Implementation of Dreambooth (https://arxiv.org/abs/2208.12242) by way of Textual Inversion (https://arxiv.org/abs/2208.01618) for Stable Diffusion (https://arxiv.org/abs/2112.10752). Tweaks focused on training faces, objects, and styles.
Regularization Images cell takes self_generated_files_prompt for generating images, but also uses this variable as dataset later on. If the prompt has spaces, this ruins the directory name when the folder is being made. Added a small fix to have the dataset variable = self_generated_files_prompt with all spaces replaced with _
Runpod Notebook Only
Regularization Images cell takes
self_generated_files_promptfor generating images, but also uses this variable asdatasetlater on. If the prompt has spaces, this ruins the directory name when the folder is being made. Added a small fix to have thedatasetvariable =self_generated_files_promptwith all spaces replaced with_