#🍥|anime
1 messages · Page 63 of 1
@nova remnant Here's an example of the setup for inpainting in Krita. Open your image, create a new layer, I've named that "Mask Layer" here, but the name doesn't matter. Use a soft edged brush such as the airbrush with the color set to white to draw the mask that you want inpainted. Use the rectangle selection tool to draw a rectangle around the masked area with as much or as little context as you want to give the AI. With the mask layer selected in the layer menu, click "Start inpaint" and it will create a new layer for each inpainting attempt so that they can easily be hidden or viewed to compare. If you ever need to fix something in the image, you can also do that and the inpainting will take into account any visible layers when you click "Start inpaint".
I understand, it looks very useful, thanks, but I think the problem is not to do inpaint, but to do upscale and I can not do upscale and then use inpaint because the vram is not enough, since it is 2k
Is there a photoshop one?
What I'm saying is that the size isn't a problem with this method because you are only feeding the area inside the rectangle selection to the AI.
i don't have the inpaint window
Go to Settings > Dockers > Inpaint and make sure it has a checkmark next to it.
I get this
Might be a conflict with one of your extensions. I get errors related to adetailer and tiled diffusion, but it still runs. Some extensions don't handle errors as well though.
Dynamic Thresholding would be my first guess. I have most of the other ones. I don't have canvas zoom either, but that doesn't seem like it should cause any problems. You can disable extensions by unchecking them in the extensions menu and clicking Apply and restart UI.
https://youtu.be/19ZNiw6sgBw
Is this plugin like the krita one?
Stable art is a new FREE Photoshop plugin that allows you to use your local Stable Diffusion installation inside Photoshop! It is really easy to install and use. Although the plugin is currently pretty basic, it has all the base options you need to start painting inside Photoshop with the help of Stable Diffusion models. In this video, I will sh...
i have removed the dynamic thresholding
but same happen
"anime model"
heh it was never an anime model 😛
dont' make counter that with zettai ryouki absolute thiccness 😛
I need to make a model that only really delivers visual novel stuff, or at least a lora that modifies it to
Lqdies and gentlemen, conway twitties
@wheat falcon Can you delete messages?
there was an attempt to anoy everyone
Ye, only staff can ping everyone 
I reported, but seems like everyone is offline or afk
Oh thats pretty cute :o
just figured out lora merging too
I can't wait for when WD is really good and we get good models with good clip :o
Fair fair :o
I should try wd with comfy later
they going on to the final epoch now around next month
ye go ahead
Thats pretty sweet, I hope it becomes like a new base for models :O
only 2.x models can be merge with 2.x models
Yeee
I think thats also why its better, cause I think it uses BLIP instead of CLIP which I think is better
wassup guys, newbie here
Got it, ty
Ayyy, thanks fruit, you da best

evenin'
for tagging they used both blip and wd tagger and for the Text encoder 2.x come with openclip instead
Ohhhh, thats fair then, that is pretty sweet 
It got fancy new stuff
I tried something neat with comfy today, looped a second clip box on the hires
same city?
omg, how i can do this? 
you already got A1111 installed?
new here, installing...
a new addiction frfr
ever since i started ive basically done this everyday during my free time
Same
i put on a show and let SD do its thing
or even worse, start spending extreme amounts of money on your pc
when it's done i put a new prompt, watch my show, rinse repeat
letsgo!
nice!
got a 3060 two months ago i can relate
lemme collect the prompt, as it's not local
Same, I got a 3060 for AI 
got a 3080 10gb about 8 months back... now i have a 3090 -> i needed the vram
my pc have one 1650 4gb, can handle?
Though I also did need a upgrade 
Yeee, thats good
i5 3 generation 😦
limited to 768 images with --medvram though
is installing now, im back when finish
yeah dw you'll still be able to do it, just slower
You need to upgrade that CPU, works tho :o
watch a show like me
I mean
With tiled he could do 1024 res
I mean
I'll be right back guys
is installing 
3060 12GB here. Has done pretty well on all this.
except for finetuning sd 2.x models
i need more
I just make LoRA 
bbl
nah you'd be surprised
Big beefy ladies?
@tepid geyser help me plis
@daring ice i'd suggest get the thing set up, and load up the first model 🙂 i can always highly recommend my model, it gives great image quality with little effort
i can second eface's model
Send me

i generated an image with just that as the prompt and i was gonna send it but its a little bit..
LMAO
it didn't work before
Fair XD
Dont work?
Let me try
it doesn't work?
https://civitai.com/models/64843 this one should work
Easy image quality that has a good context awareness! This model was made with the intent of having great image quality, and good context awareness...
God damn
the original version was uh VERY nsfw

I've tested Darkalfa Prototype2 and i think i'll post it tomorrow
it's a bit worse on the hands, but i don't think i can fix it 😦
YAYAYAYAYAYA
how many minutes to generate this?
however, the vastly increased concept/context awareness and general knowledge pretty much outweighs the cons
Oh, let it shine!
Oh, let it shine my friend!
So, let it (shine:1.3)!
So, let it fly!
Don't look back again
You remember, purest heart?
And the (beauty :1.1) of the (colors:1.0)
They ran in the rain!
They ran in the (rain:1.4)!
(best quality:1.3), (highres:1), perfect, perfection, amazing, incredible, medium shot, (dance:0.5), (joy:1.2), surrealism , splash art, swirling colors, art by studio mappa style and by artist Takehiko Inoue style and by artist ghibli style and by artist yoji shinkawa style, illustration, playful elegant layered summer(night:1.2) dress, (abstract:0.6), art by studio ghibli style, her rainbow hair blends into the colors in the sky, running through the (city:1.1), (jump:0.2), brenizer```
there's the prompt for that image 😄
yeah fr just make it hide the hands behind their back or out of frame LMAO
who would win
pure muscular goodness
or
cat girl with a SCAR-H
in a fight?
i got with my beefy lady
yup
but im biased
i gonna be specific with gun models ig uess
I dont know too haha
Naruto in tokyo cyberpunk City eating pizza🤣🤣
i don't think you use highres upscale, right?
player 3 has entered the game.
i use highres 4x ultrasharp, 20 steps, 0.5 denoising, and x1.55 upscale
and the og steps?
she has so much muscle it doesnt even matter tbh
that image was 30 but i use up to 50 sometimes.
does your model get a lot of difference from much higher steps?
depending on samplers yeah
unipc 50 should be pretty much good enough for most situations
gonna be a pretty uneven match though unless the muscle lady also have a gun
this was the whole prompt, didn't even try
eface's model kinda fire
Awesomee
holy shit, it actualyl did that?!
nah she's like that muscle guy from My Hero Academia she'll like launch muscles or smth
yeah
just a sentence
"Naruto Uzumaki in cyberpunk Tokyo eating pizza"
Show me your Best creation
well, that's what i want my model to do -> be able to do work from sentences
I prefer the muscle bunny lady 
my best is only mid compared to other peoples bests! but i'll try to find some
mirko? is her name?
yeahhh that's why i love using yours the most
easiest
Yeee
@wispy canopy show me your best
lol, filter by username / has image
i'm posting a LOT here 😛
😎
I mean, I think my best was my one POW submission
now now @wispy canopy it's basically "our" model 😛 Unless you've made significant updates
these are just a few
nothing crazy though. like i said, my best ones are mid compared to some of the crazy things people on here generate!
da p2 has yet again vastly increasted conceptual knowledge, and can do the 1920x1280 upscales pretty decently
I've never had any major issues with 2k to 4k upscales, that's with multidiffusion though
this has to be one of my favorites 😄 (Basicaly, it's lyrics, and i'm using the lyrics to 'glitch' the prompt encoder)
It is ye
I've kinda lost interest in model making, merging etc. I only really worked hard on DG3 cause I wasn't sure I was going to be alive for long
and I didnt want to leave it unfinished
Hope you'll be able stick around for a while though 🙂
Yeah if it doesn't come back again, I really hope it doesn't
Naruto and Sasuke got reaaaaaally close after the series ended...
stay the fuck away
the alternate ending
the CANON ending
The good ending (?)
Only in anime does a war criminal and a genocidal mass murderer get to live on "peacefully"
being forgiven
Is Naruto a war criminal? :o
No...they do that in fanfiction too
sasuke
yet he's free
O, fair then :O
well we can bring in gaara too but since he's a kid that was possessed by a tailed beast, that's kinda mitigating circumstances
someone give me some prompts to try
this should have png info
hm nice
i gotta go to bed, cya later lovelies ❤️
bye eface!!
water goddess made of crystal clear and reflective turquoise water rises and towers over a handful of 19th century Frigates as she commands the ocean to break and swallow the frigates whole along with the crew. She releases cyclones from her hands and we see her from the point of view of a crewman, looking up and seeing the destruction around us
There's a narrative one
thank's i'll try it right after current generation
Awesomeee
@tepid geyser you create only in anime style?
watermarked outputs

at the moment, yeah
so use a neg prompt to stop them
text, watermark, logo, words, word, trademark, copyright, title, header, footer, subtitle, written, name
thanks!
still generating with words/attributions
i'll just try a different seed
or run it through photostop, mark it with selection tool and fill with the tool I forgot its name
Krita should have something similar
databook entries with strange language
https://youtu.be/LKXl3zR7GwI Random meme time, friend sent me this the other day
"I'm Khamsin, a.k.a. the desert storm."
Original Video: https://twitter.com/tuxedopato/status/1560259972230811649?s=21&t=qabmzRg4tjKG3YdtblrACw
Song: https://www.youtube.com/watch?v=3AarFtHRqNg
I want to upscale this properly
@tepid geyser can you create something dark, will be my wallpaper 😎
want people in it?
i was already generating this when you asked me
LoRA is done training, time for the fun (or tragic) part 
make a party in this background >:)
good luck!!
or a explosion
I hope it good :o
this would be cool with a background explosion
I tried using different LR setting
Of the dataset you showed me earlier?
for a second i see link from zelda here
I don't play Zelda games, I wouldn't know 😛
at least hear the ost is really good!
try ocarina of time title and great fairy fountain
kinda hard to keep the exact same image but add an explosion, im not experienced enough
im trying
i undestand you
sometimes my worst enemy still the hands
different image with explosion
@tepid geyser but accept my little animal train
fantastic : D
i have this another but the first is really more fun
has six fingers though
Boss Fight!
no hands = good hands!!
a party of otherworldly dark adventurers
this!
good work
Yeee 
boss fight? fight with my boss >:)
1e-4 Learning Rate 1st, 5e-5 Learning Rate 2nd
gib 
@keen goblet can you generate a image for me?

is for a work
Sure :o
i send you dm okay?
also apparently training in FP32 works
Alrighty
for boosting quality
you dont know what fp32 is?
Like, here?
yup
@magic wave send you dm
but gonna be taking up more vram though
Oh, I'll have to tell Derrian to add that as a an option in the commented part then 
Oh
Well
I might be able to do it?
2nd one is better, more details
😦
I'll have to test more
i dont think there is anything to test, 2nd one is way better
you have a problem bro
I should rewatch march comes in like a lion
Unprompted husbando
Well, I have to test which is closest to the style I want
he very cool though so I wouldn't complain
Thats some screenshots of the dataset
Only 1 pic I can send without screenshooting cause they're all NSFW 
does anyone know how to resolve?
@supple raptor Might know :o
which extension are you trying to install?

stable difusion, first installation
some guy suggested opening cmd or something and doing
pip install pipwin
pipwin install scikit-learn
sorry
no need to say sorry, googling is a helpful skill you should develop
as should anyone
sorry idiot questions, im new
if you struggle finding stuff on google you could ask
Also holy shit this LoRA is good with scenery
give taste?
wow
can you send me in full hd?
i'll try
Though I guess I did have a bunch of scenery pics in the dataset
That is the full HD
Are you familiar with working in python virtual environments?
Are you getting LoRA mixed up with Dreambooth?
fp32 vs fp16 shouldn't matter much
If not, The simplest solution is delete the venv folder that's inside stable-diffusion-webui, close your cmd window and launch it again. It should install scikit-image automatically and likely just encountered an error while downloading it last time.
at least a lot less than pruning
oh yeah
man
bWm always makes better suggestions than me fr
Fair 
theoretically though. fp32 has more "precision" than fp16 so it should be better I suppose
but like it's not noticable
I think
I mean, fair, though bf16 is better than fp32, so is it the same?
For finetuning, fp32 is pretty significantly better than fp16. No idea how much it matters for LoRA training though.
@magic wave just ask people on projectai
inference is generally where fp16 vs fp32 doesn't really matter.
Fair, they may have been getting confused :o
they are all practically experts at this point
Too big expert tho, I need a babysitter to guide me through stuff 
XD
Might want to see a dermatologist about that 😂
I just read through some messages there that they had about it
salt seems to say that you should always train with fp32
not sure though
Same
Hmmmm 
Wtb nubby, 200gp
guys, how do I remove B?
Salt also has the hardware to train at fp32. It will increase vram usage. By how much? IDK
Wts (want to smash) nubby, 500gp
Kek
ctd is struggling to do it
so I guess too much
for us plebians
I mean I saw a comparison someone sent (I think it was salt again), difference pretty small, I would say bigger than xformers level of difference but like not too substantial. now you should think if it's worth it for you to like rent a gpu or something just so you would have a slightly better model/lora
at least that is my opinion
Tis what Distro said :o
Grand Prix motorcycle racing?
Seems way too intensive for a marginal difference 
.
Fair xD
1981: Marco Lucchinelli wins the 500GP world title with his Suzuki RG500 Gamma.
ok if I wanna actually move towards fixing my trash sleep schedule I need to go to sleep fr
so I go off computer like the responsible man that I am not
👋
but your anime waifus
O, fair, good nite 
Tru, rip the waifus 
Tell that to him


woofer
im sorry poor pupper but the waifus
POV you stopped taking your meds for a month and a hot thirsty goth girl appeared inside your house and she says shes been your gf for 7 years
Nice
And there is another 
Stop teasing
I am halfway in bed
Let me sleep
quick post all the waifus while kazoo tries to sleep!
still tryin to install, python is stressing me
She went to didney worl

Nsfw
I LOVE this
dont worry the shockwave will kill her
Someone challenged me to make the prompt in here maybe a couple of months ago and I just revisited it with my current mixture of loras
Whats the prompt? 
its ok its a guy
1girl, lofi girl, headphones, brown hair, hair up, studying, focussed, profile, sitting at a small table on a terrace, (a nuclear explosion in the distance), in a futuristic city, neon sign, highres, absurdres, extremely detailed, intricate <lora:NAI-bluethebone:0.3> <lora:WhiteMelancholy:0.45> <lora:loraslayernoiseoffset:0.45> <lyco:lucky768-2:0.45>
scary
I have no idea which lora is making the black sclera actually work well all of a sudden, but I'm here for it.
Imma try that but with my LoRA 
Gm chat
evenin'
She's watching a Kurzgesagt video
Or taht, true
Now I think shes in more trouble 
rot, entropy, 1girl, long hair, (red hair:1.1), sidelocks, darkness, highly detailed face, mid shot, full body, somber expression, looking down, dark energy, colorful, vibrant colors, portal to another world, highres, superb, 8k wallpaper, extremely detailed, intricate, (horror \(theme\):1.1), haunted, creepy, spooky, ghost, (lens flare:0.8), bokeh, pile of skull and bone <lora:NAI-bluethebone:0.3> <lora:WhiteMelancholy:0.45> <lora:loraslayernoiseoffset:0.45> <lyco:lucky768-2:0.45>
Imma try that 
Also tried a 16:9 pic :o
1girl, long hair, (red hair:1.15), sidelocks, dark background, (mid shot, full body:1.2), the blood moon rises over an apocalyptic landscape, ({nixeu_extra|wlop_style}:0.7), flat color, flat shading, ultra realistic, highres, superb, 8k wallpaper, extremely detailed, intricate, limited palette, blue, (highly detailed face:1.2), freckles <lora:NAI-bluethebone:0.3> <lora:WhiteMelancholy:0.45> <lora:loraslayernoiseoffset:0.45> <lyco:lucky768-2:0.45>
Time to try 
This could be like a movie poster I feel
what happened to those new folks from earlier?
Sleeping I suppose?
Also tis one is pretty dope
Sacrificial altar looking thing :o
Also tis what I'm using, am yoinking the "neutral" parts of the prompt and slapping em in, using my own LoRA stuff tho :o
❤️ controlnet's open pose
makes me think of Streets of Rage
UGH FINE i'll start up stable diffusion
sigh...
Girl from Earth (probably)
I used a Lora to generate it.
gm
good morning eface
Google chrome hair
my model can do rainbow colored hair on command ^^
which i think is pretty damn neat
heterochromia with the actualy specifiec colors too 😄
ok, lets make it even harder 😄
ah,i think i got it to fail 😛
Cutoff 
eh... it's not a pass was going for 'dragon eyes'
but... it did try a valliant attempt
them fluffy ears though ❤️ was trying cat eyes, instead, got cat ears
i see she has a booty hidden away 😄
hidden safely away
she can probably breathe fire so be careful
damn even with negative prompts to block nsfw and ecchi the model keeps wanting to remove more and more clothes
it IS based on my accidental hentai model 😛
so it's still pretty damn horny at times 😄
hahahaha true true
CAT EYES: SUCCESS
naisu
Tfw (cat eyes:2.0) 
girl, rainbow colored hair, cat-like eyes, vertical pupils, yellow eyes
or smth like that
training from behind is something done very rarely i suppose?
what do you mean?
that there is probably very little training data in the original models which have a tail or wigns shows from behind
ah yeah, im having a hard time making proper wings and tails
if they're behind her, usually not an issue, but in front -> very much so
i can semi-consistently get a tail but hard to get 2 symmetrical good looking wings
this is your new high school teacher. would you be able to pay attention?
there's this lora for back wings, but I haven't test it personally https://civitai.com/models/24554?modelVersionId=54418
Most models will draw the wings behind the character regardless of the direction they are facing. This LoRA is designed to put the wings on their b...
thanks! I'll check it out
Have there been any major updates or extensions in the past week? Haven't been here much
i suppose the biggest would the controlnet refernce?
thanks, I might check it out when I can
it's a wool shirt, but reflects like rubber?
this specific quote reminds me of a song!
https://www.youtube.com/watch?v=59pyYi-dDug 12s into the song 😄
Provided to YouTube by TuneCore
Mega Marital Problems · Starbomb
Starbomb
℗ 2013 Starbomb
Released on: 2013-12-13
Auto-generated by YouTube.
Oh, hey! Long time no see. How've you been?
the hands are great but the book is now the problem LOL
oh believe me, those are GOOD books
this would be happening so often on all models
ah so i was lucky to get good books the first two times!
openend books are dangerous 😄
wait is this... perfection??????????????
perfects hands perfect book????
only the words in the background might give it away
except hair in the coffee 😄
i have cat-hair in coffee, regularly 😄
escape pod?
she says "Hi Eface, how are you? Would you care to sit and chat?"
maybe the pilot seat of a mecha!!!!!!
fun part, this is what my LLM came up with after asking it for something to the lyrics of starlight brigade High above Earth, a sleek silver spaceship hurtles through the darkness of space. Its hull glows softly with blue and purple lights, casting eerie shadows across its surface. Inside the ship, a young pilot sits strapped into their seat, hands hovering over the controls as they prepare for lift-off. Outside, the stars twinkle like diamonds scattered across the velvet blackness of the universe. As the ship accelerates towards escape velocity, the pilot feels a surge of adrenaline and excitement - tonight, they embark on a dangerous mission to defend humanity against alien invaders
which... well, go watch the video 😄
hands are still the enemy
until someone figures how to fix it for real, it will remain that way
a fast and easy way to get wings
is to Photoshop them on and then run a control net over your character
not the worst hands I've seen
bruh
just in case it has to be said, nobody click that link. don't be stupid
damn... almost perfect. Her right hand is all wonky though
@dull jackal hope you're actually around. We got a scary (likely scammer) link
Lol, I just got back in here and saw it.
agaaain?
nice
nice hands!
thank you! took several iterations
woahh
bat girl
the dark knight
Lol I didn't even notice the bat symbol on her chest
I suppose that is one way to interpret dark knight that I hadn't considered.
Reverse sword grip
damn now the bat symbol is just a cut to the chest 
i officially hate hands
Nah, that's just the hand fetish taking hold and causing you to be extra critical of bad hands 
left hand just had to go and be fucky
i prompted "the dark knight" uhh... yeah. that's him alright
the others were a bit more of a miss... but it got the job done
lets add male to pos, and cat to neg
got batman without the batman 😄
Neat throne and she's 50% better at hugging than the average waifu.
ketchop thron
used necromancy to splice an extra arm on
ketchup throne, evil wife
ok... might have overdone it on the creepy part. rip legs on #2
blonde hair, two girl
Currently, there is a public bot on the server that generates images available as a research beta for SDXL, you can find the current status of the bot in #1047610792226340935. There are plenty of ways to use Stable Diffusion such as the official https://dreamstudio.ai/ website or by running Stable Diffusion locally using your own hardware - check out #1080946152318443610 for more details! You can also stop by #1025467151206854736 for any issues you experience while using DreamStudio or #🤝|tech-support for any problems you encounter while installing it locally!
mustard throne, benevolent wife
LOL I WENT WITH MAYONAISE TOO 😄
Finaly finished the hair drawing the facial features now
Thunder throne, intense wife
what the
it said my image is too nsfw but it definitely isnt
damn it was good too
that's clyde doing it's thing
it was emerald throne, jealous wife hehehe
sapphire throne, sullen wife
bruh again
it's not nsfw smh
Finished the sketch now to do the dreadful process of colouring it in
see you in 2 hours
nice 4/4 😄
i would let her beat me up
yeah, this prompt does so much better in landscape mode
oof those pikachus are borked
how many sampling steps do you do?
50(low)->25(high)
wait what do the low and high refer to?
lowres (initial txt2img), highresfix
time for some flaming hot coffee!
clean
Damn, you blow up the Eiffel Tower?
nah i just covered it in grey clouds
fire mage go!
Man...looking at you folks is making me want to generate an image.
I've been editing half the night. Might as well open up the Webui.
hahaha yessss
Here, I've got something for ya. lol
EXACTLY! 😄
lol
space suit without legs LOL
is magic spacesuit 😄
ah yes of course
ah yes. sexy cyberpunk elves 😄
bruh how can you upload that but i cant upload a fully clothed woman
it... happens
Clyde must be drunk tonight.
isn't he always?
Pretty much....
nooo, i need to somehow increase the intelligence of my model yet again! it falls just short of being able to do what i want now... talk about moving goal posts
Water. Earth. Fire. Air. Long ago, the four nations lived together in harmony.
what were you trying to do?
so, i got this lyricsprompt that made those gorgeous semi-half undead cyborgs right?
yeah you sent a couple yesterday i think
i could literally just keep prompting this
however, now i wanted to add a cat to it... and it has a hard time combining it
add a cat to the scene? or make a undead-cyborg-cat creature?
ooh oooh i wanna try
hmmmm I just had an idea on how to manipulate a models language prompts
though it's time consuming
probably as time consuming as my method was
my other
FML i had cat in negative prompt XD
merging Lora's with captions of natural language for things generated and making sure to generate random things and be consistent in how you form sentences and which order etc might manipulate a model into changing how it responds
you know.. i'd be rightfully VERY scared if that thing comes chasing me
so first you just generate random prompts, then you write a caption for each of them, make a lora, merge the lora and repeat
5th leg why 😢
crazy cat lady? are you going to tell her?
it's great when the anime model can do more than just anime 😄
yeah for real
u want me to press the button?
do it
Do I wanna know what happens if she presses the button?
more waifus, probably
it's like a meeseeks box but for waifus
the fifth leg looks like an AI error, but actually it's added to increase this machine beast's speed
dang 😄
hands are the worst thing to ever exist. 
I KNOW
control net open pose is like next level
it is!
ive been playing with it a lot lately defo my favorite thing now
i've again abandoned all controlnet and lora's as always... pure prompting 😄 (no need to change settings, lets go mode)
what is control net open pose?
same hereeee it's just so much more convenient
i just type whatever and it works
you are not using controlnet? 😮
controlnet openpose: get the actual poses you want from your characters
that's a thing?????????
yes lolll
you could also control the compositing
photoshop and controlnet can you get you amazing ai renders
oooooofffff that's the one bug i know my model has 😄 ITS FETISHISHM WITH DOORS/MIRRORS
ill take the left one😳
no no dont worry, mirror was part of my prompt hahahahaha
is controlnet like a lora?
or
care with that tho -> doors and mirrors make the model go bonkers
this one turned out pretty damn good i think, the reflection matches the subject
it's basically an extra way to control your stable diffusion, being able to guide it to what it should make, with better accurary
it does 🙂
Once you learn the power that is ControlNet....your world will change forever.
more cyberpunk elf 😄
wait im trying to figure out how to get
youtube is your friend
the thick outlines look really nice
Hold up...Lord Eface made a model that we can download?
in Vergil's voice Give that to me....
it's very unorthodox prompting but it works really well
much easier to use than many models
hes also gonna make a amazing change in the community and use his smart brain to fix hands forever! 😄
Easy image quality that has a good context awareness! This model was made with the intent of having great image quality, and good context awareness...
Ahem, well...regardless of my rude Devil May Cry gesture, thank you kindly for this link.
give me a prompt idea to try
like a concept for an idea
Hm...a bipedal goldfish? LMAO
A bipedal goldfish blowing up a balloon. Random shit, I dunno. Haha
GODDAMN HORNY MODEL, you grab EVERY change you can get to do cameltoe!
lets test your model eface
my kind of model 
i'm still goign through my backlog of images, there's a very specific prmopt i want for my 1.1 update
At least it isn't spewing another kind of content down there....
it will if you let it...
What the fuck?! XD
that's a model that doesn't have enough contextual awarenesss...
yesterday someone wanted to see naruto eating pizza in a cyberpunk city
Naruto Uzumaki in cyberpunk Tokyo eating pizza
that phrase was enough to generate this image
pretty cool
Damn...
correct
Ahem....bit...too many eyes.
Good evening
Good morning. XD
Time zone do be like that
Bro I'm fucking dead. I want to post this image but I don't even know if it's allowed. Haha
do it
There is literally ZERO sexual content or nudity though.
then do it
Damn she's dangerous 💀
I was trying to make her resemble Shego from Kim Possible. lol
Cirno 
it's great!
that feeling when you land on an amazing prompt and just keep changing the seed to get really cool variations
Here, have a Sith girl. lol
I can't believe I got a decent lightsaber and a somewhat decent hand on it...
oh fuck... i was working on DA1.0 instead of 1.1 😦
RIP

Hmm...now that I have my sith girl. Time to try out Lord Eface's model.
I'm sure I will be impressed.
Welp, first image generated from Dark Alfa. Pretty damn neat.
Okay....all I gotta say is damn....
it do be like that at times yeah
how she get up there
that's what i was gonna say!
just below her is a hillside and some stairs down
durarararararararararararararararararara
rararararaararararaaa
banger opening
Gm
@tepid geyser good morning
model?
I believe that one was Darksushi, the colorful one.
NEW VERSION IS UP! https://civitai.com/models/64843/darkalfa-effortless-high-quality
Easy image quality that has a good context awareness! This model was made with the intent of having great image quality, and good context awareness...
i know what i want to do next, and again, not sure how to do it 😄
i'll figure something out in the end 😛
what extension do you need for the BREAK thingy?
none, it's a A1111 thing
Beware this hat
morning! although im going to sleep now
did you get it up and running?
not yet 😦
good luck setting it up!
yeah… sometimes it’s finicky
hopefully #🤝|tech-support can help you!
and when you get to that stage, ask eface about their model. imo it’s the best for beginners
what are you guys talking about?
setting up stable diffusion
change your browser to render in software mode
that saves up a lot of vram, espcially on 4gb cards
how to do this?
download the pic and zoom into her face... wow
gud shit!
depends on browser
thx
@daring ice, each browser has their own config, which after having been set, you need to restart the browser,
did you try googling it?
or do you not know what to google to get the right result?
the reason i'm asking... this will happen a LOT in your journey with SD
you WILL have to look up a lot of things, so it's important to become good at it
article which explains how: https://www.lifewire.com/hardware-acceleration-in-chrome-4125122
dont know how to ask 😦
very clean ❤️

wrong type of teacups though, but i suppose you can't argue with that 😛
Too ez
Have you tried to shorten the folder name: renaming stable-diffusion-webui to simply webui
has solved tons of python related issues for me
Since when new PFP?
about a week
I'll try thanks
It is a fingergun
Does anyone use the photoshop plugin to connect stable diffusion?
https://www.youtube.com/watch?v=NEPrGCaBtoI&ab_channel=SebastianKamph
This one
When I hit generate in photoshop, the plugin pretends to load and creates layers with mask, but it doesn't create anything, neither in inpaint
ControlNet in Photoshop as Stable Diffusion plugins get better each day. In this tutorial I'll show you how it works.
https://github.com/AbdullahAlfaraj/Auto-Photoshop-StableDiffusion-Plugin
FREE Prompt styles here:
https://www.patreon.com/posts/sebs-hilis-79649068
Support me on Patreon to get access to unique perks! https://www.patreon.com/...
might have to take a look for any system errors .. but my guess is that it fails to contact the API
Wich model?
and this lora at 0.2 strength
https://civitai.com/models/24833/minimalist-anime-style
Thanks







