About stable-diffusion-webui-Layer-Divider

Layer-Divider, an extension for stable-diffusion-webui using the segment-anything model (SAM)

j

Published by

jhj0517

Visit View Profile

README.md

View on GitHub

Layer-Divider

This is an implementaion of the SAM (Segment-Anything Model) within the SD WebUI.

Divide layers in the SD WebUI and save them as PSD files.

screenshot1

screenshot2

About segment-anything-2

If you want to try segment-anything-2, there's a new dedicated webui :

https://github.com/jhj0517/stable-diffusion-webui-Layer-Divider

The web ui supports prompt and video segmentation, I recommend using the new web ui.

Installation

git clone https://github.com/jhj0517/stable-diffusion-webui-Layer-Divider.git to your stable-diffusion-webui extensions folder.
or alternatively, download and unzip the repository in your extensions folder!

Notice ( Read if you face an error during installation )

Some packages are problematic to install programmatically when starting webui.
So you need to manually activate venv and install these packages before running webui.

Open the terminal in the WebUI and activate the venv
```
C:\YourPath\To_SD_WebUI>venv\Scripts\activate
```
Then it will display (venv) in front of the terminal like this.
```
(venv) C:\YourPath\To_SD_WebUI>
```

In this state, run

pip uninstall -y pytoshop
pip uninstall -y packbits
pip install git+https://github.com/jhj0517/forked-pytoshop.git
pip install packbits

How to use

Adjust the parameters and click "Generate". The output will be displayed below, and a PSD file will be saved in the extensions\stable-diffusion-webui-layer-divider\layer_divider_outputs\psd folder.

Explanation of Parameters

Parameter	Description
points_per_side	The number of points to be sampled along one side of the image. The total number of points is points_per_side**2. If None, 'point_grids' must provide explicit point sampling.
pred_iou_thresh	A filtering threshold in [0,1], using the model's predicted mask quality.
stability_score_thresh	A filtering threshold in [0,1], using the stability of the mask under changes to the cutoff used to binarize the model's mask predictions.
crops_n_layers	If >0, mask prediction will be run again on crops of the image. Sets the number of layers to run, where each layer has 2**i_layer number of image crops.
crop_n_points_downscale_factor	The number of points-per-side sampled in layer n is scaled down by crop_n_points_downscale_factor**n.
min_mask_region_area	If >0, postprocessing will be applied to remove disconnected regions and holes in masks with area smaller than min_mask_region_area. Requires opencv.

Todo

[ ] Migrate to SAM-2

stable-diffusion-webui-Layer-Divider