Others

Installing Stable Diffusion & Nudifying Inpainting Guide

файл на civitaiстраница на civitai

If you enjoy the work I do and would like to show your support, you can donate a tip by purchasing a coffee or tea. Your contribution is greatly appreciated and helps to keep my work going. Thank you for your generosity!

https://ko-fi.com/gswan
https://www.buymeacoffee.com/gswan
------------------------------------------------------------------------------------------
This guide aims to provide general information.

DISCLAIMER: Please be aware that the following guide is still advanced and technical. It might not cover all possible cases or errors, which can lead to frustration and wasted time. It's vital to proceed with caution and have a good understanding of the subject matter before attempting the guide.

0. Introduction

This guide will provide steps to install AUTOMATIC1111's Webui several ways.

After the guide, you should be able to go from the left image, to the right,

Screenshot-from-2023-02-08-14-04-39.md.png



Recommended Requirements:
Windows 10/11, 32 GB of RAM and a Nvidia graphics card with at least 4-6 GB VRAM (not tested by me, but reported to work), but the more the better.
Windows 11 (64 bit): Home or Pro version 21H2 or higher.
Windows 10 (64-bit): Home or Pro 21H1 (build 19043) or higher.

For AMD GPU cards check out https://github.com/AUTOMATIC1111/stable-diffusion-webui/wiki/Install-and-Run-on-AMD-GPUs

Non-GPU (not tested, proceed at own risk auseChamp: )
It does work without a Nvidia graphics card (GPU), but when using just the CPU it will take 10 to 25 minutes PER IMAGE, whereas with a GPU it will take 10 - 30 seconds per image. (images with a resolution of < 1024px, for larger pictures, it takes longer)

I've chosen to use docker, since this will make it far easier to install the webui and minimal fiddling in the command prompt.


1. Installation

Spoiler: (Windows) Install it natively, the lazy version (not tested)

Spoiler: (Windows) Install the native way (bit technical, not tested)

Spoiler: (Linux) Install the native way for the Linux-heads

Spoiler: (Windows) Install via Docker

Spoiler: (Mac) Install it natively (not tested)


Check the file for downloading.


1. Installation

Spoiler: (Windows) Install it natively, the lazy version (not tested)

1. Stable Diffusion Portable

Spoiler: (Windows) Install the native way (bit technical, not tested)

1. Installing AUTOMATIC1111 Webui

This is the same guide as on the AUTOMATIC1111 git repo.

1.2. Automatic Installation

  1. Install Python 3.10.6, checking "Add Python to PATH"

  2. Install git.

  3. Download the stable-diffusion-webui repository, for example by running (in CMD or Powershell)

    Code:

    git clone https://github.com/AUTOMATIC1111/stable-diffusion-webui.git

  4. Or you can download by going to https://github.com/AUTOMATIC1111/stable-diffusion-webui , clicking "Code" and then "Download ZIP"

  5. Run webui-user.bat from Windows Explorer as normal, non-administrator user.

Spoiler: (Linux) Install the native way for the Linux-heads

1. Automatic Installation on Linux

1.1. Install the dependencies

Code:

# Debian-based:
sudo apt install wget git python3 python3-venv

# Red Hat-based:
sudo dnf install wget git python3

# Arch-based:
sudo pacman -S wget git python3


1.2. Install A1111

To install in /home/$(whoami)/stable-diffusion-webui/, run:

Code:

bash <(wget -qO- https://raw.githubusercontent.com/AUTOMATIC1111/stable-diffusion-webui/master/webui.sh)


Run webui.sh

Spoiler: (Windows) Install via Docker

1. Installing Docker Desktop

1.1. Requirements

Windows 11 (64 bit): Home or Pro version 21H2 or higher.
Windows 10 (64-bit): Home or Pro 21H1 (build 19043) or higher.

Docker Desktop requires one of the following,


Guide is available at

https://www.itechtics.com/enable-hyper-v-windows-10-home/


This requires several steps, available at https://learn.microsoft.com/en-us/windows/wsl/install

1.2. Hyper-V File Sharing

When using Hyper-V, you need to enable File Sharing in Docker Desktop > Settings > Resources > File Sharing.

The File sharing tab is only available in Hyper-V mode because the files are automatically shared in WSL 2 mode and Windows container mode.

1.3. Installation

Go to https://docs.docker.com/desktop/install/windows-install/ , then download and run the installer.

1.4. Installing Docker Container of the Webui

Go to https://github.com/AbdBarho/stable-diffusion-webui-docker/releases , download the latest source code release (ZIP) under Assets.

Unzip it somewhere.

1.5. Open Command Prompt (CMD) or Powershell

Go to the directory where you unzipped the archive, and run

Code:

docker compose --profile download up --build


this will download 12GB of pretrained models. Wait until it is finished, then

Code:

docker compose --profile auto up --build


More details are available on the project's wiki: https://github.com/AbdBarho/stable-diffusion-webui-docker/wiki/Setup

https://github.com/AbdBarho/stable-diffusion-webui-docker/wiki/Setup

Spoiler: (Mac) Install it natively (not tested)

Source: https://github.com/AUTOMATIC1111/stable-diffusion-webui/wiki/Installation-on-Apple-Silicon

  1. If Homebrew is not installed, follow the instructions at https://brew.sh to install it. Keep the terminal window open and follow the instructions under "Next steps" to add Homebrew to your PATH.

  2. Open a new terminal window and run brew install cmake protobuf rust python@3.10 git wget

  3. Clone the web UI repository by running git clone https://github.com/AUTOMATIC1111/stable-diffusion-webui

  4. Place Stable Diffusion models/checkpoints you want to use into stable-diffusion-webui/models/Stable-diffusion. If you don't have any, see Downloading Stable Diffusion Models below.

  5. cd stable-diffusion-webui and then ./webui.sh to run the web UI. A Python virtual environment will be created and activated using venv and any remaining missing dependencies will be automatically downloaded and installed.

  6. To relaunch the web UI process later, run ./webui.sh again. Note that it doesn't auto update the web UI; to update, run git pull before running ./webui.sh.

2. The WEBUI!

Wait until you see

Code:

 Running on local URL:  http://0.0.0.0:7860
To create a public link, set `share=True` in `launch()`.


Then go to http://localhost:7860/ in any browser.

You should see something similar to this:

Screenshot-from-2023-02-08-14-20-39.md.png



The 'hardest' part should be done.


3. Inpainting model for nudes

Download one of the following for 'Nudifying',


4. Embeddings / Lora

4.1. Embeddings

Add the following files in the 'data/embeddings'

Check My textual Inversions, under same names.

Allow you to add 'breasts', 'small_tits' and 'Style-Unshaved' in your prompt, and provide better quality breasts / vaginas. The first one is more generalized, the latter is well.. yes.

Check the Discord link in the 'Additional Tips', ppl post additional embeddings on there.

4.2. Lora

I have not tested all of these, so not sure if they work in combination with inpainting.

Add any of the following in 'data/Lora'. (CivitAI account required for NSFW)

To include Lora models click on the following icon at the right side, under the "Generate" button.

1678546760597.png


This will show an extra panel at the left, including a "Lora" tab. Clicking on any of the Lora's you have will add them to the prompt. You can move it to the front or back of the prompt to give it more 'focus', and you can also increase/decrease the number at the end "<..:1)" to give it more (e.g. 1.5) or less (e.g. 0.5) focus.


5. Loading Model

In the webui, at the top left, "Stable Diffusion checkpoint", hit the 'Refresh' icon.

Now you should see the uberRealisticPornMerge_urpmv12 model in the list, select it.


6. Model Parameters

Go to the 'img2img' tab, and then the 'Inpaint' tab.

In the first textarea (positive prompt), enter

RAW photo of a nude woman, naked​


In the second textarea (negative prompt), enter

((clothing), (monochrome:1.3), (deformed, distorted, disfigured:1.3), (hair), jeans, tattoo, wet, water, clothing, shadow, 3d render, cartoon, ((blurry)), duplicate, ((duplicate body parts)), (disfigured), (poorly drawn), ((missing limbs)), logo, signature, text, words, low res, boring, artifacts, bad art, gross, ugly, poor quality, low quality, poorly drawn, bad anatomy, wrong anatomy​


If not otherwise mentioned, leave default,

Masked content: fill (will just fill in the area without taking in to consideration the original masked 'content', but play around with others too)​

​

Inpaint area: Only masked​

​

Sampling method: DPM++ SDE Karras (one of the better methods that takes care of using similar skin colors for masked area, etc)​

​

Sampling steps: start with 20, then increase to 50 for better quality/results when needed. But the higher, the longer it takes. I have mostly good results with 20, but it all depends on the complexity of the source image and the masked area.​

​

CFG Scale: 7 - 12 (mostly 7)​

​

Denoise Strength: 0.75 (default, the lower you set this, the more it will look like the original masked area)​

​


These are just recommendations / what works for me, experiment / test out yourself to see what works for you.

Tips:



7. Images

Stable Diffusion models work best with images with a certain resolution, so it's best to crop your images to the smallest possible area.

You can use any photo manipulation software, or use https://www.birme.net/

Open your preferred image in the left area.


8. Masking

Now it's time to mask / paint over all the areas you want to 'regenerate', so start painting over all clothing. You can change the size of the brush at the right of the image.

I've experienced that painting a little bit outside the 'lines / borders' is the best for the model, else you will get visible remnants / distortions at the edges of the mask in the generated image.

Tips / Issues:



9. Inpainting, finally!

Hit generate. Inspect image, and if needed, retry.


10. Optimizations

When the model is generating weird or unwanted things, add them to the 'Negative Prompt'. I've already added some, like hair, water, etc. In some cases you need to add 'fingers', since Stable Diffusion is quite bad at generating fingers.

Sometimes it's needed to add extra words to the (positive) prompt, i.e. sitting, standing, ..


11. Caveats

I'll try to answer any questions regarding this guide you may have, but I'm no expert.

If you hit an error, or have a problem, add a comment. And don't forget to add a screenshot to the ouput of the docker container. It should provide detailed errors.

Whenever you see 'OOM (Out of Memory)' in the docker container output, it most of the time means that you do not have enough (graphics) VRAM to run it.

I don't have a lot of data, so if it works for you with less than 6 GB (graphics) VRAM, let us know in the comments.

This guide is written in a short time, so it probably needs some updates.

12. Additional Tips


13. Extra

13.1. Upscaling

When you have generated an image, and you find it's too small, you can use the 'Send to extras' button below the result. For Upscaler 1, select "R-ESRGAN x4" and for Upscaler 2, select "Nearest". Depending on the original image, use a Resize value of either 4 (times 4), or 2.
It's not perfect for very small images, but works decently enough for images with width between 512x and 1024x.

13.2. Docker Desktop

To check the status of the running containers, open Docker Desktop you should see something like this,



Here you can click on 'Containers' in the menu and see all containers installed, at the right of each container you have start/stop icons. This allows you to manage the containers.

13.3. Docker Internal

Sometimes (e.g. when training) you need to access files inside of the docker container.

Use the following commands to copy the files to the 'outside' /data folder.

Code:

docker exec -it <container> /bin/sh
# and then for example (for embeddings):
cd /stable-diffusion-webui/textual_inversion/
# list files / folders
ls
# then copy it to /data/train/ so you can access it via Windows explorer (or similar)
cp -r * /data/train/



14. Bugs

14.1. Cannot add middleware after an application has started

Around 2023-02-10 the developers of AUTOMATIC1111 introduced a bug. When you run build/start the docker container, you get the following error:

Code:

RuntimeError: Cannot add middleware after an application has started

To fix this, open the file services/AUTOMATIC1111/Dockerfile (in the project folder) with a text editor. Go to line number 75 and after it add:

Code:

pip install fastapi==0.90.1 --upgrade

The result looks like this:

Code:

RUN --mount=type=cache,target=/root/.cache/pip <<EOF
cd stable-diffusion-webui
git fetch
git reset --hard ${SHA}
pip install -r requirements_versions.txt
pip install fastapi==0.90.1 --upgrade
EOF


Then rerun the docker container, and then follow from #4.

Code:

docker compose --profile auto up --build


14.2. Clicking 'generate' loads additional model / Checkpoint b3f1532ba0 not found; loading fallback

Seem to have mixed results, it seems to load an Stable Diffusion 1.5 inpainting model that I have in the 'StableDiffusion' folder.
Not sure what the reasoning is behind this change.
To be safe, remove ALL files from the data/StableDiffusion directory and only leave the uberRealisticPornMerge_urpmv12-inpainting.* files.
And then click on the refresh icon at the top.
This also speeds up the generate process, since it doesn't need to load the extra model.[/TOC]

14.3. Inpainting doesn't work, weird results / output

Be sure to check that the webui has access to the 'Canvas' element, and that this is not blocked by your browser (e.g. Firefox) or by some extension (e.g. for privacy / anti-fingerprinting reasons).

Also, when using AMD graphics card, you can't use the standard install, you have to use a special install just for AMD cards: https://github.com/AUTOMATIC1111/stable-diffusion-webui/wiki/Install-and-Run-on-AMD-GPUs

Тэги: nudeinpaintinginpaintguidefor dummiesnudify
SHA256: 0E94A1A33C47C4E75D9B4C53F63C4A4F77DF8D624427C3544D26ED218CF93AB8

Link (BOTW/TOTK) so-vits-svc

файл на civitaiстраница на civitai

dont take it seriously :) I was just curious what would happen if I trained the neural network on the sounds Link makes in BOTW and TOTK
Example:

Тэги: characterthe legend of zeldabotwlinkso-vits-svc-forktotkso-vits-svc
SHA256: 685EA500EF4895EC3E5D8946C6290CEAF6E421879B8D90D486DED9791A17BF64

Civitai shortcut

файл на civitaiстраница на civitai

"The loading of the shortcut file has failed or it is empty." is not an error message. It is a message indicating that there are no registered shortcuts.

However, many people have found it confusing, so I have made some changes to the content. Please update it accordingly.

*****

2023-05-20

There has been an issue due to changes in the model information specification in Civitai. Data update is required. I have uploaded the fix urgently. To update the data, please perform "Manage -> Scan and Update Model -> Update the model information for the shortcut," and it will "download and retrieve the latest information.

You need to update Civitai Shortcut first.

extensions->installed->check for updates click -> apply and restart ui

If possible, please restart SDUI

*****

V 1.5

* The information tab and the saved model information tab in the Civitai model have been merged. The Civitai model information, which used to fetch real-time information from the Civitai site, has been removed. Instead, the shortcut information registered during Stable Diffusion startup will be updated. You can enable or disable this feature in manage->setting->option.

* The Assistance tab has been added. Furthermore, the classification feature has been moved to the Assistance tab. Additionally, a new feature called Prompt Recipe has been added.

* The Prompt Recipe is a feature that allows you to register and manage frequently used prompts. You can register a prompt by directly writing it, uploading an image file, or using the 'Send to Recipe' button in the Image Information section of Model Information or User Gallery.

* In the Classification and Prompt Recipe sections, the 'Create' button now appears only when [New Classification] or [New Prompt Recipe] is selected, clearly distinguishing between the Create and Update states.

* The 'sc_saved' folder, which was used to backup the registration URLs for shortcuts, is no longer in use and can be deleted. Instead, the registration URLs are now stored in the CivitaiShortCutBackupUrl.json file in the format {url: name}. You can re-register the shortcuts by uploading this file in the Upload section.

This file is automatically generated and updated when performing 'Update the model information for the shortcut' in Manage->Scan and Update Models or when the automatic update feature is enabled.

* A folder named 'sc_recipe' is created to store the images for the Prompt Recipe.

* Bug: There is an issue in the prompt recipe where the saved image loses the generated image information. It appears to be a problem with the Gradio Image component.

* A new option 'Disable Classification Gallery Preview Mode' has been added under 'Manage -> Settings -> Options'. This is an experimental feature and will be removed once it becomes an official functionality in Gradio

Civitai Shortcut

Stable Diffusion Webui Extension for Civitai, to download civitai shortcut and models.

Install

Stable Diffusion Webui's Extension tab, go to Install from url sub-tab. Copy this project's url into it, click install.

git clone https://github.com/sunnyark/civitai-shortcut

Install url : https://github.com/sunnyark/civitai-shortcut

Тэги: tool
SHA256: BB1C085FAFB4E2959EF81695F56FAB4AF32A20E33F582FEE2D32461BFF562CE4

Hatsune Miku so-vits-svc

файл на civitaiстраница на civitai

based on some acapellas from her songs
example:

Тэги: characterhatsune mikuso-vits-svc-forkso-vits-svc
SHA256: E4028CEB42BC84CFE1AE94E12F53E75F92EA43019E0EBB37D084936D8CF54620