Ailment and Debuff Formulas

[leaf]

Except... I'm pretty sure I've had eclipse drop dullahan's atk even without that, with only one set of jupiter djinn in the party. In fact, from what I remember, it's a fairly consistent drop.

Are we sure that we're not missing something here?

[Salanewt]

Wooh, I knew there was a way to lower his attack! Thanks for pointing out the Jupiter djinn on Jupiter adept thing!

It's hard to say really. Do you remember other summons landing their effects more often? Since, I mean, if strategies against Dullahan that use Eclipse are as common as I think, then this formula wouldn't really make a whole lot of sense in that context.


[spoiler=Theoretical Rebalance]So uh, yeah. I get the feeling that, if one were to divide luck by 4 rather than 2 and buff the right luck values a bit, the ailment formula would be pretty balanced. Ability effects would be more useful without actually changing a whole lot, and super bosses like Dullahan could be given higher luck values to lower the impact of debuffs and ailments slightly. In fact, you could go a step further and increase the amount of luck needed for the poison/curse/death ailment immunity without worrying a whole lot about breaking them.


Example:
(((((((14 - 10) - Floor(96 / 4)) * 3) + 60 + (25)) * 100) / 100)  >= rnd())
= ((4 - 24) * 3) + 85 >= rnd()
= -60 + 85 >= rnd()
= 25% success rate.[/spoiler]

[Aile~♥]

It's possible that Eclipse's secondary effect chance is hardcoded to be different, or that the Vulnerability value is different for summons. Or that we have some parentheses in the wrong places. Or that secondary effect rate diffusion (the unk_arg value) is different depending on damage type, and may be set to something like 120, 125, or even 150 for the primary target of a summon.

It's also possible that the Vulnerability changes the multiplier after (eMod - Luck / 2) is calculated, so instead of a *3 it might be a *2.5 or even *2. I don't know anything about this stuff, so you'd have to ask Teawater.

@Sala: Dividing Luck by 4 seems a bit much, as you just demonstrated. 96 Luck and you still have a 25% chance to ailment him. Yes, it's a vulnerability, but that's still 96 freakin' Luck! Maybe divide it by 3?

[Salanewt]

I don't know, either or could probably work. The downside to dividing by three is that keeping the same luck rate as the previous incarnation drops the rate down to 1%. Plus it is Dullahan.

Lowering luck to about 75 (which seems like a fair value for a bonus boss), we get:

[spoiler=/3](((((((14 - 10) - Floor(75 / 3)) * 3) + 60 + (25)) * 100) / 100)  >= rnd())
= ((4 - 25) * 3) + 85 >= rnd()
= -63 + 85 >= rnd()
= 22% success rate.[/spoiler]

Which isn't terrible, but then it still drops substantially if you don't stockpile all of your djinn onto your mono-elemental adept either way. If assuming vanilla classes and you go into Pure Mage, however:

[spoiler=/3 Vanilla Pure Mage](((((((5 - 10) - Floor(75 / 3)) * 3) + 60 + (25)) * 100) / 100)  >= rnd())
= ((-5 - 25) * 3) + 85 >= rnd()
= -9 + 85 >= rnd()
= -5% success rate.[/spoiler]

Then we're more or less at the same point as before, just with typically higher rates for common enemies. If we redo the /4 formula for Pure Mage (with a lower luck value):

[spoiler=/4 Vanilla Pure Mage](((((((5 - 10) - Floor(72 / 4)) * 3) + 60 + (25)) * 100) / 100)  >= rnd())
= ((-5 - 18) * 3) + 85 >= rnd()
= -69 + 85 >= rnd()
= 16% success rate.[/spoiler]

Then using Dull on Dullahan actually becomes possible, as would delude and whatnot.  Plus, if it weren't for that +25% vulnerability, then all four of these examples would fail anyway.

[Daddy Poi's Oily Gorillas]

It looks like there's a part of the code where if defender's luck is 40+, about 8 of the effects are given a chance of 0%. (For example, x18 and x19 (May Inflict Poison/Venom)

Q: I saw you sleep-lock the Avimander and King Scorpion, I thought you couldn't do that?
A: Aha! The GS wikis will have you believe that 40 luck makes bosses immune to status conditions (and thus only the Chest Beaters and Moapa's knights are able to be afflicted with status conditions) but this isn't quite correct. The 40 luck = immunity seems to hold in the original golden sun, but in GS2 an amazingly helpful guy by the name of VanishMantle (http://www.twitch.tv/vanishmantle) discovered that there's actually separate status vulnerability portions of code for each of the bosses in the game. What this means is that bosses can be weak to certain status conditions or be susceptible to status conditions despite having 40 luck.

A great example of this is Moapa who has 40 luck, his knights which have 37 and the Avimander who has 41. It isn't hard to work out that Moapa is immune to all status conditions, while his knights (with 3 less luck) are rather prone to stuns and sleep. Very clearly, the Avimander is susceptible to sleep as well despite having the highest luck rating of all three. This illustrates that status vulnerability has other factors than just luck (otherwise Avimander would be immune) and the discovery that VanishMantle made.

^So I was browsing the pastebin, saw this post, and remembered that I never documented the first part of that function, did I? Maybe we should go and list the exceptions when the 40 luck cap/formula isn't used.

080B0782 = 100% chance to Ignore Defense for enemies in 080C6B18
080B06C8 = 100% chance to Trident for enemies in 080C6B76
080B07A8 = 0% chance if effect revives and you have HP.
080B07BE = 0% chance to cure poison if you're not poisoned. (Status effect 0x3)
Branch to 080B0808.
----
080B07D6 = 0% if not infected with Stun, Sleep, Psy-seal, or  ( 02000B91 = Turns until downed. ; Death Curse) (If any are 0, branches to 080B0832)
080B0808 = 0% if not infected with Delusion, (?), (?) (Status effect 0x4) (Will branch to above if all 0)

080B0832 = 0% if not infected with effects that are for status effect 64. (0x40 ; "Remove all status ailments")
080B08AC = 0% if ( 028=May inflict Death Curse) is 1 turn...

And then after that... is where the luck cap comes in.


---
It could be that this function isn't called for some bosses, though? (Either that, or it is called, but the boss weaknesses stuff could just be inserts in the damage function?) ; I'll need to do some research on the damage function, most likely. And I shall list them here, I guess.

[Plexa]

Very interesting, I look forward to your result. Unfortunately I dont have any experience with the Aqua Hydra as he's a skippable boss :) I strongly suspect that all bosses in the Eastern Sea aside from Serpent/Poseidon can be afflicted by status conditions, or at least some status conditions.

[Daddy Poi's Oily Gorillas]

Before I do a more thorough scan, though, can we be 100% sure that VanishMantle was correct about there really being separate sections of code for boss vulnerabilities?

The eight effects that get a 0% chance when the target has 40+ luck were these:

(Number in decimal so people can look them up in Atrius's Editor.)
18 = May inflict Poison
19 = May inflict Venom
26 = May inflict Haunt
27 = May inflict Death
28 = May inflict Death Curse
34 = May reduce HP to 1
80 = May inflict Death Curse
82 = May force target out of battle

Sleep and other status ailments/effects do not use the 40 luck cap.

You can set up to three status effects to be more vulnerable in the enemy editor of Atrius's Editor. (The ones labeled as unknown, but the fourth "unknown" in that row isn't used.) Avimander has one for Sleep, so he's 25% more vulnerable to sleep,

[VanishMantle]

Hey just figured I would chime in about this. It is something I noticed more with TLA and balancing things for a difficulty patch, Risen Star, that I was working on. I noticed that a number of bosses no matter what I set their luck value to would still get hit with a status once inflicted they seemed like others hit them more easily. My original thought is that in TLA something else must factor into  what makes something immune or not. Thus the there must be something else that controls it. Which was my "there could be something else that controls it" I did not mean for it to a be a defined but that there could be since I could not explain why it was so.

My theory from hours of testing and scanning was that even if something is at "0" there is no pure zero and the boss could still be hit with a status affect. Again all pure conjecture on my part with hours of testing and such. Along with hours of frustration as playtesters destroyed bosses in my mod.

Again I wasn't sure because on bosses that should have been immune to poison I somehow could poison or could hit with other ones that should be a 0% chance.

Chances are I was mistaken about the formulas and code

[Daddy Poi's Oily Gorillas]

Interesting. You can poison bosses that have 40+ luck?

Definitely worth looking into... Might be faster if someone provided a savestate of where a status effect would work when it is not supposed to. (Preferably before the "attack".)

[VanishMantle]

I sadly don't have any save states but I know a lot of my play testers did some rather odd things to bosses. Again I am not sure if it was just a patch issue, or how the game handles RNG. IE if it stores the chance of success in a specific number then it looks to see if it matches that but again I am not 100% sure.

I just know it was a nightmare trying to make TLA difficult with the bosses without the status affects screwing with stuff. If it can happen I would say it will be a while for me to replicate in any sort of way that could be useful.

As stated though I could just be derping super hard and not have checked the boss's Luck was at 40+ or whatever.

[Plexa]

In my experience, in TLA I haven't been able to poison anything with 40+ luck (then again there are surprisingly few things that can inflict poison in the speedrun route). Nor have I been able to inflict instant death, which I've done a lot of in GS1.