Superkawaii

🎨 Overview

A specialized LoRA designed to replicate the distinct, cohesive Superkawaii art style. Optimized for anime-style base models with consistent, predictable outputs.

🧪 Training Information

• Dataset: 128 high-quality, manually filtered images with strict style uniformity

• Training Strategy: Low learning rate + extended training steps to capture nuanced style details while minimizing overfitting

⚙️ Optimal Generation Configuration

• Trigger Word: Superkawaii

• LoRA Weight: 0.6–1.0

• CFG Scale: 4–5

• Resolution: 704 × 1024

• Recommended Positive Prompts: kawaii, illustration, anime style, digital painting, artbook, pastel colors, clean lineart, soft shading

💡 Pro Tip: Adjust the weight between 0.6–0.8 for subtle style integration, or use 1.0 for maximum style fidelity.

这个模型用于创建Superkawaii的绘画风格！

使用128张人工修剪的风格高度统一数据集进行训练。

推荐的数值：

宽度：704

高度：1024

权重：0.6-1

CFG：4-5

触发词：Superkawaii

推荐标签：kawaii, illustration, anime style, digital painting, artbook, pastel colors

该模型经过荒谬的超高轮次低学习率训练，可精准创建示例的画风。