Type | |
Stats | 379 2,680 123 |
Reviews | (28) |
Published | Mar 11, 2024 |
Base Model | |
Training | Steps: 10,000 Epochs: 10 |
Usage Tips | Clip Skip: 1 Strength: 0.5 |
Trigger Words | Aesthetically_Pleasant_Woman, Pleasant:Blonde, Pleasant:Redhead, Pleasant:Asian, Pleasant:DarkTone, Pleasant:BrownTone |
Hash | AutoV2 7A5D29E95C |
This LORA aims to help reduce tendencies of certain base models to prefer one specific ethnicity over others. This is also aiming to provide more aesthetically pleasant woman subjects with appropriate hair color, skin tone based on desired race or ethnicity of the subject. Simply use these keywords to gain proper aesthetics for your subjects -
Aesthetically_Pleasant_Woman, Pleasant:Blonde - for Blondes
Aesthetically_Pleasant_Woman, Pleasant:Redhead - for Redheads
Aesthetically_Pleasant_Woman, Pleasant:Asian - for Asians (Chinese, Japanese, Thai etc.)
Aesthetically_Pleasant_Woman, Pleasant:BrownTone - for Females with Brown or Fair skin tones with Brown or Black hair (Brunettes, Arab, South Asian, Hispanic etc.)
Aesthetically_Pleasant_Woman, Pleasant:DarkTone - for Females with Dark skin tones (Ethiopian, African Americans etc.)
Depending on the base model you are using you may have to adjust the LORA strength accordingly, in some models strength of 1 may work very well, for others you may have to use lower values (0.35 to 0.65) or may have to higher values (1.25 or higher).. also the weight would have to be adjusted when using with other LORAs or Embeddings. While this is not trained with NSFW data it may produce some NSFW stuffs.