
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: kimminju 
- 
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)
- 
1st Adetailer model: face_yolov8n.pt Inpaint denoising strength: 0.25   
- 
2nd Adetailer model: mediapipe_face_mesh_eyes_only Inpaint denoising strength: 0.25   
- 
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 
 
- 
描述:
训练词语: kimminju
名称: Kimminju_v1.2.safetensors
大小 (KB): 147575
类型: Model
Pickle 扫描结果: Success
Pickle 扫描信息: No Pickle imports
病毒扫描结果: Success
                    
                
                    
                
                    
                
                    
                
                    
                
                    
                
                    
                
                    
                
                    
                
                    
                
                    
                
                    
                
                    
                
                    
                
                    
                
                    
                
                    
                
                    
                
                    
                
                    
                

 
                 
                 
                 
                 
                