https://github.com/appenz/sdshell

Simple script to reformat and rename for Stable Diffusion Dream Booth learning
https://github.com/appenz/sdshell

Last synced: 15 days ago
JSON representation

Simple script to reformat and rename for Stable Diffusion Dream Booth learning

• Host: GitHub
• URL: https://github.com/appenz/sdshell
• Owner: appenz
• License: apache-2.0
• Created: 2022-10-31T02:02:32.000Z (over 3 years ago)
• Default Branch: main
• Last Pushed: 2023-02-20T18:44:54.000Z (over 3 years ago)
• Last Synced: 2025-12-13T13:02:48.136Z (7 months ago)
• Language: Python
• Size: 11.7 KB
• Stars: 0
• Watchers: 1
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

          
# sdshell

This is a simple tool to format images for for learning with the dreambooth trainer

for Stable diffusion. Specifically it resizes the images to 512 x 512 by finding the centered largest 

square and scaling up or down as necessary. It also renames them to a consisten name of 

"NAME (seq).jpg" where NAME is the parameter passed via --iname and  is a sequence number.

 This is useful for taking a directory full of images and preparing them for learning with 

 a Dream Booth script.

 

## Running

 

```python3 sdshell.py --srcdir  --traindir  --iname ```

 

## Command line options

```Usage: python3 sdshell.py [OPTIONS]

Options:

  --srcdir TEXT          Source directory for the raw, unformatted images.

  --traindir TEXT        Storage directory for the resized training images.

  --trainwidth INTEGER   Optional training image width, needs to be same as height. Default 512.

  --trainheight INTEGER  Optional training image height. Default 512.

  --iname TEXT           Name of the new concept that the model should be

                         trained for.

  --help                 Show this message and exit.

  ```