This repository contains strong traces gaslighting, along with a few files for; machine learning, containers, and ultimatly some poorly written scripts.
This script may help with collecting captions from known websites that give useful; titles, descrptions, exif, tags, for the media using gallery-dl to create metadata files. The script currently takes information from a useful but randomly organised json file and creates a text file next to it based on the same name as well as an appended file.
The contents of the file will be: Title, Description, EXIF data, Tags
For example: The Fat Cat Barnaby, Barnaby is always very sleepy and looks drunk, ISO: 200, Focal Length: 100.0, Model is NIKON D850, Aperture: 4.5, Fat Cat, Lazy, sleeping
If the json doesn't contain a title, description, EXIF, or tags, it won't get included in the file.
Commandline options available with "python json2txt.py --help" Config file also available to set image directory.
TODO: Test additional websites.
This script may help with doing a replacement of specific words within your captions recursivly when stored in text type files.
Usage example: python captions2remove.py --captiondir c:\captions --caption-remove "Monkey Magic is great! --caption-replace "Tripitaka is the best!"
".txt" files are the default file extension but this can be set to any file extension with the extension argument. e.g. python captions2remove.py --extension ".json"
This script is less useful for general images but can be useful in identifying plants that you may have close ups. The script uses Plant NET API and you need a developer account with API key. They have a free account that allows up to 500 API calls a day. Commandline options available with "python plant2caption.py --help"
This is a simple script that will concatenate all text files (.txt) into a single file. This can be useful when you want to have all available words used with captions in a single text file. Useful for wordclouds and creating a summary of words use (--dedup) Commandline options available with "python txts2txt --help"
This script may be useful to get an idea of creating word clouds with words / tokens / text files you have used in creating captions for images. Other examples are available in the examples folder with batch (.bat) and shell (.sh) scripts.
Slow Stable Diffusion and Very Slow Stable Diffusion are both modified Colabs located in the Colab folder and based on TheLastBens Fast_Stable_Diffusion_AUTOMATIC1111 Colab. Testing only but seem to work fine.