Sign In

JAV Yuko_Ono

78
646
23
Verified:
SafeTensor
Type
LoRA
Stats
646
0
Reviews
Published
Mar 2, 2024
Base Model
SD 1.5
Training
Steps: 1,550
Epochs: 10
Usage Tips
Clip Skip: 1
Strength: 0.8
Trigger Words
yukoono_jav
breasts
long hair
brown eyes
Hash
AutoV2
0C3341F495
Created on Civitai

Imitation model. Modelled after Japanese AV star Yuko Ono (小野夕子 [2020]; also previously known as Aoi 葵 [2014]).

The actress reasonably aged quite abit during her reappearance since her first debut - to address that, the training data includes both images/photographs when she debuted under 葵, as well as 小野夕子, with more than 75% on the latter. hopefully this gives the model more flexibility in retaining some of her youth (ie body figure), with a hint of her more refined mature facial features (see examples).

So instead of modelling her on-stage appearance static within the activation token, this approach allows for better flexibility to recreate versions of her with and without makeup.

I hesitated in deciding which epoch to pick, since some models performed better with one epoch over the other; but decided to go with this since the average performance is the best over most asian models.


Recommended LoRA strength: 0.8 (0.6-1.2 are plausible range dependent on checkpoints; higher values lead to more refined and sharper facial features)

Main Prompt:

yukoono_jav, breasts, long hair, black hair, brown eyes

variations: hair buns, brown hair,

Checkpoints that work well for this LoRA:

1) Realisian, or any of the Magicmix variants. I personally find Realisian's render being closest.

2) RealisticMix, or asian-based/eurasian checkpoints.

3) PerfectWorld, or other photoRealistic Hentai models. Not too bad, captures facial detail correctly, but body ratio can be off in some seeds.

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 ~0.35 to 0.45 denoise strength for close-up, and 0.45 to 0.55 for full body.

If your checkpoint do not respect makeup prompts as much, try using something like this in face Adetailer:

face of yukoono_jav <lora:JAV_Yuko_Ono-000008:0.8>, mature makeup, lipstick:0.2, eyeshadow

Recommended approach: is to not use HiRes, but upscale it with img2img instead.

Use a denoise strength of 0.25-0.35, and turn off Adetailer.

Issues:

This actress has a fairly petite body (stunning, perfect proportions, but short..160-ish cm). So if you render her in "western" model checkpoints, her proportions might be off (I experienced that with PerfectDeliberate and Photon). Either re-seed for a better generation, or artificially tweak this with Weight/body size slider LoRAs.

Happy prompting. And do share creations~