GLORY Fast and automatic nudifying in stable-duffusion-webui (No Money Required)

[INFINITY]

[ Hidden Content! ]

Here is a new method to take off clothes just in a few seconds!

1. First you need to install automatic1111's stable-duffusion-webui if not already:

Spoiler:

Installation and RunningMake sure the required dependencies are met and follow the instructions available for:

 

• NVidia (recommended)
  
• AMD GPUs.
  
• Intel CPUs, Intel GPUs (both integrated and discrete) (external wiki page)
  

Alternatively, use online services (like Google Colab):

 

• List of Online Services
  

 Installation on Windows 10/11 with NVidia-GPUs using release package

1. Download sd.webui.zip from v1.0.0-pre and extract its contents.
   
2. Run update.bat.
   
3. Run run.bat.
   

 

For more details see Install-and-Run-on-NVidia-GPUs

 Automatic Installation on Windows

1. Install Python 3.10.6 (Newer version of Python does not support torch), checking "Add Python to PATH".
   
2. Install git.
   
3. Download the stable-diffusion-webui repository, for example by running git clone https://github.com/AUTOMATIC1111/stable-diffusion-webui.git.
   
4. Run webui-user.bat from Windows Explorer as normal, non-administrator, user.
   

 Automatic Installation on Linux

1. Install the dependencies:
   

# Debian-based:
sudo apt install wget git python3 python3-venv libgl1 libglib2.0-0
# Red Hat-based:
sudo dnf install wget git python3 gperftools-libs libglvnd-glx
# openSUSE-based:
sudo zypper install wget git python3 libtcmalloc4 libglvnd
# Arch-based:
sudo pacman -S wget git python3
 

1. Navigate to the directory you would like the webui to be installed and execute the following command:
   

wget -q https://raw.githubusercontent.com/AUTOMATIC1111/stable-diffusion-webui/master/webui.sh
 

1. Run webui.sh.
   
2. Check webui-user.sh for options.
   

 Installation on Apple SiliconFind the instructions here.

If you have <= 8GB vram, I recommend to use 

Code:

--medvram

 commandline flag. If you have <= 4GB vram: 
 

Code:

--lowvram

. Also I recommend to use 
 

Code:

--xformers

Check is it works correctly with regular generations

2. Install sd-webui-replacer extension. Go to Extensions tab -> Available -> Click Load from -> Search "replacer", and click "install"

Then search "segment anything" extension, and install too. Do not confuse with Inpaint Anything!

Then you can either stop stable diffusion, or go to Installed tab -> click Apply and Restart UI

3. Installing models
 

• Download sam_hq_vit_l.pth and move it inside 
   
  
  Code:

  extensions/sd-webui-segment-anything/models/sam
   folder. If you don't have enoght vram, you can download more lightweight models, see sd-webui-segment-anything's readme
  
• Download EpicPhotoGasm - Z - Inpainting and move it into 
  
  Code:

  models/Stable-diffusion
• Download lcm-lora-sdv1-5, rename file 
   
  
  Code:

  pytorch_lora_weights.safetensors
   into 
   
  
  Code:

  lcm-lora-sdv1-5.safetensors
   and move into folder 
   
  
  Code:

  models/Lora
  . It will be used for hires fix
  
• The athor of EpicPhotoGasm recommends to use 4x_NMKD-Superscale-SP_178000_G model for hires fixes. Put it into 
   
  
  Code:

  models/ESRGAN
   folder, then you can select it for hirs fix
  
• Additionaly I can recommend to download two embeddings: small_tits and Style-Unshaved (Requires free registration on the website). Putting them into negative or positive prompt you can control tits size and hair on pussy. Just words "small/big tits", "hairy/shaved pussy" work bad. Move them inside 
   
  
  Code:

  embeddings/
   folder
  

4. Setting up
 

• Open webui and go to Settings tab -> Replacer section. And set up this options:
  

Override Detection prompt examples

Spoiler:

clothes
swimsuit
trousers
t-shirt
sweater
shorts
jeans
skirt
dress

Override Positive prompt examples:

Spoiler:

photo of naked girl

You can ajust other quick templates for prompts what you like
 

• Click Apply Settings, and Reload UI
  

5. It's all! Now go to Replacer tab, select 
 

Code:

epicphotogasm_z-inpainting

 model in the header, put your desirable photo into Source section, click Run. It works nice in about 70% of cases for me


Advices:

You can select any specific part of clothes next to "Detection prompt" field. If model cannot recognize clothes, you can specify color of it

If the result is blured, I don't recommend to change resolution inside "Advanced options". In produces mutations. Just use "Apply hires fix" button

In "Batch process" and "Batch from Directory" tab you can process few images in a row. I recommend for it open "Advanced options" and increase "batch count" to at least 2. It increases a propability of good result

Also for batching, if your images have big resolution, you can select option "Upscaler for img2img". I recommed 4x-Ultrasharp (put inside 
 

Code:

models/ESRGAN

), or ESRGAN-4x from defaults. But it significally increase time of generation

Not working!This leak has been rated as not working 1 times this week. (1 in total)