#🍥|anime
1 messages · Page 38 of 1
wait how does your UI look like this
your doesnt look like that?
no
well mine looks like that from webui update and extensions
this is what mine looks like
did you get an extension that adds those 3 other bars at the top?
or is my UI just outdated
it said it was up to date when it loaded 
did you do git pull?
either that or editing in settings
is that your settings too?
whats the one for hypernetwork
lemme check
sd_model_checkpoint, sd_hypernetwork, sd_hypernetwork_strength, CLIP_stop_at_last_layers, sd_vae
this is mine
@stone onyx can you try this prompt in your style
best quality, solo, concept art, (1girl:1.3), reference art, vtuber, vtuber character sheet, best quality, face details, eyes details, highly detailed, (yellow eyes), tall, tall legs, standing, long high ponytail, (dark green hair:1.2), (sleeveless kong fu shirt:1.2), black pants, fingerless gloves, gorgeous, beautiful, chill eyes, (dragon horns, dragon tail), neck fur, hair between eyes, good anatomy, (fur around neck), gorgeous, smile, (buff arms), (buff legs), solo, 1girl, slender, cloth around waist, smile, very long hair, full details, highly detailed face, wide eyes, very long hair, (full body), from front, white background
i am trying to get results like this
but no good results so far
Can I try? 🙂
my pleasure
Seems that control net can help
no helping my friend make a live2d for his oc
My latest style obsession.
im getting this with no calibration
thats mega cute
Currently deleting about 200gb of models
lol thats cute
This is awesome
I have so many duplicates thanks to civitai helper
can you screenshot the generation page
could be better, i just opened a new site and copyd in
I'll dm you the meta data if that's fine
sure
could you send me too?
yessir 

lol cute style
lol revamp
from 750gb to 477gb
pink works surprisingly well, also i like the dragon logo lol
yeah I'm surprised as well, pink is very clean for this kinda prompt
dragon logo is kinda crazy 
Yummy
its incredible
ey not bad
btw its your positive prompts
yea my oc lol
i´m using revAnimated as model btw aslong they are same settings then Strix sended also to you
i love it
strix? not sure what you mean by that
I should go back to working on that Visual Novel engine I made...
ya'll need something to do with these images 😛
thats what i mean
i used same meta data like she sended you
i just use the revAnimated model
thats the only difference
did you use Dishwasher1910 Style too?
Hmm
that hair is whack in tile 1
what model is that

AlfaHentai merged with AOM3 and something I don't remember lol
i see
@simple mango
what model is that, really clean 👀
AlfaHentai merged with some other models by me
It is a model that works very well with "storytelling" prompts
You can even use lyrics and it will output something decent
How is merging working
do you have any tut for merging?
Well, just find some models you like, and try mixing them like making a cocktail
and how i mixing them in 1 runtime
what do you mean
i can just choose one model or not?
Yes, why not
bc you said you using alf and some other merged models
you do you run multiple models in once
No you don't
i´m completely confused now
Me too
@scarlet kelp
You do multiple merges
Yes
thank you now i understand for what this tab is
You can keep merging more and more on top on one model one by one
how is that working, like if you done merge once you get on package with the data of all before?
Lyrics prompt is so fun
IF you merge model A and B with weighted sum of 0.5 then you get model C that is 50% of model A and 50% of model B.
Then you can merge model C with model D and now you have merged 3 models into one
ohhhhhhh myyyyyyyyy thank you thats awesome and help me alot
Gotta run to the buss
uii awesome

multidiffusion upscale is nice for a quick fix
considering this is where we started lol
Just a heads up, model C (expmixLine_v2) isn't being used in this screenshot at all. Weighted Sum only merges models A and B. Model C is only used for added Added Difference merging and in that case, it is subtracted from model B before model B is added to model A.
Right, I have no idea what I am really doing when merging lol
I use supermerger mashing things at 0.75 and .5 when mixing more than 2 models
If you look right above where you enter the models, it gives you the formula for the merging method you have selected.
same seed
left is green hair, right is dark green hair
@stone onyx the second try and i´m stuck again owo
saved
its dark green but a bit lighter than that
i think its stuck again but why
what are your hires settings?
queen of hearts
omg he is still progressing XD
i´m now on 53%
just 2 more hours i got the picture xD
weird.... what's ur gpu?
2070
redo of an earlier pic
how many vram

kek jthey took down the announcment
that should be fine
idk then honestly, try to upscale via img2img using ultimate SD upscale see how that goes for u
ram or vram?
maybe i should restart my pc
vram, my ram is 64 gb xD
something like this, see how that goes for u
Hmmm
Does it get stuck and like go into infinite numbers? 
what is ultimate sd scale? i dont have that
he was just super slow
something happened he is really fast now again
idk what i´ve done
upscaling extension 👀
https://github.com/Coyote-A/ultimate-upscale-for-automatic1111
ree this loopback + superimpose script i found on rentry isnt working
glad you liked her lol
really cool prompt ngl 
masterpiece, best quality, ultra-detailed, illustration, 1girl, solo, outdoors, camping, night, mountains, nature, stars, moon, bonfire, tent, anime artstyle, danbooru, artstation
Deepfloyd
yea its specific
so good
once Im out of the hospital I'm going to make a few lora's to merge into my model to make side-releases. Like maybe dark fantasy demon/angel war, Apocalyptic Dystopian "Skynet" thing, a chibi model that makes everything chibi, everything. Even buildings
i really liek your style
but I hate hospitals, hate em, I want outa here so Im gonna work through the pain, go poop cause they really want that. Get the pump taken out, get a doctor to come alive to do something.
and the pain, and unable to eat cause they had to take some small intestine with the tumor
ohh sounds exhausting, get well soon!
reminds me of mori calliope lol
nice i like the purple
now I cannot unsee it

first two colors are wrong 
reminds me of Sagiri from Hell's Paradise
i've literally just been shoving the names of doctor who episodes into the prompt
this one was "the eaters of light"
this one was "the crimson horror"
i'm doing batch count of 4 with hires fix on 2, so it takes like 8 minutes to get my results
i also can't do hires fix with an upscale of 2 on any size other than 512x512
not even 512x768
could be vram limitation
yeah i get a cuda ouf of memory error
how much vram do you have
8 GB
I assume less than 8
try --medvram in command arguments, could help
MIX Pro v4.5 + ColorBox
your mix?
nope, it's on civit
even on 512x512, the vram estimator says i'm using more vram than i have

mmmm i wonder if i should add colorful to my model, thats rlly nice 
with the same settings but at 512x768 it chomps all my vram 
what's that 
does generating an image and then upscaling it with the extras tab use less vram?
i notice that when you use hires fix even with upscale set to 1, it still noticeably improves the image
oh I take it back, extras tab sucks poo poo
i normally use hires fix set to 1.5
colorful? a model i saw a vid last night and seeing colorbox made me think of it, ill see if i can find and ill link
img2img with ultimate SD upscale is the way to go 
i don't use the image2image tab because idk how 
for now enjoy dis
might look a bit overwhelming but you get used to it super quick
@nova remnant try it out see if it works
and/or --medvram argument
see which works

bagoure
BAGMURB
Man
jesus, that face
LOL
haha since you got remembered
The first one is just showing Ariel's reaction when she met the talking crab for the first time.
did you check the enable box
try increasing the control weight
its happening again @stone onyx

yeah i guess restart pc is good option too
Aaaaaaaa
First attempt at AI anime filter video, I saw a video recently where it looked way more anime and less realistic but I have no idea how they did it. The closer you get to anime the less consistent it becomes between frames. Can't find any instructional videos that don't feature a realistic cartoon/anime aesthetic.
Also trial and error for this process is hell because at 20fps for a 3 second video takes over an hour to generate
1080 ti
I saw a video recently like this and thought it was neat. Would like to learn how to do it.
let me see if I can find it
this is it... https://www.youtube.com/watch?v=q74qxAYhUHM
Transform real person to anime character using stable diffusion and multiControlNet
#aiart #animeedits #stablediffusion #visualart #aiart #metaverse #outfitreel #creativevideo #creativity #viralreels #visualauthority #collectivelycreate
I watched a lot of videos and I did get a decent result, but I specifically saw one recently where the face didn't have a "nose" if you know what I mean. It looked way more anime with much more solid colors and less shading. It was also very smooth, had little to no jitter/flickering.
I think someone posted it here in the animation channel actually
That video is still not what I'm talking about because it is realistic anime
I'll find the video I'm talking about
Yeah, but I still like it. I want to do that. It says "multi controlnet"... whatever that is. 😄
multi controlnet just means you can use multiple controlnets at once. controlnet is necessary for making these videos

https://www.reddit.com/r/StableDiffusion/comments/12y7z7i/tiktok_girls_hot_dancing/ hopefully this isn't too saucy for this discord but this was the video
my best attempt at making a character fly
try --medvram
I sometimes use it when I'm genning stuff while watching videos
as it doesn't hiccup my pc as much
use ultimate upscaler
not enough vram 
you need xformers and tiled vae
--xformers --medvram
not medvram for 8gb
really?
yea
what should i do now?
do you have xformers?
nope
if not then edit the webui-user.bat and add --xformers at the line:
Commandline_ARGS=
do i need to dl anything?
no
and the vae
actually take 10 sec for a full image with highres fix
thanks
np 🙂
thats an extension you can install, its called multidiffusion with tiled vae.
Good stuff from everyone today! 
Was lurking all day...

okay and what to do with it? i still get the error after xformers using
your settings might be to high still
what exactly are you trying to do?
i wanted to try to upscale it 4times instead of 2
with highres fix?
yea then you need the extension
or use img2img sd ultimate upscale script
what extension do i need for using high res fix
the multidiffusion with tiled vae
heyoo, what do you think about this?
wait you can get hi-res results from img2img?
yea thats the way if you want to make wallpaper or higher res
Oh dang. I thought about it for a sec about using it for AI video but I imagine it has long render times
how do i use it? just set up the script or something else?
you just enable tiled vae and set the tile size to 1024
yea it tiles the image and upscales evry tile on its own, that takes longer
Wait so this is already built into the latest version of auto1111?
yea sd upscale script in img2img
there is also the ultimate sd upscaler extension
This is quality. 
"My computer won't work!"
"--Sir, have you tried turning it on?"
Thanks for this wonderful information. I don't usually use img2img so my knowledge of it is limited but I recently have started to have to use it because I'm trying to do AI videos with controlnet
When my sister was little she couldn't open my laptop, so asked me to open it so she could watch shows on it. She had been trtto open it from the hinge
oh my god
i tried it today and didnt work
so thats why

omg
to be fair it doesnt make sense
enable what
a bomb?
you expect it to be automatic
Doin' ultrawides is a hobby
It's a little shy of 3840x1080 because it's just a little easier, but sometime's i'll run higher
The workflow depends on the size and image type.
wasnt there more ships on left side?
You lost me. lol
I've got one more cookin'
its really nice
did you just changed the width and weight while creating or how do you create those wide ones?
For this one I just started with wide proportions. Sometimes you have to merge images (or crop) to get the right aspect ratio.
yeah it's fun
I do a lot of ultrawides
at some point I'll curate my faves and post online
though i kinda feel like people looking for wallpapers is dead now
kinda yeah since steam wwallpaper is that puplic
and decoder 128 helps for me
ohh is this faster?
i am on directml repo so 512x64 breaks the image, 1024x128 only way it works
OHHHHH
@keen hollow noticed something in picture
Sorry i dont understand "you lost me" 😄
now it is o.k.
That's the picture. I dunno. I posted all 3 back to back in the characters channel.
ive yet to experiment with very wide ar, may you share your workflow for this?
thats good too wow
The wallpaper engine thing? I haven't used it but know about it.
same for me. But i know many people used it
f yeah my image is broke too now i´ll try the 128 now

I love this one even more. Much more of a focused shot, great layout and sense of depth.
Yeah I liked the water splashes
But yeah I do a lot of ultrawides. They can be a little finnicky with character proportions and such. Takes a little messing around.
I tend to like the simpler ones as they make for better wallpapers.
🔥
@keen hollow Hey. i liked your ultrawide work. I tried some myself but I have potatoe PC. Do you work on segments or generate large image out of the box?
hmmmmm
Both. LOL. Depends on the image really.
Sometimes I have luck with prompts right at the intended aspect ratio.
Then I can just img2img them (or hires) to where they need to be.
Sometimes I have to mix in other techniques like inpainting or SD Upscales.
And even a few times I've had to stitch together two images. But that's cheating and not as fun. 😄
@keen hollow What kind of hardware do you need to gen large images like those? I am going to rent some GPUs from cloud servers but I would like to know more or less what I can expect (how fast/how many) at this res
woah, this looks like something straight outta an anime movie
The lighter whispy-er stuff tends to make for better wallpapers so you can see the stuff on your desktop.
ayo fr its looking like a anime movie
I use an RTX 3060 12GB.
Using multidiffusion upscaling it in small segments at a time is my go to for my amd potato
You gotta be careful because if your subject ends up centered like this, then for dual monitors it looks like crap. 🙂
what im workin on rn
yeah testing a model that I have trained now
what programm do you used for training
Here's an example one where I generated at a more normal aspect ratio and then basically just did an img2img to "fill in" the right side. It just filled in some generic wall.
woahhhh alright I love this
dreambooth
Here's one that had two images stitched together. This is generally easier if you are doing two separate characters.
not my best work LOL.
you werent spose to see till i posted the final upscalingsss
ohh great i´m about to learn it too then ext days
Everyone knows I do a lot of Tifas... This is one of my faves 🙂
any hints for beginner?
yes the 3rd version is on general chatt with images @keen hollow
Do you know about those hairs and ship?
great to hear
see what? did you post something I wasn't supposed to see? 
hm?
The wip i posted, i thought you were afk 
?
yeah some artificating... you'd have to inpaint those out or something.
wrong person. 😄
In my journey to do AI videos I have stumbled across EbSynth and Temporal Kit and my god they have absolute bare minimum instructions on how to use this stuff, and the instructions say things that are clearly out of date from the current version. So frustrating
no problem. I am still on full hd only 🙂
oh me oh my
I was being sarcastic 
dw I'm even MORE excited now

Rx 6600 kinda limits me to doing it small batches but it works that all i care about
are you using SHARK or auto1111?
I dunno how SHARK is nowadays, but back then it didn't have that many options
Auto, i couldnt live without my extensions and stuff on auto that shark doesnt offer yet
understandable, I was on an RX 580 around 3 months ago
Nvidia now?
Yeah i was always amd for gaming but now i wanna swap ti Nvidia just cause the tech is better supported
despite it being slower, it was 20 times better thanks to everything else u get
this tbh was the only reason why I went with nvidia
and a 3060 12gb over anything else
Very very understandable 

so if i get a 4090 with 24gb i can create very high res pictures=?
i mean yeah lol
The more vram the better off you are.
but there are tricks to getting higher res images with less.
you can go up to around 4000x4000 with 12gb and tiled vae
the multiple vae kills my images so yeah idk
However you have to keep in mind how resolutions are exponential, 1080p to 4k is 4x the pixels, so for example a 1024x1024 is 1/4 the pixels of 2048x2048, so you will reach a limit pretty quickly and need to use other methods to extend your mileage
you use the wrong one
you should use tiled vae not tiled diffusion
ok different approach
i do but its not working
can you show an example?
thats the settings
dont check "move vae to gpu"
okay anything else ?
the rest is okay
okay lets try again
may get the error in terminal that its not enabled and wont actually kick in so theyll have to pay attention in terminal
what's the command to update stable?
Shameless plug and reminder to check out my image browser for sd generated images. Local Windows program, lightweight, easy. 🙂 https://github.com/GarlicCookie/PNG-SD-Info-Viewer
@stone onyx
oh my godddddddd


what other extensions youre using?
looks like the nightmare fuel i get with latent coupling
its everything not enabled
i just enabled tiled vae
spicy
ohhh hello Darling
try my settings
mine or strix?
nice one
Thanks 🙂
Will try keying in on the character a little more, but liked this one.
yes of course
oh well then lol img2img that shit out idk
by any chance you know how it would look with a wolf ?
thought was like messed up vae or smth
i will, lets go
Try it. Inpaint one in there! 🙂
How is one makin wide image
my model have issues with animals and proportions 🥲
Talked about this a little ways up. I use varying methods. Right now, I'm mostly sticking to just generating images at the wide ratio, selecting favorites, and doing img2img at higher res.
generate wide aspect ratio images then upscaling a bunch/fixing details through img2img
is wehat im doing atleast
eevryone's workflow is diferent
Yep, you can use stitching of multiple images or SD upscales, too.
git pull
my gpu would be like

it was for the joke lol, my pc deny me if I do that
probably something like this.... (low res, concept only...)
the wolf looks good
but would the split be better vertically or horizontally 
😄
no same result like before @trim current
try with it off
still weird would be better someone else with more knowledge helps lol
beautiful 
then i cant create the image
You can get some very high resolutions with consumer hardware. This one is 6144x2048 and was generated on a 3080 12GB. I could go much higher by using SD Upscale as well. The workflow for this was txt2img with Hires fix to go from 1536x512 to 1024x3072 then tiled diffusion with tiled vae to go from the Hires fix result up to the final resolution. At this point I could run normal upscaling through the Extras tab or the SD Upscale script from the img2img tab if I wanted to make it even higher res.
how to set it up that my "old" image is upscaled like almost same then before
haha that gif loool
You mean the Tiled diffusion and Tiled vae setup? One sec. I'll launch SD so I can get a screenshot of my standard settings for that.
hey mis can i help you?
no the img to img
the tiled is not working for me
would you say at the 2nd step using tiled diff is better than ultimate sd upscale?

I'm working on generating the img2img (tiled diffusion + tiled VAE) for this right now, but here are my txt2img settings, along with the image with PNG info.
Yeah, tiled diffusion allows significantly higher denoising without getting seams between the tiles. I can usually run 0.39 denoising on tiled diffusion without a problem as long as I change the prompt correctly.
hmmm I see, imma give tiled diff a shot then, never used it before
can also use img2img with only tiled vae, which allows for much higher denoise
tiled vae/diff is fun
Here is my tiled diffusion + tiled VAE setup
oh 128 huh, ill tryu that
If you have less than 10GB of VRAM you will probably need to reduce the Latent tile batch size. That job used 8.9GB VRAM for me.
oh yeah i will then prolly keep it at 96x96




Now I'm going to try to SD Upscale that last image by 2x to bring it up to 12288x4096.
I need
fear not my friends im making a few example photos and uploading
So for this if yoh can help me understand. Lower values will save more or use more ram?
For tiled vae/diff
lower tile sizes and/or lower Latent tile batch size will lower vram usage.
Aaaa tor some reason i thought the opposite. For now for vae i have to use 1024 x 128 otherwise i get broken segments
I think its a directml bug
just tried it out, I think it was a bit slower compared to ultimate SD upscale BUT the result is quite a bit better + MUCH less vram usage, ty for the info senpai 
Have you tried loopback? Its an interesting method of improving an image with some randomness
12288x4096 
Dear christ

what is the latent tile overlap, should i be matching that toi the other values>?
That is the amount of overlap the tiles will have which helps them to be more coherent with each other. I've found 36 works well for me and I don't really ever change it.
ill give that a shot, been using just default settings cos i dont quite understand the parameters lol
Fair, there's a lot of knobs and switches there. I haven't tried all of them myself.

Nice!
I oughtta make some stickers
what model did you use for that btw? its super cute
its 2gb? what is it gonna do exactly
my own lora plus my own model 😅
Oooooooo
im gonna upload this emote lora soon
the model is already on civit
Ill give it a go be sure to tag me,
under same name?
this is the model
the lora will be called emotemaker but ill tag u
tag me as well 👀
me as well
will do ^^
me too

haha u guys are cuties

how many models did you use to get that if you dont mind me asking
1 lora i custom trained on style and 5 merges , including my mega mix
my mega mix basically was refined until i got perfect hands
and its flawless 98% of the time
damn
this one
so you mixed megamix with few other to make animepunk?
yes!
anime punk also has the crazy expressions lora in it
so u can easily make yandere
nice i will check animepunk later since its the latest one
if youll see theyre a week apart cuz im sick so
but im kinda addicted to making models and loras
i just find something i have a problem with make a lora or model to solve said problem
very lazy
lol the ai gave her a name
Burile
Burile
now u have to go by that name for ur OC
damn great art
Heavy relatable. This has been my hyperfixation for the last month straight
lol i was planning ryoku
the smartest trad girl
okay i think its enough for genning
for me ?
nyo for the emote make lora previews
where is her hand going
https://civitai.com/models/53294?modelVersionId=57660 @trim current @stone onyx @supple raptor @simple mango
Example Prompt: twitchemote , 1girl , green wavy hair , blue eyes , black witch hat , :D , thumbs up, okay , highly detailed , side view, fangs <...
there you go guys ^^

i recommend using my animepunk model as it was trained with that
great work
but otherwise excited to see what yall come up with
thank you , your work is incedible too
i plucked a huggingchat chibi prompt in there with your lora/model lol
now time to make it actually pouting
awww
many thanks!
i find it works better without the chibi keyword

i provided an example prompt in the models description if u need extra help let me know ^^
not necessarily the power of controlnet
the power of will
im loving the fact that youre loving it
XD
LOOK HOW CUTE IT IS
Now i need to figure out how to make it into emotes
ez photoshop remove background
Which version of ps? I cant find the ai quick remove background i used to have
i use latest
thats adorable!
right!?
yes!!!
wut
can someone tell me if there is a option to fix that eyes 😄
what tool do you used for background remove?
ohhhh
hello

@trim current @scarlet kelp you can also use that same ABG extension for free on huggingface so you don't need to wait for the model to load on your own system https://huggingface.co/spaces/skytnt/anime-remove-background
ohhh nice thats helpfull
oh very good to know thansk nubby ❤️
i´m creating some sticker for discord, any hints to to remove the little white stuff around or is this the handmade part?
Depends on the image. Usually I just manually erase those parts.
yeah like me, waiting for new creations and editing the already done ones
Yep, exactly
smol
same buddy same 
nawww i need to find a wife with pink hairs XD
was there ever someone with pink hairs?
someone like this? 
married
both fine
sorry, reposting old stuff 
old but gold
why are always my eyes are broken ._.
if the hands are perfect the eyes shit
Can anyone give me anime recommendations?
is that a extension?
Hellsing Ultimate
but i´m using text2img
D-frag! for comedic
???
🤨 🚔

HOW
this is good as hell
wha wha happened
oiii did he postet more 😮

uhhhm with a bunch of loras and settings
thats wonderful!
i have a lot of jinx chibi arts
fuck just came out of a 1 hour pain attack, took so much oxy IV. Because the Epi pump ran out and the pain spiked....god damn that hurt
like being disemboweled
What program are u using to make that?
Stable diffusion
pqni

panic no

yup yup, I'm using SD 
pain
I did not mean that
Hahahahaaha
ohh thats not better :c
extreme pain, the kind that makes you incapable of movement and cry out
well, I'm using the auto1111 webui?
is that what you are talking about? 
Is that the name of the AI that you use?
okay i hab stickers
Ps is amazing for it
good god im putting laptop back down I cant type
ahhh, if you are talking about model, then it's the Holokukiv2
I neeed to lie like a statue to avoid pain
Quick 320x320 export and 128x128
good night
Thanks ❤️
Redos of the concept pics protagonist for my story. Right one her uniform as a Magpie, a magical thief who steals knowledge and artifacts for the Goddess of Secret Knowledge
with 4 gb of vram I can only generate images at 640 pixels 😔
Ouch I'm sorry 😭
I want to see a picture of Duke Nukem but loli
does SD even know Duke Nukem?
Idk 😔
it's not close, but kinda something
nawwww thats me before 20 years
as a little cheeky boy
what about this?
It was not what I expected but I am satisfied
👍
uhhhhm sexy but sus,m could be me either in youth years while training
it's definitely kinda sus
any good settings to use?=
Buy a 4090
actually i looked today for parts for new pc XD
for inpainting, usually your inpaint area should be only masked, your width and height should be 768 or 512, and your denoising strength to around 0.1 to 0.4
if you can buy it accompany it with an i9 13900K
uwu
your denoising strength will differ a lot depending on what you wanna do, so do your own testing on what works


I have taken over your role as the Chibi master

👀 I feel like I missed something
I'm kidding, I'm not that good at chibi 
A dude had l+ratio
i made some really nice ones
but more emote and sticker stylie
i will try owo
yeah i will get the i7
idk how ai knew to gib me a potion bottle for sip
make sure it is the K model that is used for overclocking
delicious
everytime, i building pc´s since im 9 years old 😄 i just do overclocking
I think they are more expensive than other models
yes like always
but i mean if i spend 3 or 4k for the pc it doesnt matter XD its over 3k already so yeah gg
ez
Friday vibe everyone https://www.youtube.com/watch?v=U6n2NcJ7rLc
Now do a drip version
It's friday vibe tho 
It is Friday my dudes

Mood
its saturday already in my time
its still thursday here 💀
Poggers
you must be from asia




















