---
license: cc-by-nc-4.0
tags:
- text-to-image
- stable-diffusion-xl
library_name: diffusers
pipeline_tag: text-to-image
---

![sample](https://huggingface.co/judgeou/pvcgine-xl/resolve/main/00159-3246781111.jpg)

这是用一个用一些pvc塑料小人图片与 [animagine-xl-3.0-base](https://huggingface.co/cagliostrolab/animagine-xl-3.0-base) 微调训练出来的模型，从一个大的数据集里面随机抽取了1000张进行训练，非常感谢群友提供的数据集。

使用 chibi 大概率可以得到粘土人。

## 更新

* [2024-02-22] pvcgine-xl-v20240222.safetensors 降低了 arb 桶分辨率，防止过度拉伸图像，提升了清晰度

![sample](https://huggingface.co/judgeou/pvcgine-xl/resolve/main/xyz_grid-0043-2468089033-1.jpg)
![sample](https://huggingface.co/judgeou/pvcgine-xl/resolve/main/xyz_grid-0053-3505932595.jpg)
![sample](https://huggingface.co/judgeou/pvcgine-xl/resolve/main/xyz_grid-0057-198065193.jpg)
![sample](https://huggingface.co/judgeou/pvcgine-xl/resolve/main/xyz_grid-0068-1541593507.jpg)