Jihan

Hello it's OKGC, kpop girls/actress lora maker. I speak Mandarin, English, and a little Korean. Follow the info below to get better result. Feel free to ask for help.

Recommended checkpoints:

• supashy mix - https://civitai.com/models/156202/supashymix

  • final version works best (what I'm using), get it on supashy's gumroad to support him please - https://supashy.gumroad.com/l/supashy_mix_final?layout=profile

• Moonfilm Reality 3.0 - https://civitai.com/models/43977?modelVersionId=73071

• majicMIX realistic v7 - https://civitai.com/models/43331?modelVersionId=176425

Recommended parameters: (works best in my opinion)

• Trigger word: hanjihyo

• Sampling Method: Euler

• Steps: 50

• Hires. fix Upscaler: 4x_NMKD-Superscale-SP_178000_G - https://huggingface.co/gemasai/4x_NMKD-Superscale-SP_178000_G/tree/main

  • Put it in folder models/ESRGAN

• Hires upscale: 2

• Hires steps: 15

• Denoising Strength (Hires): 0.25

• Clip skip: 2

• CFG scale: 7

• Size: 512x768

• vae: vae-ft-mse-840000-ema-pruned - https://huggingface.co/stabilityai/sd-vae-ft-mse-original/blob/main/vae-ft-mse-840000-ema-pruned.ckpt

  • Put it in folder models/Vae

• Eta noise seed delta (ENSD: 31337)

  • Find it in SD webui/Settings

    

    

If the face comes out badly, use adetailer to fix it.

https://github.com/Bing-su/adetailer

Recommended parameters: (works best in my opinion)

1. 1st Adetailer model: face_yolov8n.pt

   Inpaint denoising strength: 0.25

   

   

2. 2nd Adetailer model: mediapipe_face_mesh_eyes_only

   Inpaint denoising strength: 0.25

   

   

3. If the face detection of adetailer detects the face of other people in the background,

   set Detection model confidence thereshold: 0.8~0.85

   

Used embeddings:

(how to use?: https://github.com/AUTOMATIC1111/stable-diffusion-webui/wiki/Textual-Inversion)

• ng_deepnegative_v1_75t - https://civitai.com/models/4629/deep-negative-v1x

• badhandv4 - https://civitai.com/models/16993?modelVersionId=20068 (add disfigured in negative prompt may also helps getting better hand)

• bad_pictures - https://civitai.com/models/23178?modelVersionId=66043

Highly recommended tools: Lora Block Weight - https://github.com/hako-mikan/sd-webui-lora-block-weight

• How to use?

  <lora:lora name:1:lbw=IN02>  

  <lora:lora name:1:lbw=1,1,1,1,1,1,1,1,1,1,1,1,0,0,0,0,0,0,0,0,0,0,0,0,0,0>

• Recommended parameters (name whatever trigger word you want, it's all about the block weight)

  • Body:

    WINGS:1,0,0,0,0,1,0,1,1,1,0,0,0,0,0,0,0

    BODY:1,1,1,1,1,1,1,1,0,0,0,1,1,1,1,1,1

    BODY0.5:1,1,1,1,1,1,0.2,1,0.2,0,0,0.8,1,1,1,1,1

  • Face: (face shape, hair style, eye shape, eye colors, etc.)

    EYE:1,0,0,0,0,0,0,0,0,1,1,0,0,0,0,0,0

    MOUTH:1,0,0,0,0,0,0,0,1,0,0,0,0,0,0,0,0

    FACE:1,0,0,0,0,0,0,0,1,1,1,0,0,0,0,0,0 (which I use for my lora)

    FACE0.5:1,0,0,0,0,0,0,0,0.8,1,1,0.2,0,0,0,0,0

    FACE0.2:1,0,0,0,0,0,0,0,0.2,0.6,0.8,0.2,0,0,0,0,0

  • Hands:

    HAND:1,0,1,1,0.2,0,0,0,0,0,0,0,0,0,0,0,0

  • outfit: (with tags)

    CLOTHING:1,1,1,1,1,0,0.2,0,0.8,1,1,0.2,0,0,0,0,0

  • pose: (with tags)

    POSE:1,0,0,0,0,0,0.2,1,1,1,0,0,0,0,0,0,0

  • palette/color style: (with tags)

    PALETTE:1,0,0,0,0,0,0,0,0,0,0,0.8,1,1,1,1,1

  • Character:

    KEEPCHAR:1,1,1,1,1,0,0,0,1,1,1,1,1,1,1,0,0

  • Background:

    KEEPBG:1,1,1,1,1,1,0.2,1,0.2,0,0,0.8,1,1,1,0,0

  • Reduce overfitting (same as OUTALL）

    REDUCEFIT:1,0,0,0,0,0,0,0,1,1,1,1,1,1,1,1,1