Type: LoRA / Style　Base model: Illustrious (trained on WAI-Illustrious-SDXL v17)

Trigger word: chdstyle

Recommended weight: 0.35 (0.25–0.5). 0.7 applies very strongly.

What it does

- Softens skin into smooth, matte planes and tones down white highlights, wet shine and harsh contrast.

- Bold outlines and large hair masses, with only a few thin inner lines (no dense hair-thin hatching).

- Gentler, more mature faces: a shorter lower face and a rounded chin instead of the typical pointed one.

- Keeps faces, skin and hair midtones readable even in night scenes.

- Designed to work at a low weight (around 0.35), so the base model's composition, hands and anatomy stay intact.

Dataset

- 100% original. Every training image was made from scratch with an image generator as an original adult woman.

- No existing artwork, no copyrighted characters and no artist names were used.

- The set was deliberately varied in skin tone, hair, face shape, face angle and framing (close-up, upper body and full body), so the style does not pull toward one face or one composition.

- Captions include the trigger plus full content tags (hair, clothes, place and body type), so clothing and hairstyles still follow your prompt.

Known tendencies (why I retired it)

- The training images exaggerated softness so that the LoRA would work at a low weight. As a result, figures tend to come out fuller (bust and hips) and colors lean warm.

- At 0.7 the LoRA can override prompted clothing colors.

- Lowering the weight to 0.25–0.35 reduces both effects.

Comparison

- Same seed for every column. Left: no LoRA / Middle: 0.35 / Right: 0.7.

- Settings: WAI-Illustrious v17, euler_ancestral, 28 steps, CFG 5, 896×1152.

日本語（短く）

- 肌の柔らかさと、包容力のある大人の雰囲気を出す画風LoRAです。学習用の絵は全部オリジナルで、既存の絵や絵師名は使っていません。

- 推奨は0.35です。体型が豊満寄り・色が暖色寄りになりやすいので、気になる場合は0.25〜0.35に下げてください。