/sdg/ - Stable Diffusion General

Old Thread: >Local install
GPU: rentry.org/voldy
CPU: rentry.org/cputard

>AMD GPU Guides
Docker: rentry.org/sdamd
Onnx: rentry.org/ayymd-stable-diffustion-v1_4-guide
Docker on Arch: rentry.org/amddockerarch

>M1 Supported Fork
github.com/lstein/stable-diffusion

>Cloud Hosted Install
Colab: colab.research.google.com/drive/1Iy-xW9t1-OQWhb0hNxueGij8phCyluOh
Paperspace: rentry.org/865dy

>SaaS Sites
txt2img: huggingface.co/spaces/stabilityai/stable-diffusion | beta.dreamstudio.ai | dezgo.com | (NSFW) pornpen.ai
img2img: huggingface.co/spaces/huggingface/diffuse-the-rest
inpainting: huggingface.co/spaces/fffiloni/stable-diffusion-inpainting | inpainter.vercel.app/paint

>Krita Plug-ins
github.com/sddebz/stable-diffusion-krita-plugin
flyingdog.de/sd/en

>Other Guides
Inpainting/Outpainting: rentry.org/drfar
Upscaling images: rentry.org/sdupscale
Textual inversion: rentry.org/textard
Training: rentry.org/informal-training-guide

>In-Depth Usage Info
Any Forums wiki: wiki.installgentoo.com/wiki/Stable_Diffusion
SD Akashic Records: github.com/Maks-s/sd-akashic

>Tools & Resources
Big index: pharmapsychotic.com/tools.html
Artist styles: rentry.org/artists_sd-v1-4 | urania.ai/top-sd-artists | sgreens.notion.site/sgreens/4ca6f4e229e24da6845b6d49e6b08ae7
Prompt sharing: lexica.art/ | krea.ai/
Dataset: rom1504.github.io/clip-retrieval/ | laion-aesthetic.datasette.io/laion-aesthetic-6pls/images

>Finetuned Models
rentry.org/yrpvv

>Related
/h/ /sdhb/: /vg/ /aids/: /mlp/ /ppp/: Official: discord.gg/stablediffusion
NSFW: discord.gg/2ZhBzDGf8u
Training: discord.gg/xVsyrmhQWS

Stability.ai updates: twitter.com/EMostaque

Attached: 00022.jpg (2048x3072, 909.01K)

READ THE FUCKING GUIDE BEFORE YOU ASK A STUPID QUESTION. THE NEXT FAGGOT TO SAY "CAN THIS RUN ON (INSERT OS) OR HOW MUCH VRAM DO I NEED TO COOM WITH THIS" GETS THE ROPE.

please give me prompts to make my anime waifu real, i'm currently using dreamstudio to "learn".
I love the painted realistic cute girl style so recommend me good promps to get close to Yunyun

Attached: dsasdsad.png (843x1193, 1005.76K)

I like snakes!

Attached: 01778-331692306-a girl with pink hair holding the body of a brown snake, colorized.png (448x512, 247.3K)

yeah but can this run on windows xp?

Can you run this on insert OS?

How are you guys getting realistic faces?

like

Rad the kind of things you can change with img2img solely via the prompt with no inpainting

Attached: ruined.jpg (1664x1792, 747.27K)

Why thread early?

>108 images
>new thread!

>add the medvram parameters
>now able to generate 512x512 at a batch size of 5
>could only do 1 at a time before
what wizardry is this

How much VRAM do I need to coom with this?

someone generate photorealistic asian catgirl feet pics please

muslim man eating and enjoying delicious bacon, realistic, sharp

Attached: 00044-1206311759-muslim man eating and enjoying delicious bacon, realistic, sharp.png (512x512, 322.68K)

Attached: 00415-140524663-Alphonse Mucha, Eugène Delacroix, carne griffiths, Franz Xaver Winterhalter, Lilia Alvarado, Sophie Anderson, ((perfect face)),.png (896x1280, 1.29M)

How to use a .pt file all text inversion stuff I've had so far comes as 4kb .bin files and this is a 33kb .pt file?

Settings, select Code Former, then turn Restore faces on in txt2img

unironically tell the AI to not do bad faces
also chack the restore faces box

are you suppose to copy this in git bash?
It said it downloaded something.

Attached: 20220916_001710.jpg (1013x379, 68.51K)

>stuck at installing torch
seems to take longer than the 10-15 minutes of the voldy guide, is it normal?
I'm using a 4 gb VRAM toaster btw.

now make him jewish

how much coom do I need to VRAM with this? also, how do I generate the pictures of memes that look like normal objects? also, how do I fix hands? how do I fix face? how do I remove censor? how do I shot web? also, PROOOOOOOOOOOOMPT?????

will it run on my amiga 500

Now add his friend who is beheading infidels standing next to him

Awesome, I really appreciate it. Getting decent anthro characters is tough.
…though I hate to ask, how do you use pt files?

What does this mean?

Attached: 1512323181.png (632x864, 142.84K)

Ok, but can this run on KolibriOS with an AMD GPU, 4GB RAM?

>KEK done
muslims praying to a statue of a pig, realistic

Attached: 00042-1005777859-muslims praying to a statue of a pig, realistic.png (512x512, 468.28K)

yes it's downloading 5GB

update the guide mentioning it

it tags images better

In lamens terms, think of it as an addon for SD that makes the AI better understand your prompt and what you want it to make for you.

learn to read first

Don't make it realistic, make it look artsy then shop some random artist name into it before spreading it around the internet

Are you suppose to copy this here?

Attached: 28d7cf05c5e2ff646cfb200fa6221101.png (568x364, 5.88K)

>faces, fingers and limbs slightly more fucked than dalle
this is still not at the same level as dalle is it?

will do

Attached: 1654725509548.jpg (600x800, 61.19K)

It means we're gonna coom even harder soon

That's wonderful, hopefully it starts understanding my prompts at dalle2 level soon.

man that is devilish

read a book nigger

No, but it isn't limited like dalle is either.

>Some artists gets bombed 1 week later.
Whoops!
???!??
Profit!

Someone summon Voldemort, he needs to add this shit

if it takes like over 30 minutes and your internet isn't shit delete venv and repositories folders and try again
it's a decent few gigs but it shouldnt take forever
yes
what guide is that even

Where do I find the guide?

CLIP is about the connection between images and words. a more powerful CLIP model means the ai will be better at connecting words to concepts, rather than just image tags. it makes the ai model better at general purpose tasks, at working with images it wasn't trained on, at recognizing objects from different angles or obscured behind things, etc. it's a big deal for developers, but less impressive for end users right now

Cursed be those that share their images but do not share their prompts.

Check inside your anus.

yes

So, when do they drop it? Sounds great, I hope we'll finally get proper horse-riding astronauts and cockroach cola

Mein Gott!

Attached: 1663280530753.png (512x704, 329.96K)

img2img -> classic princess maker prompt -> SD upscale
holy fucking shit guys

Attached: 00477-3372965820-19-year-old young [[[flat-chest]]] auburn-haired girl, (Albert Lynch), Carne Griffiths, Franz Xaver Winterhalter, Lilia Alvarado.png (1024x1024, 1.26M)

>horse-riding astronauts and cockroach cola
or cock-riding astronauts and horse cola

based as fuck, share prompt?

Attached: 1663090020829536.png (1024x1024, 1007.31K)

I can make a gigabloated """portable""" exe for the CPU-only branch. It will be several gigs in size, supposedly 6, as a zip archive. No idea why anyone would use it, but I'm curious if there's a need for it. Will include a full-blown tkinter GUI and no install needed whatsoever, can be used on Loonix with Wine.

Why is she holding a dragon dildo

This one
github com/AUTOMATIC1111/stable-diffusion-webui
It said it downloaded something
What am i suppose to do now please help i don't know where these files are

Attached: Screenshot_20220916-002429_Chrome.jpg (1080x290, 104.62K)

Emma

>running compiled python through wine to use an ML model on CPU
kill it with fire

>runtimeerror couldn't install torch

this is why I dropped out of school

Attached: 1644565481439.png (480x572, 185.31K)

I've ascended in promptuning thanks to the guy who posted artists comparisons in /h/ so thank him.

Blonde female with pigtails, (((ahegao))), standing pose, [[portrait]], bikini photoshoot, [[digital art]], pixiv, ((ilya kuvshinov)), symmetrical face, (Alphonse Mucha), ((sylvain sarrailh)), ((((hirohiko araki)))), (((osamuz tezuka))), ((a-1 pictures)), ((((hideaki anno)))), ((((sabbas apterus)))), (((kawacy))), octane render
Negative prompt: ((chubby)), old, squiggly lines, (ugly), bloody, blood, glowing, alien, deformed, mutation, specs, pointy, hand, chin, nose
Steps: 20, Sampler: Euler a, CFG scale: 7, Seed: 2417708639, Size: 512x704

REEEEEEEEEE
REEEEEEEEEEEEEEEEEEEEEEE

Do it, I will put it on my cold storage backup for posterity

No idea, it was more like a piece of emerald in the original.

Attached: 00479-2780868813-19-year-old young [[[flat-chest]]] auburn-haired girl, (Albert Lynch), Carne Griffiths, Franz Xaver Winterhalter, Lilia Alvarado.png (1024x1024, 1.18M)

That sounds like a lot of effort for a very very small group of people

model.ckpt refers to the AI model, Stable Diffusion, which you were instructed to download by Step 3

You didn't skip step 3, did you user?

Prompt for that image?

Attached: magic-8-ball.png (948x752, 493.59K)

i modified that prompt with all those magic girls with golden armor, i have to say its pretty based, wanted to make my D&D character, made a steampunk gigachad instead

Attached: 00025-2926192332-ornate intricate filigree framed, alcoholic redhead man in his 30s with a mohawk haircut, determined face, inky , swirling mist,.png (1024x1408, 2.19M)

I thought it got downloaded by
Am i suppose to manually download it?

>less impressive for end users right now
Really? The ability for the AI to better infer what the user wants seems like a big deal to me. At the moment, it really struggles with multiple concepts, one after the other, in a single prompt