Old Thread: >>88542140

Old Thread: >Web UI
(Recommended) voldy: github.com/AUTOMATIC1111/stable-diffusion-webui
hlky: github.com/sd-webui/stable-diffusion-webui

voldy guide: rentry.org/voldy

img2img guide: rentry.org/drfar

>Krita Plug-ins
github.com/sddebz/stable-diffusion-krita-plugin
github.com/nousr/koi

>AMD
rentry.org/ayymd-stable-diffustion-v1_4-guide
Updated guide: rentry.org/sdamd
huggingface.co/docs/diffusers/optimization/onnx

>Colab
voldy: colab.research.google.com/drive/1Iy-xW9t1-OQWhb0hNxueGij8phCyluOh
hlky: colab.research.google.com/drive/1AfAmwLMd_Vx33O9IwY2TmO9wKZ8ABRRa

>Paperspace
github.com/Engineer-of-Stuff/stable-diffusion-paperspace/

>In-Depth Usage
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:
sgreens.notion.site/sgreens/4ca6f4e229e24da6845b6d49e6b08ae7
rentry.org/artists_sd-v1-4

prompt sharing: lexica.art/ | krea.ai/
dataset: rom1504.github.io/clip-retrieval/ | laion-aesthetic.datasette.io/laion-aesthetic-6pls/images
Interrogator: replicate.com/methexis-inc/img2prompt

>SaaS Sites
txt2img: huggingface.co/spaces/stabilityai/stable-diffusion
img2img: huggingface.co/spaces/huggingface/diffuse-the-rest
txt2img, new accounts get 200 generations: beta.dreamstudio.ai
txt2img, no account needed: dezgo.com
txt2img, NSFW focused: pornpen.ai

>Finetuned Models
huggingface.co/hakurei/waifu-diffusion
huggingface.co/naclbit/trinart_stable_diffusion_v2

>Related
/h/ /sdhb/: /vg/ /aids/: /mlp/ /ppp/: Official: discord.gg/stablediffusion
NSFW: discord.gg/2ZhBzDGf8u
Tech support: matrix.to/#/#info:conlaboro.xyz
Stability.ai updates: twitter.com/EMostaque

Attached: 1662751181402-7.png (448x832, 514.21K)

Other urls found in this thread:

vimeo.com/748114237
rentry.org/zfawb
twitter.com/SFWRedditImages

>photorealistic render of link from the legend of zelda fused with princess zelda, trending on artstation, ray tracing, octane render, full body portrait
how tf does "fused with" work

Attached: 1660867600281.jpg (4000x999, 496.01K)

Attached: grid-0033.png (1024x1024, 1.54M)

Attached: file.png (1021x510, 957.31K)

with fused it's trial and error, not all fuses look good, so u have to find characters that fuse well, and then just stick to that and its variations

Anyone else enjoys typing random shit and looking at what it generates?

Attached: 00003-98062848.png (1024x1024, 1.78M)

sooo this exists

vimeo.com/748114237

any possibility of getting that functionality in the web front end?

Apparently not very well

Attached: 1632582948376.png (1536x1536, 1.22M)

did you guys prune the waifu model? I get "no module found for torch" line 2. I just installed python as per the guide. how do I get torch on windows? I dont see pip installed anywhere

God damn it I just asked in the old thread right before the new one was made
How does this run on Ubuntu with ROCm? Is it ez enough for a retard like me? Can't get a normal Endeavour or Manjaro running on my external drive

Have you tried reading the fucking info in the OP?

How good is inpainting now, can it be used for porn yet

the insatiable rapist Charlie Brown has struck again

gay

How do you avoid getting blob-faces with the waifu model?

Attached: 00287-2701965240-beautiful_demon_girl,_art_by_slugbox_and_sakimichan_and_anikiki.png (512x512, 348.94K)

Since it's all Conda & Python it's pretty distro indifferent.

blessing this thread with more pancake elephants

Attached: file.png (576x640, 655.75K)

How many steps should I bother with before diminishing returns?

I wrote this up, may be easier to understand. You may not have added it to PATH.
>rentry.org/zfawb

Attached: 00007-137038136-gigapixel-art-scale-2_00x.png (1024x1536, 1.89M)

Sometimes you just want vegetables

Attached: download.png (512x512, 375.26K)

This is what I been experimenting with so it's far from being realistic, but even this image makes me sick cause the moment you zoom in, their eyes look veeeery weird and the body doesn't make any sense anymore. It looks like early waifu upscaler stuff.

Attached: normal furry.png (512x512, 644.53K)

skill issue

Attached: 00207-2624419583-(cat_girl_artstation_fused_with_emma_watson)_with_cat_ears_and_cat_tail,_french_maid_costume,_(((full_body_portrait))),_vivid_te.png (512x704, 467.08K)

Wholesome

(reposting for new bread)

the cputard guide people should be aware that windows users have to track down a copy of an out of date version of visual C build tools, a large iso that you have to log in with a microsoft account to even download, in order to pip install the openvino repo's requirements

hardly anyone is going to bother proceeding past that step, and even if your attitude is "lol windows" you should probably at least mention in the rentry that windows users are going to need to do that if you're intending it to be a guide for retards

I was able to figure it out but a lot of people wouldn't be, especially people who are following a guide purporting to be for the non-technically inclined

or this... depending on my mental state

Attached: horror furry.png (512x512, 496.68K)

yeah

Attached: 000047.3996446223.png (512x512, 634.94K)

Yep, I think that's enough for one night. This is far more disturbing than the ones covered in blood from earlier.

Attached: grid-0284.jpg (2048x2048, 444.05K)

Attached: download.png (512x512, 309.8K)

I didn't have that problem. just works on my machine. Download and run.

What fucking skill issue? That image looks weird as fuck too. Eyes are fucked, everything looks soft as ass, the body looks unreal for some reason.

You might be making it worse for yourself with either of those. Fursonas are flashy, colorful, and just a lot to look at. Eldritch horrors are the same but god damn that is a pretty cool eldritch horror, nice output

Attached: patchouli nightgown.png (512x512, 333.59K)

I agree about the complexity of the guide in OP for a newcomer. I feel like an autist repeatedly posting my own, but my friend that barely knows how to move files between folders was able to follow this process without needing to ask additional questions:
>rentry.org/zfawb

Attached: 00906-1055601418.png (1088x512, 815.42K)

you're either on linux, or already had visual C dev tools installed, or you're just lying

amd radeon rx 5700

>amd

well there you go.

I'm in a strange place right now

Attached: download.png (512x512, 348.11K)

>hint of aerola shading
very good

I'm on windows 10. I am also like 99 percent sure I didn't have them downloaded beforehand since I don't do anything with visual C. Is there some way I can check?

I have the RX6800 and on stock settings i need 3.58 seconds to generate an image. I feel like something is amiss on your end.

python is definitely in my path. I can run it from cmd, but it cant find torch for some reason

I'm with user on this one, it's possible to get photorealistic quality pretty damn close. Closer than you'd expect

Attached: 00001-696500074-gigapixel-art-scale-2_00x.png (1024x1536, 1.64M)

people are not giving krita enough fucking credit
this shit's soooooooo so good
instead of just
>duh press generate a thousand times brainlessly
you can
>draw a shitty composition "a la paint" or just let the euler_a get you something with a vague prompt
> select aportion of the composition, add something with txt2img, or draw it yourself and add it with img2img
> use a transparency tool, make it blend together to help the ai merge, for the next step
>run img2img with a prompt defining what you just added, helping it blend the composition together better
>repeat
>get EXACTLY what you want, or at least a vision very close to what you want, instead of relying on gacha euler_a for a thousand outputs
>this is basically ai dungeon but better and with pictires, no years of practicing gate keeping you, your limits are only set by your imagination.

This shit NEEDS negative prompts and that weird () [] syntax PLEASE for the love of god

Attached: ndkqshdqsjdqs.jpg (1920x1080, 136.23K)

i'm using docker on windows, how do i make disk.vmdk bigger? i'm using the DockerToolbox-19.03.1 since i can't use the newer one.

Attached: Untitled.png (639x512, 25.37K)

I just followed the AMD guide and haven't changed anything other than the prompt

I agree this is a nice guide but it appears to be a guide for Voldy's, which is a repo for people running on GPU
I'm referring to the cputard guide, for people who want to run the model without a GPU

>using this for le art instead of tna
faggot

maybe it's just a windows 11 problem then, because that's what I'm on

I got it working it was just a fucking hassle needing to track down and install this ancient version of the visual C build tools just so openvino's requirements installation would succeed

please join our discord official and nsfw !
absolute state of g board

Attached: file.png (1024x1024, 225.83K)

Agreed, but look at the state of the threads for the past, what, week, and ask yourself if most anons here give a shit about anything beyond coom

local retard angry he cant generate good images, more at 11

Attached: 00214-256701652-(cat_girl_artstation_fused_with_emma_watson)_with_cat_ears_and_tail,_sexy,_french_maid_costume,_(((full_body_portrait))),_vivid_.png (512x704, 417.06K)

>skill issue
says the user whose emma has no cat tail or french maid costume

This is Any Forums. If you're simultaneously running Windows and incapable of following a tutorial of that complexity then this is probably the wrong place to be.

Attached: 02127-2144988210-crying_toddler.png (512x512, 328.5K)

last one, i think im becoming autistic.

Attached: 1640600670692.png (1536x1536, 1.46M)

>Old Thread:
ur mum old

but this is the board for consolewarring over consumer cell phone brands, there are only like 3 threads about programming on the catalog at any given time

Did you do the manual setup instead of running the .bat? Also I suppose I should also ask, are you using Voldy's or another repo?

Attached: 00001-2212565256-gigapixel-standard-scale-2_00x.png (1024x1536, 2.1M)

strawbelephant

Attached: file.png (576x640, 635.33K)