can be derived from the formula for the statistics given dice of XdN:
avg = X*(N+1)/2
stdev = [X * (N*N-1)/12]^0.5
From the second formula, the X (number of dice) that would be needed to match the observed standard deviation is:
X = 12 * stdev * stdev / (N*N-1)
I then take the the experimental average and subtract the average from this XdN, to determine what the additive constant would be given this XdN:
C = experimental avg - X*(N+1)/2
From here, given any N, I can give the estimated X and C for XdN + C, given the experimentally determined mean and standard deviation. In this case, it looks like:
Basically, any of these XdN + C would fit the observed statistics. To bound the possibilities, I also assume that:Code:X N C 320.84 2 -226.05 120.32 3 14.58 64.17 4 94.79 40.11 5 134.90 27.50 6 158.96 20.05 7 175.00 15.28 8 186.46 12.03 9 195.05 9.72 10 201.74 8.02 11 207.09 6.73 12 211.46 5.73 13 215.11 4.94 14 218.19 4.30 15 220.84 3.77 16 223.13 3.34 17 225.13 2.98 18 226.90 2.67 19 228.48 2.41 20 229.88 2.19 21 231.15 1.99 22 232.29 1.82 23 233.34 1.67 24 234.29 1.54 25 235.16 1.43 26 235.96 1.32 27 236.70 1.23 28 237.39 1.15 29 238.02 1.07 30 238.62
1) The additive constant C is non-negative (which sets bounds on the lower end of N)
2) The minimum possible in the formula is less than or equal to the minimum observed amount (which sets bounds on the higher end of N)
3) The maximum possible in the formula is greater than or equal to the maximum observed amount (which sets bounds on the higher end of N)
The first assumption would mean in this case that N is at least 3. The second assumption means that N is at most 14, since 4d15 + 221 is 225 but the minimum observed amount is 224. These are highlighted above.
Any of those N fit the data, so I simply give one that seems like a value that a person would pick. For example, generally people tend to pick even numbers or multiples of 5, rather than something like 7d13 + 183. So here, anything like 40d5 + 135, 20d7 + 175, 12d9+195, or even something like 10d10 + 200 would be likely (in fact, 10d10 + 200 looks pretty good, all nice round numbers). It won't be 2d22 + 232 though, nor 321d2 - 226, or at least I hope not.
In general though the specific dice isn't necessarily important, and it's the mean and possibly the standard deviation that matters.
Thanks for pointing out the error. I probably shouldn't try to type these things up late at night.
Last edited by Vanshilar; 11-02-2012 at 11:36 AM.
Wow! That is the most interesting thing I've ever read on these forums.
I had always wondered how the raw numbers from the Lightning Strike data converted into the final 20d20 + 400!
Incidentally, I totally agree with you that 10d10 +200 sounds like a formula the game could use.
Khyber: Aggrim (Completionist!)
In Von 3 the breakables in the Troll Ambassador optional room are slow to get to and unnecessary for ransack.
Blind insta-kills floating eye balls.
Worth noting that this conclusively proves the formula is NOT the commonly-used 200+8d20 (Incineration, Greater Incineration, Cacophony and I think one or two of the rarely seen weaker procs as well).
I would expect 200+10d10 to be the actual formula.
I don't have a zerging problem.
I'm zerging. That's YOUR problem.
Very interesting thread!
I think you've probably considered this angle, but just in case not, I thought I'd mention it: weighted dice.
In DDO, rather than fireball and lightning bolt being 1d6/level, they use weighted dice and are instead 1d3+3/level. I think all spells in DDO use these weighted dice, so rather than 1d4 we have 1d2+2, and rather than 1d8 we have 1d4+4 etc. In general, if I use the notation "wd" rather than "d" to denote weighted dice rather than normal dice, then X weighted dice of Y sides XwdY = X(d(Y/2) + Y/2) = Xd(Y/2) + XY/2. So we can accurately say that fireball and lightning bolt are not 1d6/level but are actually 1wd6/level.
As to the relevance to this thread, if I first define a new variable Y = 2N, then I can write X rolls of weighted dice of Y sides as:
XwdY = X(d(Y/2) + Y/2) = Xd(Y/2) + XY/2 = XdN + XN.
Now, the formula you are fitting your data to is XdN + C. If it so happened that C = XN, then your function would be exactly XwdY = Xwd(2N).
Looking at the results you've determined in your initial post, this appears to often be the case:
(Old) Cloudburst = 10d3 + 30 = 10wd6
Disintegration = 15d20 + 300 = 15wd40
Lightning Strike = 20d20 + 400 = 20wd40
Slicing Winds = 3*(8d10 + 80) = 3*(8wd20)
Unfortunately, even with spells there are exceptions, e.g. magic missile has an extra constant term per level. So it may well be that this way of thinking isn't always useful, e.g. the formula for Incineration doesn't match:
Incineration = 8d20 + 200 = 8wd40 + 40 = 8(wd40 + 5)
If it did happen to be the case for Greater Sunburst, then it looks like as close as you'd get to a match would be 27wd12 which doesn't look particularly "round" (though at least it's a regular D&D die size!). Oh well, this is probably an example of an effect that does have a magic-missile-like constant term.
In any case, thanks for all of the work you've done with this!
PS: A brief aside that again you're probably already aware of: Cloudburst has changed, and is now that nice exploding bubble sound and graphic on vorpal hits (I'm pretty sure only on vorpal, though am not logged in to check).
I'm curious if you've retested cloudburst since the stealth upgrade. DDO wiki attributes the proc rate verification to this thread, but now the damage seems to be higher and added a ~100damage AoE sonic effect. If true to the 9% proc that would put damage around 13.5 single target (+ AoE for the sonic) which would dramatically improve weapons like Couldburst, which would then handily outstrip a Lit II for damage output, losing 7 damage from holy and lose 2pts of base damage (and a bit from shocking blast), but gaining 13.5 from cloudburst + AoE. The improvement is even better against non-evil enemies.
So, I'm very curious to see if the proc rate has remained the same.