⚡ Update: the recommended weight is now pre-baked into the file — just drop it in and use it at weight 1.0 (no need to set 0.8). Lower only if stacking many LoRAs.

Cantarella (Wuthering Waves) LoRA

Trigger word: ctl

Trained base: waiIllustriousSDXL_v170 (Illustrious XL family)

Network: dim=32, alpha=16

Optimizer: Prodigy (lr=1.0)

Recommended settings

• Weight: 0.8 (identity metrics measured at 1.0; human quality calibration caps the recommendation at 0.8).

• - Hi-res fix: recommended (~1.8x, denoise ~0.45) for clean lines.

• - Checkpoint: passed the 5-criteria acceptance harness (identity / no-overtrain / NSFW / outfit).

Usage

1girl, solo, ctl, mature female, purple hair, blue eyes, long hair, mark under eye,
hair flower, blue flower, headdress, purple nails, white dress, white thighhighs,
<pose here>, <background here>,
masterpiece, best quality, amazing quality, absurdres, newest
```

Recommended weight **0.8**. Trigger `ctl`.

**-- Identity (baked into the trigger) --**
purple hair, blue eyes, and her signature **marks under the eyes** (`mark under eye`, `facial mark`). Keep those tags in for full likeness — the under-eye marks especially. She reads as a **mature woman**; add `mature female` if the face skews young.

**-- Trademark details (add for the canonical look) --**
white frilled **headdress** + **blue flower** hair ornament, teardrop earrings, **purple nail polish**, white dress, white thighhighs.

**-- Style note (IMPORTANT) --**
For clean anime shading (not 3D / semi-real skin), put anti-realistic terms in the NEGATIVE prompt: `realistic, 3d, photorealistic, shiny skin, oiled body, semi-realistic, render`. The negatives alone are enough. This matters most on nude / exposed skin.

**-- NSFW --**
Identity holds when undressed / nude; the under-eye marks and headdress persist through explicit scenes. Keep weight ~0.8 for coherent full-body / dynamic ero.

## Tested checkpoints

- ✅ `waiIllustriousSDXL_v170` (trained base)
- 🔍 Other Illustrious XL fine-tunes (untested; start weight 0.6-0.8)
- ❌ Pony / SDXL base (architecture mismatch, untested)

## Caveats

- On checkpoints other than the training base, the recommended weight shifts (try 0.7-1.2).
- Only outfit variations present in the dataset reproduce strongly.
- Keep the `mark under eye` tag for full likeness — it is a defining trait.