Hi all, this is my first ever lora creation. I'm sorry if it's not very good but i'm just testing things out and learning as i go, and wanted to give this top waifu a little bit of love since nobody else has made one on civitai yet.

I really couldn't quite figure out the captioning process and only made a few minor changes to the default BLIP captioning to correct some errors.

Trained on 20 images @ 100 steps. SD 1.5 base model. 512x512

I think i could remove a couple of images from the training set which were a little over exposed, and perhaps if i could figure out how to fine tune the captions properly it might yield better results. If anyone would like to colab with me or has any tips to improve this, please contact me or leave a comment.

I'd to love to finally turn this into a great model and have a history of previous versions to show an individual's learning curve for this process. I think that transparency would be quite cool and maybe benefit someone.

Trigger Word: yunaffx
Key Words: kimono, white top, blue skirt, blue dress, floral, (blue earring), (a green eye and a blue eye), (holding a staff)