Hair Library Vol.01 – Shadôws Collection
Welcome to Hair Library, a collection of modular LoRA models dedicated entirely to hairstyles.
This project was created with a simple idea in mind: instead of training one LoRA that tries to learn an entire character, each visual element is developed separately. Hairstyles become reusable assets that can easily be combined with different characters, outfits, expressions, and artistic styles.
This first release introduces two original hairstyles inspired by Shadôws Lucky. Both were trained using multiple viewing angles, including front, three-quarter, profile, and back views, allowing the model to better understand the overall structure and volume of each hairstyle rather than memorizing a single image.
For the best results, use trigger words such as hair01 or hair02 together with a portrait-style prompt. A simple prompt like masterpiece, best quality, solo, portrait, hair01 is usually enough to activate the hairstyle. You can then freely add your own descriptions, colors, clothing, accessories, or character details.
This LoRA performs best with a weight between 0.6 and 0.9, around 25 to 35 sampling steps, and works well with samplers such as Euler a or DPM++ 2M. Since it only focuses on the hairstyle, the final rendering style, colors, and overall appearance will still depend on the base checkpoint and the rest of your prompt.
Hair Library is designed as a long-term project. Future volumes will introduce many more hairstyles, including short cuts, long hair, ponytails, braids, fantasy designs, kemonomimi variations, and themed collections inspired by steampunk, cyberpunk, and gothpunk aesthetics.
Thank you for your support, and I hope this library becomes a useful tool for your own creations.
Description
Hair Library Vol.01 – Shadôws Collection
Welcome to Hair Library, a collection of modular LoRA models dedicated entirely to hairstyles.
This project was created with a simple idea in mind: instead of training one LoRA that tries to learn an entire character, each visual element is developed separately. Hairstyles become reusable assets that can easily be combined with different characters, outfits, expressions, and artistic styles.
This first release introduces two original hairstyles inspired by Shadôws Lucky. Both were trained using multiple viewing angles, including front, three-quarter, profile, and back views, allowing the model to better understand the overall structure and volume of each hairstyle rather than memorizing a single image.
For the best results, use trigger words such as hair01 or hair02 together with a portrait-style prompt. A simple prompt like masterpiece, best quality, solo, portrait, hair01 is usually enough to activate the hairstyle. You can then freely add your own descriptions, colors, clothing, accessories, or character details.
This LoRA performs best with a weight between 0.6 and 0.9, around 25 to 35 sampling steps, and works well with samplers such as Euler a or DPM++ 2M. Since it only focuses on the hairstyle, the final rendering style, colors, and overall appearance will still depend on the base checkpoint and the rest of your prompt.
Hair Library is designed as a long-term project. Future volumes will introduce many more hairstyles, including short cuts, long hair, ponytails, braids, fantasy designs, kemonomimi variations, and themed collections inspired by steampunk, cyberpunk, and gothpunk aesthetics.
Thank you for your support, and I hope this library becomes a useful tool for your own creations.