seedream(doubao)-image-generation

  }
---

# Seedream Image Generation Skill

This skill provides methods for calling the Volcengine Ark large model service platform's image generation models (Seedream).

## Select Environment

The script provides implementations for both Python and JavaScript:
- If your runtime is primarily Node.js/TypeScript or a web module, use **scripts/seedream.js**.
- If your runtime involves complex data analysis, batch scheduling, or is purely Python, use **scripts/seedream.py**.

## Supported Model Parameters (Model ID/Endpoint ID)

When invoking, the `model` parameter needs to be the deployment Endpoint ID from the Ark console (e.g., `ep-202X...`), which must be backed by one of the following foundational image generation models:
- `doubao-seedream-5-0-260128`
- `doubao-seedream-4-5-251128`
- `doubao-seedream-4-0-250828`

## Script Parameters

You must pass specific parameters to control the image generation. The available parameters are:

- **model** *(string, optional)*: Endpoint ID corresponding to the model. Defaults to `doubao-seedream-5-0-260128`.
- **prompt** *(string, required)*: Text description used to generate the image.
- **size** *(string, optional)*: Target image size. Supports `"2K"`, `"3K"`, `"4K"`.
  - `doubao-seedream-5-0` supports **2K** and **3K**.
  - `doubao-seedream-4` and `doubao-seedream-4-5` support **2K** and **4K**.
  - If not provided, this parameter is completely ignored in the API request.
  - You can also specify the desired resolution directly in the `prompt`. It is highly recommended to explicitly state the resolution in the `prompt` for the best results.
- **image** *(string | list of strings, optional)*: Local image path or image path list used for image-to-image generation. The scripts will read the local file, convert it to base64, and send it as the request `image` field.
- **watermark** *(boolean, optional)*: Whether to add a Volcengine AI watermark. Unprovided value assumes `false`.
- **optimize_prompt_options** *(object, optional)*: Configure automatic prompt optimization options. Format: JSON object. E.g., `{"mode": "standard"}`.
- **tools** *(list of objects, optional)*: Capabilities options for the model, formatted as JSON. Example: `[{"type": "web_search"}]`.
- **output_format** *(string, optional)*: The output image format. Options include `png`, `jpeg`.
- **sequential_image_generation** *(string, optional)*: Sequential image generation strategy (e.g., `"auto"`).
- **download_dir** *(string, optional)*: Local directory to save the generated images. If not provided, images will not be downloaded locally, and only the API response will be returned.

## Recommended width and height values

| Resolution | Aspect Ratio | Dimensions |
| :--------- | :----------- | :--------- |
| **2K**     | 1:1          | 2048x2048  |
|            | 4:3          | 2304x1728  |
|            | 3:4          | 1728x2304  |
|            | 16:9         | 2848x1600  |
|            | 9:16         | 1600x2848  |
|            | 3:2          | 2496x1664  |
|            | 2:3          | 1664x2496  |
|            | 21:9         | 3136x1344  |
| **3K**     | 1:1          | 3072x3072  |
|            | 4:3          | 3456x2592  |
|            | 3:4          | 2592x3456  |
|            | 16:9         | 4096x2304  |
|            | 9:16         | 2304x4096  |
|            | 2:3          | 2496x3744  |
|            | 3:2          | 3744x2496  |
|            | 21:9         | 4704x2016  |
| **4K**     | 1:1          | 4096x4096  |
|            | 3:4          | 3520x4704  |
|            | 4:3          | 4704x3520  |
|            | 16:9         | 5504x3040  |
|            | 9:16         | 3040x5504  |
|            | 2:3          | 3328x4992  |
|            | 3:2          | 4992x3328  |
|            | 21:9         | 6240x2656  |


## Usage Examples

**Python Example:**
```bash
python3 <current_skill_dir>/scripts/seedream.py \
  --prompt "A cute orange cat laying under the sun" \
  --model "ep-xxxxx..." \
  --size "1024x1024" \
  --watermark "false" \
  --output_format "png"
```

**Node.js Example:**
```bash
node <current_skill_dir>/scripts/seedream.js \
  --prompt "A cute orange cat laying under the sun" \
  --model "ep-xxxxx..." \
  --size "1024x1024" \
  --watermark "false" \
  --output_format "png"
```

python3 <current_skill_dir>/scripts/seedream.py \
  --prompt "Turn this product photo into a clean white-background ecommerce image" \
  --image "/path/to/source.png" \
  --model "ep-xxxxx..."

**Example with Download Directory:**
```bash
python3 <current_skill_dir>/scripts/seedream.py \
  --prompt "A beautiful sunset over the mountains" \
  --download_dir "/Users/vincent/Downloads/generated_images" \
  --model "ep-xxxxx..."
```

python3 <current_skill_dir>/scripts/seedream.py \
  --prompt "A futuristic cityscape" \
  --model "ep-xxxxx..." \
  --optimize_prompt_options '{"mode": "standard"}' \
  --tools '[{"type": "web_search"}]'

## Troubleshooting

If the scripts encounter issues, please check the following:
- **SEEDREAM_API_KEY**: (env, required) Ensure this API key is set in your environment. It is the core authentication for Volcengine Ark.
- **SEEDREAM_BASE_URL**: (env, optional) Ensure this is correct if you are using a custom endpoint. Defaults to `https://ark.cn-beijing.volces.com/api/v3/images/generations`.

For more detailed API documentation, please visit: [https://www.volcengine.com/docs/82379/1541523?lang=zh](https://www.volcengine.com/docs/82379/1541523?lang=zh)
For more model id list or check the newest model id if required, please visit: [https://www.volcengine.com/docs/82379/1330310?lang=zh#36969059](https://www.volcengine.com/docs/82379/1330310?lang=zh#36969059), and then update this seedream-image-generation/SKILL.md file "Supported Model Parameters".