KQM | Genshin Impact - aether-weapons-and-builds

Zajef - TF Bennett aficionadoApr 9, 2021 6:46 PM

by the time you have 32 substats, min-maxing is a hobby, not a necessity

if you want to look at 32 substats, that's fine

but it shouldn't be called realistic

PlumeApr 9, 2021 6:47 PM

I see I see

Well it was realistic to me since I pretty much have the build for Albedo

and close for Aether

So how would it go

Hmm going down to

27?

Zajef - TF Bennett aficionadoApr 9, 2021 6:49 PM

27 is pushing it a tiny bit

not as abherrent but a little too much imo

25 is what i generally use

27 isn't too far from it so it should be fine

also, reducing the field time you give should be your other priority

PlumeApr 9, 2021 6:49 PM

By how much

right now it's 1.5s per 8s

Zajef - TF Bennett aficionadoApr 9, 2021 6:50 PM

the limited empirical testing I did yesterday with Bennett, ZL and fischl as supports gave me a field time around 60%(edited)

that was without using ZL Burst

PlumeApr 9, 2021 6:50 PM

Oh that's why

our supports are completely different

Zajef - TF Bennett aficionadoApr 9, 2021 6:51 PM

i saw in your assumptions

you use like

PlumeApr 9, 2021 6:51 PM

I am an Albedo Main, and I use XQ most of the time

Zajef - TF Bennett aficionadoApr 9, 2021 6:52 PM

XL and XQ

but XL should never be used without a proper battery

her damage will be too low for her place on the team to be justified

albedo and XQ are fine, but neither of them take advantage of geo MC's biggest strength

PlumeApr 9, 2021 6:52 PM

That was just an example

I'm not using XL at all

Zajef - TF Bennett aficionadoApr 9, 2021 6:53 PM

which is that a lot of the power budget went into E, so he doesn't NEED as much field time as other carries

while i understand that you can set up scenarios that would lead to 80% field time

That's basically like using range-stance childe

you're gimping yourself

PlumeApr 9, 2021 6:53 PM

That's biased, because that would mean using his burst setup

Zajef - TF Bennett aficionadoApr 9, 2021 6:53 PM

Not exactly

PlumeApr 9, 2021 6:53 PM

I want people to be able to choose

Zajef - TF Bennett aficionadoApr 9, 2021 6:55 PM

you can go a hybrid route

which is what I expect your results to be if you change the field time

2p petra is kinda

cuz it's only 15%

so if damage is split ~50/50, it's less of an increase than gladiator's

I expect the actual damage split in most comps to be something like 55-60 geo, 40-45 phys

which would likely bring us to 4p glad with geo goblet

PlumeApr 9, 2021 6:56 PM

So I did take 6/8th of the current physical attack number

it's still over 50% of the overall DMG

Zajef - TF Bennett aficionadoApr 9, 2021 6:56 PM

hmmm

PlumeApr 9, 2021 6:56 PM

It might not go for full geo DMG, support hybrid route

Zajef - TF Bennett aficionadoApr 9, 2021 6:56 PM

Is this not your motion values?

PlumeApr 9, 2021 6:57 PM

Right, mb, I took the full build numbers

Zajef - TF Bennett aficionadoApr 9, 2021 6:57 PM

yeah, if you have a physical damage goblet, phys damage will be boosted

PlumeApr 9, 2021 6:57 PM

obviously

Zajef - TF Bennett aficionadoApr 9, 2021 6:58 PM

but if you change that goblet to ele DMG, i expect total damage to go up

PlumeApr 9, 2021 6:59 PM

I'll dive into that a bit later, should be over in a day or 2

Zajef - TF Bennett aficionadoApr 9, 2021 6:59 PM

aye no worries

@ me with any changes!

PlumeApr 9, 2021 6:59 PM

Got it

PlumeApr 10, 2021 12:20 PM

@Zajef - TF Bennett aficionado Small question regarding my doc

About the "build list" sheets

Should I remove them? Is it too much info?

The sheets giving example of artifact substat distributions and set used

Zajef - TF Bennett aficionadoApr 10, 2021 12:22 PM

Nah, I think it's good to be able to see how the substats were distributed

I'd probably merge the 2 into one in order not to have too many sheets

PlumeApr 10, 2021 11:14 PM

Going with that

3s switch out every 6s, so 66+% sreen time, with priority on E/Q over switching

Also doubled the "human follow-up" time, from 0.1s to 0.2s

Atm running the maximization program, 23 substats used for the "realistic" part

But I also need to run the perfect optimization parts again because my skill/attack count values changed

So... not before tomorrow I think

PlumeApr 11, 2021 7:10 PM

@Zajef - TF Bennett aficionado Updated

https://docs.google.com/spreadsheets/d/1__g4Q5c4uD5sG2gWzS3IwAkQEU_sO9-ivNALB9_sQmI/edit?skip_itp2_check=true#gid=2067267121

Aether Analysis - Shared

Cover page Document Creator:,Plume Discord:,Liunel#2555 Reviewer/ProofReader:,- Date of creation:,2021.03.08 Version:,1.0 Document status:,To be reviewed Last update:,2021.04.11 GENERAL PURPOSE The purpose of this document is to share a personal structured load of data I've been working/accumul...

TL;DR being what I posted earlier: - from 0.1s to 0.2s reaction time ; - 1.5s per 8s -> 3s per 6s switch out duration ; - 32 Substats -> 23 Substats used for the realistic sheet.(edited)

I also removed the "build list" sheets. I initially put them so that people could verify the calculations, basically giving them every single parameter so that they could check the numbers by themselves... but since no one does that, I think this is just unecessary overloading the document with information.(edited)

Zajef - TF Bennett aficionadoApr 11, 2021 8:02 PM

Assumptions looking much more realistic now, gonna look it over now

Hmmm, it's weird to me to see jade cutter this low

Gonna run the numbers on my end

Zajef - TF Bennett aficionadoApr 11, 2021 8:14 PM

Yeah, running the numbers on my end, I have summit shaper shielded and jade cutter basically neck and neck, at every refinement level

You mind giving me the amount of substat rolls used for both the weapons? Gonna try and figure out if I messed up somewhere or if there's something you forgot(edited)

PlumeApr 11, 2021 8:24 PM

Which analysis

Burst/DPS, realistic/perfect

@Zajef - TF Bennett aficionado

Zajef - TF Bennett aficionadoApr 11, 2021 8:24 PM

realistic is the one i'm comparing to

but i suspect there might be something wrong with all of them, because i don't see how jade cutter can be significantly under shielded shaper

but maybe that's just me being pepega

who knows

PlumeApr 11, 2021 8:39 PM

Only giving R1 since I'm copy pasting. Refinement level shouldn't affect substat distribution as the crit stats are already optimized and completely independant from refinement for these weapons

Zajef - TF Bennett aficionadoApr 11, 2021 9:06 PM

assuming any HP or HP% rolls on your non-optimal rolls for jade cutter?

I think i see the problem

you gave jade cutter a crit rate circlet

so even by maximizing crit damage rolls, you still get REALLY far from 1:2 ratio

@Plume

PlumeApr 11, 2021 9:08 PM

Right

I had to put crit circlet as there's no possible "if" constraint in VBA

Zajef - TF Bennett aficionadoApr 11, 2021 9:08 PM

i see i see

PlumeApr 11, 2021 9:08 PM

Yes that's probably the issue

Zajef - TF Bennett aficionadoApr 11, 2021 9:08 PM

might have to do that one manually then

PlumeApr 11, 2021 9:08 PM

Yes, but np it won't take long

Zajef - TF Bennett aficionadoApr 11, 2021 9:12 PM

my calcs show a perfect distribution of the crit stats you have right now, rather then the initial distribution, creates an ~18% in DPS

wait no i'm pepega

hmmm

the difference isn't that large

PlumeApr 11, 2021 9:13 PM

Better indeed

wait that's R5

Zajef - TF Bennett aficionadoApr 11, 2021 9:13 PM

yeah

hmmmmmm

PlumeApr 11, 2021 9:13 PM

This is weird

Zajef - TF Bennett aficionadoApr 11, 2021 9:13 PM

yeah that doesn't

change that much

PlumeApr 11, 2021 9:13 PM

SS is still better?

Zajef - TF Bennett aficionadoApr 11, 2021 9:13 PM

i'll keep going

PlumeApr 11, 2021 9:13 PM

What do you have on your end?

Zajef - TF Bennett aficionadoApr 11, 2021 9:19 PM

maybe i'll find something else

but yeah that increase makes sense

shoul be ~1.3% increase

what ATK, CR and CD do you get with both?

(this is with your initial values)

wait no

forgot 2p glad

PlumeApr 11, 2021 9:39 PM

Wait there's an issue with my ATK%

No actually there's none

PlumeApr 11, 2021 10:02 PM

For the SS part, just remove 1 CR enhancement and put it into CD

raises the overall DMG a little bit

Also for SS part you forgot the 2nd weapon stat, 49.6 into ATK%

Wait you did not

that's weird

23.5 from substats, 46.6 from sand, 24 from inherent ATK%, 80 from effect, 49.6 from weapon 2nd stat, 18 from 2GF

241.7%

Zajef - TF Bennett aficionadoApr 11, 2021 10:08 PM

wawit

why is that

r5 ss

on the right(edited)

but r1 PJC

on the left

@Plume

PlumeApr 11, 2021 10:11 PM

Oh I doubled SS, crap

PlumeApr 11, 2021 10:27 PM

Ok I'll finish that tomorrow then

The Main DPS realistic analysis is done if you want to compare (light blue one)(edited)

I'll update the other ones tomorrow

if I have the time

Zajef - TF Bennett aficionadoApr 12, 2021 12:27 AM

alright no worries, take your time!

I'd probably take a look at the black sword to make sure crit distribution is good

PlumeApr 12, 2021 9:14 PM

Should be ok?

I quickly changed everything, I hope I didn't mess anything up

Zajef - TF Bennett aficionadoApr 12, 2021 11:38 PM

Hmmmm

aquila looking kinda sus

could it be you're assuming passive proc off-cd?

PlumeApr 13, 2021 8:03 AM

Nop

It seems that both my SS and my AF sub effect got doubled. I checked with all other weapons and they're the only 2 that got that buggy thing

Ima remake the AF value too

later

Zajef - TF Bennett aficionadoApr 13, 2021 11:16 AM

aye no worries

ping me with the results

PlumeApr 15, 2021 10:46 PM

I think it should be good (I hope). Sorry for the late response, I was extremely busy for the past few days.

I remade the calcs for SS/AF/Skyward weapons

And modified some graph titles

PlumeApr 15, 2021 10:57 PM

Actually, I'll have to remake Skyward once more

I've been told it's a pike sword

So I'll have to remake the calcs, I've counted its effect as a phys-bonus-like effect

Zajef - TF Bennett aficionadoApr 15, 2021 11:40 PM

alright, ping me again when it's done

and as i said earlier, no worries, take your time

PlumeApr 16, 2021 10:26 PM

Should be over

I hope

~~pls let it be~~

Good thing the next artifacts aren't useful for GMC

Also haven't checked Alley Flash yet, but tbh it doesn't look good at all on him

I'll add it someday

Zajef - TF Bennett aficionadoApr 17, 2021 8:12 AM

I think I've seen the sheet too many times, I'd rather ask someone else for a new pair of eyes to look at it

denmaxApr 17, 2021 10:58 PM

Since I'm late at the party, I would just like to ask a couple of questions with regards to the builds. - I think the usual obvious one is the switching of CR% circlet to CD% circlet for weapons with a CR%-stat weapon, so I guess I'll get that out of the way - Though I see the total number of substat rolls, I'm trying to look somewhere in the sheet explaining how the substat rolls are to be distributed, but I couldn't find any (maybe I'm just looking at the wrong place

). Bar the first point above, do all the other builds have very similar substat distributions? Are the substat distributiong catered towards the weapon? The reason why this is rather important is because I don't want the whole debacle to repeat about people overvaluing then undervaluing Black Sword for the first few months of the game. Just to keep it short, there was a time where some people didn't really take into account ATK% substat rolls much because of how much it usually stacks anyway, thus some builds completely ignored it in calculations. However, in a weapon that doesn't provide any ATK% stats, each ATK% substat provides much more damage comparatively to an ATK%-stat weapon. Likewise, a weapon with CR% stats are affected much, much less by CR% substats compared to weapon with ATK% - If the distributions are catered towards the weapon, what kind of 'internal rules' did you implement for yourself? Did you try to keep the stats as close to 1:1 as possible? Did you use some sort of source (i.e. one of those reddit posts some time ago saying to maintian an ATK% of around 120% - not saying it's proven, but if you're using that as a rule, why not)? As an example when I was doing weapon comparisons, I have these rules shown below (note, I'm not saying you should follow the same set of rules, nor to implement them, I'm just asking what you are doing with your substat distributions)(edited)

-- Try to maintain a 1:1 ratio between equivalent ATK% (Additional/Base) and Crit value% (product of CR% and CD%) -- 25 substat roll distribution for either ATK%, CR%, CD%, including initial stats. Magnitude of rolls are average -- The same substat cannot re-roll three times in a row in a single part -- At least the substat the contributes the least to balancing the 1:1 ratio should be rolled at least once per part --- To support the above two rules, substats are to be distributed evenly to each part first, and is not done on a per artifact part basis (i.e. I'm not distirbuting stats to the Circlet first, then moving to Goblet - I will provide one substat to Cirlcet, then Goblet, then Sands, etc)

PlumeApr 17, 2021 11:02 PM

Right, since I removed the build list examples, there's no explanation of the "perfect build" stat distributions

denmaxApr 17, 2021 11:03 PM

Correct. I think your idea of ridding the build-list for compressing the document is alright. However, I think it necessitates at least an explanation on your side how the builds were done

PlumeApr 17, 2021 11:05 PM

It's actually a 7/8/8/8/7 distribution (38/45 rolls), distribution made in order to maximize the end output. Left graphs on these sheets are using tier4 stats, right graphs using tier 2

Yes I was just a little forgetful

Anything else I need to add?

denmaxApr 17, 2021 11:07 PM

Alright. How were they distributed further? I.e. in those 7 rolls, which of them is ATK%? CR%? CD%? Is this fixed per weapon, or did you cater it per weapon? If it is catered to the weapon, what logic did you use to distribute those stats like that?

The earlier questions with regards to how those 7 rolls are distributed - you don't need to answer that directly

But I think you get what I mean

PlumeApr 17, 2021 11:12 PM

There's no real internal rule aside from the in-game constraint, aka at least 1 roll into atk and each crit stats, no main = substat, max 6 stats (initial + 5 enhancements), and enhancement sum per artifact equal to 7/8 depending on the artifact. As for their distribution, I just really used a nonlinear GRG method, which will maximize the output with a set error margin (1% for me) in which I added said constraints.(edited)

Rules such as 1:2 CR/CD ratio will automatically be taken into account since they are necessary in order to optimize the result, without any input on my end

So I didn't need to set this kind of rule

denmaxApr 17, 2021 11:14 PM

Ah alright. nonlinear GRG has its limitations, but I think that should be sufficient

Understood - that should be enough

Looking back at your doc(edited)

I guess it was mentioned

Apart from the roll distribution, yeah I think what you have right here now is gucci

PlumeApr 17, 2021 11:16 PM

Well I also made some manual checking after running the optimization. Because part of that 1% error is usually reflected in crit stats distribution. For instance an artifact with 3 rolls into CR and 4 into CD, where 4CR/3CD would have a higher overall input.

It's obviously not perfect, but checking manually was reassuring for me

denmaxApr 17, 2021 11:19 PM

I'll look into this further one more time and see if it's all good

You can close this ticket for now if there's nothing more to add. We'll re-open if we see something

Yuehai Ticketing SecretaryBOTApr 17, 2021 11:20 PM

Ticket Closed by @Plume

Save transcript

Reopen Ticket

Delete Ticket

denmaxApr 17, 2021 11:57 PM

Skimming through it again, I think this is a great doc as it is. But with how large the entry is, I think it would be best to have another editor or intern second the motion for approval. zajef already voiced his approval earlier but I want to confirm if that is still the case now, just to be sure

Yuehai Ticketing SecretaryBOTJun 4, 2021 1:04 AM

Transcript Saving