AruAstyle - standard

AruAstyle

LORA
原创


更新

AruAstyle by Aru on Tensor.Art
AruAstyle by Aru on Tensor.Art
AruAstyle by Aru on Tensor.Art
AruAstyle by Aru on Tensor.Art
AruAstyle by Aru on Tensor.Art

📌 About This LoRA This LoRA is designed to give your generations a pure, clean 2D flat anime style with crisp line art and neat cell shading.

During the dataset curation, I intentionally removed all images with heavy lighting, bloom, and complex realistic rendering. As a result, this LoRA acts as a perfect "Base Style". It does not have baked-in lighting effects, which means it will not clash with your prompts and works incredibly well when mixed with other specific Lighting or FX LoRAs!

📦 Versions Available I am uploading 3 different versions based on the training epochs. Try them out and find the one that best fits your workflow!

  • Standard : Balanced. Highly recommended as your starting point. It offers great flexibility and style adherence.

  • Strong : Enhances details and reinforces the specific 2D flat art style more aggressively.

  • Lite : Simpler, cleaner, and strictly flat. Great for minimalist cell-shaded anime styles.

⚙️ Recommended Settings & Weight Guide After extensive testing, here is how the LoRA responds to different weights:

  • 0.4 ~ 0.6 (🔥 Sweet Spot): Amazing synergy with most anime checkpoints. It naturally applies the 2D flat coating without breaking the base model's anatomy.

  • 0.6 ~ 0.8: Strongly forces the flat 2D style while remaining highly aesthetic.

  • 1.0 ~ 1.2: Very close to the original training dataset style. (Might be a bit stiff depending on your prompt).

💡 Pro Tip: Use this LoRA at 0.4 ~ 0.5 weight as your foundational style, and combine it with your favorite lighting LoRAs (like moonlight, sunset, or cinematic lighting) for the best results!

版本详情

Illustrious

项目权限

    使用权限

  • 在吐司在线使用

  • 在 吐司 作为在线训练的底模

  • 使用时无需注明出处

  • 用于模型融合

  • 分享融合模型时使用不同的许可

    商用许可

  • 生成的内容用于商业用途

  • 作为生成服务来商用

  • 转售模型或出售融合模型

相关帖子