Imitation model. Modelled after Japanese AV star veteran Saki Okuda (奥田咲)
Recommended LoRA strength: 0.8 (0.6-1.0 are plausible range for most scenarios; with 1.0 leaning towards more exposed clothings)

Note for achieving good resemblance:

Most of the tested checkpoints designed for asian faces tend to smoothen facial features and make characters appear younger. To address that, one way is to add "mature" tags, I particularly find age adjustment LoRAs (including anime ones, at very low weights useful: https://civarchive.com/models/230410). Unfortunately, I am unable to share the results here as it tend to come with more exposed clothings. But do experiment, responsibly.

Advice to use 'makeup prompts' as always: makeup, (eyeshadow), defined cheekbones, for better resemblence.

Check with your model characteristics, some models tend to create gothic looks when prompted with "eyeshadow" and "red lipstick". Adjust accordingly.

Face ADetailer optional, only needed for HiRes or full-body images - use at ~0.45 to 0.55 denoise strength.