fooocus-api，得到一些训练图片的过程

## 弹簧器与阻尼器


![image.png](/static/img/921b0cfc0ad4a908d377b685a928727e.image.webp)






![image.png](/static/img/a8fd4f795eb3076bc882a9c547a279cd.image.webp)


![image.png](/static/img/cdc209d1a0aa967375fb9710b62f31e9.image.webp)

## 并联和串联


![image.png](/static/img/db6d41598e007f02df3c42ba38d5a40a.image.webp)


![image.png](/static/img/ecdb921342f51c50c1f199856ab3f0dd.image.webp)


![image.png](/static/img/8b90da26acbdb5304f288ff03f52e630.image.webp)


并联结构的系统方程包含质量、阻尼、刚度项，适用于描述自由振动和强阻尼系统，振动衰减较快。

串联结构的系统方程主要依赖位移和速度，通常用于柔性或低频振动系统，振动衰减较慢。

## 题目1



![image.png](/static/img/b277b208393f57ef6f2ab141b6405c8b.image.webp)


**解答：**

我们需要求解机械系统的传递函数 $G(s) = \dfrac{X_o(s)}{X_i(s)}$。为此，按照以下步骤进行分析。

### 1. 引入辅助点并建立力平衡方程

在阻尼器 $f_1$ 和弹簧 $K_1$ 之间引入一个辅助点，设其位移为 $x$，方向向下。系统的结构如下：

- **点 1**：位移为 $x_i$
- 阻尼器 $f_2$ 和弹簧 $K_2$ 并联，连接点 1 和点 2
- **点 2**：位移为 $x_o$
- 阻尼器 $f_1$，连接点 2 和辅助点
- 弹簧 $K_1$，连接辅助点和地面

### 2. 建立力平衡方程

对于点 2，根据力平衡，可得：

1. **从点 1 到点 2 的力：**

   - 弹簧 $K_2$ 的力：$K_2(x_i - x_o)$
   - 阻尼器 $f_2$ 的力：$f_2 \dot{(x_i - x_o)}$

2. **从点 2 到辅助点的力：**

   - 阻尼器 $f_1$ 的力：$f_1 \dot{(x_o - x)}$

因此，点 2 的力平衡方程为：

$$
K_2(x_i - x_o) + f_2 \dot{(x_i - x_o)} = f_1 \dot{(x_o - x)}
$$

对于辅助点，根据力平衡，可得：

- 弹簧 $K_1$ 的力：$K_1 x$
- 阻尼器 $f_1$ 的力：$f_1 \dot{(x_o - x)}$

由于弹簧 $K_1$ 和阻尼器 $f_1$ 连接在一起，力相等，因此有：

$$
K_1 x = f_1 \dot{(x_o - x)}
$$

### 3. 对方程进行拉普拉斯变换

假设初始条件为零，对上述两个方程进行拉普拉斯变换：

1. 对第一个方程：

$$
K_2 [X_i(s) - X_o(s)] + f_2 s [X_i(s) - X_o(s)] = f_1 s [X_o(s) - X(s)]
$$

2. 对第二个方程：

$$
K_1 X(s) = f_1 s [X_o(s) - X(s)]
$$

### 4. 解出 $X(s)$ 与 $X_o(s)$ 的关系

从第二个方程解出 $X(s)$：

$$
\begin{aligned}
K_1 X(s) &= f_1 s [X_o(s) - X(s)] \\
K_1 X(s) + f_1 s X(s) &= f_1 s X_o(s) \\
X(s) [K_1 + f_1 s] &= f_1 s X_o(s) \\
X(s) &= \frac{f_1 s}{K_1 + f_1 s} X_o(s)
\end{aligned}
$$

### 5. 消除 $X(s)$，建立 $X_i(s)$ 和 $X_o(s)$ 的关系

将 $X(s)$ 代入第一个方程：

$$
K_2 [X_i(s) - X_o(s)] + f_2 s [X_i(s) - X_o(s)] = f_1 s \left[ X_o(s) - \frac{f_1 s}{K_1 + f_1 s} X_o(s) \right]
$$

简化右侧：

$$
\begin{aligned}
f_1 s \left[ X_o(s) - \frac{f_1 s}{K_1 + f_1 s} X_o(s) \right] &= f_1 s X_o(s) \left[ 1 - \frac{f_1 s}{K_1 + f_1 s} \right] \\
&= f_1 s X_o(s) \left( \frac{K_1}{K_1 + f_1 s} \right) \\
&= \frac{K_1 f_1 s}{K_1 + f_1 s} X_o(s)
\end{aligned}
$$

因此，方程变为：

$$
\left( K_2 + f_2 s \right) [X_i(s) - X_o(s)] = \frac{K_1 f_1 s}{K_1 + f_1 s} X_o(s)
$$

整理得：

$$
\left( K_2 + f_2 s \right) X_i(s) = \left( K_2 + f_2 s + \frac{K_1 f_1 s}{K_1 + f_1 s} \right) X_o(s)
$$

### 6. 求解传递函数 $G(s)$

传递函数为：

$$
G(s) = \frac{X_o(s)}{X_i(s)} = \frac{K_2 + f_2 s}{K_2 + f_2 s + \dfrac{K_1 f_1 s}{K_1 + f_1 s}}
$$

### 7. 化简传递函数

首先，将分母中的分式合并：

$$
\begin{aligned}
\text{分母} &= K_2 + f_2 s + \frac{K_1 f_1 s}{K_1 + f_1 s} \\
&= \frac{(K_2 + f_2 s)(K_1 + f_1 s) + K_1 f_1 s}{K_1 + f_1 s} \\
&= \frac{K_2 K_1 + K_2 f_1 s + f_2 s K_1 + f_2 f_1 s^2 + K_1 f_1 s}{K_1 + f_1 s}
\end{aligned}
$$

将分子也乘以 $K_1 + f_1 s$：

$$
\text{分子} = (K_2 + f_2 s)(K_1 + f_1 s)
$$

因此，传递函数变为：

$$
G(s) = \frac{(K_2 + f_2 s)(K_1 + f_1 s)}{(K_2 + f_2 s)(K_1 + f_1 s) + K_1 f_1 s}
$$

### 8. 展开并整理

展开分子：

$$
\text{分子} = K_2 K_1 + K_2 f_1 s + f_2 s K_1 + f_2 f_1 s^2
$$

展开分母：

$$
\begin{aligned}
\text{分母} &= K_2 K_1 + K_2 f_1 s + f_2 s K_1 + f_2 f_1 s^2 + K_1 f_1 s \\
&= K_2 K_1 + (K_2 f_1 + f_2 K_1 + K_1 f_1) s + f_2 f_1 s^2
\end{aligned}
$$

### 9. 写成标准形式

将分子和分母写成关于 $s$ 的多项式：

$$
\begin{aligned}
\text{分子} &= f_1 f_2 s^2 + (K_1 f_2 + K_2 f_1) s + K_1 K_2 \\
\text{分母} &= f_1 f_2 s^2 + (K_1 f_2 + K_2 f_1 + K_1 f_1) s + K_1 K_2
\end{aligned}
$$

### 10. 归一化处理

为了进一步简化，我们将分子和分母同时除以 $K_1 K_2$：

$$
G(s) = \frac{\dfrac{f_1 f_2}{K_1 K_2} s^2 + \left( \dfrac{f_1}{K_1} + \dfrac{f_2}{K_2} \right) s + 1}{\dfrac{f_1 f_2}{K_1 K_2} s^2 + \left( \dfrac{f_1}{K_1} + \dfrac{f_2}{K_2} + \dfrac{f_1}{K_2} \right) s + 1}
$$

### 11. 最终的传递函数

得到机械系统的传递函数：

$$
G(s) = \frac{\dfrac{f_1 f_2}{K_1 K_2} s^2 + \left( \dfrac{f_1}{K_1} + \dfrac{f

_2}{K_2} \right) s + 1}{\dfrac{f_1 f_2}{K_1 K_2} s^2 + \left( \dfrac{f_1}{K_1} + \dfrac{f_2}{K_2} + \dfrac{f_1}{K_2} \right) s + 1}
$$

### 总结

通过引入辅助点并建立力平衡方程，利用拉普拉斯变换，将位移 $X(s)$ 表示为 $X_o(s)$ 的函数，进而消除 $X(s)$，建立了 $X_i(s)$ 和 $X_o(s)$ 之间的关系。经过代数化简，得到了机械系统的传递函数 $G(s)$。

## 题目1的问题

**解答：**

您的问题主要是关于如何确定力的方向，以及为什么力的表达式是那样的。让我一步步解释。

### 1. 力的方向如何确定？

在机械系统中，力的方向取决于元件的特性和位移（或速度）的方向。一般来说：

- **弹簧力**：弹簧力的方向总是试图恢复弹簧的原始长度，即阻止弹簧被拉伸或压缩。它的大小为 $F = K \Delta x$，方向与位移的增量 $\Delta x$ 相反。

- **阻尼器力**：阻尼器的力与两端相对速度成正比，大小为 $F = f \Delta \dot{x}$，方向与相对速度 $\Delta \dot{x}$ 相反。

### 2. 为什么弹簧 $K_2$ 的力是 $K_2(x_i - x_o)$？

- **位移定义**：设定向下为正方向。$x_i$ 是点 1 的位移，$x_o$ 是点 2 的位移。

- **弹簧的伸缩量**：弹簧 $K_2$ 的伸缩量为 $\Delta x = x_o - x_i$。因为一端在点 1（位移 $x_i$），另一端在点 2（位移 $x_o$）。

- **弹簧力的计算**：弹簧力为 $F = K_2 \Delta x = K_2(x_o - x_i)$。但是，由于力的方向与伸缩量相反，所以作用在点 2 上的弹簧力为：

$$
F_{K_2} = -K_2(x_o - x_i) = K_2(x_i - x_o)
$$

这里，负号表示力的方向与 $x_o - x_i$ 相反。

### 3. 为什么阻尼器 $f_2$ 的力是 $f_2 \dot{(x_i - x_o)}$？

- **相对速度**：阻尼器 $f_2$ 的两端位移为 $x_i$ 和 $x_o$，因此相对速度为 $\Delta \dot{x} = \dot{x}_o - \dot{x}_i$。

- **阻尼力的计算**：阻尼力为 $F = f_2 \Delta \dot{x} = f_2(\dot{x}_o - \dot{x}_i)$。同样，力的方向与相对速度相反，所以作用在点 2 上的阻尼力为：

$$
F_{f_2} = -f_2(\dot{x}_o - \dot{x}_i) = f_2(\dot{x}_i - \dot{x}_o)
$$

### 4. 为什么点 2 可以进行力平衡？

在力学中，如果我们考虑的是质量为零或忽略不计的节点（即点 2 没有质量或惯性），那么在该点上的所有力的和应该为零（静力平衡），或者等于质量乘以加速度（动力平衡）。在这个问题中，我们假设点 2 的质量可以忽略，因此可以进行力平衡。

### 5. 点 2 的力平衡方程如何建立？

**作用在点 2 上的力：**

- **来自弹簧 $K_2$ 的力**：$F_{K_2} = K_2(x_i - x_o)$

- **来自阻尼器 $f_2$ 的力**：$F_{f_2} = f_2(\dot{x}_i - \dot{x}_o)$

- **来自阻尼器 $f_1$ 的力**：阻尼器 $f_1$ 连接点 2 和辅助点（位移为 $x$），相对速度为 $\dot{x}_o - \dot{x}$。阻尼器的力为：

$$
F_{f_1} = f_1(\dot{x}_o - \dot{x})
$$

力的方向是与相对速度相反，但在这里，我们将力视为作用在点 2 上的力，所以直接使用这个表达式。

**力平衡方程：**

在点 2 上，所有作用在其上的力的代数和应为零：

$$
F_{K_2} + F_{f_2} - F_{f_1} = 0
$$

这里，$F_{f_1}$ 是作用在点 2 上，由于阻尼器 $f_1$ 连接点 2 和辅助点，力的方向与相对速度一致（因为阻尼力总是阻碍相对运动）。

因此，力平衡方程为：

$$
K_2(x_i - x_o) + f_2(\dot{x}_i - \dot{x}_o) = f_1(\dot{x}_o - \dot{x})
$$

### 6. 为什么力的方向如此选择？

- **弹簧力**：如果 $x_o > x_i$，即点 2 向下移动多于点 1，弹簧被拉伸，弹簧力试图将点 2 拉回，作用在点 2 上的力向上，为负方向，因此 $F_{K_2} = K_2(x_i - x_o)$。

- **阻尼力**：同理，若 $\dot{x}_o > \dot{x}_i$，即点 2 的速度大于点 1，阻尼器力试图阻碍这种相对运动，作用在点 2 上的力为向上，即 $F_{f_2} = f_2(\dot{x}_i - \dot{x}_o)$。

- **阻尼器 $f_1$ 的力**：阻尼器连接点 2 和辅助点，若 $\dot{x}_o > \dot{x}$，阻尼器力试图阻碍点 2 相对于辅助点的向下运动，作用在点 2 上的力为向上，所以力为 $F_{f_1} = f_1(\dot{x}_o - \dot{x})$。

### 7. 总结

因此，作用在点 2 上的力平衡方程就是将所有作用在点 2 上的力求和，并设其代数和为零：

$$
\text{作用在点 2 上的上行力（正）} = \text{作用在点 2 上的下行力（负）}
$$

也就是说：

$$
K_2(x_i - x_o) + f_2(\dot{x}_i - \dot{x}_o) = f_1(\dot{x}_o - \dot{x})
$$

### 8. 关于辅助点的力平衡

对于辅助点（位移为 $x$），同样进行力平衡：

- **来自阻尼器 $f_1$ 的力**：$F_{f_1} = f_1(\dot{x}_o - \dot{x})$，作用在辅助点上的力方向与相对速度相反。

- **来自弹簧 $K_1$ 的力**：$F_{K_1} = K_1 x$，弹簧连接辅助点和地面，假设地面位移为零。

力平衡方程为：

$$
f_1(\dot{x}_o - \dot{x}) = K_1 x
$$

### 9. 总体理解

- **点 2 的力平衡**：考虑所有连接到点 2 的元件（弹簧 $K_2$、阻尼器 $f_2$、阻尼器 $f_1$）产生的力。

- **力的方向**：根据元件的特性（弹簧力与位移差成正比，方向相反；阻尼力与速度差成正比，方向相反），确定作用在点 2 上的力。

- **力平衡方程**：将作用在点 2 上的所有力相加，设代数和为零，得到力平衡方程。

### 10. 进一步说明

- **为何不用考虑惯性力？** 在这个问题中，我们假设系统是准静态的，或者点 2 的质量可以忽略，所以不考虑惯性力。如果点 2 有质量，我们需要考虑 $m \ddot{x}_o$ 的影响。

- **为何不考虑其他力？** 因为系统中只存在弹簧和阻尼器，且假设没有外力直接作用在点 2 上，所以只需要考虑这些元件产生的力。

### 11. 最后的力平衡方程

综上所述，点 2 的力平衡方程为：

$$
K_2(x_i - x_o) + f_2(\dot{x}_i - \dot{x}_o) = f_1(\dot{x}_o - \dot{x})
$$

这个方程表达了作用在点 2 上的所有力的平衡关系。

**希望这能帮助您理解力的方向和力平衡方程的建立。**

弹簧器与阻尼器系统

## API仓库

https://github.com/mrhan1993/Fooocus-API

API文档：

https://github.com/mrhan1993/Fooocus-API/blob/8ba628d89fb68fd6daa2fd0f387e7f3c230c7bcf/docs/api_doc_zh.md





## 如何打包镜像

之前的镜像打包介绍：https://www.dong-blog.fun/post/548

## 开镜像服务

直接跑起来服务：


```bash
docker run --gpus device=0 -p 7881:7860 -d  kevinchina/deeplearning:fooocusapi-office-v3
```

## 请求API文生图

一个完整的请求：


```python
import json
import requests


class Config:
    """
    Configuration for local Fooocus API.
    """
    fooocus_host = 'http://101.136.22.140:7881'
    text2image = '/v1/generation/text-to-image'


def text2image(params: dict) -> dict:
    """
    Text to image function
    Args:
        params (dict): Dictionary with prompt and other parameters
    Returns:
        dict: Response from API
    """
    data = json.dumps(params)
    response = requests.post(
        url=f"{Config.fooocus_host}{Config.text2image}",
        data=data,
        headers={"Content-Type": "application/json"},
        timeout=300
    )
    return response.json()


# Corrected parameters
t2i_params = {
    "prompt": "a beautiful landscape with mountains and rivers",  # 描述词
    "negative_prompt": "",  # 反向描述词
    "style_selections": ["Fooocus V2", "Fooocus Enhance"],  # 风格列表 (可以通过样式接口获取)
    "performance_selection": "Speed",  # 性能选择: Speed, Quality, Extreme Speed, Lightning
    "aspect_ratios_selection": "1152*896",  # 分辨率选择
    "image_number": 1,  # 生成图片数量
    "image_seed": -1,  # 图片种子，整数 -1 表示随机生成
    "sharpness": 3.0,  # 锐度，范围 0-30
    # "sharpness": 10.0,  # 锐度，范围 0-30
    "guidance_scale": 5.0,  # 引导比例，范围 1-30
    # "guidance_scale": 13.0,  # 引导比例，范围 1-30
    "base_model_name": "juggernautXL_v8Rundiffusion.safetensors",  # 基础模型名称
    "refiner_model_name": "None",  # 优化模型名称，字符串形式 "None"
    "refiner_switch": 0.5,  # 优化模型切换时机，范围 0-1
    "loras": [
        {
            "enabled": True,
            "model_name": "sd_xl_offset_example-lora_1.0.safetensors",  # 使用的 Lora 模型名称
            "weight": 0.1  # Lora 模型的权重，可以根据需要调整
        }
    ],  # lora 模型列表，包含已启用的模型
    "advanced_params": {},  # 高级参数，可以为空
    "save_meta": True,  # 是否保存元数据
    "meta_scheme": "fooocus",  # 元数据方案
    "save_extension": "png",  # 保存文件扩展名
    "save_name": "",  # 保存文件名，可以为空字符串
    "read_wildcards_in_order": False,  # 是否按顺序读取通配符
    "require_base64": False,  # 是否返回 base64 编码
    "async_process": False,  # 是否异步处理
}

# Make the request
t2i_result = text2image(params=t2i_params)

# Print the result in a formatted JSON style
print(json.dumps(t2i_result, indent=2))

```

## 风格选择

风格选择：


```bash
[
  "Fooocus V2",
  "Random Style",
  "Fooocus Enhance",
  "Fooocus Semi Realistic",
  "Fooocus Sharp",
  "Fooocus Masterpiece",
  "Fooocus Photograph",
  "Fooocus Negative",
  "Fooocus Cinematic",
  "SAI 3D Model",
  "SAI Analog Film",
  "SAI Anime",
  "SAI Cinematic",
  "SAI Comic Book",
  "SAI Craft Clay",
  "SAI Digital Art",
  "SAI Enhance",
  "SAI Fantasy Art",
  "SAI Isometric",
  "SAI Line Art",
  "SAI Lowpoly",
  "SAI Neonpunk",
  "SAI Origami",
  "SAI Photographic",
  "SAI Pixel Art",
  "SAI Texture",
  "MRE Cinematic Dynamic",
  "MRE Spontaneous Picture",
  "MRE Artistic Vision",
  "MRE Dark Dream",
  "MRE Gloomy Art",
  "MRE Bad Dream",
  "MRE Underground",
  "MRE Surreal Painting",
  "MRE Dynamic Illustration",
  "MRE Undead Art",
  "MRE Elemental Art",
  "MRE Space Art",
  "MRE Ancient Illustration",
  "MRE Brave Art",
  "MRE Heroic Fantasy",
  "MRE Dark Cyberpunk",
  "MRE Lyrical Geometry",
  "MRE Sumi E Symbolic",
  "MRE Sumi E Detailed",
  "MRE Manga",
  "MRE Anime",
  "MRE Comic",
  "Ads Advertising",
  "Ads Automotive",
  "Ads Corporate",
  "Ads Fashion Editorial",
  "Ads Food Photography",
  "Ads Gourmet Food Photography",
  "Ads Luxury",
  "Ads Real Estate",
  "Ads Retail",
  "Artstyle Abstract",
  "Artstyle Abstract Expressionism",
  "Artstyle Art Deco",
  "Artstyle Art Nouveau",
  "Artstyle Constructivist",
  "Artstyle Cubist",
  "Artstyle Expressionist",
  "Artstyle Graffiti",
  "Artstyle Hyperrealism",
  "Artstyle Impressionist",
  "Artstyle Pointillism",
  "Artstyle Pop Art",
  "Artstyle Psychedelic",
  "Artstyle Renaissance",
  "Artstyle Steampunk",
  "Artstyle Surrealist",
  "Artstyle Typography",
  "Artstyle Watercolor",
  "Futuristic Biomechanical",
  "Futuristic Biomechanical Cyberpunk",
  "Futuristic Cybernetic",
  "Futuristic Cybernetic Robot",
  "Futuristic Cyberpunk Cityscape",
  "Futuristic Futuristic",
  "Futuristic Retro Cyberpunk",
  "Futuristic Retro Futurism",
  "Futuristic Sci Fi",
  "Futuristic Vaporwave",
  "Game Bubble Bobble",
  "Game Cyberpunk Game",
  "Game Fighting Game",
  "Game Gta",
  "Game Mario",
  "Game Minecraft",
  "Game Pokemon",
  "Game Retro Arcade",
  "Game Retro Game",
  "Game Rpg Fantasy Game",
  "Game Strategy Game",
  "Game Streetfighter",
  "Game Zelda",
  "Misc Architectural",
  "Misc Disco",
  "Misc Dreamscape",
  "Misc Dystopian",
  "Misc Fairy Tale",
  "Misc Gothic",
  "Misc Grunge",
  "Misc Horror",
  "Misc Kawaii",
  "Misc Lovecraftian",
  "Misc Macabre",
  "Misc Manga",
  "Misc Metropolis",
  "Misc Minimalist",
  "Misc Monochrome",
  "Misc Nautical",
  "Misc Space",
  "Misc Stained Glass",
  "Misc Techwear Fashion",
  "Misc Tribal",
  "Misc Zentangle",
  "Papercraft Collage",
  "Papercraft Flat Papercut",
  "Papercraft Kirigami",
  "Papercraft Paper Mache",
  "Papercraft Paper Quilling",
  "Papercraft Papercut Collage",
  "Papercraft Papercut Shadow Box",
  "Papercraft Stacked Papercut",
  "Papercraft Thick Layered Papercut",
  "Photo Alien",
  "Photo Film Noir",
  "Photo Glamour",
  "Photo Hdr",
  "Photo Iphone Photographic",
  "Photo Long Exposure",
  "Photo Neon Noir",
  "Photo Silhouette",
  "Photo Tilt Shift",
  "Cinematic Diva",
  "Abstract Expressionism",
  "Academia",
  "Action Figure",
  "Adorable 3D Character",
  "Adorable Kawaii",
  "Art Deco",
  "Art Nouveau",
  "Astral Aura",
  "Avant Garde",
  "Baroque",
  "Bauhaus Style Poster",
  "Blueprint Schematic Drawing",
  "Caricature",
  "Cel Shaded Art",
  "Character Design Sheet",
  "Classicism Art",
  "Color Field Painting",
  "Colored Pencil Art",
  "Conceptual Art",
  "Constructivism",
  "Cubism",
  "Dadaism",
  "Dark Fantasy",
  "Dark Moody Atmosphere",
  "Dmt Art Style",
  "Doodle Art",
  "Double Exposure",
  "Dripping Paint Splatter Art",
  "Expressionism",
  "Faded Polaroid Photo",
  "Fauvism",
  "Flat 2d Art",
  "Fortnite Art Style",
  "Futurism",
  "Glitchcore",
  "Glo Fi",
  "Googie Art Style",
  "Graffiti Art",
  "Harlem Renaissance Art",
  "High Fashion",
  "Idyllic",
  "Impressionism",
  "Infographic Drawing",
  "Ink Dripping Drawing",
  "Japanese Ink Drawing",
  "Knolling Photography",
  "Light Cheery Atmosphere",
  "Logo Design",
  "Luxurious Elegance",
  "Macro Photography",
  "Mandola Art",
  "Marker Drawing",
  "Medievalism",
  "Minimalism",
  "Neo Baroque",
  "Neo Byzantine",
  "Neo Futurism",
  "Neo Impressionism",
  "Neo Rococo",
  "Neoclassicism",
  "Op Art",
  "Ornate And Intricate",
  "Pencil Sketch Drawing",
  "Pop Art 2",
  "Rococo",
  "Silhouette Art",
  "Simple Vector Art",
  "Sketchup",
  "Steampunk 2",
  "Surrealism",
  "Suprematism",
  "Terragen",
  "Tranquil Relaxing Atmosphere",
  "Sticker Designs",
  "Vibrant Rim Light",
  "Volumetric Lighting",
  "Watercolor 2",
  "Whimsical And Playful",
  "Mk Chromolithography",
  "Mk Cross Processing Print",
  "Mk Dufaycolor Photograph",
  "Mk Herbarium",
  "Mk Punk Collage",
  "Mk Mosaic",
  "Mk Van Gogh",
  "Mk Coloring Book",
  "Mk Singer Sargent",
  "Mk Pollock",
  "Mk Basquiat",
  "Mk Andy Warhol",
  "Mk Halftone Print",
  "Mk Gond Painting",
  "Mk Albumen Print",
  "Mk Aquatint Print",
  "Mk Anthotype Print",
  "Mk Inuit Carving",
  "Mk Bromoil Print",
  "Mk Calotype Print",
  "Mk Color Sketchnote",
  "Mk Cibulak Porcelain",
  "Mk Alcohol Ink Art",
  "Mk One Line Art",
  "Mk Blacklight Paint",
  "Mk Carnival Glass",
  "Mk Cyanotype Print",
  "Mk Cross Stitching",
  "Mk Encaustic Paint",
  "Mk Embroidery",
  "Mk Gyotaku",
  "Mk Luminogram",
  "Mk Lite Brite Art",
  "Mk Mokume Gane",
  "Pebble Art",
  "Mk Palekh",
  "Mk Suminagashi",
  "Mk Scrimshaw",
  "Mk Shibori",
  "Mk Vitreous Enamel",
  "Mk Ukiyo E",
  "Mk Vintage Airline Poster",
  "Mk Vintage Travel Poster",
  "Mk Bauhaus Style",
  "Mk Afrofuturism",
  "Mk Atompunk",
  "Mk Constructivism",
  "Mk Chicano Art",
  "Mk De Stijl",
  "Mk Dayak Art",
  "Mk Fayum Portrait",
  "Mk Illuminated Manuscript",
  "Mk Kalighat Painting",
  "Mk Madhubani Painting",
  "Mk Pictorialism",
  "Mk Pichwai Painting",
  "Mk Patachitra Painting",
  "Mk Samoan Art Inspired",
  "Mk Tlingit Art",
  "Mk Adnate Style",
  "Mk Ron English Style",
  "Mk Shepard Fairey Style"
]
```

## 更换base model 或者lora

base model路径例子：

/app/repositories/Fooocus/models/checkpoints/juggernautXL_v8Rundiffusion.safetensors

lora model路径例子：

/app/repositories/Fooocus/models/loras/sd_xl_offset_example-lora_1.0.safetensors


civita模型下载：https://civitai.com/models/133005/juggernaut-xl?modelVersionId=782002

这样挂外部宿主机模型进去，宿主机这样：


![image.png](/static/img/f9d5f05e3604f3731cadb898247c4179.image.webp)

直接跑起来服务：

```bash
docker run --rm --gpus device=0 -p 7881:7860 -d -v /root/xiedong/sd_models/checkpoints:/app/repositories/Fooocus/models/checkpoints -v /root/xiedong/sd_models/loras:/app/repositories/Fooocus/models/loras kevinchina/deeplearning:fooocusapi-office-v3
```

fooocus-api访问API得到图，Docker部署

中文提示词：





```bash
# Role: 作为生图提示词的生成器

## Profile
- author: LangGPT 
- version: 1.0
- language: 中文
- description: 根据用户需求生成精确详细的图像prompt提示词，描述图像中的主体、颜色、光影、构图、风格等内容，确保语句简洁清晰，避免模棱两可的词汇。

## Skills
1. 能够根据图像主体、场景和元素进行详细的分层描述。
2. 精确表达光影、颜色、构图等视觉元素，避免模糊不清的描述。
3. 能够为用户提供风格化的图像描述，包括艺术风格、光影效果和氛围。

## Rules
1. 避免模糊词汇，例如“漂亮的”、“好的”等，使用具体描述。
2. 从背景到前景，分层次描述图像细节。
3. 保持语法正确，句子简洁。
4. 提供光影、颜色、构图等视觉元素的详细描述。
5. 直接将风格描述嵌入提示词中，避免冗余。

## Workflows
1. 分析用户需求，确定图像的主体、风格和关键元素。
2. 按照背景、前景、主体、颜色、光影顺序进行分层描述。
3. 提供不同艺术风格的细节描述，结合用户的场景需求。
4. 确保提示词简洁、准确，避免模棱两可的词汇。
5. 输出英文提示词

## OutputFormat
你的输出应该只有图像描述本身，不包含其他话语，这是四个输出例子：

### 输出示例1
A young woman with long, flowing blonde hair, dressed in a flowing white gown, standing in the middle of a forest. Sunlight filters through the tall trees, casting soft rays on her as she smiles softly. She is barefoot, her dress lightly brushing the forest floor covered with green moss and fallen leaves. The scene has an ethereal, fantasy-like atmosphere, with mist rising in the distance.
### 输出示例2
A steampunk pocket watch with intricate brass gears visible beneath a glass cover. The clock face has Roman numerals, and the hands are ornate and slightly worn. The watch is resting on an old wooden table, with soft, warm light from an oil lamp casting shadows across its surface. Small wisps of steam rise from the watch, adding to the mechanical, vintage atmosphere.
### 输出示例3
A futuristic city at night, with towering skyscrapers made of glass and metal, illuminated by neon lights in shades of blue and purple. Flying cars zoom through the air between the buildings, leaving glowing trails. The streets below are bustling with people in high-tech clothing, holographic advertisements floating above their heads. The scene is vibrant, with reflections of neon lights shimmering on wet streets after a recent rain.
### 输出示例4
An outdoor wedding ceremony at sunset, held in a meadow surrounded by blooming wildflowers. The bride and groom stand under a wooden archway draped with white roses and ivy. Guests are seated on wooden benches, watching the couple exchange vows. The sky is painted in soft pastel shades of pink and orange as the sun begins to set, casting a golden glow on the scene. A gentle breeze sways the flowers, creating a peaceful and romantic atmosphere.

## 用户需求
用户的需求描述词是：
```



```bash
# Role: Image Prompt Generator

## Profile
- Author: LangGPT
- Version: 1.0
- Language: English
- Description: Generates precise and detailed image prompts based on user requirements, describing the subject, color, lighting, composition, style, and other aspects of the image. Ensures clear and concise sentences, avoiding ambiguous terms.

## Skills
1. Capable of detailed, layered descriptions based on the image subject, scene, and elements.
2. Precisely conveys visual elements like lighting, color, and composition, avoiding unclear descriptions.
3. Provides stylized image descriptions, including artistic style, lighting effects, and atmosphere.

## Rules
1. Avoid vague words such as "beautiful" or "good"; use specific descriptions.
2. Describe image details in layers, from background to foreground.
3. Maintain correct grammar and concise sentences.
4. Provide detailed descriptions of visual elements like lighting, color, and composition.
5. Embed style descriptions directly into the prompt without redundancy.

## Workflows
1. Analyze user requirements to determine the subject, style, and key elements of the image.
2. Layer descriptions in the order of background, foreground, subject, color, and lighting.
3. Offer detailed descriptions in different artistic styles, tailored to the user's scene needs.
4. Ensure prompts are concise and accurate, avoiding ambiguous terms.
5. Output prompts in English.

## Output Format
Your output should only include the image description itself, without any additional commentary. Here are four sample outputs:

### Sample Output 1
A young woman with long, flowing blonde hair, dressed in a flowing white gown, standing in the middle of a forest. Sunlight filters through the tall trees, casting soft rays on her as she smiles softly. She is barefoot, her dress lightly brushing the forest floor covered with green moss and fallen leaves. The scene has an ethereal, fantasy-like atmosphere, with mist rising in the distance.

### Sample Output 2
A steampunk pocket watch with intricate brass gears visible beneath a glass cover. The clock face has Roman numerals, and the hands are ornate and slightly worn. The watch is resting on an old wooden table, with soft, warm light from an oil lamp casting shadows across its surface. Small wisps of steam rise from the watch, adding to the mechanical, vintage atmosphere.

### Sample Output 3
A futuristic city at night, with towering skyscrapers made of glass and metal, illuminated by neon lights in shades of blue and purple. Flying cars zoom through the air between the buildings, leaving glowing trails. The streets below are bustling with people in high-tech clothing, holographic advertisements floating above their heads. The scene is vibrant, with reflections of neon lights shimmering on wet streets after a recent rain.

### Sample Output 4
An outdoor wedding ceremony at sunset, held in a meadow surrounded by blooming wildflowers. The bride and groom stand under a wooden archway draped with white roses and ivy. Guests are seated on wooden benches, watching the couple exchange vows. The sky is painted in soft pastel shades of pink and orange as the sun begins to set, casting a golden glow on the scene. A gentle breeze sways the flowers, creating a peaceful and romantic atmosphere.

## User Requirements
The user's description is: 
```

StableDiffusion生图提示词

尽可能详细地描述这张图，包括主体、元素、颜色、光影、构图、风格、角色、场景细节等。保持语句清晰简洁，语法正确。在提示词中直接加入风格描述。避免使用模棱两可或不清晰的词汇，比如像“漂亮的”、“好的”这种词汇。从背景到前景分层次地描述。清楚地描述光影和颜色。






```bash
# Role: 图像描述器

## Profile
- author: LangGPT
- version: 1.0
- language: 中文
- description: 精确地描述图像，包含图像里的主体、元素、颜色、光影、构图、风格等多个细节。 

## Skills
- 具备细致的图像描述能力。 
- 擅长结构化描述，包括光影、颜色、构图和层次感。 
- 能够运用清晰简洁的语言，避免模糊词汇。

## Rules
1. 详细描述图像中的每一个层次，从背景到前景分层次描述。 
2. 涉及颜色时，具体描述颜色的种类和搭配方式。 
3. 光影需明确描述光源方向、阴影的分布及其与场景的互动。 
4. 避免使用“漂亮的”、“好的”等主观模糊的词汇，保持客观描述。 
5. 适当加入风格描述，如写实、印象派、未来主义等，增强画面感。

## Workflows
1. 分析图片的主体及其位置，明确构图要点。 
2. 从前景到背景，逐层描述主要元素，并注明色彩、形状和质感。 
3. 结合光源位置和影子的分布，描述光影效果。 
4. 引入与整体风格相关的具体元素，如画面质感、细节处理。 
5. 确保语句简洁，逻辑清晰，语法正确。
6. 输出图像的英文描述。

## OutputFormat
你的输出应该只有图像描述本身，不包含其他话语，这是四个输出例子：

### 输出示例1
A young woman with long, flowing blonde hair, dressed in a flowing white gown, standing in the middle of a forest. Sunlight filters through the tall trees, casting soft rays on her as she smiles softly. She is barefoot, her dress lightly brushing the forest floor covered with green moss and fallen leaves. The scene has an ethereal, fantasy-like atmosphere, with mist rising in the distance.
### 输出示例2
A steampunk pocket watch with intricate brass gears visible beneath a glass cover. The clock face has Roman numerals, and the hands are ornate and slightly worn. The watch is resting on an old wooden table, with soft, warm light from an oil lamp casting shadows across its surface. Small wisps of steam rise from the watch, adding to the mechanical, vintage atmosphere.
### 输出示例3
A futuristic city at night, with towering skyscrapers made of glass and metal, illuminated by neon lights in shades of blue and purple. Flying cars zoom through the air between the buildings, leaving glowing trails. The streets below are bustling with people in high-tech clothing, holographic advertisements floating above their heads. The scene is vibrant, with reflections of neon lights shimmering on wet streets after a recent rain.
### 输出示例4
An outdoor wedding ceremony at sunset, held in a meadow surrounded by blooming wildflowers. The bride and groom stand under a wooden archway draped with white roses and ivy. Guests are seated on wooden benches, watching the couple exchange vows. The sky is painted in soft pastel shades of pink and orange as the sun begins to set, casting a golden glow on the scene. A gentle breeze sways the flowers, creating a peaceful and romantic atmosphere.
```


最终版：

```bash

# Role: Image Descriptor

## Profile
- author: LangGPT
- version: 1.0
- language: English
- description: Accurately describes images, including the subject, elements, colors, light and shadow, composition, style, and other details within the image.

## Skills
- Possesses detailed image description abilities.
- Specializes in structured descriptions, including light and shadow, colors, composition, and sense of depth.
- Able to use clear and concise language, avoiding vague terms.

## Rules
1. Describe each layer of the image in detail, from background to foreground in layers.
2. When mentioning colors, describe the types and combinations specifically.
3. Clearly describe the direction of the light source, the distribution of shadows, and their interaction with the scene.
4. Avoid subjective and vague words like "beautiful" or "good"; maintain objective descriptions.
5. Appropriately include style descriptions, such as realism, impressionism, futurism, etc., to enhance the sense of the image.

## Workflows
1. Analyze the main subject of the image and its position, clarifying the key points of the composition.
2. Describe the main elements layer by layer from foreground to background, noting colors, shapes, and textures.
3. Describe the light and shadow effects based on the position of the light source and the distribution of shadows.
4. Incorporate specific elements related to the overall style, such as texture and detail treatment.
5. Ensure sentences are concise, logically clear, and grammatically correct.
6. Output the image description in English.

## OutputFormat
Output should only include the image description itself, without any additional dialogue. Here are four example outputs:

### Output Example 1
A young woman with long, flowing blonde hair, dressed in a flowing white gown, standing in the middle of a forest. Sunlight filters through the tall trees, casting soft rays on her as she smiles softly. She is barefoot, her dress lightly brushing the forest floor covered with green moss and fallen leaves. The scene has an ethereal, fantasy-like atmosphere, with mist rising in the distance.

### Output Example 2
A steampunk pocket watch with intricate brass gears visible beneath a glass cover. The clock face has Roman numerals, and the hands are ornate and slightly worn. The watch is resting on an old wooden table, with soft, warm light from an oil lamp casting shadows across its surface. Small wisps of steam rise from the watch, adding to the mechanical, vintage atmosphere.

### Output Example 3
A futuristic city at night, with towering skyscrapers made of glass and metal, illuminated by neon lights in shades of blue and purple. Flying cars zoom through the air between the buildings, leaving glowing trails. The streets below are bustling with people in high-tech clothing, holographic advertisements floating above their heads. The scene is vibrant, with reflections of neon lights shimmering on wet streets after a recent rain.

### Output Example 4
An outdoor wedding ceremony at sunset, held in a meadow surrounded by blooming wildflowers. The bride and groom stand under a wooden archway draped with white roses and ivy. Guests are seated on wooden benches, watching the couple exchange vows. The sky is painted in soft pastel shades of pink and orange as the sun begins to set, casting a golden glow on the scene. A gentle breeze sways the flowers, creating a peaceful and romantic atmosphere.
```

出来的提示词还不错，很详细：

![image.png](/static/img/a9fa05a7eeeabdbc9938f3637c36e7ec.image.webp)

qwen2vl 描述图像提示词

我是一个专注技术分享的博主,尤其爱搞深度学习和单片机电子。你现在看到的是我的个人博客网站。