Priscilla (Re:Zero) — Illustrious SDXL Character LoRA

中文说明 | English (current)

An unofficial character LoRA for Priscilla Barielle from Re:Zero − Starting Life in Another World, trained on waiIllustriousSDXL v17. Publishes formal E1–E7. Recommended default: E6 (priscilla_rE5S755-000006.safetensors) at strength 0.8. Dataset: abbuibuibui/Priscilla-Dataset.

Fan-made derivative; character rights remain with their owners.

Files

Label File Role
E1 priscilla-000001.safetensors Alternate. Early.
E2 priscilla-000002.safetensors Alternate. Early.
E3 priscilla-000003.safetensors Alternate.
E4 priscilla-000004.safetensors Alternate.
E5 priscilla_rE5S755-000005.safetensors Alternate. Matrix candidate.
E6 priscilla_rE5S755-000006.safetensors Recommended default at 0.8.
E7 priscilla_rE5S755.safetensors Alternate final formal epoch (resume).

priscilla.safetensors (unique-SHA leftover from interrupted first run) is not published.

E6 SHA-256: ee91697bfe4685ff3878be2e1a0499d9a24a031b0c064c41775df6880163c1c7

Showcase

Direct E6 @ 0.8 outputs. Euler a, 28 steps, CFG 4.5, CLIP skip 2, 832×1216.

Portrait Red dress full body
Face Main

Night

Trigger Words and Settings

Role Token
Identity priscilla_barielle (blonde hair, long hair, orange eyes)
Outfit priscilla_red_dress (40/40)

Recommended strength 0.8. Example:

<lora:priscilla_rE5S755-000006:0.8>,
masterpiece, best quality, very aesthetic, absurdres, anime screencap,
1girl, solo, priscilla_barielle, priscilla_red_dress,
blonde hair, long hair, orange eyes,
full body, looking at viewer, outdoors

Why E6

Seven formal epochs screened @ 0.8 (face + red dress). E5–E7 entered the weight matrix (9 scenes × 0.4/0.6/0.8/1.0). E6 best overall balance at 0.8. No alt outfit in this set.

Full notes: evaluation/横向测评记录.md.

Public Horizontal Evaluation

Round 1 @ 0.8

Front portrait Red dress
R1 face R1 dress

Round 3: E5 / E6 / E7 × weights

Face Three-quarter
01 02

03

04

07

08

09

10

11

Training

  • 40 stills, single outfit folder; captions start with priscilla_barielle, priscilla_red_dress, …; keep_tokens=2, repeats=4.
  • LoRA dim 32 / alpha 16; AdamW8bit; TE + UNet LoRA; bf16, 1024 buckets.
  • 7 epochs (first run interrupted mid-E5, resume priscilla_rE5S755 to completion).

Related repositories

Compatibility and Limits

Best on waiIllustriousSDXL / Illustrious family. Single outfit token.

License

CreativeML OpenRAIL-M for weights. Character © original rights holders. Non-commercial fan use.

Downloads last month
54
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Dataset used to train abbuibuibui/Priscilla-Lora