💬 Discord — join here · ☕ Ko-fi — support the factory

Hsin — Wuthering Waves 鸣潮

Black-tipped fox ears, red claws, and a smile with a fang in it.

Character LoRA for Hsin, the white fox lady of Kuro Games' Wuthering Waves. The trigger hsinwuwa carries her whole identity block — fox ears and tail, white hair with black tips, red-rimmed amber eyes, the red glyph on her forehead, her chest and arm markings, and the long red claw-nails. All of it was stripped from the captions, so the trigger is the description — you never describe her face.

She trained on 165 images from 141 different artists, so her wardrobe stayed steerable: the canon black dress (67) with detached sleeves (27) is the default, with a trained alternate pool on top — alternate costume (18), swimsuit / bikini (10). Her signature fox hand sign is in 27 of them.

🧩 Version

• 🔵 v1.0 Anima — for Anima / Qwen-Image DiT checkpoints. ✅ Works in the Civitai on-site generator. Weight 1.0 (usable range 0.85 – 1.0).

• 🟢 Illustrious — for SDXL / Illustrious / Pony / NoobAI checkpoints. Weight 1.0.

📋 Prompt

Canon kit — her default look:

hsinwuwa, 1girl, solo, black dress, detached sleeves, bare shoulders, hair ornament, cleavage, cowboy shot, looking at viewer, smile

The fox hand sign — her signature gesture:

hsinwuwa, 1girl, solo, fox shadow puppet, upper body, smile, fang, looking at viewer

Off-duty — the trained alternate wardrobe:

hsinwuwa, 1girl, solo, alternate costume, black bikini, navel, sitting, looking at viewer

NO quality tags needed. Keep prompts LEAN — long tag stacks dilute the trigger.

🌸 Krea 2

The Krea version speaks prose, not tags. Same trigger, hsinwuwa — it carries her whole identity block on its own (fox ears and tail, white hair with black tips, the red glyph, the fang, the claw manicure). Ask for the kit in plain words:

hsinwuwa, full body shot of the subject standing in red and white robes with gold trim

Runs on Krea 2 Turbo: 8 steps, CFG 1.0, euler/simple, no negative needed. Showcases stack Semi Real Anime Soft Light 2.5D at 0.8 with its trigger semireal.

💡 Tips

• Give her somewhere to be. 59 of 165 sources are on a plain background and only 10 are full body — scenery is her weakest area. Name the room and the light; upper body and cowboy shot are safest.

• Fan-pool junk negative. 20 of 165 images are chibi, memes or captioned art — if a render goes chibi or grows text, add chibi, english text, text, logo, watermark, character name, border.

• Chest bias: large breasts is on 126 of 165. medium breasts is trained (11) and pulls it down; small breasts is not in the set at all.

• Her tail is baked but thin — visible in only 30 of 165 sources, so it shows up mainly at full-body framing.

• The canon dress is sticky. For another outfit, lead with the garment, drop every canon word, and negative black dress, detached sleeves if it creeps back.

More of my LoRAs: Techannel on Civitai — likes & Buzz keep the factory running ❤️