Based on <a target="_blank" rel="ugc" href="https://civarchive.com/models/795765/illustrious-xl">illustrious-xl v0.1</a>Thanks for all the work done by the<a target="_blank" rel="ugc" href="https://civarchive.com/models/795765/illustrious-xl"> illustrious-xl team</a>!基于 <a target="_blank" rel="ugc" href="https://civarchive.com/models/795765/illustrious-xl">illustrious-xl v0.1</a>感谢 <a target="_blank" rel="ugc" href="https://civarchive.com/models/795765/illustrious-xl">illustrious-xl 团队</a>所做的所有工作!<h3 id="https:civitai.commodels795765illustrious-xl-zmqb5n3ts"><a target="_blank" rel="ugc" href="https://civarchive.com/models/795765/illustrious-xl">https://civarchive.com/models/795765/illustrious-xl</a></h3>Recommended Positive Prompts / 正向提示词:<pre><code>best quality, high quality, masterpiece, aesthetic, absurdres</code></pre>Recommended Negative Prompts / 负向提示词:<pre><code>worst quality, low quality, bad quality, lowres</code></pre>Recommended CFG / 建议 CFG:<pre><code>5~7</code></pre>Sampler 采样器 :<pre><code>Euler a</code></pre><h3 id="v0.1.3-log-rvtxpfg0b">v0.1.3 log</h3><pre><code>Thanks for your support!

Illustrious-Flying v0.1.3 was trained on 896 images. Of these, 60% to 63% (at least 560 images) were generated by me and contributed to the model's core style. If you find the style similar to certain images generated by NovelAI3 (or similiar model), that was my intention.

Thanks to the foundation laid by Illustrious v0.1, this model can generate images of most famous anime characters up to (and possibly beyond) January 2024.

Version 0.1.3 is quite overfit in style; therefore, some artist tags may not work well. I'm still considering the training direction for future versions. If you have any recommendations, please message me! :D

I am grateful to all those who inspired me and laid the foundation for my work, and to those who generously shared their knowledge and experience. Without them, I would have wandered in the cyber darkness for a very long time.

谢谢您的支持！

Illustrious-Flying v0.1.3 使用 896 张图像进行训练，其中 60% ~ 63% (最少 560 张图片) 由我自己生成，塑造了模型的核心风格。如果您发现该风格与一些 NovelAI3 (或相近模型) 生成的图像相似，那是肯定的，是我有意往这方面训练。

得益于 Illustrious v0.1 的基础，该模型能够生成截至 2024 年 1 月（或可能更远）大多数著名动漫角色的图像。

v0.1.3 的风格过度拟合，因此某些艺术家标签可能效果不佳。我仍在考虑未来版本的训练方向。如果您有任何建议，请随时联系我！:D

我非常感谢所有启发我，为很多工作奠定基础的人们，以及那些慷慨分享知识和经验的人们。没有他们，我将在网络的黑暗中迷失很久。
</code></pre>

Files with SHA256: eb1dd549849af0d5c555a4d718d59aedbc1205e5b0c7eeb32b21e24e7bfa00d3