[X] My Assistant

This was missing in the research thread: data from sssss2, for which I'm very grateful.

If you care about that kind of data, I have some extras to add on:

CODE

Stocking Stuffers (49)
10.2 % [5] Legendary
89.8 % [44] Magnificent

Shrine Fortune (14)
7.1 % [1] Legendary
92.9 % [13] Magnificent

Festival Coupon (72)
1.4 % [1] Peerless
18.1 % [13] Legendary
80.6 % [58] Magnificent

Annoying Gun (36)
5.6 % [2] Peerless
19.4 % [7] Legendary
75.0 % [27] Magnificent

Trophy Tier 5 (81279)
0.0 % [9] Peerless
0.5 % [415] Legendary
5.0 % [4024] Magnificent
47.5 % [38601] Exquisite
47.0 % [38200] Superior
0.0 % [30] 1x Peerless Voucher

Noodly Appendage (4181)
0.1 % [3] Peerless
2.1 % [86] Legendary
20.0 % [836] Magnificent
77.9 % [3255] Exquisite

A while back I discovered something interesting about Random Encounters. They can persist past the dawn!

I've tested it a couple times, and it can allow you to trigger two REs within minutes of each other if you trigger one right before dawn, don't start it, trigger dawn, and then start the first RE (from before dawn) shortly before it expires, and then trigger another RE 30 minutes after dawn.

Not exactly earth shaking discovery, but it's not mentioned in the RE wiki entry. The dawn wiki entry Says "the next encounter cannot be triggered until 30 min. after the dawn event was", but doesn't mention that you can still start REs triggered in the half hour before dawn.

The wiki says "During battle, monsters gain 10%? of MP and 5% SP per active turn."

I tested in battle just now and it seems like on IWBTH difficulty, monsters gain 5% SP per turn (20 Attacks) and 4% SP per turn with Haste active (25 Attacks).

On PFUDOR difficulty, monsters seem to gain roughly 8% SP per turn (12~13 Attacks) and 6% per turn with Haste active (16~17 Attacks). Given my crude testing method it's reasonable to guess, and visually consistent with what I saw in my various trials, that the true gain could have been 8% 7.5% SP per turn.

These tests were performed in isekai at level 344 with 0% Attack Speed Bonus. I used a caveman method of hitting monsters that visually appear to randomly start with close to zero SP. Perhaps it would be better to test by waiting for a monster to execute an SP attack, however I'm not sure they won't regenerate a tiny bit of SP that can't be visually seen during that same turn anyway.

When using Imperil, it seemed that the amount of Imperils needed was roughly double the amount of Attacks, so monster SP regen seems tick-based, as expected. (I think I checked this at PFUDOR and with Haste on).

Question: It seems odd that Haste would only slow down a monster's SP regen by a factor of 1.25, we should expect a slow down factor of 1.5 right? Actually Haste is a multiplying factor of 1.49254 to your speed as I discovered before here but that's trivial and not the reason for this discrepancy. (Also, even if we mistakenly consider and use Haste as being in the "denominator" then that is a mathematical factor of 0.330 which can't be twisted to get the encountered SP performance either).

Perhaps my SP tests are wrong? Anyone else care to check or investigate in some other way, or have an alternate mathematical explanation?

I didn't check monster MP regen but it does appear to be roughly twice as fast as monster SP regen.

(note: This investigation is a precursor to something else, but I don't know if I will ever get around to writing about it.)

Question: It seems odd that Haste would only slow down a monster's SP regen by a factor of 1.25, we should expect a slow down factor of 1.5 right? Actually Haste is a multiplying factor of 1.49254 to your speed as I discovered before here but that's trivial and not the reason for this discrepancy. (Also, even if we mistakenly consider and use Haste as being in the "denominator" then that is a mathematical factor of 0.330 which can't be twisted to get the encountered SP performance either).

Did I really not post about this anywhere?

As of update 0.74 time units taken for an attack or spell are calculated as:

CODE

time_units = minmax(20, 10000 / (100*action_speed_multipliers), 500)

In the simple case of autocast Haste (1.5) with abilities that gives you a time unit cost of 66.666... per regular attack.

(I confirmed this is still the case with Tenboro in PMs when I was trying to work out whether there was a speed factor I hadn't accounted for - it turned out to be the extra 1.25x from spell proficiency.)

I don't think monster regen is normalized around your time units spent at all - just tick based like you mentioned - and I think the numbers you've got are probably pretty close.

That link was useful, but I dont' think it answered my Question: or rather, mysterious problem, that I encountered above. I expect that having Haste on should slow down apparent monster SP regen by a 1.5 factor, since that is what Haste gives. But the SP regen is only slowed by 1.25, which is weird.

I did another test just now to make sure my isekai Haste is working (with full Abilities) and it seemed to increase the duration of my 1H Overwhelming Strikes as expected, from 4 to 6 turns, or a 1.5 factor.

That link was useful, but I dont' think it answered my Question: or rather, mysterious problem, that I encountered above. I expect that having Haste on should slow down apparent monster SP regen by a 1.5 factor, since that is what Haste gives. But the SP regen is only slowed by 1.25, which is weird.

I did another test just now to make sure my isekai Haste is working (with full Abilities) and it seemed to increase the duration of my 1H Overwhelming Strikes as expected, from 4 to 6 turns, or a 1.5 factor.

It may be that a monster's regen is applied whenever the monster gets a turn, so monsters that have any attack speed (ie, PFUDOR +25% and chaos tokens) would seem to be less affected by haste in that manner.

As far as I could tell, all monsters regenerate SP at the exact same percentage rate, so that's probably not the answer. It could still be possible.

In any case, I just did a wild haymaker test and my first guess as to the answer may have been correct. (I already hinted at it above)

I removed my Ability that improves Haste, dropping the improvement of my Haste from 1.5 to the default 1.25, then retested IWBTH both with and without Haste. I got the exact same results as previously, Haste slowed down monster SP regen by a 1.25 factor.

So the answer would seem to be that Haste is partially broken. The Ability that improves Haste, does not improve your speed at all for the purposes of monster MP/SP regen timing. The Ability that improves Haste does work for extending buff/debuff durations.

I don't know for sure that the Ability works for the speed/ticks of your Attacks and enemy Attacks, but I presume it does, or someone would have discovered it by now. Actually now that I think about it, this is not a good assumption. Someone needs to test this. It needs to be tested in a non-1H style, and without any Attack Speed or Casting Speed to mess up the test.

This is creepy because it means there are two timelines for the world. In reality it's probably just bizarre coding. Or maybe the Ability for Haste accidentally affects monsters just as well as yourself.

P.S. - In all these tests that I recently did, I sometimes used Haste on IA and sometimes hardcasted Haste without having it on IA, just to make sure there were no surprises. There were no surprises regarding this, it's always the same.

This morning's tests were helpful.

I tested Scroll of Swiftness (1.6 times boost to action speed and no Ability reliance) and the result was the same as Haste. Enemy SP regen slowed by a 1.25 factor, and there was no observable difference with regular Haste.

I also tested Slow and MagNet (with slowing Ability slotted) and there was no difference to monster SP regen (it took 19 Attacks to fill the SP bar on IWBTH difficulty, it's 1 less Attack because casting Slow/MagNet has a long cast time about equal to 1 Attack).

So everything in my previous post remains confirmed. The additional thing learned is that the 1.25 slowing multiplier to monster MP/SP regen is not because of the Ability being glitched, but rather is probably an intentional mechanism. Apparently a 1.5 times slowing of monster SP regen due to Haste was deemed unfair.

This also means that the Ability improvement to Haste (from 1.25 to 1.5 times action speed) probably works for everything else, such as reducing the number of incoming enemy Attacks relative to each one of your Attacks. Though it wouldn't hurt for someone to confirm it (a script would be ideal for that particular test, and I'm not much of a scripter).

There is another interesting question to test now: How do your Attack Speed and Cast Speed affect monster MP/SP regen? These tests should probably be initially carried out with Haste off. Any mage or DW user would be ideal to perform this. Script is not necessary, just find a random enemy who starts with close to zero SP and count the number of hits it takes to fill his SP bar, like I did.

(note: This investigation is a precursor to something else, but I don't know if I will ever get around to writing about it.)

Although there are still things to investigate above, here's Part 2 of this discussion. I don't know if anyone mentioned this before but TL;DR Banshee weapons are good in the Tower because they will suck out the Spirit of the enemy being targeted. That pesky enemy that you couldn't stick Weaken/Silence on won't be able to use its Spirit attack on you. After more tests, Banshee weapons seem to work as follows:

• Siphon Spirit proc chance works as stated, is not affected by critical hits, and goes up to 30% on a Peerless weapon
• Siphon Spirit proc effect is 0.87% base drain on my Magnificent Rapier, this scales up with player level and equipment quality, like proc duration
• Siphon Spirit proc effect amount does NOT work as stated, in an obvious way
• On my Magnificent Rapier, the log states the amount drained each time is 1~6 SP. It drains as a fixed ratio of what the target monster's currently filled SP is.
• The drain SP written in the log is correct for what is added to your own Spirit bar. Occasionally it's off by 1 SP, so I think actual SP is fractional and rounded in the log.
• I get 1 SP (display log) if the monster has zero SP, all the way to if the monster has a decent amount. I only get 6 SP (display log) if the monster SP is perfectly full.
• The monster loses 2% of its max SP times the number of SP you got. If the log says 6 SP, then the monster loses 12% of its max SP.
• All target monsters seem to be affected the same, regardless of their stats, level, or PL, and including even schoolgirls and Flying Spaghetti Monster.
• Difficulty does not affect any of these Spirit Siphoning amounts
• The amount of damage you do, or activating your own spirit stance, do not affect these Spirit Siphoning amounts
• On IWBTH, siphoning 2 SP is enough to keep a monster's SP bar unchanged that turn. Because it's fractional (?) sometimes 2 SP will let the monster fill slightly.
• On PFUDOR, siphoning 3 SP is enough to keep a monster's SP bar unchanged that turn. Because it's fractional (?) sometimes 3 SP will let the monster fill slightly.
• If your siphon attack is a deathblow, the dead monster will have lost more SP than usual, this is because it won't regenerate its own SP on that turn

I will not say my tests and these results are complete, rigorous, or precisely correct, but perhaps they could be correct to see how it works. I am switching to an Illithid Rapier now. I suspect it will work the same, but because monster MP fills twice as fast and needs half as much to activate (and together with the drain being proportionately less when the monster bar is less full) I think it will not be enough to freeze monsters like Banshee. Even if a monster has already filled its SP bar, all you need is one lucky Banshee strike and you will buy yourself at least two whole turns of no monster Spirit attack.

Just confirming my observations regarding my Illithid weapon before I forget and retire from the season.

• Siphon Magic prof effect is 0.93% base drain on my Legendary Rapier, this scales up with player level and equipment quality
• On my Legendary Rapier, the log states the amount drained each time is 1~8 MP. It drains as a fixed ratio of what the target monster's currently filled MP is.
• My isekai character has about 1530 MP and 1170 SP at around level 350.
• I theorize that the upper maximum of the HP/MP/SP drained could be the stated ~1% percentage times your character's base HP/MP/SP.
• On PFUDOR, even siphoning 8 MP will not reduce a monster's MP bar, it still wants to fill slightly.
• On IWBTH, siphoning 8 MP will reduce a monster's completely filled MP bar just a little.

The performance with MP drain was even slightly worse than the previously hypothesized poor result, for unknown reasons. I probably don't have the general method to calculate the monster HP/MP/SP loss exactly right. We know that they lose an order of magnitude, roughly ten times more than we'd expect. It's amazingly helpful on Banshee weapons to silence monsters, but not enough on Illithid weapons even though it works the same. Vampire weapons might be less laughable than people realize, if the HP drain to monsters is major. This would be difficult to test because we can't see the amount of HP sucked out of monsters' health bars, and we can't trust the log as it only states the HP that we gain.

It's also possible that your other stats might somehow affect the HP/MP/SP loss to monsters. What if proficiencies, Demonic weapons, or your Intelligence and Wisdom stats, increase the damage of Vampire weapons? (similar to the Drain spell)

Theoretical Analysis of Bleeding Wound versus Penetrated Armor, both pre-update and post-update

(I decided to move this from the isekai Update 91 thread, to avoid cluttering it)



Not sure if we experimentally tested or calculated this before, but I'll begin with a beating down Schoolgirls race. Apparently I once tested that Schoolgirls have almost the same PMit as Elementals. They don't get chaos upgrades, which makes a big difference. I'll estimate a level 500 Schoolgirl has 66.6% PMit.

PA x1 would do (1 - 50%) / (1 - 66.6%) ≈ 1.5x damage
PA x2 or Imperil would do 2x damage
PA x3 would do 2.5x damage
PA x3 plus Imperil would do 2.75x damage

BW x1 adds 8% adb per turn that ignores mitigation, or 3 * 8% / 2.434 = 0.099x damage per turn, taking into account elemental strikes and averaged crit
BW x5 adds 0.493x damage per turn ≈ 1.5x damage per turn
BW x5 plus Imperil would do 2.493x damage per turn

Conclusion, it is an extremely close contest! Even closer if you consider that Peerless Shortsword of Slaughter warrior has about 4.3% more adb than Peerless Rapier of Slaughter warrior. And note that is for single targeting a Schoolgirl. If you fought 7 Schoolgirls at once and spreaded them, old BW x1 adds 56% adb that ignores mitigation per spread cycle, thus doing about 1.7x damage, in other words stronger than Penetrated Armor. And that is without Imperil.


For reference, I redo the above calculation and comparison for normal 80% PMit monsters in the old Persistent version of the game.

PA x1 would do (1 - 60%) / (1 - 80%) = 2x damage
PA x2 or Imperil would do 3x damage
PA x3 would do 4x damage
PA x3 plus Imperil would do 4.5x damage

BW x1 adds 8% adb per turn that ignores mitigation, or 5 * 8% / 2.434 = 0.164x damage per turn, taking into account elemental strikes and averaged crit
BW x5 adds 0.822x damage per turn ≈ 1.8x damage per turn
BW x5 plus Imperil would do 3.822x damage per turn ≈ 4x damage if also adjusting for Shortsword higher adb

That is for single targeting in general. For spread targeting, let's say swarms of 7, old BW gets 7 times stronger, so old BW x1 does about 2.15x damage, which is marginally stronger than Penetrated Armor. And this is not considering Imperil. It should also be noted that Bleeding Wound takes effect 1 turn (or entire spread cycle) earlier than Penetrated Armor, which may do absolutely nothing if it wears off before you come back.

Conclusion, old BW versus old PA was properly balanced. I've always said it has been a misconception that Bleeding Wound was weak, and that it even needed Imperil. In practice, players did find that 1H rapier was faster than shortsword without Imperil usage, but if I recall the tested results were not as different as traditional belief held. (That is the reverse of what I just claimed for spread targeting, so what was the difference? Counter attacks would be my guess. They are only improved by PA).


The tremendously increased 6x power of new isekai BW is still worrisome. Let's revisit the calculations in isekai, starting with schoolgirls:

PA x3 is still 2.5x damage
PA x3 plus Imperil is still 2.75x damage
PA x4 is 3x damage against schoolgirl, when you get fresh proc which may be over half the time at high level

BW x1 adds 50% adb that ignores mitigation, or 3 * 50% / 2.5 = 0.6x damage
BW pseudo-maxed I'll say is 165%, so adds 3 * 165% / 2.5 ≈ 2x damage, 3x total, actual cap is 200% but hard to reach
BW pseudo-maxed plus Imperil is thus 4x damage


Against regular monsters in the new version:

PA x4 is 5x damage

BW x1 adds 50% adb that ignores mitigation, or 5 * 50% / 2.5 ≈ 1x damage, our same prior result, note a more generic crit portion because the system is reworked
BW pseudo-maxed adds 5 * 165% / 2.5 ≈ 3.3x damage, for 4.3x total
BW pseudo-maxed plus Imperil is thus 6.3x damage

We've already said that one stack of new BW grants 2x damage for single-targeting. For spreading, one BW stack adds 50% + 40% + 32% + 25.6% + 20.48% + 16.384% + 13.1072% ≈ 197.57%, which is 5x damage. That's exactly the same as one broken new PA hit. Hmm, I guess that's fair! New BW may have a lead in the Schoolgirl race, but that's where I did the most fudging (the 165% came out of my ass, and there is another mathematical flaw in my general approach).

I still think the following tweaks might help to balance new BW versus new PA better.

1) Change Imperil to stack additively/subtractively with Penetrated Armor. Currently, PA loses half its normal efficacy when stacked with Imperil.
2) After doing that, perhaps we can lower the first turn PA proc to half mitigation, as Tenboro suggested earlier. As it would still hit 0% PMit when stacked with Imperil.
3) Fiddle with and lower some of BW's 200%/50%/40%/20% numbers to once again try to make PA versus BW perform the same in the 3 major situations: single-targeting Schoolgirls and/or Tower 200 monsters, single-targeting regular monsters, spreading regular monsters (mostly in theory, as in practice rapier will just OHKO everything).

It's fine to forget all that, the balance is decent now, except that PA is ultra bombastic while BW may actually still be stronger. Tenboro (or Nezu) probably have some way to do sandbox simulations against Schoolgirls to get the real answer.