Tanabe Zenjirou from Yumemiru Otome

caution

This Lora model is recommended for use with https://civarchive.com/models/3007711/moss-image-sft-deband-fix-color-banding-or-int8-or-fast-and-quality-versions?modelVersionId=3412867

I don't speak English and I don't have much time.

Therefore, please understand that I cannot answer questions or requests.

Lora learning uses kohya-ss/sd-scripts [Many thanks to the creator, kohya-ss, and all the contributors who support the project!]

The training model used is the official Base model: https://civarchive.com/models/3005406/moss-image?modelVersionId=3409868

Due to the specifications of MOSS image, it should not work with tuning models other than the recommended/base model.

I don't understand the technical details, so please ask Google AI or others why it might not work.

The end of the file indicates the training epoch.

(I believe the best epoch and Lora intensity will vary depending on how it's used... this is just my experience, so I might be wrong.)

The training uses a restrained art style, focusing mainly on the character's face and hairstyle. Clothing should be considered an afterthought.

It should be controllable via prompts.

The learning settings remain the same in the metadata, so please refer to that.