Posted May 12, 2025 by Mashed Mice
#tool #ai #dataset #image #editor #crop #stable #diffusion #indie
Tonight I decided to upload some free stuff!
Here's a dataset preparation tool used for handling dataset batches; scanning them for duplicates, spelling errors, corrupted files, low or high resolution (compared to a set uniform target resolution for the entire set) and quick cropping to the set target resolution for the dataset. It also has a randomize tool to give your files more variation in the matter of naming in case you plan to use a filename-as-prompt solution for a single-themed concept dataset (Frida Kahlo's art, Monet-artstyle, Cartoon-style images etc).
It's designed to quickly handle massive folders, since a dataset may contain just as much content as you need for your training.
I decided not to have you pay for it since its quite simple in comparison to the actual trainer (https://mashed-mice.itch.io/mashed-mice-ai-trainer-for-stable-diffusion), but may be helpful to many others no matter what trainer you're going for, since its all handling your local files and you may apply them to anything - maybe not even datasets, sometimes you'll just need square-sized images in general.
Enjoy!
/Joel Sandström, developer and moon-howler