This is a LoRA designed specifically for the Anima model, aiming to replicate the "3D-rendered-to-look-like-2D" (cel-shaded) art style of Custom Udon.

这是一个专为 Anima 模型打造的 LoRA，旨在复刻 Custom Udon 的三渲二的艺术风格。

Base model（基础模型）

Anima (anima-base-v1.0). NOT compatible with SDXL / Illustrious / Pony / Flux / Krea.

Anima (anima-base-v1.0)家族。与 SDXL / Illustrious / Pony / Flux / Krea 不兼容。

How to use（使用方式）

Trigger word: udonstyle (put it right after your quality tags, before the subject)

触发词：udonstyle（请将其紧跟在画质标签之后、主体描述之前）

Example（示例）:

udonstyle, masterpiece, best quality, score_7, 1girl, solo, upper body, looking at viewer, smile, simple background, white background

Negative（负面提示词）:

worst quality, low quality, score_1, score_2, score_3, artist name, blurry, jpeg artifacts, chromatic aberration

Combining with character LoRAs（结合角色 LoRA）

This is a style LoRA — it can conflict with character LoRAs at high weight.It may encroach upon the style of character-specific LoRAs, effectively overriding their style; you can try adjusting the weight or using the LoRA Block Weight plugin to inject the LoRA into specific layers.

这是一个风格类 LoRA——当权重较高时，它可能会与角色类 LoRA 产生冲突。它可能会出现侵占角色类LoRA的风格情况，导致“吃”角色LoRA的风格，您可以尝试调整权重，或使用LoRA Block Weight插件在不同层注入

Training details（训练内容）

Base: anima-base-v1.0

55 images · rank 64 · dim 64 · alpha 64

lr 4e-5 · 5280 steps · bfloat16