#🍥|anime
1 messages · Page 133 of 1
holy fuck X2
TEACH ME
you have a lot of prompts at high weight, so your other prompts would have decreased weightage
I AM A STUDENT TEACH MEEE
im blushing
fr fr fr teach me
I can give ya my prompt and show ya some stuffs
ALL HAIL BOOBA
yes yes
teach us master The Lord of Archives
Sure thing
((reference sheet, full body:1.5)), (concept art:1.3), (a watercolor painting of a girl, (Agostino Carracci)), ((different perspectives:1.3)) masterpiece, extremely detailed, page border,
BRAKE
,(a cute girl, gothic art image of Systemy girl with pale skin, Antonio J. Manzanedo, Ashley Wood:1.2), full body, (midriff), (makima:1.2) close up, (makima_eye:1.2), (glowing_eye:1.2), scary, detailed_eyes, , (duality_style:0.3), (line_style:1), (minimal_gradient:0.7), (nixeu_basic2:0.7), (nixeu_extra:0.7),(nixeu_soft:0.7),(nixeu_white:0.7), (dark_fantasy:1), (flame_surge_style:1), bloody, evil, dark,moody, spooky background, villian, colorful, beautiful, braided, somber expression, looking down, dark energy, colorful, vibrant colors, (a cute girl, image of Systemy girl, Antonio J. Manzanedo:1.2), Massive tree of white stone city at the base, forest, storm clouds Jean-Baptiste Monge, (Arcimboldo, Utagawa Hiroshige,) masterpiece, best quality, CG, wallpaper, HDR, high quality, high-definition, extremely detailed
Negative prompt: (((multiple girls))), lowres, text, error, missing fingers, extra digit, fewer digits, cropped, worst quality, low quality, jpeg artifacts, signature, watermark, out of frame, extra fingers, mutated hands, (poorly drawn hands), (poorly drawn face), (mutation), (deformed) breasts), (ugly), blurry, (bad anatomy), (bad proportions), (extra limbs), cloned face, flat color, sketch
Steps: 25, Sampler: DPM++ 2M Karras, CFG scale: 5, Seed: 668708869, Size: 960x640, Model hash: bb32ad727a, Model: darkSushi25D25D_v40, Version: v1.3.2
so here is all that is needed to make what im making I dont really use Lora cus it makes it harder for people to replicate whatever i make
that is beautiful
So the way i have been doing it the passed two days is like this
have ((reference sheet, full body:1.5)) and all that other stuff seporate from the real discription of the people im trying to make
something like this?
nope
i forgot to adjust the size
demon ladys are hawt
it made me a elden ring charcter sheet
but anywho I set the discription apart from the whole ((reference sheet, full body: bla bla bal so i can just add new prompts for what i want the charater or whatever to look like coming out
thats cool
((reference sheet, full body:1.5)), (concept art:1.3), (a watercolor painting of a girl, (Agostino Carracci)), ((different perspectives:1.3)) masterpiece, extremely detailed, page border,
BRAKE --- this then whatever you want after that
for it to look like the ones ive sent so fart
far lol
turned out really nice
i have been taught by the master
your to kind
bless youlord of archives
stuff like this one is also cool
sure let me find it and ive sent this one here before but something you might like
alice in wonderland type of vibes
https://civitai.com/models/24779/dark-sushi-mix-mix this one is more or less identical but the one I am using seems like the one i got vanished for the time being
hey, how do you do the controlnet img2img upscale technique? where you can get high denoising, but still get it to look a lot like your original pic
cute
using my ai to enhance my ai
thats awesome i love the natural science textbook type layout
I use ultimate SD upscale script and the controlnet tiles model
I think you download the ultimate one as an extension
Not sure if it's default
what about just tiled diffusion? or is that too many tiles
Tiled diffusion is better tbh
or should I just.. double the res, use tiled vae
I dunno how to use tiled diffusion
Best is probably loopback for initial upscale to add detail
I don't use it that much though since it requires a lot tweaking 
do you use "Upload independent control image"? or will it use the img2img as default
Default
seems like it needs independent image for tiled controlnet, I got randomness without it
Looks like midjourney/niji
good ol' sd 🙂
looks like samyang
I dunno what niji is 😐
😮
yeah, there's some of that in there 🙂
niji is midjourney's anime model
never used midjourney
Anime offshoot of (mid)journey
unfortunately MJ is discord only
and quite expensive
dalle3 is more impressive now imo
haven't used that either
search "bing create" and you can do some neat ones
Are you using the script together?
oh turns out I forgot to set the model on CN
Oh
I'm finding at certain folds, it is wanting to generate too much detail and getting mad wrinkly
oh hmm, yes I guess I did try that not too long ago
How much denoise are you using
0.9

hehe
I thought that was the main point of the CN upscale
to let you get mad denoising
Why so much
I think it previously existed but was Dall-e 2 which wasn't very good. It was recently updated to Dall-e 3 and is surprisingly good if you look past the annoying prompt blocks.
I never go above 0.5 for an upscale, usually less 
I'll try 0.5 and set controlnet to more important
Is your controlnet set to tile?
usually I'd do 0.35 for upscale but with CN Tile I thought the aim was go high
yeah
it's definitely working
maybe my tiled diffusion is too small of tiles
0.9 is almost changing the entire picture
hmm, my lora isn't showing up
oh, interesting, ill try it
Cnet tile just keeps the cohesion so it doesn't hallucinate your prompt into individual tiles
Yeah, new dall-e not bad. Random 'robot chicken'
yeah, which is why 0.9 denoising can work
I guess I might be expecting too much for certain small details in upscaling
it also understands "half-robot" quite well, which is surprising
slow as crap though 😄
You can get some crazy stuff like 3 or 4 characters all doing a specific thing in the prompt
It does images really fast
We get the slow part because of wait time
Sometimes it generates 4 images in 10 seconds for me
At 0.9 it's gonna add too much detail while only keeping general shapes, even with cnet tile
oh, maybe auto1111 doesn't support sdxl loras?
This dall-e thing blocks like... everything.
Booba
oh.. pff.. I didn't have an sdxl model loaded. : facepalm :
wow... dall-e pretty good.
Sometimes.. too good...

That Pikachu generation image is stuck in my brain
interface is buggy as crap though
man I wish I had this local 😛
it just absolutely does not want to create Tifa
You did it
yeah it's ridiculous, but somehow 4chan still making hitler gens
I think it's just square for now
i was using it to gen concepts, then I img2img in SD to change dimensions or lewdify
what model in SD to line up?
any
This is pretty nuts... would struggle to get some of these with SD...
doesn't matter
are you kidding me?
I'm almost out of "coins" or whatever
now I'm farting around in this and SD at the same time. 😛
lol yeah gives you double the gens to do
I had 100 coins to start
idk if they refill daily or what
Continue creating with weekly boosts!
When you run out of boosts, you can still create. However, image generation may take longer.
holy cow
I wish I could "send to img2img" without controlnet disabling itself
is this dalle3?
wish i could run both servers at the same time lol
I like it
Running SD and Dall-e at the same time, they just be crankin' out. 😄
getting neat characters
I wish I can post some SD anime but I am having issues with Stable diffusion and I don't know where to go to get help

those are crazy my mind always assumes those are just phptoschopped i forget sdxl can do stuff like that
my hands hurt
that was the new dall-e 3. 🙂
ooo
Jesus christ no wonder they hurt only got 4 fingers
aa
im ok i guess just been having life problems feeling overwhelmed but otherwise ok i guess idk, how about u?
honestly pretty much the same 😄

i cheated on the test
ive been addicted to watching matt dillahunty videos
whozat?
bonk all you want you know i'm right
why not just image2image outpaint?
i've never had much luck with outpaint
wait a sec why are her legs covered 
What do you mean

There
big wings
would u show me how 2 do graffiti
got cloud wings this one time
Neutral prompt extension
I just prompt for a girl and whatever + graffiti
And add AND_SALT Large wings
So the extension adds wings in it as it sees fit
im listening carefully
Many secrets Booba is always the secret
im trying too
this dall-e thing has like.. ruined my expectations
if you know who that is 10 points to gryffindor
Is that Odin when he had both of his eyes?

Yeah, Dall-e 3 is pretty good. All of these are from simple prompts with no editing.
I ran outta generations 😦
Sadge. I don't think I've let myself drop down below 30 so far. You'll get more tomorrow.
i think they moved it to weekly
it gives popups talking about weekly but mine refreshed last night
and each night before that

take it bigger
If you're seeing things distorted go rest
the important things are not distorted 
Is that 6 toes on one foot?
i love the beach
👍
guys all of you
I asked for multiple tails.... I guess I got them.
average anime villain
thems a lotta tails
yesh
It's a model trained on Matcha and a LoRA that's supposed to turn everything into fire. It's bound to do the funni
lets wear a corset in the beach
night omnia 
now look at those shoulders
The pointing gun pic is crazy good with the hands 
I always have to fix the hands when generating gun pics in SD 
SDXL is pretty good with the general outline and how they hold it but the fingers are still hit or miss
The problem rn is that it's always down for me

Some of my favorite styles
now make her pregnant 🫃
I could do that 

But I wont 
allow me

🫃
dont do it
that's dalle pregger?
yea dalle
lol
bro I'm getting censored by bots trying to make meta comments about censorship by bots
I've been enjoying dalle3 as an img2img base
im planning to img2img a drawing my friend made. heres the drawing im gonna crop it a bit and maybe ask for a better angle, but how should i prompt it
maybe abstract pattern art
ok
i wish i could get the infinite zoom addon working good this would be acool candidate
I'll try this too
Robocat
idk what happened here
here's a few tries from me abstract art, colorful, vector art
the 4th is most like their actual art
my new form
awesomee
Need to know every blocked prompts though
yeah it should not consume my token imo
ah yeah
i asked them for a dead on angle shot but they havent sent it yet
im gonna send them those and see how they like em thank u
im cookin rice and a frozen borrito
😮
tell them someone named Lotta Lewds created it
and that I hid a lewd image inside it somewhere
lolok
best crossover
lmao
I also hope that it stays free lol
so thats where pokeballs come from
omg
'millions of teeth' 'too many teeth'
🍑
I've spent the last few days trying to get Kohya to work, after hundreds of errors I finally get it starting and get hit with a Cuda Out of Memory error...
this one omg
needs more thicc
Is that still from dalle
Dall3
Or something I forgot the name lol
I actually have tried creating stuff from there
the compositions are so good
more teeth
damn that's cool
wwhats kohya
vampire girls probably rock with this
First try, not bad
Tried to gen werewolf girl, for this instead
sharing my Full AI music video using Stable Diffusion with anything-v3-fp32-pruned + my own trained Lora https://www.instagram.com/reel/CxFhN_TxgAM/
in there - music video [out now] link in bio 🫶🏽
so so happy to collaborate with @izitheartist on this. ⸜(。˃ ᵕ ˂ )⸝♡ he spent months researching how to train personalized AI models, then made one of me to make this music video 🤯
so cool to see this come to life cus 'in there' was inspired by a convo we had about an anime show ᵔᵕᵔ
#aiart #aiart...
159
Kohya is a tool used for training Lora's

please let's go easy on the nsfw stuff
we're getting reports and 4 were just you hehehe, we can be creative without being too explicit
models*
and preparing datasets
well depends on which ui you choose (or none)
technically speaking kohya is a guy who codes
but usually when people say stuff like "use kohya" they either mean the scripts (https://github.com/kohya-ss/sd-scripts) or the GUI written/managed by bmaltais (https://github.com/bmaltais/kohya_ss)
no need to thank me for this useless info
Yes you're right lmao, not just Lora but models in general
Didn't recognize it was named after a guy though lmao
Still though, the problems I'm having still exists... I did find some info when I was at work (just got back home)
Gonna try a few things and hope it works
What the
Lmao
ye
Didn't know that either
I did find the name odd, and it does appear like a username
Just never really thought about it
both comfy and auto1111 are here
in this server
tho automatic1111 is a weird creature
and primarily uses 4chan
was banned from here like twice
and comfy is cool ig
If that's the way things are, I'll name something after myself! Perhaps the anime model I'm trying to work on... Eh, think I should use an alias or something like Rai Soft or something... PlayStateWolf actually has meaning in it's name
or name it like how I named my old anime merge
Saucity Sauce
has a lot of sauce indeed
Lmao
Well, I haven't decided on a name yet
Also, I think I figured out why my Lora isn't computing
Seems like I got a lot more work to do to get it working
if you need help I just spent 4 weeks trying to train a good lora 💀
if you want to know how things work or affect the output
Hewwo people
Gangster Miku
neuron activation
I'm sorry, I'll post better 🙇♂️
The problem I'm having is getting started, not perfecting Lora making, just trying to get it to run at all is a challenge
Hewwo 
Hai
ohhhh okay
I have 8gb vram and figured out how
but if u still need help 🫡
I figured out something
Wait I just realized tis was you, how you make this? With what? 
Cheeky
I believe I know why it chugs on memory
Fair 
Prompt ? 

1girl, leaning forward, hand to own mouth, :d, one arm behind back, camisole, skirt, chibi, dark-skinned female, short hair, brown hair, ponytail, yellow eyes, half-closed eyes
I believe it's due to the sheer number of individual buckets. I have a bit of understanding regarding interfacing with GPU's such as writing shaders and stuff (game development)
I believe I need to crop all my images into a standard format such as 512x512
I think it would reduce the strain on the GPU to allocate less buckets
Interesting. I trained on about 500 images with 50 buckets but I don't know if that's a lot
I just kinda didn't crop any and let it figure itself out
I am using extremely minimal settings, only using 50+ images, and all setting at bare minimum
I tried this but
Cuda errors
It's the only thing I can think of
I don't even know, I think 4GB
💀 oh that's why
I've been flipping between AdamW8bit and AdamW
I'm pretty sure you need 8gb minimum. maybe I would keep trying but I doubt it
I can't tell though
mine runs about 7.8 g of memory at adam8bit which is supposed to be low usage
To train loras, usually people use 12~ but can work with 8
@past junco have you tried collab for loras
Cyberpunk 
allocated; 0 bytes free; 3.46 GiB reserved in total by PyTorch)```
It acts as if I only have 4GB vram
I mean, I'm not sure if LoRA training plays too nice with Intel GPUs
No and I'm not particularly sold on colab
I'm actually using Nvidia, wait, maybe that's it, maybe the Nvidia GPU uses 4GB
Hmm

So you have two GPUs?
I made one on Collab using an automated guide
Not sure if it's still working cause I keep hearing it's dead D:
dxdiag on windows search
I see
And second tab has info on gpu
Yeap
Damn...
Hello there 
I really thought my laptop was high in specs
Guizmussssss, we missed ya 
Hai
True, a lotta peaches 
In my around 2 years in this community (SD)
Anime has been the horniest one, and also the most qualitative one I've come across
You guys put me through some sweats, I'm not gonna lie
I mean, people are weebs, and horny, its a match made in heaven 
Fair 
Hey I don't complain
WAAAAAAAAAAAAAAAA 


Happy to see the anime-chan still alive
BAH, my emoji doesn't wanna work
Clari would agree but I've lost my credentials lol
Nice to meet you though Guizmus, I'm kind of new(?)
Oh ooof, gotta make Clari 2 xD
I've been a mod around for a long time, on leave for a while
I've came back and asked for less responsability
Ah I see
Also I just got a really dope android generation 
Can't handle being the NSFW police anymore
Fair fair, lotta stress? 
I help with #📝|prompting-help when I can
Too much 
Also were you on vacay? Had fun times? 

Honestly? Yes, but it's my fault. It's never been asked of me, but I take it as I would take a full time job and make it hell for myself
I'm currently contemplating if SD is really worth getting an upgraded setup
You'd need a PC right
Just been out and around. I've been dealing with tough shit since it came on this server (mother died), and I've been up or down, depending on the days. I've just been dogsitting 4 dogs for a full month though, so that's a thing I guess
I don't know
Ohhh thats fair then, I think I getya tho, I think I would have done the same and taken it too like... formal or serious and would have eventually burnt me out too, so I think das just normal 
But I wouldn't go messing with it personally
Oh damn that sucks, sorry for your loss, and I see, you've been just really busy then :o
Yeah it sucks... I lost someone who basically was a mother to me
Depends on your personal interest in the matter. I build a custom top of the line last year just before the 40xx series dropped and I don't regret it, but really depends on your cash/personal interest in AI
Cash isn't a problem, but I can't go spending crazy especially for a toy
I mean, at the start of the year I had a GTX 960, and I upgraded to 3060 12 GB because I was using the same card for like 5 years, and for AI 
Would those portable GPU setups work for sd
Forgot the name exactly
I mean, a GPU is good to have, and will last you a while 
Most of what I do is gaming/AI/streaming/programming related so I went for it. But yeah, it's complicated to say. My personal advice is, stay on low hardware, even use online services like paid rented GPU to run it as long as you can, and build a full setup when it becomes clear you are in it for the long term.
The problem is
When I started posting here and kind of been around frequently
I considered I wanted to take SD seriously
Actually do stuff
Both for myself and others
I really feel like I hit the jackpot with what I've done with SD in the short time I've been using it
The art style I've picked up is something to be admired
So I really am considering putting money towards it, just got to plan around it somehow
I'm a game developer lmao
To be fair, my current laptop isn't even a year old
So I think I should off and just put time on it
There will be a day
I'm not gonnag give up though
I still have lots of ways to generate my art
wow that's a good one
That 3050 is a fighter
Tell me more! Gamedev is an interesting topic to me. 🙂
3060 12GB here. It's been really solid for just having fun with SD.
Looks like Dall-e 3 might be down.
I went a little overdrive, built for training models and future proof a little
Went 64GB ram, 3090TI and just about 13TB of data storage
and my credits didn't refill over night for dalle either... dang
I think my next build will be whenever nvidia drops 5000 series
feels too late in the game to get into 4000
40xx series didn't feel the most interesting to me in the end... the benefits for AI weren't that high, and VRAM didn't improve from what I understood
Starfield didn't hit it off with me at all...
It's definitely got some jank, but I've been enjoying it so far. 🙂
But definitely prefer the Fallout world.
I felt disconnected starting from the first quest itself, and uninstalled once I met constellation
meow 
maybe you can revisit it later and try again
the worst version of a bethesda game is the day one version, i will see how it looks in a few years after all the dlcs and needed mods
I sure will, once it gets expended upon, and I've had more time since playing BG3.... it was a constant comparison in my mind, and the quest forcing me in a single path was already a big problem for me, while it wasn't that much a problem in the past
BG3 is supposedly really good. Havne't tried that one.
I did love BG3, and it did change a little how I view RPG, in the sense that I could decide for myself what to do, and the option worked. Like first thing I did in starfield was decline the ship and becoming a part of constellation. I didn't feel worth it. It was my first day as a miner and I just tripped balls on a mineral. Being refused that liberty was a little immersion breaking.
In the past, before playing BG3 that gives too much liberty, it wasn't. And I can't really ask such liberty from any RPG. But it was already a turn off.
Loading screens was the turning point
Larian are truly masters of their craft. Divinity: Original Sin 2 and Balder's Gate 3 are both masterpieces.
I still need to try original sin 2. Haven't had the time. Problem with Larian is that their games are hundreds of hours long, and quite the mental commitment too
I wanted to play through persona 5 royal recently too... like I loved that game, but it took me 300h the first time
The only problem I had with bg3 is that if you play with a full party of players, you miss out a shit ton of stuff
had to play it 2 times in order to fully enjoy it
Yeah, they would be a tough sell to a casual gamer who only plays maybe an hour or less per week.
They do have this problem of lots of content being locked behind using a character, even in solo given the number of characters. It's a problem and a good thing too, brings replayability
Yeah, they're best played with 1 or 2 people IMO.
There's definitely some "main" quests they will shoehorn you into.
yeah bing/dall-e is just down for me. They must be working on it.
Guess I'll head back to SD for some gens. 🙂
Need to figure out a good new workflow for ultrawide.
I can tell you all about it!
For some main they do. I didn't hit those points, but was able to refuse some main ones like the Raphael arc, and didn't even understood it could be that big.
Anyway, sorry, I'm still quite the fanboy it seems
What are you building in?
I built a VN engine in Unity that would use AI art a little while back. I guess Unity has really burned a lot of bridges recently.
Sorry I missed that point, really interested too, tell us more !
I mean, BG3 for me is like, the peak of RPG
So, I am very familiar with both 2D and 3D game dev. Did a lot of stuff with Unity back in the day but lately I've been into 2D game dev.
I actually studied at university, but dropped out because of money problems...
You don't buy their reversal recently ? I'm not sure what to build on recently
I dunno, it isn't the first time they've made questionable choices. So many people have bailed, so not sure what their future will be.
I'm only a hobbiest... I don't have the bandwidth to create and publish anything real. So it doesn't really impact me much.
If SD ever gets to a point of truly building full animated assets, maybe I'll have a chance some day... 😛
Well problem doesn't seem SD but the world for this to happen. steam rejecting games using AI made assets is a real problem for a start
Oh damn I didn't know Steam did that
it was mostly a reaction to spam i think, it will get better over time when people start seeing ai as a other tool like photoshop
Sorry again, lots of things happening here where I am
“As the legal ownership of such AI-generated art is unclear, we cannot ship your game while it contains these AI-generated assets, unless you can affirmatively confirm that you own the rights to all of the IP used in the data set that trained the AI to create the assets in your game,” their first warning letter stated.
Then, a week later: “we reviewed [Game Name Here] and took our time to better understand the AI tech used to create it. Again, while we strive to ship most titles submitted to us, we cannot ship games for which the developer does not have all of the necessary rights. At this time, we are declining to distribute your game since it’s unclear if the underlying AI tech used to create the assets has sufficient rights to the training data.”
https://www.google.com/amp/s/techcrunch.com/2023/07/03/valve-responds-to-claims-it-has-banned-ai-generated-games-from-steam/amp/
Not the most reliable source but I'm on mobile
Msft is backing copyright claims for those that use their copilot services. Maybe we'll see more of that.
why is it that the last "3 seconds" of an sdxl generation seem to take like 12 seconds?
vae is the heaviest part of generation.
dalle3 then uspcaled in SD
Okay so, most games I've successfully made are game jam games, however most games I've ever worked on were failed or put on infinite hiatus (such is the life of a game developer)
I like to think that I'm really good at it now, but the one thing I've always wanted to do was do it full time and actually sell games
the industry is pretty tough, that's for sure
Oh damn, thats pretty good :o

The game market is extremely oversaturated with the rise of easy game development tools
And what is that? :O
I'm a bit of an oldschool, hands dirty kind of person
Oh damn that sucks x.x
So these days I write my games in HaxeFlixel (and when I have the time to learn it, Godot)
What ?! You don't get unreal assets and make a game full out of them ?

Nice
I kind of been going through a dry spell lately, but my friends believe in me and they want to help out when they can
Australia has a crap ton of laws and taxes and stuff so I've been focused on free games
Until someone can literally tell me how to sell the games I make lol
I wouldn't mind learning one of those low code platforms like Construct or GDevelop.
Funny story, when I first started learning programming about 25 years ago, it was on my calculator, using Basic. I managed to do a 3d rendering engine using the Doom kind of camera. It was fun but like, one frame every 5 seconds. And not enough memory on the calculator to store more than 1 level
TI-83
I'm not sure anymore but maybe
O, on a calculator? 
LMAO
Yeah, typing code was a pain
Damn that is crazy 
I started with QBASIC 😐
people who don't post images with png info when I specifically like them
True and real 
... actually, i might be guilty of that
i copy/paste
I think that drops it
But anyone can ask any time! 😄
What about people that post PNG info, no picture. New meta ?
The true anonymous image posting
Can't get flagged for nsfw.
i'm thinking about redoing my SD image viewer to include parameters from some of the other generators like comfy...
Oh shit! This is de wae for people with no/weak GPUs 
Wel....
True, I can post my (groin:1.5) that way 
People using comfy tend to use custom nodes more in my opinion, could be hard to follow
Yeah maybe. The parameters within the image aren't too tough to dump. Can be a challenge to parse, but not sure I care. Might do a program that just displays. I dunno.
I'd need to use an image library to do the new thumbnailing I'd like to do. And I just... I don't know if I really wanna do it. 😛
maybe I can figure out how to do a plugin to file explorer or something
It needs like... 4 LoRAs
Add detail (civitai) + Midjourney (Civitai) + one that is fully in chinese (tusiart) + Light and dark (Civitai)
Params are ok to parse. But the workflow itself is part of the params and could include lots of models and process going in lots of directions, compared to the more straight forward automatic way.
Send links 
You could take a look at what they've done here. For Comfy gens, it just extracts the workflow.json from the image so that can be plugged into comfy easily. https://github.com/sALTaccount/PromptInspectorBot
If ya'll had a cool, vertical image, that you wanted to expand to an ultrawide ratio, what method would you use?
man 
this one #🍥|anime message
I've been listening to the same song for about an hour trying to prompt something 
help
I have them download for a pretty long time alrd, It'll take a while for me to find it lol
Or send the names/hash, I can search by hash on Civit 
And the chinese one I can probably find by looking it up
Though I got the add detail one, don't worry about that one 
O, pog
https://civitai.com/models/24584?modelVersionId=29405
I'm using V1.0
Alrighty, I'mma grab that 
not with sdxl 🙂
yeah... I don't use SDXL for several reasons. You could check out InvokeAI's outpainting. I think they might have SDXL support now.
no it is not, it is my dnd character
I know this is nowhere near anime but isn't it just beautiful?
Adrik Baldirk
what reasons? compatibility stuff like this?
tell us the reasons master nubby
well you need god's pc for a start for sdxl
and their nips purple
what do i do
put purple nipples in negative?
like this it keeps doing this
You need the unrealsed 6099TI to run half of sdXL they say in the street
I WANT HIM CLOTHED
add "ecchi, hentai, nude, naked, half naked, bare chest, eroge" to negative
will do
Add pieces of clothing in the description. Names of types of clothes from different inspirations. It should help
compatibility, generation speed, lack of models that know a broad range of anime concepts and characters, and difficulty of training.
ill try that out
even with my 3080 12GB the speed sacrifice to run XL just isn't worth it.
time to play some starfield, hogwarts and then start playing CPunk 2077 Phantom Liberty
starfield to pahntom liberty is a big leap
reminds me of this gen i did
starfield is a bethesda game like old bethesda game
Got it
It's weird cause SD 1.5 can more or less do the same things with the latest models
Tonight is cyberpunk for me too I think
I have 200+ hours in it, I know what Starfield is
dont skip straight into it
Oh, also the limitation of aspect ratios that can actually work decently.
W
I've made a new character, and have been at the main quest for a few days, just started the dlc 2 days ago, I still have to meet half of the cast
I don't want to make a new game to play phantom liberty, I put so many hours into it and I would force myself compulsively to do everything again
while the game is awesome and I hear even more awesome now, I don't want to repeat the story again
well yes
but I have so much loot stashed away for any build type
is there a NG+? If there is, does it let you keep loot and level? 😛
afaik, loot has been largely reworked anyways. game seems to be much less of a looter shooter now.
no new game plus
but restarting is so worth it
everything is pretty much redone
especially cyberware
I only played through the epilogue at launch and the game didn't convince me to buy it in that amount of time so I moved on. After seeing the recent changes I'm waiting for it to go on sale again and I'll probably pick it up.
the story isn't redone
50 steps 50 hires 1080-
so thus, I would be playing through the same stuff. Something I used to love doing with kotor etc, but there would need to be significant story beat changes. The game also can't be 100+ hours for a compulsive completionist
150 steps 150 hires, 1080P
same stuff but better
It's just so much I have to do to get to phantom liberty if I replay it, that I might burn myself out and lose interest 😛
thats why u take time with it
I can't really play that many games to get my downtime removed inbetween
I wouldn't know how to keep track of it all
oh yeah thats fair enough
so playing like 2-3 hours a day ain't gonna work
I'm bedridden a lot of the time either post surgery, for infection or during chemo (whenever next time is)
so I don't do a lot
so games are my escape
model with .7-1.0 denoising
im normally at home taking a year off from school waiting to go to college, i relate to playing games as an escape
I literally can't do much else in physical activity etc, so there's not many options
ugh why can I not just BUY dalle credits? this marketplace booster thing is so stupid
you wha
The Bing thing is just a preview. I assume the plan is to get people interested in it so that they can transition people over to ChatGPT and have them pay for gens.
which of these do ya'll like best?
touché
40 steps has the least artifacts, but 35 steps has a subtly different eye shape that I like better.
It is a nice looking lora
45 long hair hawt
she got sock on her hand though 
35 for me
she does sock puppets for the local kids
Some models work best with set amount of steps 35 has nice hands and looks like a nice upscaling base
exactly
I think my new checkpoint merge is gonna be fire
let's see it
eta 1:50
Different faces: Ya'll like the stylized, softer version more? Or the more detailed but somewhat genericized version?
oh my
how do i make my thangs pop
Do you know the power of this symbol ":"
LESS GOO CAT WITH 4GB OF RAM
@worthy delta hello good morning day night evening month
idk what it does
bro i slept for HOURS
I like the right one more
(that darker looking colors)
Here is its power| (white:pants), (orange:hair), (wearing:dress), (lightning:background)
the power of semicolon
THE POWER OF A GOD
Do you need an extension for that ?
nope
Here is another superpower
Use numbers to add the power that it takes over sertain things, etc you will see
what is a good upscaler?
kono powa
right on
(balding:1.4)
chaining semicolons can make some awesome resulits
my super power: "AND_SALT lightning" to stuff 
(black hair, balding, depression:1.4)
like this | (anime male:wearing tuxedo:golden), (black:red:lightning:city:background)
(brown hair, still as dense as when I was 20, a fivehead, resting bitch face, major depression:~~)
Anyway you asked to see my new checkpoint merge ive been working on,its name is....Pentellion
Super Pentellion to be exact
Pro-tip: don't ask Dall-e to make a sign that says headpat me in french
am french 🥖

to be fair we do love baguettes. I even saw some baguettes vending machine once
finally finished my darn lora
if anyone is interested ill give link i think
funny that the new disgaea game was released 3 days ago in english
and i just finished this. LOL
does it make big eyes ? 
probably
oh
its very flexible and really doesnt change the composition much so
disgaea is a game
LOL
yes
I'm having a stroke trying to type that in google
elp
disgaegae
I need a prompt to use with this style... I'm having a creative block
My checkpoint will drop in 27 mins
mine? it does 1girl and character design pretty well but itll do anything
maybe a monster or demon girl



