lacoz Asian girl Face LoRA (SFW NSFW) Part 4
세부 정보
파일 다운로드
모델 설명
The recommended LoRA weight is fixed at 0.7~1
It would be very helpful if you could read all the information below
☕ Support
If you find this LoRA useful, your support is always appreciated:
https://buymeacoffee.com/lacoz
🔹 Trigger
Trigger word: lacoz
When using the lacoz trigger, the LoRA’s identity becomes stronger and more defined.
Even without the trigger, the face blends naturally into the base model and remains consistent.
Feel free to use the trigger—or not—whichever works best for your workflow.
🔹 Test
I tested the following models :)
aimormixRealisctic_v10,
asian realism,
asian top beuaty,
AsianVibes_NSFW,
HirosakiMix＿NSFW,
JunX Asian Dreamix,
real asian beauty,
RuyanRealistic_E_AsianBeauty,
SD Realistic Asian Girls,
SEX Sexy Eastern Experience v3,
Sultry Grace Asian,
WildCardX -Asian, WildCardX REAL NSFW SFW
Try other models as well :)
Positive
lifestyle fashion photo,beautiful woman sitting at a modern coffee shop table,modern interior, wooden furniture, cozy atmosphere, afternoon sunlight,bust shot, chest-up, upper body only,
wearing a fitted white lace blouse,delicate embroidery, puffed sleeves, tight bodice, body-hugging silhouette,paired with a form-fitting high-waisted midi skirt, slim waist, hourglass figure,medium-sized breasts, full bust, prominent cleavage,sitting gracefully, hands delicately holding a coffee cup or resting on the table,gentle smile, soft expression,classic feminine vibe, warm vintage mood
Negative
(worst_quality:2.0) (MajicNegative_V2:0.8) BadNegAnatomyV1-neg broad hands cartoon, cgi, render, illustration, painting, drawing, muscular, athletic build, bodybuilder
🔹 Notes
Facial distortion may occur if the latent image resolution is low, or if the face appears too small within the latent image.
In such cases, using FaceDetailer is recommended. 🙂
Face distortion issues?
→ Fix Face Distortion in ComfyUI (FaceDetailer Guide) | Civitai
🔹 Technical Info
Training Environment: LoRA training using kohya sd-scripts.
Kohya-trained LoRAs work just fine in A1111 :)
Runtime UI: ComfyUI
Recommended Nodes:
LoraLoaderModelOnly
LoraLoader
CLIPSetLastLayer