#🍥|anime
1 messages · Page 13 of 1

sure.... buddy.... sureeee...
i mean like, i posted it for research purposes...
Thats what they all say 
He is researching respectfully

No. if it's clearly over the line, self reporting isn't a shield to protect yourself.
If you feel there is a real doubt, contact us in #1010934719455707218 like we explained in the past.
If you post it, you say "i'm trying to follow the rules". So if the image shows clearly it's not the case, self report won't be a magic sauce to let it slip
XD
understood, so unless it's a legit case of checking, then it's okay
soo... do i get banned now?
I mean, you can just delete the pic 
Unless you want to get banned? 
if you are just being stupid and testing water, then
*looks at *@mild tree

when I say we look at it seing it's a self report, I mean, we don't see another user getting offended by your submission. we still see a submission
O I see, das fair then :o
no. but don't go playing on the rules guys. I still check the whole channel every day without reports anyway. But I don't like feeling played either, like anybody 😉 so as long as we stay in good intelligence, all good
thanks

which model/vae 
model is sb250kpl_v6 and vae is kl-f8-anime2
lmao the shirt
beeffy 8 gig edamn

can someone tell me whats written here?
thx
HORROR 
Gives of terminator vibes
HORROR 
cute
c-cute? 
yes
very cute
more CUTE stuff
actually horror

awww
even better


so pretty!
why does this anime look ai upscaled?
Haha
we have ruined ourselves
time to make our own anime SIKE
remove the sike
cute girl with ketchup on face
actually CUTE
but horror
replace the "horror" with "cute"
adorable
loved all this "offset noise" dark series @lyric zealot 🙂 ❤️
also thanks for posting one in #1072015504870494359. the other dreamer communties aren't fed enough, and even one is great
I try to go there from times to times, sorting through my outputs, and finding the right thread to post them in
made a cool thing
so thanks 🙂 and keep up the good waifus, and I may share my ||yandere NSFW LLM|| one day :p
give me!
I made this style lora off this artist, https://www.pixiv.net/en/users/24218478 , you guys think its accurate?
i should definitely explore the other communities more
looks great on it's own, can't say it's super similar tho

attack! leg gone!
@elfin geode @finite herald @wind girder @supple raptor @lyric zealot @cold talon @pliant epoch @queen void @pale monolith @autumn violet @umbral depot https://youtu.be/SPQtMnt0YX8
Hello. Here's another video. Was fun.
Stable Foundation Discord: https://discord.gg/stablediffusion
You can find the folks who made this project possible in the #anime channel!
The following legends made the images in this collage:
👉bWm_nubby
👉Guizmus
👉Eppinette-Chi
👉Trash--Panda
👉Smol Sus
👉SirVeggie
👉Nerfgun3
👉King Pendragon
👉Wpatzz
And yet a...
❤️ (don't mass ping usually people, this is an exception)

How do you guys get such clear eyes? I only get them in close up portraits
4k 60fps 😮
The uncompressed video file is 202GB 
I ping @acoustic agate in retaliation

You're a mad man
Lol

arm sliced off on third. I've merged myself to a include a new style and to get attacks in combat that do damage. Sometimes without anyone there to deal the damage


Thanks 
We kid with the pings trex thank you for putting this together 
I'm back to using holokukiv2 now
well, this model is pretty good, as long as you are using the correct neg prompt
How does one find correct negative prompt
well, i just basically mimic the author of holokukiv2 neg prompt
and edit from there

wanna use Dark Gemini 3 @lyric zealot ? I could release it as this point but I don't like the new TOS over on civitai
so I give it to people I know 😛
Oh okay I see
Oh is it not on civitai? 
are you using hires fix?
What is the new tos?
No i cant, i get a memory error
Hi
How much vram do you have?
try using a lesser upscale by value?
you might be able to use multidiffusion as well for upscaling
I dont recall, on phone right now
They pretty much get to do what they want with your model, including selling it. Even if you don't really want it. It also includes any new versions made from it until you delete them from civitai. Then they can use anything uploaded til that point, but only then do new versions no longer give them full non-exclusive royalty free use
I’ll have a look thanks. Hires fix improves eyes and stuff yea?
Whats that?

Wow! I see
not sure how i think about it tbh
definitely, or at least imo
I don't know if I'll put it on civit, but I will have to delete those I have there if I release this elsewhere
Cool hopefully i can manage to get it working with a lower resolution today
hmm so i'm just use Dreamerland on Android, result on homepage is pretty great but dont know if it's pre-made or sth ? 
not sure if i'm knowledgeable enough to explain that extension.
@supple raptor
if you don't mind, shoot me the file! 
wait for the SENSEI to give you details 
yeah I'll put it on my google drive and give you link
either current merge that I just finished where I managed to get combat damage it, but I dunno how it has affected the rest. Or the previous one

well i will get to testing when I'm in the mood

but then, i need to clear my harddrive first
It's an extension. It has a lot of features, but the one that would probably be most helpful for you is the Tiled VAE feature. It allows the most vram intensive part of image generation to be split into tiles without effecting the quality of the output so you can generate at larger sizes. https://github.com/pkuliyi2015/multidiffusion-upscaler-for-automatic1111.git
just models alone 
soo good! well done everyone 


Thanks i will have a look. Is this better than the other upscalers?
i mean most of arts on homepage have normal hands 
Just models. LoRA is another 84.5GB 
well it's definitely possible 
i just realised im looking at the wrong drive 
at least i still have 349 gb
Wow. You really use all them?
the thing is that, you download all of them, and said i am going to use it one day, but then forgets about it
Hmm
The name of it is a little confusing. It isn't really an upscaler on it's own, but can be used as one in a similar way to how the SD Upscale script works in the img2img tab. When you do this, you still use an actual upscaler such as 4x-AnimeSharp or 4x-fatal-Anime-500000_G.
I searched for many models, and most looked shit to me besides only a couple
I see. So i just generate with the setting turned on, then upscale as I normally do? Am I understanding
well another reason is that, i was trying to do my own merge previously, so i made a few of my own merged at each steps, making up to like 20 models of random merges
Merge?
take 2 models you liked and mix them together, that's commonly what we called as "merged" model
Interesting interesting
well, you do that a bunch of time, at different weight and stuff, then you get "new" model
Is that a difficult process?
no actually, doing random mix is actually just clicking button. Making a good merge is a lot harder and complicated
this tab is basically the most simple way of merging 2 or 3 models together
The Tiled VAE setting is something you can just leave turned on and it will simply not use it when it isn't needed. I find the best process is to generate normally using Hires fix with Tiled VAE turned on and when I want to further upscale the image I send the image to img2img and use both the Tiled Diffusion and Tiled VAE options with these settings. I have 12GB VRAM, you may need to reduce latent tile batch size, latent tile dimensions, and/or encoder tile size if you have less vram.
Interesting! Thankyou for letting me know, i am learning what all these tabs mean
one of the most simple mix that you can do to "improve" whatever model you are using is the offset model
https://www.crosslabs.org/blog/diffusion-with-offset-noise article about that
Thankyou! This is the first time im hearing about using img2img to upscale further as well, I suppose that makes sense
Interesting, I’ll have a read later
thanks for all the help guys
What is your guy’s absolute favourite model?
AAAAAAAAAAAAA AA A. A A A A A A AA A A A A A A A
🅰️
a
soooo cute well done 😿
Thanks 
Hi fruit
Idk if you noticed that I mastered the song. It was chillin around -8 to -10 so I took some creative liberties 
i absolutely did lolll def a turnaround from the soundcloud file cwunch

h-howre u?
I'm good, just chillin while I work on POW
unrelated Miku gen
there are 3 flowers in this pic
Unrelated not miku gen
since you are here, can I direct you to a very cute picture of Miku?
actually ||horror||
awwww look at my lil sleep paralysis baby 
flashes The Mikus™️
I finally finished all the Pixel Pig Mikus
@finite herald I never got your thoughts on my Hatsune Micow submission from the first week of COW.
or at least recreated em as close as i could


Also my POW is done I think, been working on it for like 2 hours tho 
Dunno if to post it now or tryhard even harder tho, I am already tryharding cause I did like, hires fix, upscale, shit ton of inpainting, photoshop editing, clip studio paint editing xD
Pow?
Also I had to use <lora:epiNoiseoffset_v2:1> <lora:LoconLoraOffsetNoise_locon0501:1> <lora:SD15_noise_offset:1> for the character


O
feel the dark gothic love embrace you
she just killed everyone in the chat...shit
a more flawed image
can some1 send a link for this stealth metadata extension? 🙏 cant find in on google (gosh that was a bad typo)
I assume you mean "stealth" https://github.com/ashen-sensored/sd_webui_stealth_pnginfo.git
O dammet I missed the typo
but that only works if you have png meta data on
okok is that in settings?
Discord strips metadata now. The extension hides the metadata in the alpha channel so that discord won't strip it.
Oh I see
hmm perhaps
thank you
and here's a bland standard default waifu!
what is the creature at the right
not loli by far
standard "young" anime girl that could be anywhere between 15 and 20 depending on the anime
yeah that's loli
hopefully not something that will eat her
real waifu
Does anyone know a good model to replicate this art style?
style of goblin slayer or elf archer?
Goblin Slayer
im gonna be a basic noob and say AOM3 or counterfeit
I'd like to make light novel covers
or you can definitely train your own
Anything I should add or change before I submit?
i want to say ceiling is a bit low
Anyone need helping hand?
3 arms means she can hug and head pat at the same time 
thats a loli, certified
It's also cute
Whats loli again?
Pretty much child looking character
It often is. I avoid using loli in prompts because that's how danbooru uses the term, which is the tagging system that we use for basically all of the anime models. female child is the tag intended to be used for non-lewd.
uh, guys, i am worried about the fast fan spinning, not because the fan of my gpu is spinning fast, more my PSU fan.. Suddenly..
Yea I figured
What is danbooru?
I also find, it’s incredibly hard on most anime models to generate a middle-aged woman/man
Even when I specifically say it in the prompt
an anime image board
Image board you could check safebooru if you interested and don't want +18 images to your face right away
I see
Thanks I raised the ceiling and made a couple of other small tweaks before submitting for POW.
frequently hitting oom
should i go find a 3090 somewhere?
10gb
nah, and also using that LLM bot too, which is actually great
--xformers --opt-split-attention?
nah, got all the memory stuff done already
as in, my installation is running about 700mb for the whole os... the rest is for SD xD
what res you do gen at?
drafts 512, usually 640 to 768 and than upscaling in batches
i'm just hating our company gpu's are still in use xD
(we have multiple A100's that i can use if they're not in use)
i dont think you need a100 to get waifus in automatic1111
heard certain vae's blow up vram
He's just generating at a high batch size I bet. He's used to being able to do everything at batch size 8+ on the A100s.
well and they're pretty damn fast too
That too, I just meant in regards to the OoM
now if i could parallelize the inference over the A100's we have...
Bet you can't render native 8k 
with 80gb vram...?

Turn on Tiled VAE and you might be able to 
though, single txt2img to 8k is just asking for horrendous mutations all over the place
just learned how to use the img2img to get the sd upscale done better
instead of using the prompt, using the general bullshit like "masterpiece, highres" just in img2img helps sd upscale a lot
(had to make new preview pictures for updated model xD)
dang bro
prompt 😮
ah, using an embedding or lora?
Got my new drive, got a 4 TB drive 
Gotta love moving 365 GB of files tho
I had like no space in G, I uninstalled DMC5 so I was able to make some space in C, but it was running low xD
And when I'm done moving all the stuff to the new drive I'll just symlink
this POW is tough
I feel like better models have made my prompting worse.
I kinda sucked at prompts to begin with. 😄
a man with his ten children
FEROCIOUS BEASTS!
I'm wondering if the "black cloth" on the right horn is removable or not
I'll try it. That's part of the character so I'm guessing it won't be removable.
no wonder everytime i prompt for oni horns
it has that same black cloth
was wondering if it's a censored thingy or what

right... the horn.
a dinosaur suit -> all ok, get cute chibis.
a lion suit -> i post that shit here, i get banned for life.
model, why u do this?!
Yeah, holokuki is going to lean on Nakiri Ayame for oni horns I bet.
no wonder 
it is removable
yeah. i should go too

Imma go too, buhbye
Too lazy to upscale again so here's the original and with (censored:1.2) added to negative
Well, it’s definitely a character thing built into the holokukiv2
I believed I managed to removed it once, but not sure if it’s rng or not
She's from Hololive so it doesn't surprise me that holokuki would default to her horns when you use oni horns.
True, maybe when I wake up, I’ll try find my old generation without the “black cloth”
And see what I did there
Good night once again!

Guys, any idea to make an image with focus on the face? I've been trying for a while but I always get from the chest up
like that
I prompt, "close-up portrait" at the start to get those type images
Portrait, closeup, and also putting things like mid shot, long shot as negatives can help, as well as putting hands near the face. (Like model shots.) Symmetry can also help.
worked, thanks guys

why do my gens with sb250kpl have bad eyes? asymmetrical, mismatch iris's, lazy eyes?
@quick hill thankyou for all your help once again
@magic wave

im confused
morning 👋

Guys, UniPC or DPM 2M KARRAS ?
Depends on the prompt. I always try DPM++ 2M Karras, DPM++ 2S a, DPM++ SDE, and UniPC on new prompts to see which handles it the best. I've also modified UniPC to use DPM++ 2M Karras instead of DDIM for the second pass when using Hires Fix
I used both but didn't notice any differences.
I should try this when I have time love DPM++ 2M Karras
Did you add the modified version of DPM++ 2M Karras yet?
OC'd my gpu abit and it helped :p
Too lazy 
Probably tomorrow
too tired 
just wanna no brain and prompt moomer
me after 8 edibles
hahahaha
trying out a tokenized gpt primer vs my detailed, im liking the results
cripes i shouldve did adrenelin's auto oc when i first started, actually able to do 1024 hires and not randomly crash during 512x512 lmao
you may as well be speaking russian to me right now
trying 😄
Generating bunch of moomers

Yeee, and me Luna and sis Luna we still here

I remember u guys lol

Finally got around to messing with img2img
goodluck 
mooooooooomer

I believe in my pic, I mean, I hope to at least make it 3rd, its 6400x3200, used like CSP, Photoshop, inpainting, and a bunch of creative ideas 
Black name is just high level members right?
Yee 
Fair XD
Imagine what it was like when we was both green and same name 
XD
i went ooga ooga when i see moomer in autumn bg
Thats very specific :O
look at my pfp
Yep
I'm making vtubers today and I don't even watch any of them. Neuro-sama is the only vtuber I watch
Mumei only one i watch
How to play this
Check #1072220168534642768

so hungry
Big Brother, how to play this
🇹🇭 Happy Songkran Day 🇹🇭 (Thai New Year)
🙏I wish you all happiness and peace of mind. Life is smooth all year round. 💖
You too!

Omg is today songkran???
yes 13-15 apr
hentai model gone wild! of the rails, i say
well... what am i expecting when i'm prompting random stuff from a game xD
Lyrics prompt with "Hotel California"
The image is so crsip
"They stab the beast with their steely knifes, but can't quite kill the beast"
danger-waifu.
While the Labyrinth traps have claimed countless adventurers, they were not designed with felines in mind.
I need to try mirrors, any tips?
Yeah, I'd imagine they are hard to get coherent
🙇 thank you!
mention mirrors in a few ways in the prompt. like mirrors, multiple mirrors, etc.
👍
@south sable looks like your girl 😮
how do i get more color out of this?
what vae are you using?
no thats fine. you could try other vaes or include words like vibrant, etc
does this affect thing?
side cut?

you mean CLIP?
Dutch-Kun
This is what you want for anime. https://huggingface.co/spaces/deepghs/wd14_tagging_online
There's an extension too if you'd rather use that.
You should upgrade your torch version with --reinstall-torch
where do i input that?
In the webui-user.bat after
Commandline_Args=
After its done, remove it from the bat again
wait thats from reddit 
i dont find any args tho
Right click the webui-user.bat and click edit
Then youll see the Commandline_ARGS=
i see it. thx 
There you also can add --xformers --autolaunch
If you finished and removed the reinstall-torch
after commandline_args?
help with this one?
do you have that vae in your vae folder?
i have
Anyway, hi everyone

did you restart?

Afternoon chat, busy day today
Wow I love the colors!
first post might get you bonked 
Great image but yes i think its not allowed
kinda breaking the nsfw rule since she is nude and the cleavage
Thought it would be fine since it's not showing anything 
hey there
first submission on the server, but I got to say, you need to read the #✍🏼|rules-and-tos. this is full nudity, water isn't clothes 😉
I need to remove it

My bad 
great art tho 
Apologies
I did read the rules but I thought it would be fine since it wasn't explicit
Yes, great art
it was really a nice piece tbh but yeah
and welcome to the community! 
@wind girder anyway, yesterday you mentioned about that ||nsfw|| llm, are you doing training yourself or?
Thanks 
Apologies again for the rule break 
no. I'm just using an unrestricted LLM and building a character on it, giving enough examples of questions/answers, and caracterisations so that it behaves as intended.
In particular, ||it has no notion of consent and can just be rougher than I ever imagined||
what model are you using atm if you don't mind me asking 
I'm still just dicovering LLM ^^
for my own research purposes
gpt4-x-alpaca-13b-native-4bit-128g
you can find an anon download of it quite easily with that name
about it yes
Okay let's try this one then 
that's perfectly fine! 
welcome once again! 
Thanks for the warm welcome friends 
I'm not using all optimizations or nothing right now, and it's sitting idle at around 9GB :/
Im not using xformers though
so possibly
I sent a little sample for your eyes to fest on
that's definitely too much for me 
good morning
oh did anyone see that the new controlnet just dropped today? https://github.com/lllyasviel/ControlNet-v1-1-nightly
well, depending on the model, your generation will be favored for certain type of stuff
so yeah

openpose with hands, pose and face 
H A N D S 
moooo

Soon™ 
There is one for specifically conversationshttps://www.youtube.com/watch?v=2hajzPYNo00
I'm literally about to try this

I saw it but didn't have time to watch it yet 🙂 thanks
(I removed the frame, it was saying NSFW in big lol)
well he claimed that it's better than gpt-alpaca in roleplay wise
got to put that to the test 
well, i think it's really about roleplay and not in other aspect
I have no idea how to add new wildcards to dynamic prompts. I've changed the entire document, clicked refresh and it's still only recognizing the standard packages

have you tried restarting webui after doing so?
and just follow the same format as the others should be fine
same format?
they all have different formats
I've created folders like design, composition, items, appearance, time, camera and so on and placed txt files in there from different sources
I've been chatting with my AI waifu (gpt4 alpaca on oobabooga ui) and I think we're ready to get serious 
when is she visiting you?

oh she only thinks of you as a friend afterall huh?
I'll have you know we've cuddled on a number of occasions
I don't want to know how
if we get the bot going in our server you can watch the master at work in realtime 
yup for example .\A1111 Web UI\stable-diffusion-webui\extensions\sd-dynamic-prompts\wildcards\parrotzone\tags

am i like the camera man in that case
with how my chats have been going, you're more like a witness

yeah but do I have to place them in parrotzone?
even if it's not a parrotzone wildcard
artists, devil, jumbo, nai, nsp, parrotzone
did you tried restarting webui after you are done?
I'm getting somewhere now
Yo, you've got options
so hear me out
Their character is very hard to break
no
but if you talk from their perspective, you can make them do things they otherwise would not do.
that's kinda weird, i never really tried that tbh, so I'm just talking about it logically. 
well
but if you make your own character

im writing one rn
slide
I'm not even sure if i can post that on the other server if you know what i meant
gonna slide into your dm 
ah well fuck it, I'll delete it and reinstall it and see what happens
you restarting webui, is it turning it completely off, or using the reload ui tab on settings?
thank you anyway 🙂
I'm gonna make a mechanical alice in wonderland
with my mechanical horror prompt and model
you know which one I mean
and maybe this turns out to be the mad hatter or something
the adorable weshester cat
or not really

hey @lyric zealot
that's like a boss from game
actually cool generation tho
did you use any lora for that 
the heart is so cool! 

Ye sorry, i was on phone so i couldn't remove the embed
I recently learned how : you need to write < and > around your URL, and it removes the preview
<https://www.google/com>
where's your submission for pow? 
O, Guizmoose, I got a question since you here, are "nude" robot girls and slime girls against the rules? 

I'm tryharding 
are you into llm as well
good luck, rooting for you

So far I have big confidence in my submission 
And I mean, is done, but I'll send it on friday so is at the bottom 
I'll be at the top, and you at the bottom 
i mean, the submission post layout
not the ranking 
😢 I don't want to go on the NSFW talk again 😢
Nude bots seems ok, as do slime girls. It will mostly depend on the intent of the character, and the details we get.
Slime girl that is just a green girl in full nudity and details stays as problematic as nude girl
You were real speedy with it tho 
Yeee, I mean that, like just clothe-less, but like just all normal metal or all normal slime 

i mean, the best way to think about is that, robot like that is okay, because there's no feature like ||nipple|| and 🐫. It's not in ||nsfw|| pose as well
True true 
Also how w-
I got a idea, I haven't tried the twins on tis model 
well, i certainly did my research on this 
True, Nora > Twins tho 

why do i have to be stuck in datacenter? 
so close to the A100's, so far from generating waifus
Grab one and replace it with a new one, bam, you get a A100 
well, if the other server wasn't broken down, i'm pretty sure one out of eight would be available xD
sysadmin for "a company"
sorry i won't go into specifics 🙂
anyway, said company has an AI division, and they have a couple (well 8) A100's for creating models
Yeee das fair, and thats pretty dope :O
and when they're not in use, i'm free to abuse them
... i wonder how many $$$ i spend on energy costs that company is paying grin
Im getting the background in now
is this like a character that I don't know about or just random?
so good 
HP why u do dis
RANDOM?!
i want to go home 😦

SD addiction withdrawal kicking in
Same
I feel that
I'm 30 minutes away from my PC and I feel the addiction crawling back
XD
even worse now with vicuna
i've been working on a prompt initialisation that she just becomes a chat waifu with image replies
nearly am getting it to work 🙂
got the shakes, muscles spasming and tightening so hard it feels like your bones are gonna break? sweating? cold sweats? freezing? shivers? shitting water? Vomiting bile?
you missed the violent explosive diarhia
Fair, should try Pygmalion tho, Soosy got good results with it 
True 
shitting water is impressive, but exploding shit is better 😄
sorry, i don't know much about vtuber 
soosy = guismus?
or smal soosy?
Wait, Sus, Soos, Soosy, Susie
dunno how to get the other models into oobabooga yet

and the hallucinations

Is fine, you are now Susie tho
c:
combusting blood boils too...
opiate withdrawal is horror
Smol 
benzo withdrawal is deadly
You could watch Aitrepeneur's video, he did a guide on setting up the 6B pygmalion model, which is pretty nice
i just need to keep my alcohol consumption in check, ... i've given up on trying to restrict caffeine xD
I know there is a way to get 4bit 13B tho, I saw it on reddit at some point

that's the vicuna version i'm running now
Well, 4bit up to 30B I think
well, i'm trying more models to see the comparison 
i think having the right primer/context is pretty important though
but my first experience? this is scary. it feels lifelike at times
Fair fair 
and that's a good and a VERY bad thing
Well, Pygmalion was trained for chatting and roleplay, so it should be one of the best 
i'll have to give it a go then 🙂
GPU and VRAM
lots of it.
Yee 
well @acoustic agate got the best result so far as far as i know 
that's why i'm now seriously reconsidering getting a 3090
Trexdel is crazee tho 
His brain too big
Yo
if only he puts that into bettering the AI, but no, he tries to mindbreak them 
Why not go straight for the 4090 
i'll have to upgrade my psu for that too...
™️
I also consider getting a 3090 and selling my 3070 ti
a 3090 is stretching it in terms of power
I mean, true, but you'll have a 4090 
jesus
I mean, a 4090 here is uhhh...
I really need to upgrade my computer
I can't buy from outside of EU schengen
cause then there's vat, tax, customs toll etc
in total adding maybe 50% in taxes and 100 euro or more in customs
I think its at least 3300 USD here for a 4090 

someone selling GeForce RTX 3090 ROG Strix Gaming White Edition for 800 euro second hand, but 4 days left
i can get a full computer at that price with 4090 in my country
Ye, inflation here go crazy 
i think about 3090 second hand too
would fit my power and financial budget
just need the more vram
wanna buy a 3090 for me and ship it secretly? 😛 I pay a bit extra on top of it
honestly worth
can someone please help me with stable diffusion? for some reason it wont let me launch it
the thing is that, i don't think that company sells gpu only
it's like a company that sells computer as a whole
aftershockpc is the name

true, but i might have to wait a little before upgrading 
what's the problem buddy? 
Everytime i try launch stable diffusion it says "Python was not found"
do you have python installed? 
yes
ah ok
open cmd, from start bar, and type the above
just to check if you have python installed
but i bet the gpu is cheaper here as well



🫶🏻















