PVCXL v1.0 LoRA for AnimagineXL 3.0 / AnimagineXL 3.0 base

HuggingFace: https://huggingface.co/p1atdev/pvcxl-v1-lora

PVC style LoRA trained on cagliostrolab/animagine-xl-3.0-base.

Recommended Settings

LoRA strength

• default: 1.0

• heavy: 1.25

Generation settings

• CFG scale: about 6.0

• Inference steps: 24~

• Sampler: Euler A

Prompts

If you want strong pvc style, start the positive prompt with pvc figure of.

Also, putting flat color in the negative prompt makes it makes more like the texture of pvc figures.

Positive

pvc figure of 1girl, upper body, hoodie, window, curtains, sunlight leaks, brown hair, medium hair, shadow, expressionless, looking outside, eyes, masterpiece, best quality, newest, late

Negative

flat color, nsfw, lowres, bad anatomy, bad hands, text, error, missing fingers, extra digit, fewer digits, cropped, worst quality, low quality, normal quality, jpeg artifacts, signature, watermark, username, blurry, artist name

Training details

Dataset

Curated high quality 2.5k pvc figure images from p1atdev/pvc

p1atdev/pvc-quality-swinv2-base was used to curate images.