Armor and Defense

19 replies [Last post]
Kalaina-Elderfall's picture
Kalaina-Elderfall

I recently conducted a test aimed at determining what effects armor has on damage received, including defensive UVs and trinkets. There have been a few such tests done before, but I was not satisfied with the results. So I conducted my own; my findings are here.

Method

I had come into possession of a Gunslinger hat with Max Pierce defense which I decided to make into a Justifier. Justifier is the defensive equivalent of Skolver. I also have a full set of Nameless, the defensive equivalent of Vog Cub, a Justifier hat with a non-Pierce UV, and a 5* Pierce reduction trinket. This provides me enough flexibility to mix and match armor to hit several different levels of Pierce defense.

Unlike previous tests, testing on Pierce provides me a very controlled source of damage which, combined with an elevator pass, allows me to conduct as many tests as I want for free. That is, spikes. I chose to do the tests in an arena (depth 19), where I have access to both spikes and a healing pad, using an alt account to allow me to leave and join as many times as I want.

Before testing for damage, I had to make sure that spikes deal pure Pierce damage. I did so by equipping two separate pieces of armor with no Pierce defense and vastly differing levels of Normal defense. I took the same amount of damage; as such, I concluded that spikes deal pure Pierce damage.

Findings (Part I)

As I was doing the test, I noticed one thing: the damage which I received when wearing any given armor configuration followed the same pattern every time. Specifically, if I took 4.5 bars of damage on the first hit and 5 bars of damage on all other hits, then after healing to full and taking damage a second time, I would again receive 4.5 on the first hit and 5 on all other hits. From this, I can conclude that damage is nonrandom, much like the damage we deal to enemies is similarly nonrandom.

Secondly, I also noticed that the amount of damage received by any given attack does not follow any logical pattern regarding any rounding of damage to the nearest fourth-bar. Since damage is nonrandom, I can conclude that knights actually have numeric HP, much as monsters do, and that the pips on our health bars are merely a rounded display for easy viewing rather than anything that meaningfully affects damage received. Specifically, they are rounded up to the nearest half-pip.

One further conclusion to be made from this is that when recording damage received from an attack, it's important not to simply include the low and high points (i.e. I received 4.5-5 damage), but rather their relative frequency to each other (i.e. I received 4.5 damage once and 5 damage five times), to more accurately determine how much damage is actually being received.

Results

Nameless                                    4.5-5, 4.5 once, 5 all other times
Trinket                                         4.5
1x Justifier                                   3.5-4, 4 rarely
1x Justifier & Max UV                   3-3.5
2x Justifier                                   2-2.5, 2 once, 2.5 all other times
2x Justifier & Max UV                   1.5-2, 1.5 once, 2 all other times
2x Justifier & Max UV & Trinket     1.5-2, slightly closer to 2

Findings (Part II)

These results are consistent with the displayed Pierce defense bar in my knight's equipment loadout. I therefore conclude that the defense bar is accurate. One thing to note is that my Max Pierce Justifier actually exceeds the length of its own Pierce bar by a few pixels. These pixels are seen in the combined loadout, so it can be assumed that there is no defense cap on any one piece of equipment. Whether there is a combined defense cap in the entire loadout remains to be seen. Method for measuring bars here: http://wiki.spiralknights.com/User:Antistone

Spiral Knights is a video game, and therefore need not be subject to rigorous statistical analysis, as video games produce very deterministic results. All values should fit exactly into the damage formula as it is with no deviation at all beyond that produced by rounding. The rounding to half-bars which is done by the game provides a layer of uncertainty to us as to how much damage we are really receiving, so we cannot precisely say what the damage formula is, but we can get pretty close.

[Removed this portion of analysis as it turned out to be incorrect, read later posts for details]

Monster Defense Guide (At a Glance)

Data types taken from wiki:

Wolver - Normal/Pierce
Chromalisk - Pierce

Jelly/Ice Jelly - Pierce
Minis/Blast Cubes/Rock Cube/Imposto - Normal
Lichen - Normal (tackle), Pierce (spikes)
Polyp - Pierce
Oiler - Pierce
Quicksilver - Pierce (tackle), Elemental (zap)

Gun Puppy - Normal/Elemental
Lumber - Normal/Elemental
Mecha Knight - Normal/Elemental
Retrode - Normal (swipe), Elemental (beam)
Scuttlebot - Normal

Gremlin Melee - Normal
Gremlin Bomber - Normal
Gremlin Scorcher - Elemental

Devilite - Normal/Shadow
Greaver - Shadow
Trojan - Normal

Howlitzer - Shadow
Spookat - Normal/Shadow (bite), Shadow (bullet)
Zombie - Normal/Shadow
Phantom - Shadow
Grimalkin - Normal/Shadow
Slags - Normal

Spikes - Pierce
Elemental Grates - Elemental
Explosion Boxes - Elemental

Further Questions

  1. What percentage of dual-type monster attacks is Normal damage? Does it vary by monster?
  2. What percentage of dual-type monster attacks is Normal damage in tier 2? What about pure-type monster attacks?

    - Hypothesis: All typed damage is converted into 50% normal, 50% type.

  3. How is armor defense scaling done across tiers?
  4. Base damage from attacks increases with each successive depth. How is this done?
  5. Shield defense: how is it done?

    - Hypothesis: Shield defense works the same way armor defense does, including UVs.

  6. How much HP do knights have per bar?
  7. Analysis for Lockdown.
Huhbum's picture
Huhbum
great work

This is a remarkably detailed analysis, thank you.

Hopefully this adds more information to the collaborative efforts of various forumers to unlock the mysteries of SK's game mechanics!

Could you elaborate a bit more on how you came to your conclusions regarding your Damage Formula, though? I'm a little confused about that part(I'm not much of a mathematician, I'm afraid).

Also, that hypothesis about Gremlins and health capsules to determine a more accurate player HP....if that actually worked, that would be something very significant.

Kalaina-Elderfall's picture
Kalaina-Elderfall
My selection of the damage

My selection of the damage formula was somewhat arbitrary. I'm reasonably confident that the form is right (Damage = Base * Coefficient^Defense), because it's simple and yields the kind of results we see ("diminishing returns"), but I haven't done any rigorous analysis to "prove" it or anything.

My selection of the coefficient was just picking numbers to get one that fits with my observations. 11/12 is a clean fraction and it fits. I was sort of just eyeballing the bar values anyway (i.e. 1.5-2, slightly closer to 2 sounds like 1.8 or 1.85). I divide defense by 25 simply because it divides into 25 and 100.

I find that the results from this formula are pretty accurate for defense values from 0-265 regarding Pierce defense against d19 spikes, and the rest just comes from my intuition about video games. At any rate, a formula that yields pretty good results up to 300 (scaled) points of defense is going to be good enough for most people, regardless of whether or not it's the true formula. And unless I'm missing something important, this yields good results.

In short, accuracy isn't my goal so much as describing the trends and providing good rules of thumb.

Kalaina-Elderfall's picture
Kalaina-Elderfall
I want to add the value of

I want to add the value of Heart Pendants here, which is coupled with a description of hits to kill.

In Tier 3, a Penta-Heart Pendant adds 6 HP. Given that you have regular armor (+10 HP) and a +12 Vitapod, you're going to have 27 HP. The Penta-Heart Pendant brings you up to 33. So, if you were taking 3 damage per hit, you'd be at 9 hits to kill without the pendant, and 11 hits to kill with it.

Therefore, in this scenario the value of the pendant is a 22% increase in hits to kill. If, however, you were taking 6 damage per hit, you'd be at 4.5 hits to kill without the pendant, and 5.5 hits to kill with it. Again, that's a 22% increase in hits to kill.

If you have a +15 Vitapod instead, for 3 damage, it's 10 hits without, and 12 with. A 20% increase. So, rather expectedly, the true value of HP decreases the more base HP you have. Which is why Vitapods are so important; without a Vitapod you have 15 HP. For 3 damage, you're dying in 5 hits. With even a +6 Vitapod or one Penta-Heart Pendant, you're already up to 7 hits, which is a 40% increase. +12 is an 80% increase.

Heart Pendants have two other advantages. Firstly, you get HP right off the gun without needing to find a Vitapod. Secondly, adding HP defends against ALL types of damage rather than just one.

Heart Pendants have one notable disadvantage. When using HP instead of defense, your healing is effectively slowed as each heart you recover is a lower percentage of your overall health.

You could argue that the value of armor increases as you get more HP, as well. With 200 points of defense, your damage is reduced by 50%, so that's a 100% increase in hits to kill regardless of what your HP is. But if you have a lot of HP and therefore a higher "base hits to kill," then doubling your hits to kill will provide a bigger advantage (i.e. going from 3 hits to 6 hits is not as big of a deal as going from 6 hits to 12 hits). On the other hand, this is largely situational.

Antistone's picture
Antistone
Not So Fast!

I'm very pleased that my work has inspired others to perform their own experiments! Your data isn't consistent with my previous leading theory (you suggest that defense is about 50% more effective than I did), but truth arises more easily from error than from confusion--that could be because traps are special, or because the presence of poison in my test affected the results, or something else we haven't thought of. This is a very useful experiment--thank you for carrying it out, and for posting your results!

But I'm afraid I have some quibbles with your analysis.

First, you seem to present several assumptions as if they were conclusions. Importantly, you say that adding fixed defense reduces damage by a fixed percentage--what I called "exponential reduction"--but you don't seem to have performed any analysis to rule out other possibilities. Remember that my results were not conclusive on this point!

Secondly, you seem to have glossed over a couple important details in the math. Your data point for a trinket alone gives only one number (4.5) rather than a range--I presume this means that was the only number you observed, but unless you tell us how many consecutive trials you performed, we can't calculate the uncertainty in that number, so we only know that the true value is between 4.0 and 5.0 (not as helpful as it could be). But more importantly, you haven't given any precise formula that is consistent with all of your data!

To quote your own reasoning, "video games produce very deterministic results. All values should fit exactly into the damage formula as it is with no deviation at all beyond that produced by rounding." That means if you measured damage in the range of 1.5-2.0, there is NO possibility that the true damage is really 2.1, or even 2.001, because then we'd never see 1.5 as an experimental value. Our value for the true damage isn't exact, but the endpoints of the range are.

With that in mind, (11/12)^(defense / 28px) doesn't fit your data for any "base damage" value in the interval of [4.5, 5.0]. For example, if we assume a base damage of 4.9, then your formula predicts 3.481 damage for your Justifier test (you got 3.5-4) and 2.144 for your 2 Justifier + Max UV test (you got 1.5-2)--since that's too low a value for one data point and too high a value for another, multiplying by a constant (i.e. changing the base damage) can't fix the problem.

In fact, no exponential formula fits your data. Unless one of your data points is flat-out wrong, or there is some additional variable we have failed to account for, defense CANNOT be exponential!

But if we assume a base damage between 4.96 and 5.00, then the flat damage formula from my original write-up actually DOES fit your data (except for the 4.5 damage for the trinket alone, but that's probably not truly exact, and we're within a quarter-bar on it).

This is actually very informative! Not only did you rule out one of my two theories, you've found new data that directly supports the other one, which greatly increases our confidence. It means we were both probably wrong about defense being exponential, but that was really a philosophical guess, not a conclusion supported by direct evidence.

FLAT DEFENSE takes the lead!

On a side note, I think your analysis of heart pendants is seriously misleading--healing is far too important a factor in Spiral Knights to neglect. For example, if you begin a Vanaduke fight with a +15 vitapod and three ultra health capsules, you are effectively starting that fight with 66 health, not 30; and in a close fight, I'd guess you'll be collecting at least 10 hearts from slags. That puts the penta-heart pendant at more like 8%, not 20% (though the percentage rises if you use a lot of energy revives).

If the flat defense theory is correct, then a defense trinket basically equals a penta-heart pendant if you get hit 18 separate times between full heals (by attacks that are at least partially the corresponding damage type), with the defense trinket becoming better the more times you're hit. I would hazard the appropriate defense trinket is probably better for most "regular" levels (with lots of small fights and few or no opportunities to fully heal), though the penta-heart pendant has a decent chance of being better in arenas and boss fights, and the defense trinket doesn't stand a chance in Lockdown.

Kalaina-Elderfall's picture
Kalaina-Elderfall
Yeah, seems like I was a

Yeah, seems like I was a little hasty in my analysis. The formula I gave is accurate enough for rule of thumbing normal armor values but it's not working out to match everything, particularly the higher values (and doesn't match your higher value data either).

I've nearly got enough tokens for a second Piercing trinket now (and some Pentas) so I'll be happy to perform another test with more precise data logging sometime soon. When doing this test, I was concerned about finding out whether or not damage is random, whether the bars or accurate, whether damage is rounded to the fourth bar, and whether spikes deal pure Piercing. Now that I'm reasonably confident about the answers to those things I'll focus more on the numbers.

The biggest thing to note, I think, is that 2x Justifier seems to reduce damage by 50% here. I'll run the tests on multiple floors to get varying base damages and see if that affects anything. I'm really not seeing how defense could be flat, though, but we'll see.

Also, my guildmate has been nice enough to provide me some gremlin healing numbers. They scale by floor, so I don't know about their utility in discovering knight HP, but they might provide some insight into how damage scales. Don't have much data yet, but I'll post it here.

D5: 54 from mender, 47 from capsule
D7: 70 from mender
D7 Danger Room: 107 from mender
D9: 80 from mender, 80 from capsule
D10: 92 from mender
D11: 103 from mender
D12: 114 from mender, 107 from capsule
D12 Danger Room: 137 from mender

Also, Ancient Plate seems to have roughly 200 scaled points of Normal defense, so it may be a good testing candidate for very high values.

Antistone's picture
Antistone
Scaling

Don't forget about equipment scaling issues--we don't know yet whether 5* armor is fully effective at depth 19.

Suppose that defense is flat, but that damage and defense both scale up by 20% when you go down to depth N. Then your tests at depth N will look just like your current tests except all the numbers will be multiplied by 1.2, which means each armor set will reduce damage by the same percentage as it does in your current tests.

If you want to get more evidence regarding whether defense is flat or not, testing at other depths doesn't really help; you'd need to test different amounts of defense or different amounts of base damage at the SAME depth.

Though if defense DOES scale up at lower depths, that would make defense trinkets more attractive...

Markosmesh's picture
Markosmesh
I read through your findings,

I read through your findings, and just want to comment on one thing:

"Since damage is nonrandom, I can conclude that knights actually have numeric HP, much as monsters do, and that the pips on our health bars are merely a rounded display for easy viewing rather than anything that meaningfully affects damage received. Specifically, they are rounded up to the nearest half-pip."

This is not technically correct (and may affect interpretation of your results). The mods (Nick? cant remember exactly which post and when) have confirmed that each health pip actually has 4 segments to it, however it only visually updates for every half pip lost. So if you went into T1 with full T3 gear, and got hit by something, your health bar would not change, when in fact you lost a quarter pip of health. s four hits in a row would visually sho up as: no change, half pip, no change, half pip.

I only bring this up because of the course of a full health bar, extra quarter pips of health can make a difference. Of course with no visual confirmation, it makes is very hard to measure (or irrelevant if you want :P)

Kalaina-Elderfall's picture
Kalaina-Elderfall
That's not true, both for the

That's not true, both for the reasons I stated in the original post and because I know when I wear a piece of Ancient Plate on Depth 1 I can take about 6 hits before losing half a pip.

The most meaningful numbers will probably be on Stratum 6 if the weapons are any indication, so I'll see what happens there.

Edit: The only arena in Stratum 6 right now is one of the random floors. I guess I'll have to test in FSC.

No-Thanks
Zelda

i dont have any numbers but properly heated up ancient plate set reduces slag guard/horse damage in stratum 6 fsc by alot more than one would expect. as far as i remember the normal dmg monsters in unknown passage from shadow lair still hit me real hard when i was ther in lvl 10 set without any normal uv

actually it seems some1 has the numbers

http://forums.spiralknights.com/en/node/33027#comment-204525

"If I have time I will test Ancient Set vs Skolver + Vog for normal damage and shadow damage.

From trojan:
Full Ancient set: 3.5-4 bars of damage,
Skolver+Vog: 8-9 bars

From zombies:
Full ancient set: 7-8 bars of damage(not counting probability of getting fire status),
Skolver+Vog: 8 bars.

From spikes:
Full ancient set: 8 bars of damage.

So if you always get hit by monsters that do pure normal damage, than it is good, but I won't recommend it in FSC or arenas.

Moreover I died having 16 health bars with Ancient Set when walked into shadow fire(8-9 from elemental 3*2.5 from fire status)"

in my whole SK time, i found this experience to be the most substantial for armors

i hope this helps concluding the defense model(or w.e its called)

the thing is i rarely bothered with armors in this skill based game, but being a tank seems to be the best option at some point in the game

Kalaina-Elderfall's picture
Kalaina-Elderfall
Well looks like I'm going to

Well looks like I'm going to eat my own words here. It's flat.

D27 spikes, which deal pure Piercing (and I verified that again).

Base damage: 7.75
1x Trinket: 7.32
2x Trinket: 6.875
1x Justifier: 6.07
1x Justifier + 1x Trinket: 5.58
1x Justifier + 2x Trinket: 5.167
1x Justifier + Max UV: 5.357
1x Justifier + Max UV + 1x Trinket: 4.929
1x Justifier + Max UV + 2x Trinket: 4.5
2x Justifier: 4.364
2x Justifier + 1x Trinket: 3.944
2x Justifier + 2x Trinket: 3.222
2x Justifier + Max UV: 3.458
2x Justifier + Max UV + 1x Trinket: 2.958
2x Justifier + Max UV + 2x Trinket: 2.682

Using my scaled points (100 for armor, 25 for trinket, 40 for Max UV), that's a pretty consistent 0.43 or so per 25.

Will post full data, methods, and some shield stuff tomorrow. Also going to do further testing on D26. For now, I'm going to bed since this took about an hour and a half.

Etamalgren's picture
Etamalgren
While we're on the subject...

Do you mind testing defense and shield effectiveness?

I only ask because I put on some 1* equipment (_____break helm/armor) with a 4* shield (The shield recipe one you buy from the spiral QM) and ran through strata 1/2, which resulted in my shield breaking after 2 or 3 hits from weak attacks.
Using full 4* armor and shield resulted in the shield breaking after 5 or 6 hits.

Confirmation would be nice.

Kalaina-Elderfall's picture
Kalaina-Elderfall
@No-Thanks That's

@No-Thanks

That's interesting. It looks like the normal component from the zombie attack is being nullified almost entirely by regular armor. Definitely going to have to look into that.

@Etamalgren

That's odd. I'll look into it when I do D26 testing.

-

My data. I tried to get as much HP as possible to get a larger sample size. Then I realized I could just heal a little bit without healing to full and continue the chain where I left off (but it takes forever so I didn't do it much). Parentheses separate points where I healed.

D27 FSC (Spikes)

Proto (Heat lvl 3)
7.5, 8, 7.5, 8

Nameless + Ancient Plate (0 px)
7.5, 8, 7.5, 8, 7.5, (8, 7.5, 8, 7.5, 8, [7.5, 8, 7.5, 8, 7.5, {8, 7.5, 8, 7.5, 8}])

7.75 damage from the spikes, very little possible variance here.

1x Trinket (28 px)
7, 7.5, 7, 7.5, 7.5, (7, 7.5, 7.5, 7, 7.5, [7.5, 7, 7.5, 7.5])

2x Trinket (56 px)
6.5, 7, 7, 7, (7, 6.5, 7, 7, [7, 7, 6.5, 7])

1x Justifier (110 px)
6, 6, 6, 6, 6, 6, 6.5

1x Justifier + 1x Trinket (138 px)
5.5, 5.5, 5.5, 6, 5.5, 5.5

1x Justifier + 2x Trinket (166 px)
5, 5, 5.5, 5, 5.5, 5

1x Justifier + Max UV (156 px)
5, 5.5, 5.5, 5.5, 5.5, 5, 5.5

1x Justifier + Max UV + 1x Trinket (184 px)
4.5, 5, 5, 5, 5, 5, 5

1x Justifier + Max UV + 2x Trinket (212 px)
4.5, 4.5, 4.5, 4.5, 4.5, 4.5

2x Justifier (220 px)
4, 4.5, 4.5, 4.5, 4.5, 4, 4.5, 4.5, 4.5, (4.5, 4)

2x Justifier + 1x Trinket (248 px)
3.5, 4, 4, 4, 4, 4, 4, 4, 4

2x Justifier + 2x Trinket (276 px)
3, 3, 3.5, 3, 3.5, 3, 3.5, 3, 3.5

2x Justifier + Max UV (266 px)
3, 3.5, 3.5, 3.5, 3.5, 3.5, 3.5, 3.5, 3.5, 3.5, 3.5, 3.5

2x Justifier + Max UV + 1x Trinket (294 px)
2.5, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3

2x Justifier + Max UV + 2x Trinket (322 px)
2.5, 2.5, 3, 2.5, 3, 2.5, 2.5, 3, 2.5, 3, 2.5

-

Base damage (0px): 7.75
1x Trinket (28px): 7.32
2x Trinket (56px): 6.875
1x Justifier (110px): 6.07
1x Justifier + 1x Trinket (138px): 5.58
1x Justifier + Max UV (156px): 5.357
1x Justifier + 2x Trinket (166px): 5.167
1x Justifier + Max UV + 1x Trinket (184px): 4.929
1x Justifier + Max UV + 2x Trinket (212px): 4.5
2x Justifier (220px): 4.364
2x Justifier + 1x Trinket (248px): 3.944
2x Justifier + Max UV (266px): 3.458
2x Justifier + 2x Trinket (276px): 3.222
2x Justifier + Max UV + 1x Trinket (294px): 2.958
2x Justifier + Max UV + 2x Trinket (322px): 2.682

~0.0157 damage reduction per pixel.

-

Shields

Tested while wearing armor without any Piercing defense.

Shields: 132 health px

Crest of Almire (Piercing defense 0 px)
2 hits + break (&2.5-3 damage)

Looks to be 20.5 bars on this shield.

Dread Skelly Pierce High (23px)
2 hits + break (&0.5-1 damage)

Royal Jelly Pierce High (143px)
6 hits + break (&1 damage)

Notes

1) Defense is flat. This means that there is no such thing as diminishing returns on defense. In fact, equipping further defense of the same type actually provides increasing returns on hits to kill.
2) Shields appear to use the same formula for damage as armor, with the standard shield health being 20.5 bars (probably 20 for pretty numbers). Shield defense is not divided by two like armor defense is, however, but shield UVs are half as long as armor UVs, indicating that shield UVs and armor UVs provide equal amounts of defense. Not sure how to calculate the split on the breaking hit's damage.

- Spikes vs enemies. I doubt spikes are different, but they could be.
- Floor scaling and tier scaling. No idea how this works. Will test some on D26 for now and maybe do lower tier testing later.
- Damage splits on enemies. How much is normal and how much is the other type? Will test this too.

I have the above Pierce equipment and will soon have full Ancient Plate. I have a friend who has the three sets of armor without Normal defense (Radiant Silvermail etc) and is willing to help out. Feel free to design tests for us.

Antistone's picture
Antistone
Test Proposals

Depth Scaling
The logical next thing is probably just to repeat your current test at a bunch of other depths, so we can try to determine the scaling rules (at least within tier 3). Sounds like you already have plans along those lines.

I think the most interesting depths to test next would be 22 and 24 (immediately above and below the clockwork terminal), if you can manage it--that will give us a wider spread of data within each stratum, and might tell us something interesting about the stratum boundary.

By the way, I think the new gate that activated last night has an arena at depth 28.

Star Scaling
After that, it would be interesting to repeat the test (at some of the same depths) with only FOUR-star gear (if you have any) to see what (if anything) changes. I see several plausible hypotheses.

Split Damage Breakdown
The post by No-Thanks suggests that it is reasonably practical to get enough normal defense to completely negate the normal damage component of a monster that deals split damage (on tier 3). This means we should be able to tell how much of the damage is normal and how much is special.

1. Get a full set of matching armor with no normal defense (I believe the only options for this are the Dragon Scale or Divine sets). Measure the damage dealt by a monster that has split damage targeting neither of your defenses (for example, a zombie or devilite while wearing Dragon Scale, or a wolver while wearing Divine).

If you don't have the right armor, it may be possible to test with another armor and extrapolate the monster's base damage based on your other defense tests.

2. Equip full Ancient Plate, and measure the damage of the same monster (at the same depth) again. Try to count how many red boxes appear (if any) when you take damage.

3. Replace one piece of Ancient Plate with something that has moderate normal defense but no other relevant defense, and measure the damage again. If this matches the damage from test 2 and no red boxes appear, then we can be pretty sure we've blocked all of the normal damage, and so all the remaining damage is special. (If it matches test 2 but red boxes DO appear, then there's probably a minimum normal damage that can't be eliminated.)

4. If possible, repeat with a different monster (or different attack from the same monster) to check whether the split is the same.

5. Would also be nice to repeat this test in tier 2, to see how the results differ, but I suspect you won't be able to get enough defense to completely eliminate either damage type on tier 2 (since the normal component will be larger, and there's no equivalent of Ancient Plate for other defense types). Might still be able to place some reasonably tight bounds on the split with the right armors, though.

Once we've done all of the above tests, it will probably be possible to design some experiments to test damage scaling across different tiers, if you want.

Status resistance tests would also be interesting.

No-Thanks
Zelda

i did shield tests, ill leave the math to u again

lvl 6 ancient plate shield with no UV breaks after X hits at Y boss stratum in depth Z:
27 snarb depth 5
30 snarb depth 6
30 snarb depth 6
29 snarb depth 7

lvl 10 ancient plate shield no UV:
33 snarb depth 5
30 snarb depth 6
29 snarb depth 7

ill repeat the tests at snarbolax boss probably at least once a week(ofcourse with different shields or at least heat levels), whenever i remember it, untill my elevator pass runs out

ive noticed the weapon damage increasing with depth, not only with tier and i was hoping, that this test could contribute to calculating the scaling thingy. keep in mind that the bushes/monsters and their damage could be dependent of depth, stratum, theme, tier, boss/non-boss area, special room, w.e. the tests were only done once at the bushes next to me in the screenshots

i hope this helps concluding the shield defense model(or w.e its called)

Kalaina-Elderfall's picture
Kalaina-Elderfall
Cool, the more data the

Cool, the more data the better.

Haven't gotten around to doing the 26/28 spikes tests yet, but I'll get there.

Also worth noting, my friend recently told me that her Plate Shield takes Rocket Puppy rockets better than Owlite, so it's possible that some monsters with split damage do more normal and some do more typed.

Tralan's picture
Tralan
I'm a bit tired so I don't

I'm a bit tired so I don't know how relevant this is, but in regards to rounding on damage taken, it's not half a pip or nothing. You can easily go on T1, get hit by something, and not take any damage according to the health bar, but still be able to pick up a heart from the ground to heal.

Kalaina-Elderfall's picture
Kalaina-Elderfall
Yeah we went over that

Yeah we went over that somewhere in the huge walls of text. Definitely need to condense this stuff at some point. There's a common misconception that it's "a fourth pip or nothing" but that just isn't the case either.

I suspect that damage on knights and damage on monsters is probably comparable, so health per pip is likely 20, 25, 40, 50, or 100. Something like that, but I haven't been able to find a value that fits my data yet.

Nyan-Cats's picture
Nyan-Cats
OMG!!!1!1!!1!11

OMG!!!1!1!!1!11 YOU NOW NOT GOOD?? SWORD ONE NOW GUNS IS DAMAGE AND NO WEAPONS

Nyan-Cats's picture
Nyan-Cats
ONE HAVEN IS HOME

ONE HAVEN IS HOME YOU KNIGHT IS PLAYER