多图生成视频

通过 CometAPI POST /kling/v1/videos/multi-image2video 使用多张图片创建 Kling 视频，并可配置运动控制和生成输出设置。

POST /kling/v1/videos/multi-image2video

Create a Kling multi-image-to-video task

cURL

curl --request POST \
  --url https://api.cometapi.com/kling/v1/videos/multi-image2video \
  --header 'Authorization: Bearer <token>' \
  --header 'Content-Type: application/json' \
  --data '
{
  "model_name": "kling-v1-6",
  "duration": "5",
  "image_list": [
    {
      "image": "https://h2.inkwai.com/bs2/upload-ylab-stunt/se/ai_portal_queue_mmu_image_upscale_aiweb/3214b798-e1b4-4b00-b7af-72b5b0417420_raw_image_0.jpg"
    },
    {
      "image": "https://h2.inkwai.com/bs2/upload-ylab-stunt/se/ai_portal_queue_mmu_image_upscale_aiweb/3214b798-e1b4-4b00-b7af-72b5b0417420_raw_image_0.jpg"
    },
    {
      "image": "https://filesystem.site/cdn/20250414/chxiLc2O45zoLT8BCrQ6WQlTvGDDnK.png"
    },
    {
      "image": "https://filesystem.site/cdn/20250414/chxiLc2O45zoLT8BCrQ6WQlTvGDDnK.png"
    }
  ],
  "prompt": "The astronauts got up and left."
}
'

200

{
  "code": 0,
  "message": "SUCCEED",
  "data": {
    "task_id": "861309198188019719",
    "task_status": "submitted",
    "task_info": {},
    "created_at": 1773380633979,
    "updated_at": 1773380633979
  }
}

当一张源图像不足以满足需求，并且你希望 Kling 基于多张图像参考构建运动效果时，请使用此端点。

​

调用前准备

• 在 image_list 中准备 2 到 4 张图片
• 从 model_name: kling-v1-6 或你已验证的其他当前模型开始
• 首次测试尽量保持简单：使用一条简短 Prompt，不使用运动遮罩，不添加额外的回调设置

​

任务流程

1. 提交多图请求

   发送图像列表、Prompt 和时长，然后保存返回的 Kling task_id。

2. 轮询任务

   继续参考 单个查询，直到任务进入终态。

3. 持久化结果

   如果你需要在提供商交付 URL 的保留期之外继续保存成品视频，请将其存储下来。

在 2026-03-13 的实时验证中，已确认一个最小化的双图请求可被接受，并返回 task_status: submitted。

完整参数说明请参见 Kling 官方文档。

授权

Authorization · string · header · 必填

Bearer token authentication. Use your CometAPI key.

请求体

application/json

model_name · string · 默认值: kling-v1-6 · 必填

Kling model variant to use. See the Models page for current options.

duration · string · 默认值: 5 · 必填

Output video length in seconds. Options: 5, 10.

image_list · object[] · 必填

2 to 4 source images. Each item provides an image field as a public URL or Base64 string (without data: prefix). Accepted formats: JPG, JPEG, PNG. Maximum 10 MB per image, minimum 300×300 px, aspect ratio between 1:2.5 and 2.5:1.

Show child attributes

prompt · string · 默认值: The cat turns its head and blinks. · 必填

Text prompt describing the desired motion. Maximum 500 characters.

negative_prompt · string

Elements to exclude from the video. Maximum 200 characters.

callback_url · string

Webhook URL to receive task status updates when the task completes.

mode · string

Generation mode. std for standard (faster), pro for professional (higher quality).

cfg_scale · number

Prompt adherence strength. Higher values follow the prompt more closely. Range: 0–1.

static_mask · string

Static brush mask image as a Base64 string or public URL. White areas are frozen in place during video generation. Must match the aspect ratio and resolution of the input image.

dynamic_masks · object

Dynamic brush configuration. Up to 6 groups, each containing a painted-area mask and a motion trajectory.

Show child attributes

external_task_id · string

Custom task id for your own tracking. Does not replace the system-generated task id but can be used to query tasks. Must be unique per user.

响应

200 - application/json

Task accepted.

code · integer · 必填

Error code; specifically define the error code

message · string · 必填

error message

data · object · 必填

Show child attributes