This LoRA is trained to reproduce a mature anime-style female character with soft facial features and a gentle, calm expression. The main focus of this model is facial identity and character consistency rather than extreme stylization.
The character features silver ash grey hair, blue eyes, and a warm, approachable look. The LoRA performs best in portraits, upper-body shots, and daily-life style illustrations.
V1 Trigger word
Character :
Hayasaka, caramel skin, light warm grey hair, long straight hair, side-parted bangs, side braid detail, hair gathered at the back, long loose hair flowing behind, ribbon hair bow, small hoop earrings,
Standard Outfit:
(navy blue top:1.2), navy blue short sleeve slim-fit workwear top, collared neckline, front button placket, navy blue slim office pencil skirt, high-waisted skirt, knee-length skirt, sheer pantyhose, sheer tights, thin nylon stockings,
Leisure Outfit:
(pink sleeveless ribbed turtleneck top:1.2), fine vertical rib texture, light cream beige long pleated skirt, light cream beige ankle-length long skirt,
Massage Outfit:
(plain white sleeveless fitted tank top:1.2), (black sport bra inside the tank:1.2), low neck line, cleavage, navel, (pink loose-fit athletic shorts:1.5), contrast trim, side piping,
Massage Ponytail:
long straight hair, side-parted bangs, long face-framing sidelocks, side braid detail, hair gathered at the back, hair tied into a mid-to-high ponytail, fabric ponytail scrunchie,
Recommended weight: 0.8 – 1.1
This LoRA works best with anime-style base models. For better results, clearly describe hairstyle, expression, outfit, and mood in your prompt.
Trained on a curated dataset with structured captions separating character identity, facial traits, hairstyle, outfits, and accessories, aiming to improve stability and controllability.
Suitable for original character illustrations, portraits, and casual scenes.
Not intended for photorealistic rendering.
This release focuses on character accuracy. Refined or alternate versions may be released in the future.
Feedback and shared results are welcome.
Description
Model Overview
This is v2 (short-tags trained) of the same character LoRA.
Compared to v1 (structured-caption version), this model was retrained on the same image set, but using a compact tag-style caption system instead of long descriptive layered captions.
The goal of v2 is to provide a more prompt-efficient, flexible, and Civitai-friendly version of the character.
FAQ
Details
Files
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.