Base Model:T-noobnai3 V8 https://civarchive.com/models/823566/t-noobnai3

Trigger: tomoshibi

USAGES:

1.Download the training dataset

2.Copy all text prompts associated with the images containing the desired poses

3.Remove any negative prompts from the copied text

4.Add your preferred character's tags (e.g., the model's identifier) to customize the output

Negative Prompts: noise,watermark, too many watermarks, artist name, speech bubble, text focus, white border

To ensure the accuracy of the generated images' poses, the prompts for each image in the training set were manually typed, rather than automatically generated by a tagging tool. Since the images required for training the LORA model must each have a resolution of 1024x1024, and to maintain the integrity of the image content, I added white borders to those images whose content would be incomplete after cropping. This ensured the completeness of the training images, which is why you may see white borders in some of the images in the training set.