生成图像

把客户端 base_url 配成 https://ark.oairoute.com/doubao/api/v3，即可沿用方舟的字段名直接对接。

请求体是平铺的 JSON，没有 content 数组。字段与方舟 API 一致并原样透传，上游新增参数无需网关升级。

文生图、图生图还是多图融合，只由 image 字段决定：

model 必须写全名。seedream-5 之类的缩写会返回 400。

Point the client's base_url at https://ark.oairoute.com/doubao/api/v3 and Ark's field names work as-is.

The body is flat JSON with no content array. Fields match the Ark API and pass through unchanged, so new upstream parameters need no gateway upgrade.

Text-to-image, single-image edit, or multi-image fusion is decided solely by image:

model must be the full name. Abbreviations such as seedream-5 return 400.

同步返回，data 中每项对应一张生成的图片：

{
  "data": [
    {
      "url": "https://ark-content-generation-v2-cn-beijing.tos-cn-beijing.volces.com/seedream/xxx.jpeg",
      "size": "2048x2048",
      "output_format": "jpeg"
    }
  ],
  "created": 1757469067,
  "usage": {
    "generated_images": 1,
    "output_tokens": 15908,
    "total_tokens": 15908
  }
}

• data[].url (string) — 有效期 24 小时的图片链接，response_format 为 url 时返回。
• data[].b64_json (string) — response_format 为 b64_json 时返回的 Base64 内容。
• data[].size (string) — 该张图片的实际像素尺寸。
• data[].output_format (string) — 文件格式。图层固定为 png，底图跟随请求中的 output_format。
• data[].z_index (integer) — 图层叠放顺序，仅图层拆分时返回。底图固定为 0，图层从 1 起递增。
• data[].name / data[].description (string) — 图层名称与描述，仅图层拆分时返回。
• data[].bounding_box (object) — 图层在底图坐标系中的区域，含 absolute 与 normalized 两套坐标。
• data[].error (object) — 单图失败信息，仅组图场景中某一张失败时出现。
• created (integer) — 完成时间，Unix 秒。
• usage (object) — 用量信息，含 generated_images、input_images、output_tokens、total_tokens。

The response is synchronous; each entry in data is one generated image:

{
  "data": [
    {
      "url": "https://ark-content-generation-v2-cn-beijing.tos-cn-beijing.volces.com/seedream/xxx.jpeg",
      "size": "2048x2048",
      "output_format": "jpeg"
    }
  ],
  "created": 1757469067,
  "usage": {
    "generated_images": 1,
    "output_tokens": 15908,
    "total_tokens": 15908
  }
}

• data[].url (string) — an image link valid for 24 hours, returned when response_format is url.
• data[].b64_json (string) — inline Base64, returned when response_format is b64_json.
• data[].size (string) — the actual pixel size of that image.
• data[].output_format (string) — the file format. Layers are always png; the base image follows the request.
• data[].z_index (integer) — layer stacking order, layer decomposition only. The base image is 0; layers count up from 1.
• data[].name / data[].description (string) — layer name and description, layer decomposition only.
• data[].bounding_box (object) — the layer's region in base-image coordinates, with absolute and normalized sets.
• data[].error (object) — per-image failure, present only when one image in a group fails.
• created (integer) — completion time in Unix seconds.
• usage (object) — usage facts: generated_images, input_images, output_tokens, total_tokens.

更多状态码与重试建议见错误码与限流，常见报错的排查步骤见常见问题。

For more status codes and retry guidance see Error codes and rate limits; for troubleshooting steps see the FAQ.

以下图像模型均可通过本接口提交任务，清单只列各厂商最新一代，实际可用范围以你的账号分组为准：

Every image model below can be submitted through this endpoint; the list shows only each vendor's latest generation, and what your account group actually allows takes precedence:

同步返回，无需轮询

提交后直接在响应里拿 data[].url。链接 24 小时有效，需要长期保存请及时转存。

组图

sequential_image_generation: "auto" 一次最多返回 4 张同主题图片，例如同一角色在早晨、正午、傍晚的样子。某一张失败时该数组项只带 error，其余照常返回。

图层还原

拆分得到的图层按 z_index 从小到大叠放。绝对坐标换算为 x=left、y=top、w=right−left、h=bottom−top；任意 W×H 画布用归一化值除以 1000 再乘画布尺寸。

关闭水印

默认带「AI 生成」水印，商用场景必须显式传 watermark: false。是否允许去水印取决于你的账号分组。

Synchronous, no polling

Read data[].url straight from the response. Links last 24 hours, so download anything you need to keep.

Image groups

sequential_image_generation: "auto" returns up to 4 images on one theme, such as the same character in the morning, at noon, and in the evening. If one fails, that array entry carries only error while the rest still return.

Layer reconstruction

Stack the returned layers by ascending z_index. Convert absolute boxes with x=left, y=top, w=right−left, h=bottom−top; on an arbitrary W×H canvas, divide the normalized values by 1000 and multiply by the canvas size.

Turn off the watermark

An "AI generated" watermark is on by default; commercial use must pass watermark: false explicitly. Whether removal is allowed depends on your account group.