Attack, impact, damage, and perks (Destiny)

by General Vagueness , The Vault of Sass, Monday, December 29, 2014, 01:14 (3859 days ago) @ General Vagueness
edited by General Vagueness, Monday, December 29, 2014, 01:28

(fuckin' numbers, how do they work?)
So I did some more testing.

It turns out the level 8 spawn I was looking to use (in the Forgotten Shore) isn't consistent on top of not producing that many enemies.

I tried my Bandit Mk. 36 at level 6 enemies (Fallen, Moon, Archer's Line): 195 base and 538 precision damage
and at 6Y (knights at Hive entrance in Archer's Line, random reaver vandals): 163 base/244 precision

My Mos Gannon III on group 6 did 166 base/497 precision-- the same as at level 10. Going by the calculation suggested before, the numbers should keep scaling up until level 9 (with a more minor increase for the last level), but it stops at 5 or 6. Applying 7% increases it should be 162.930719 at level 5 and 174.335869 at level 6, which suggests if it's capped in the way suggested, but not at the point suggested, it's capped slightly above level 5.
This implied to me the damage formula proposed was wrong-- more exploration of that is below.

A new gun caught my eye: A Little Voice, rare, requires level 20, 224 attack, 94% impact, I got it as a strike reward. In other words, the same as my Painted Big Chief Mk. 45 before upgrades except for the name, the source, and some stats not related to raw damage. I checked it against groups 2, 4, 18Y, and 20, and found the same damage as my un-upgraded PBC45, and also the the same as my Regulator Mk. 56 for groups 2 and 4. This is good, it implies the same stats for the same type of weapon will do the same damage, barring any individual adjustments like the announced ones applied to most exotics.

I'm pretty sure servitors take less precision damage than regular Fallen and most enemies. Because of this, group 15 inconsistency, yellow-bar enemies seeming to take much less precision damage (looks like the multiplier is 1.5), testing showing a 2.5 multiplier for auto rifles and machine guns, and Bungie announcing they were changing the multiplier for auto rifles at one point, I stopped collecting precision damage during my last round of testing.

The armor piercing rounds perk on my PBC45 put impact at 95%. This came after an attack increase to 232. Together they damaged 18Y at 324/486. With the final attack increase to 242, plus above, level 2 damage was still at 165/493, and 18Y is at 343/514. Also, the first sight unlocked on it increased the reload bar by 1%, and the second sight increased it 2%. This was just barely noticeable, and more noticeable with Chain of Woe.
I guess the takeaway, for this along with the other perk, is that that tiny sliver of increase you see with a lot of upgrades is probably as tiny as it looks.
For damage at 242 ATK/95% IMP it acted like this:
at group 2: 165 base damage
at 4: 188
at 6: 216
at 18Y: 343
at 20: 411

This is interesting because it puts it above my legendary and how my exotic formerly was, albeit with neither fully upgraded.

I upgraded my copy of The Last Word post with Xur, resetting it and giving it 302 attack, with the reset meaning the impact is back to 68%, and I only have the first sight available, which has less impact than the second despite not changing the bar. Afterwards it acted like this:
for group 2: 127 damage
for 4: 145
for 6: 166
for 18Y: 311
for 20: 427

This is significant because it shows once again how much impact matters (PBC45 has overtaken TLW at 18Y), but also that if the enemy level is high enough, attack will end up mattering more. I have never seen two hand cannons with a bigger gap in impact than a fresh TLW and this upgraded PBC45, and yet with enough attack, and tough enough enemies, TLW can still beat it.

I thought I'd test some of the guns against level 1 enemies as a sort of base, like in the thread linked above. I used the Fallen in Dock 13, in old Russia.

MG3: 124 damage
B36: 139
R56: 154
SM41: 139
NB: 139
PBC45: 154
TDYK: 139
TLW: 119

Using these I checked the proposed formula some more.
Multiplying these by 1.07 (to see a 7% increase) gave me numbers fractionally less than the observed level 2 numbers, but continuing to multiply still ran into inconsistencies with the formula proposed before for my Mos Gannon III. For my Bandit Mk. 36, it was more accurate up to level 10, but it passed the observed level 15 value at the estimated scaling for level 11, and it supposedly would be capped a little after level 12. For the Regulator Mk. 56 and Painted Big Chief Mk. 45 it breaks down after level 10. At level 15 the formula gives about 397, which is more accurate than some of the other results it gave, but it's still way off from the observed values, especially with the PBC45 supposedly being capped just before 15.
Unfortunately it looks like the formula from that thread is too inaccurate to be of much use.

I also thought I'd give some proper testing to my Regulator Mk. 56's "kill then reload for extra damage" perk.
For group 1 it went from 154 damage to 205, for group 2 it went from 165 to 219, for group 4 it went from 188 to 251, and for group 20 it went from 346 to 461. If my math is right, that means it gives you one third (1/3) more damage, all with attendant increases in precision damage that seemed to follow the 3x multiplier for the new (1/3 greater) damage value.
I stand by my earlier statement of it not being very practical, at least for this perk showing up on this gun this time-- there's a short window to reload and then a short window where the damage actually applies, I could get a maximum of three shots off during the increased damage window. It might be practical for fighting a boss at medium range with a lot of soft targets surrounding it, or for fighting a lot of soft targets you can almost take down in one or two hits without the damage boost, otherwise, I'd use a different gun or just keep firing instead of stopping to reload every few shots.

The final thing that I would like to do, besides adding even more data, is put all this into a spreadsheet and generate a graph from it. Since I've let this go until late (note to self, don't keep working on stuff like this at 3 in the morning, 4 in the morning comes way too fast), I'll think on that and do it later, unless someone beats me to it. I kind of want to look at other weapon types too, but I'd be tempted to get a bunch that match up in impact or attack even though that might not make the data more useful, and I don't want to have more weapons I'm mostly not going to use hanging around for an extended amount of time, I already have to shuffle things around just to get something out of my vault (although I've been keeping a few like that already since I keep thinking I'll do that testing).