#🍥|anime
1 messages · Page 96 of 1
nice
I just got AnimateDiff up and running That was one of my first set of animations.
hmm havne't tried th at
ok I like it

is that what the current fork of auto looks like ? (sorry, I've been out of the game for some months)
it's a theme extension
looks slick
loads quickly too
I need to do a little round check of the tools I knew, see how they progressed. I loved comfyUI
hmm getting all kinds of errors joy
hmm seems i'm missing some models
so animateddiff is throwing error. sigh
ahh found the models I need, it really should install those when it does the ext
ok... now where do I find the gif?
oh might help if I turn it on. DOH!
Do my eyes deceive me? Is that an @wind girder I'm seeing?!
Yes it is 🙂 I connected back discord yesterday night, trying to get back into things with not too much pressure
trying to not fall back into the moderation 24/7 life I tend to dive into
how are you Nubby ?
blinks wow this is gonna take a while
necessary to solve some real life toxic situation, I was breaking appart. It's getting better 🙂
I'm getting the infos on how all this progressed around here, the server seems quite active, but the bot is surely helping on that front ^^
Yeah, I think the bot and just a general increase in interest thank to SDXL have helped. You'll be happy to know that ComfyUI is the best way to use SDXL locally btw.
I'm talking with a friend of mine that worked on SDXL comfy in an official manner from what I'm learning, yeah it seems great !
SDXL isn't far from releasing too I hope ? x)
what do you mean ? running comfy as an extension of auto ? xD
yep
AnimateDiff. Specifically this version of it https://github.com/neggles/animatediff-cli
ty!!!!
Yeah, word is that it's supposed to release on the 28th I think.
cuz im trying the official and i cant get it to work
just learned about SDXL "release" before 0.9 xD must have been some fun drama around here
not managing to install it at all
26th isnt it?
I'm probably wrong
There actually wasn't any drama. From what I heard, Stability was just impressed that it took so long to leak XD
lol to be fair, leaks are quite the thing these days.
Now I'm teased for that release !
Yeah, I'm really looking forward to seeing what the community can do with it. It looks like it's going to be really good.
🤞
ooh wow...
ok latent mirroring works wonders.
wow...
and with the latent mirroring
Miku is flying to new horizons there, those are great
thanks
finally got her out of the water
ok not sure if t his theme did this but i'm getting better results and a1111 is running better
huh.. odd
just Theme UI do process better?!
in that case i'll take 4 at once
g`night
last one
was suppose to be upside down from tree above water. that didn't work
tried cause someone ealrier said it wouldn't do it. guess they are right
loads up his animetoonpus model and tries again .. just cause :D
and yeah it came out an odd anime .. so yeah with that bed...
time for the daily check-up
Yesh. I don't usually even attempt it. That one was the product of a pose wildcard XD
and now bed... here have this
Fixed the face with adetailer fortunately
guys, if i want to generate an image of a person look at the surface of the water, with prospective from underwater, how i can do it?
i always get weird results
it's hard to generate transparent objects and fluid effects with stable diffusion
intresting idea! i think i` am must be try it first
yeah, that would look very cool, but it's very hard
but for starters, just see the references of the desired result
i think maybe kile this?
yeah, close enough
what's the prompt, i'l try to play with it
something autocorrected it
<---- dirty mind
1boy ,:o ,blue hair ,blush ,earrings ,eye reflection ,fish ,from below ,goldfish ,grey eyes ,hair between eyes ,jewelry ,looking at viewer ,male focus ,open mouth ,parted bangs ,poi (goldfish scoop) ,portrait ,raised eyebrows ,reflection ,short hair ,sidelocks ,single earring ,solo ,sweatdrop ,underwater ,wide-eyed ,highres this is just get on danbooru. and SD get this is prompt. try.
cute
but result is not the same. he jut underwater
yep, as usual
same like my
then we can ask help controlNET
nope. this it. and single reference image
juat try diffirent "control mode" for best result you like
select preview as input or save preview pic and reopen it like single image
huh.. for me best result that i think
Is this style of like drawing a Lora?
lora. sorry...
thats pretty sick ngl
why?
what prompt did you use?
because it looks good
maybe.
you can download pic, it has all metadata
oh nice
Does anyone have a google colab stable diffusion that doesn't disconnect and can put models?
Is there any model that has the same style as darkalfa but uses booru tags prompt?
nay
that look better!
Lora to give your work a pencil drawing style.
photoshop generative fill doesnt know what a rotary phone is
how much does it cost to buy a house? and where is the airship to get to? 💖
👍
hmm airships?
of course
generative fill on a green wallpaper in photoshop "THE IMAGES COULD NOT BE GENERATED BECAUSE THEY VIOLATED OUR GUIDELINES"
alright alright bruh relax photoshop
damn, isolated island, wonder how they get down to the water and ships, maybe with a rope?
haha I think it got confused
tries something different
not exactly waht I wanted
ya know thats how it should be spelled, waht, instead of what. makes more sound sense
ooooh
oops something blew up
will it float?
in space
oh noooooo
testing out dream studio with the credits I won 😛
Hello, Stastne is a cute anime character.
hello, very very nice anime, the anime character is very mega.
Hello, very very nice anime character, the picture is very very mega.
I bring uncertainty and death in times of peace, welcome me with open arms and I shall equally open mine that are so long as to stab every person! muahahaaha
She is a very pretty miku anime character.
did you prompt that?
Then here's one
She wanders the endless road buried under ages of sand She needs not water nor food to live Yets he withers She walks into a camp of strangers And is attacked for looking like a crack zombie The sky descends to eat them dead On top a mountain of lego, playdough and transformer toys
I made sure to be as random and inconsistent as I could
while still maintaining concistency
pretty sure the model im using wont understand such natural language prompting
none of my homies understand such natural language prompting
ok this then
1girl, magenta hair, glowing blue eyes, skinsuit, black leather with white vertical racing stripes, piloting, mecha kaiju, ancient beast, with cybernetic control, stands in the blackened ocean, winds blowing, cat 5 hurricane, cybernetics neon lights shine up the docks full of military tanks and police, 2.5d digital illustration, anime art style, hard lighting, dim lighting, cinematic angle, 8k, hdr, octane render
was sdxl really this good with text??
You mean like this ?
pretty sure i got some good result with words as well
on sdxl?
yup, the website version (clipdrop)
not that bad, not exactly the mecha I wanted but not bad
it made pixel art thats neat
WDXL attempt of a mining spaceship (some inpaint included)
those are awesome
So I was messing with settings and can't get this back to None instead of a LoRA selected. I've even reloaded the UI, completely closed out and reopened everything... help? 😂
might have to edit a json
Yep, it was in the config json. Thanks!!!
anyone running A1111 on WSL?
almost got a cool addams family thing going
wonder if I can use edge or chrome to download my webui as an app
heh it worked
got auto and invoke in my startmenu now as apps
When you interrupt a gen mid-upscale lol
I'm afraid you haven't seen early midjourney mid gens, they are true nightmare fuel
like wow
Oh gosh sdlfkjfdjl
there's often a lot of MJ vomit that's spread around like foam latex
now I have to decide on a night movie
something horror
im watching this black mirror bandersnatch rn its not horror exactly but its good so far
that worked awesome
saw up to the very long mid season episode
the app installation of the webui?
yeah on edge
no clucnky address bar
now having it separated from web is nice
omg the other day i was saying they needed to have lower desks characters on irl star trek and today a crossover episode game out
i img2imgd a pic of my doggie
thats so cool
conways game of life with random artist lol
witch (rate /10 please)
idk why my pc refuses yep its really good at text!
Huge
big wings
and with this i'm going back to bed
Hey i need someone to like do abit of experimenting if its okay with you guys
Prompting this
Well except curtaincall
and the fourth season. This is all...
got this on accident
It's really hard to prompt it though
Deadly limes
hmm not sure how that came about. gonna reboot. had a torch crash in auto1111 and my whole screen went dark and came back up.
and back
there we go
now let's get crazy
that went sideways
hehe and so did her face
😄 she's got a pet
accidentally activated clip skip 11, turned out OK
damn good actually
don't click that
sets clipskip 11 and his go
watis for either a masterpiece or a piece of horror
ok not as bad as I thought...
i`am sorry.I used your prompt a little
Don't worry it was already partly stolen
huh... then it's ok

ok so I still have another 24h on that cow, I forgot which day it was
Been traveling about 2 weeks... what'd I miss? 🙂
Not a grid btw
that racoon that was on the road when you were on the road home
it's still alive @keen hollow
you lost me, @cyan burrow
you asked me what you missed
the racoon
when people don't join me in on the weird it makes me feel weird
SD-XL and WD-XL beta releases and AnimateDiff (gen short, coherent gifs using your own normal sd1.5 based models)
Interesting, I'll see if i can find some get-up-to-speed video on the releases. 🙂
This is the best AnimateDiff implementation imo. https://github.com/neggles/animatediff-cli
wd-xl isn't on dreamstudio is it?
Thanks I'll scope it out
I've heard of sdxl because of the news here -- what is wdxl?
can 8gb vram run this? Cos everywhere i looked, it's 12gb vram
the team over at waifu diffusion finetune a version of their own sdxl
sdxl was just a new sd model wasn't it?
yup
damn even dreamstudio taking awhile to render my prompts
or at least that's my understanding of it
Ya'll created anything interesting with this new gif thing? Would love to see some of your creations/examples 🙂
pretty sure nubby created some stuff 
lemme see nubby!
pretty neat
she look better and better!
boil your chibi day
I don't boil my chibis only water them
model?
Huge
I don't see the appeal of breasts so big that gravity is actually going to make the skin stretch more and more each year and make them look like long burritos
darkSushi25D25D_v30
still too small

I'm not seeing objects in that world being pulled from the ground and the walls to being orbiting around her tits, do you?
make her on Mars, then they will levitate
you might be able to just barely gen at 512x512 on 8gb. I've been doing 512x768 and it uses 8.5gb on my system.
you are not using the auto1111 extension version right?
Right, I'm using Neggles' refactor that I linked. It uses xformers and some mixed precision shenanigans to be faster and more efficient than the official repo. From what I hear, the extension gives way worse outputs.
im too lazy to do that 
how was it to be able to put the clip and vae up to change?
You mean in the quicksettings list at the top of the WebUI?
In the Settings tab under User Interface.
thanks
what is unipc
how can i install this models faster https://huggingface.co/lllyasviel/ControlNet-v1-1/tree/main
without going one by one
it's another sampler that has some extra parameters that can be adjusted.
If you have git installed you can open a cmd in your .\stable-diffusion-webui\models\ControlNet then run the following commands:
git lfs install
git clone https://huggingface.co/lllyasviel/ControlNet-v1-1
This is the same thing as cloning a github repository, but requires the extra step of git lfs install due to the file sizes.
anyone messing with any new models lately?
I released a new one at the end of june https://civitai.com/models/100505/kakigori
I'm actually working on what could be the next version of it right now.
right on, ill scope it out
Heh I have every webui installed as an app now
not official app
but still
this will free up browser ram
Does it though? It's still running in your browser, just a separate instance of your browser. I would think it would create even more overhead if you're going to have other pages open in your full browser anyways.
What were your before and after ram consumption numbers?
show me
A merge of several different models, Elegant Entropy, aZovyaRPGArtistTools, mistoonAnime to create a new Merged mixed style. works pretty well I th...
hello guys
hmm might need to change name as it sorta does anime but more realistic style.. hmmm
so it's sorta realistic anime style?
shrugs to himself
cg fantasy?
puts the word Miku and uses magical random prompting ext
oh my bad dynamic prompts
panda girl
Ghibli studio inspired anime scene
the rest is a saved style prompt and then random
not what i was hoping for, but sweet, still trying to produce an image from underwater
better delete it, too realistic. it gives me creeps
genarate
switches to the anime merge and tries again
for a sec I thought that was a banana she had
same as last prompt
same see even
that was the rpg tool model
my merge
I swear if she sneezes wrong
here more, kinda anime
no?

Nice.
Your hands are making me cry
Btw, you using clip skip 1 or 2?
How about you @south sable
Go to sleep! 
2 on these
there
ok.... goes to bed
U using something for your perfect eyes? If you are pls tell
dragons are efficient. especially those who fart into the fire. give em a green medal. Produced in China but sourced in Kongo(the corresponding one with democratic in it's name, turned into components on Vietnam, Taiwan, South Africa, Chille, Uruguay, Sweden. Ship all the regs to China to put them into medals and then ship to countries where the green party needs them. Then they throw on a climate smart sticker because nobody in their country did it
oh I took a detour
but it's ridiculous how many places things come from. but then again they need to make money
when you were a kid and called to the principle
I think SDXL once it has a few good base models we'll have a new standard
cause the mergers will go nuts
idk what to make
its your? or you just want to repeat it?
i just cant think of interesting prompt today
goth outfit wildcard keeps giving me photos of feet-knees shot lol
maybe like this is
btw. main question how make a bottom black part
I'm guessing: generate castle, generate girl, cut out girl using one of the background removal extensions, put the two together in an image editing program
i am using just controlNet and make it in one pass
Nice
oh cheated a little. made the castle a separate passage
Like... generate castle, generate girl, put them together in Photoshop, run through Canny preprocessor, and make a new image based on that?
(That's what I had to do for an image I made, to fix background inconsistencies to the left/right of the character)
nope. generate castle, then use a controlNet make a pose girl and depth for castle and generate use a two Unit in controlNet
3AM bocchi be like
I see familiar character
heh.. just for the purpose of learning and gaining experience!
Greetings people of machines, I showed my friend pictures I obtained here and he asked me to prompt: Yui Yuigahama , can you?
Ah yes asking for friend 
Here's wd14 tagger output for this image I found from google
1girl, solo, yuigahama yui, skirt, shirt, jacket, white shirt, black jacket, open mouth, school uniform, brown footwear, smile, shoes, blazer, full body, sobu high school uniform, hair bun, single hair bun, collared shirt, socks, pleated skirt, looking at viewer, loafers, black background, :d, standing, plaid, black socks, simple background, plaid skirt, pink hair, outline, bangs, ribbon, kneehighs, red ribbon, dress shirt, border, long sleeves, neck ribbon, white outline, hand up, open jacket, brown eyes, short hair, bow, open clothes, red bow, single side bun
👍
it wasnt really intentional, it might have been from the artist or art movement token
Maybe. but it was interesting to get such a result using the functional SD
and.. Sensey`s who know why do I get this result and not inverted as written in the perprocessor
or black Backgroung and white line is a normal?
rofl signed up for a adult dating site, 5 minutes later 10 messages. Either its botted to hell or Swedish Women are more desperate than I thought
i can see you are looking for a catfish
i can be your catfish for 1000$ month
why looking for mediocrity when you can give money to a pro? lol
i will even provide for you daily realistic ai generated photo of an inesistant woman
ill do it for 3/4 that price
silence, you are not a pro lol
ok 5/4ths
only i0m certified
😦
an aunty
working a new style
Will you marry Eula?
fear me
thats great
how did u get the fish eye lense type look
i don't think there's really fish eye effect here tho

you can try prompting for fish eye if you want tho 

any suggestions for best multidiffusion upscale sampler for semi-photorealistic stuff?
mangaka-sensey workroom!
alight tring something
tell me please who is she?
You talking about my oc?
maybe yes. "OC"?!
omg...
she really kawaii!
then it should be promoted!
One more 
😂 why?
hi .....
some model and/or prompt to do something like this in txt2Img .... and not in Img2Img
lineart recommendations will be greatly appreciated 🙏
dont blink
timey wimey wibbly wobbley gobbly goobley
has u seen ai video in any tv shows or movies yet
i keep expecting ot see some in like a black mirror episode or something
theyve had that in black mirror years ago
and ai movies have been around for years to o
a tv show all about AI ran for 5 seasons
Persons of Interest
i wanna make vibeo
try maybe like that (Glowing neon:1.1) (lineart:1.3)
(dark background:1.2), (darkness:1.2)
@gloomy scroll I didn't want that much fanservice. i`am so sorry
with sundress I normally use see-through in negative
I will take this into account
I wonder if babydoll image would be too much for this server 
I'll compromise with sundress
I hope no\
Nickelback is...the terminator
Tmt: Have you seen this photo?
Barte: Uh yeah that's you
Tmt: What do you think? I've been mewing a lot
Bartender: the fuck is mewing?
Tmt: does mew face
Bartender: Looking emanciated is what mewing is? Okay. Go see Lenny, he mews 24/7
goes to Lenny
Lenny: Damn government took my dollar store and replaced it with fashion and toys...the fuck am I gonna do with fashion throws toy into tent
Tmt: Look at this photograph!
Lenny: That's you, nice headshot.
Tmt: Tell me how you obtained this mewing?
Lenny: the fuck?
*lift Lenny into the air
Tmt: ARgjhiiiuuuu I must know NAU
Lenny: I've been starving and homeless for 17 years. I can barely eat.
Tmt: So starvation is the key then.
hands Lenny his headshot
Tmt: LOOK AT THIS PHOTOGRAPH
Is there any model that has the same style as darkalfa but uses booru tags prompt?
anyone know how to write the prompt to generate such boots
Is there any model that has the same style as darkalfa but uses booru tags prompt?
no
ankle boots would be the booru tag for them, but the folded over cuff isn't something that has a tag. footwear ribbon might be helpful for getting the prominent bow at the front. Adding touhou along with those might work too.
how did you do this with that effect?
Private merge
https://civitai.com/models/13941?modelVersionId=18445
https://civitai.com/models/58390?modelVersionId=62833
Using these two.
merged with auto1111?
supermerger
hello
I'm working on a new version of Kakigori 👀
oo
in Img2Img
that was an error when copying data from a previous image controlnet was activated... with a strong guide based on lineart processor old preview
bonk her neck to normal
there , all except the text is generarated
even then I used cooltext.com for that so it is too 😄
Thts better
is this for a banner?
was just an idea I had
ahh, its a cool one then
i do!
their name is ✩
so was Prince's name for a period of time...
i'm just grumpy this evening don't mind me
didn't like that one either and thought it pretencious and full of it
shrug
well that turned out interestin
I love this magical prompting ext
ok it's actually called dynamic prompting
what she said
model?
Mixing This model is a continuation of my old Buttonmix, mixed using the scientific process of "Inebriated Mixing". I don't remember what exactly w...
thanks bbg
- this lora recipe " <lora:[LoConLoRA] pseudo-daylight偽日光 Concept_style:1> lora:DiscoElysiumStyle_v1:0.53 lora:Aetherion_style:0.33 lora:cutesexyrobuttsStyle_cutesexyrobuttsAnything:0.35 lora:Bloodborne-style:0.35, lora:vyvanLeAsianInstagram_v2:0.3 lora:himikoToga_v10:1 "
wow what a king 🙏
Lycanthorpe for a webnovel I'm writing.
not really anime but this is pretty cool
victorika de blua?
AI keeps turning hair into cloth
Sensei`s have a question. how i can uninstall extentions in a1111-sd-webui? not just uncheck box.
unbox them and than delete the directory in stable-diffusion-webui\extensions\
🙇
ghibli model?
pretty probable
SDXL 0.9
no this is the base image, unrefined even
the refined gets worse, then the upscaled gets better but loses the really nice film grain/blur aesthetic this has.
wow
this is a successful result through upscale
must be retouched here and there, but is spettacular
in a way its cheating because these shapes are eaier for it to do in general compared to people but its still pretty good for SDXL
this one is interesting
I've seen artists that do this. Take a blurry photo and draw a mech
that wasan't a real photo? i remember something similar
right?
i mean im sure its based like a million photos or something
probably the summation of everything we thought we remembered about gundam lol.
weird but still cool
cute, but from where comes out the umbrella?!!!!!!!!
i don't even know how to replicate the details of this image in another prompt, it's so damn random!
unexpected result 🫣
SD realized her mistake.
you should add the car in the prompt

just like me when I broke everything in SD
perfectly conveyed emotions on the face and in the pose
btw. what the mean superb in prompt?
armhair lmao
Optical illusion
how did you get that backdrop?
bruh dis amazing how u do dis? ik the model
i `am using a very cool and kawaii model @supple raptor https://civitai.com/models/100505/kakigori
About the model: My goal with this mix was to create a mostly 2d anime style model that responds well to varied prompts, including a relatively wid...
everything else in png info
based
yes
dawg 😭
sorry. I did not want! true true! it almost always goes like this....
lmao its ok it generated a pretty good one jus now 💀

been coloring chainsaw man panels
I didn't know is this Included as anime but whatever
She smiled at me and teleported away

👋
then we will consider her school time exclusively fan service
superb just means "very good" on some models it improves images a little bit. Lots of my prompts have it in them.
I just realized where the car came from ... This is the Skoda Superb
🤗
Synthwave shot in the car.
Update ETA?

metadata
Hey. This is amazing. Which model and Lora did you use?
I make goblin 
i like them too


O wao das a really cute style 
Kakigori got some upgrades 
Kakigori? Das a new one to me 
Doesn't know my model 
Lol I don't blame you. My model isn't exactly the style you typically gen either
Hey Which model and Lora did you use here?
I mean true, it do be cute tho
It's all in PNG info
I think I got it updated to the latest
the latest that supports sdxl?
Ohhh, the dev branch? I haven't gotten that yet :o
https://www.reddit.com/r/StableDiffusion/comments/158zob5/automatic1111_updated_to_150_version/ pretty sure it's pushed to the main branch
O wait 1.5.0
O no I don't have that
I should get that :O
I got 1.4.1 :O
Imma update now :o
good luck
heard you need to remove the lycoris extension since it's supported natively now
you don't have to remove it. but you can
oh really? You at the latest version?
I think Auto said that in that in the changelog :o
Not yet, but i've been running 1.5 from the rc branch for a while
Oh, fair then :O
im scared of updating 
Before the SDXL support was released it said that the extension didn't need to be removed :o
oh it just happened quite by accident. and used a wonderful kawaii model @supple raptor
https://civitai.com/models/100505/kakigori
About the model: My goal with this mix was to create a mostly 2d anime style model that responds well to varied prompts, including a relatively wid...
@lyric zealot Everything seems to be fine, haven't disabled any extensions and I just updated 


imma take a leap of faith 
Yis, it seems fine 
Hai Nerf 




