This version is Hashimoto Kanna

's first online release of LORA. I purchased Hashimoto Kanna's photos at her own expense

Naturel

based on some of the images in this

photo for Lora's upper body training for the face and upper body. The face has a good degree of restoration. Running locally will have a better experience than online generation

Friends who have configured SD locally are

advised to download the local running map

and use the enhanced version of Malt Orange v2

The

trigger

word for basic training is qbhn, and the initial size is 512 x 768

The guide word coefficient

is 7, and the redrawing amplitude is increased to 0.2

Lora

's rights are important to around 0.8 and can be adjusted by yourself

After multiple rounds of testing

the sampling methods dpm++sde and dpm2 a karras

The two sampling methods

produced the best test images were Dpm2 a karras

This model is not optimized for the whole body when drawing

a full body map

Your face will break down and look like it. Stay tuned for updates to Lora, which is optimized for the whole body, in the future!

I hope everyone will actively return to the picture and ask questions!!

(Note: Using the model to draw is a personal act. The purpose of drawing should be personal appreciation. The model must not be used to carry out any kind of commercial activity, and images generated by the model must not be used to spread or post false information)