简介:

尝试对装饰主义和华丽质感插画进行学习而来的LoRA模型.

训练集41张.

触发词(trigger):

<taiki>, fantasy art, highly detailed 

或是

<taiki>, fantasy art, highly detailed  (要包含尖括号)

底模:

训练使用ReVAnimated, 例图为ReVAnimated生成. 建议使用笔触或者金属细节比较强的模型, 可以是各种现实系模型, 搭配其他二次元模型也可能有不错效果

推荐参数:

• 生图尺寸在 512*512 - 1024*1024 之间均可, 长宽比尽量一致.

• LoRA weight不要超过0.9, 在0.5-0.9之间会有不错效果

• step 25, 不要过低

• CFG 6-8

• Denoise: 0.6-0.7

提示词建议:

使用 royal_crimson, Fantasy_Inkwash 这类质感提示词会有不错发画面效果

使用声明

CUC AI STUDIO 出品
欢迎大家使用这款 LoRA 自由创作！我们很期待看到你的返图作品~
本模型仅供个人学习和非商业创作使用，严禁商用哦！

​​Introduction:​​

A LoRA model trained by studying Art Deco and ornate-texture illustration.

Training set: 41 images.

​​Trigger words :​​

<taiki>, fantasy art, highly detailed

or

<taiki>, fantasy art, highly detailed (Must include angle brackets)

​​Base model:​​

Trained on ReVAnimated. Examples shown were generated with ReVAnimated. Recommended for use with models featuring strong brushstrokes or metallic details (various realistic styles), or it may also pair well with other anime models.

​​Recommended Parameters:​​

• ​​Resolution:​​ 512 * 512 to 1024 * 1024, keeping aspect ratios consistent.

• ​​LoRA weight:​​ Do not exceed 0.9. Best results between ​​0.5-0.9​​.

• ​​Sampling Steps (Step):​​ ​​≥ 25​​ (Do not set too low).

• ​​CFG Scale:​​ ​​6-8​​

• ​​Denoising strength (Denoise):​​ ​​0.6-0.7​​

​​Prompting tips:​​

Including texture keywords like royal_crimson or Fantasy_Inkwash in prompts often yields excellent visual results.

​​Usage Policy:​​

CUC AI STUDIO Production​​

Welcome to use this LoRA freely for your creations! We eagerly await seeing your feedback and artworks!
This model is provided for ​​personal learning and non-commercial creative purposes ONLY​​. ​​Commercial use is strictly prohibited!​​