Guide Installing Stable Diffusion Webui & Nudifying (Inpainting) Guide

[arbuz2855]

Please, Log in or Register to view quotes

do i need to check out some extensions? i have it like this:

Please, Log in or Register to see links and images

console output the same, except that line "Applying cross attention optimization (InvokeAI)."and a different config(the one from the guide) "Creating model from config: C:\Users\***\stable-diffusion-webui-directml\models\Stable-diffusion\uberRealisticPornMerge_urpmv12-inpainting.yaml"

thanks a lot for trying, even if it wont work lol.

 

[animefag]

Please, Log in or Register to view quotes

Those built-in extensions are fine, technically stable diffusion shouldn't use them if you won't tell it to (at least stuff like Lora), but sometimes it gets buggy, and well... weird stuff happens.
If you don't have any additional files with textures inversions or loras put into models/* you are 100% fine and if you do It will always tell you when it uses one of those in the console.
So that part shouldn't be a problem. Sorry, don't know what's wrong.

 

[arbuz2855]

Please, Log in or Register to view quotes

yeah its weird. thanks a lot anyways

 

[animefag]

Please, Log in or Register to view quotes

I just read the guide

If you are using those embeddings from the guide, try to remove those from ` data/embeddings` (and restarting the app)
it's a wild shot but who knows.

 

[arbuz2855]

Please, Log in or Register to view quotes

tried that and didnt do anything sadly. tried using your settings, and with masked content on original it does nothing, just the same picture again lol

 

[arbuz2855]

Please, Log in or Register to view quotes

the image is 1080 x 1247. will try to do it with a higher resolution one.

i mean i am using the "uberRealisticPornMerge_urpmv12-inpainting.safetensors" one so that should be right.

 

[softarchitect]

Please, Log in or Register to view quotes

That should be fine, with "low resolution" i meant < 256 pixels (height/width).

It's weird that the result was that blurry.

Yeah, I thought animefag 's response was correct since you gave it a like :)

 

[arbuz2855]

Please, Log in or Register to view quotes

yeah, might try to install it on linux + docker. seems like a stretch but who knows.

ah i see, i liked for the effort :D

 

[softarchitect]

arbuz2855 which GPU do you have ?

I'm going to update the guide with some other ways to install.

I installed A1111 in docker on linux because I have a lot of python environments (conda) and it was just easier.

And that's on what I based my guide.

But, as animefag correctly stated, it's a bit heavy for some users who are already RAM constrained. And the additional hurdles (Hyper-V, etc) make it a bit cumbersome for the non-techhead folks.

 

[arbuz2855]

Please, Log in or Register to view quotes

i have a radeon rx 7900xt, overall a pretty solid pc.
yeah wanted to with a docker, but cant because of the amd gpu i think.

 

[softarchitect]

Please, Log in or Register to view quotes

Ah.. AMD.. yeah, no experience with those.

Have you followed

Please, Log in or Register to see links and images

? Like installing pytorch with rocm ?
Not sure if that happens automatically.

 

[arbuz2855]

Please, Log in or Register to view quotes

yeah i did the first one at the top. since the ones after that are for linux.

 

[softarchitect]

Please, Log in or Register to view quotes

Try following the guide on the wiki:

Please, Log in or Register to see links and images

Have not tested this, but it seems thorough.

 

[softarchitect]

Please, Log in or Register to view quotes

Ah, yes, you are right, ROCM is Linux specific library.

The forked version with DirectML doesn't seem to be used by a lot of ppl, and several "doesn't work" tickets.

If you don't see similar ticket as yours, it's always useful for the developer of that repo to have extra info. Obviously don't post NSFW images there.. but generate some "safe" pictures and create a new ticket on that repo.
Who knows it'll get fixed eventually.

But for now, your best bet is indeed Linux. But even on that, google first, not sure if it's guaranteed to work perfectly.

 

[Kurgan]

Very good guide. Its a complex install, but this guide handles it well. I did have to google an error, which turned out to be due to extensions in firefox. I simply used Edge instead. Many thanks!

 

[chuckyboiiii]

I'm having some issue in removing the clothes off of some pics, sometimes it works fine and other times instead of the girls clothes being taken away, they are morphed into different similar color clothes, doing the opposite of what i was hoping for. Any way around this to ensure it'll remove the clothes?

 

[softarchitect]

Please, Log in or Register to view quotes

Most of it has already been mentioned in this thread.

• Mask more, go outside of the edges of the clothes. If you only mask clothes, the generator doesn't know about any skin colors, so it's best you include parts of the unclothed body (e.g.. neck, belly, hands, etc).
• Use another masked content option (experiment)
  • Fill: will fill in based on the prompt without looking at the content behind the masked area
  • Original: will use the masked area to make a guess to what the masked content should include. This can be useful in some cases to keep body shape, but will more easily take over the clothes
  • Latent Noise: will generate new noise based on the masked content
  • Latent Nothing: will generate no initial noise
• Denoising strength: the lower-the closer to the original image, the higher the more "freedom" you give to the generator
• And sometimes: you forgot to mask some (tiny) area of the clothes.. this is esp. difficult to find with clothes that are dark. But look at the "in progress" images that are generated during the process.. you should normally see where it is.

 

[divya02]

Any tutorial to install it on Mac OS?

 

[Michaelcrann]

I have downloaded stable diffusion webui following a different guide than yours and it works. But the docker was not included on that guide. Am I able to just turn off the process or make it stop running without having to download the docker ?

Also a bit of a stupid question, but how do I know if it’s running in the background or not?

 

[softarchitect]

Please, Log in or Register to view quotes

Added to guide. Not tested.

Please, Log in or Register to view quotes

Normally you should have a CMD or Powershell window open.. you can just close it and it'll stop the process. But depends on the process / guide you followed. It shouldn't run in the background normally.
I've changed the guide to include different installation methods I know of. Check if one of them includes the same steps your guide used.