v1.0 base

本lora基于anima base v1.0 本lora基于100张莫奈的画作训练了2000步，我认为其在风景上表现尚可，但在人物肖像方面表现较差，尤其是特写肖像。 本lora可能会生成作者签名，请务必在负面提示词中加以抑制！ 建议权重在1左右。建议配合例图中的tag使用。

这是我的第二个lora。其生成效果可能不稳定。

Base Model: Anima Base v1.0

Training: 100 Monet paintings, 2000 steps. Performance is decent on landscapes, but not recommended for portrait subjects, especially close-ups.

⚠️ This LoRA may generate artist signatures. Please add "signature,text,watermark" to your negative prompts!

Suggested strength: ~1.0. Refer to tags in example images for best results.

This is my second LoRA. Results may vary.

v1.1 base

该版本使用120张莫奈的图像训练。皮肤风格较上一版有变化。很难说哪一版本更好，请依据情况使用。不同的是，该版本建议权重为0.8到1.0

This version is trained on 120 images by Monet. The skin style has changed compared to the previous version. It's hard to say which version is better; please choose according to your needs. The difference is that this version recommends a strength of 0.8 to 1.0.