The dataset contain roughly 2k of anime Screencap data and 8k of scrapped danbooru illustration data.
This is the proccessed version of the dataset meant to be used for my personal finetuning practice project, please visit my RicemanT/Background-Finetuning repo for the raw unprocessed data that you can process yourself.
The dataset have two minor type of processing being done… See the full description on the dataset page:
https://huggingface.co/datasets/HappyHenAi/Anime-Background-Finetuning-V1.1.