Page 1 of 1

Possibility of Mob-Specific Evasion Rates

Posted: Mon Oct 03, 2022 12:02 am
by Honua
Were there any additional meaningless (eg 'garbage') numbers for many mobs in the data dump last year used in the codex?

For example, did the <Abyssal> Ahib Condemner (https://bdocodex.com/us/npc/21928/) have a '10' value anywhere near DP/Evasion?

I'm increasingly thinking that certain mobs may have an Evasion Rate stat. This would align with differences in BigAndShiny's PvE hitrate calculator for Oluns & Ash Forest (https://www.reddit.com/r/blackdesertonl ... calculator), both of which should have identical evasions. Oluns/Ash/Crypt seem like the mostly likely candidates, but some others like Bloody Monastery may have it too.

The DRs from the data dump are spot-on, though the DR breakpoint (place where damage starts scaling linearly with AP) is actually above the DR due to quirks of the damage formula. I've been able to solve for the full damage formula on Global Labs and have verified that the codex DRs are accurate, though most Elvia mobs obviously have to be adjusted based on the DP increases from the Elvia rework patchnotes. I would expect Evasions to be similarly accurate, but the standard (0.25*Accuracy - 0.2*Evasion + Rates) formula simply doesn't work with the Evasions in the codex.

This is all speculative, and I'll continue evasion testing on Global Labs regardless. I'm just looking for anything that could explain a large discrepancy in hitrates between mobs with the same Evasion.

Re: Possibility of Mob-Specific Evasion Rates

Posted: Mon Oct 03, 2022 10:21 am
by Kiriak
The only 10s for this mob are limiters for max attack/move/casting speeds.
We can only guess, how the game can adjust the mob's stats. For example, it can apply certain bonuses if mob belongs to a specific mob group, has a title or belongs to a certain tribe. :unk:

Re: Possibility of Mob-Specific Evasion Rates

Posted: Tue Oct 04, 2022 4:51 am
by Honua
Thanks! Just knowing that there doesn't appear to be that 10% Evasion Rate helps a lot. This means something else is at work. It's even possible that the general accuracy formula itself isn't fully correct. I'll keep exploring!