Imitation model. Modelled after retired european-like Japanese AV star Tia (ティア, aka Tia Bejean).
Retired back around 2014, some of you might know this one.
For some context on the actress’s background, she is technically not ‘Hafu’ despite her slightly euro-centric looks. This former actress was one of the chart-toppers when she first debuted; but clearly died down since then. Her vibe is more of a cute-highschool girl type, but she does well in genre that require more mature looks and elegance too. Whilst she’s technically not mixed-race, the LoRAs do reveal fun dynamics that where different models weighs different facial dynamics differently! Experiment for yourself~
~~~~~~~~~~~~~~~~~~~~~~~~~~
Recommended LoRA strength: 0.9 (0.7-1.2 are plausible range dependent on checkpoints; higher values lead to more refined eyes, freckles and rough skin texture)
Main Prompt:
tiabejean_jav, breasts, eyeshadow, long hair, brown hair, blonde hair, freckles:0.4, eyeshadow, makeup,
variations: single hair buns, black hair, tied hair, ponytail, and of course twintail
facial variations: thick makeup, teeth, parted lips, juicy lips*
*I do not find the need to use the facial prompts in most checkpoints, unless your checkpoint have a bias towards minimal makeup
The reason why we have brown and blonde hair, is for SD to create shades that sits between. For the pros, you can use [brown hair|blonde hair:xx], which has a similar effect.
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Checkpoints that work well for this LoRA:
1) PerfectWorld, or other photoRealistic Hentai models (European model works well for this LoRA). This time, PerfectWorld does very well, again.
2) Realisian. Tend to generate minor-asian bias, which is understandable. the “eyeshadow” prompt can address that.
3) RealisticMix. Similar to Realisian actually.
4) ClearRealismMix-Eur, or asian-based/eurasian checkpoints. Not perfect, but still better than pure-European checkpoints.
~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Note for achieving good resemblance:
Use Adetailer for baseline generation (at 512, 768 etc) only, try to control LoRA strength at 0.9-1.0, adjust your denoise accordingly. For this LoRA, recommended to use Adetailer at ~0.4 denoise strength for close-up, and 0.45 to 0.55 for full body.
For this LoRA, you dont need to set the prompts for Adetailer separately. I prefer to use just the main prompt, and turn ADetailer on. But this works too if you insists:
“face of tiabejean_jav {insert_lora}, eyeshadow, freckles:0.3, thick makeup, long hair, brown hair, blonde hair, blush”
for good resemblance, you only need to pay attention to 2 things:
1) the shape of the eyebrows, and
2) her smoky eyeshadow.
Recommended approach: is to not use HiRes, but upscale it with img2img instead. Though in retrospect, HiRes isn’t too bad for this LoRA as well.
Use a denoise strength of 0.25 -0.3 (anything higher will overcorrect face during the process), and turn off Adetailer.
~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Issues:
in img2img upscaling, sometimes the shape of her eyes can be alittle off – I guess you can try running Adetailer again with very low denoise to try and fix that. What I didn’t manage to show is that this LoRA works very well with nigh gown/evening dresses etc (as it reveals too much cleavage – pushing the limit of the CivitAI TOC), but try it out yourselves, responsibly.
Post-production edits/Reproducibility:
Some preview images underwent inpainting manually for clothing and badhands.
Happy prompting. And do share [SFW] creations~