README.md

---
license: cc-by-nc-4.0
---

* 本模型仅作学习交流使用，禁止用于任何商业活动或侵犯他人版权、肖像权等权利的行为。
* Study and communication only. Any commercial activity and violation of rights like portrait and copyright is prohibited.

几个自己常用的融合的模型，成分写在末尾了，不过比例不记得了。

Some of my merged models, the ingredients are at last of readme, though I forgot the proportions.

*推荐参数*

（其实也没什么推荐的，想怎么用怎么用，这只是我常用的设置）

*Recommended parameters*

(Anyhow, use it as you like. Here are my parameters)

Sampler: DPM++ SDE Karras, or DPM++ 2M Karras

Steps: 20

CFG scale: 8

Clip skip: 2

负面推荐使用EasyNegative，推荐使用高清修复。
EasyNegative recommended. hire.fix recommended.

## ReedMix_A

偏厚涂画风的模型。脸部画得比较好，背景需要较多tag才能画得比较细致。**容易出nsfw**。

Impasto-like model. Good at face, but many tags needed for detailed background. **Prone to nsfw**

![](./images/A_1.png)
![](./images/A_2.png)
![](./images/A_3.png)

## ReedMix_B

比A的背景细节更多，但脸部略不如A。

More details in background than A, while a bit worse at face.
![](./images/B_1.png)
![](./images/B_2.png)
![](./images/B_3.png)

## ReedMix_C

以counterfeit 2.5为基础融合的模型。

Model based on counterfeit 2.5.

![](./images/C_1.png)
![](./images/C_2.png)
![](./images/C_3.png)

## ReedMix_P

偏平涂画风的模型，全身图肢体不易崩，背景细节也较多。

Flat-color model, good at anatomy and detailed background.

![](./images/P_1.png)
![](./images/P_2.png)
![](./images/P_3.png)

### Acknowledgement

感谢这些作者发布的好用的模型

* 7th_anime_v3_A (https://huggingface.co/syaimu/7th_Layer)
* pastel-mix (https://huggingface.co/andite/pastel-mix)
* counterfeit-v2.5 (https://huggingface.co/gsdf/Counterfeit-V2.5)
* PileDream (https://civitai.com/models/20255)