Davinci Magihuman
Generate a human-centered video from an existing reference image and a text prompt using MagiHuman. A valid reference image is required even when audio is provided. Optional audio drives lip synchronization and cannot replace the image. If a local reference image already exists, call ae_file_service_get_upload_addr, upload its bytes to uploadURL, and pass downloadURL as image. Existing public image URLs do not need re-uploading. If no reference image exists, obtain one from the user or generate one before submitting this tool. A text prompt and audio alone are not sufficient. The tool returns task_id; use ae_get_result to obtain progress and the final output.
Pricing
| Resolution | Charge |
|---|---|
| 256p | 24 credits/second |
| 720p / Not specified | 48 credits/second |
| 1080p | 72 credits/second |
An omitted duration is charged as five seconds. Starting price applies to Resolution: 256p.
Parameters
Optional parameters use the platform defaults when omitted.
| Parameter | Type | Required | Description | Default |
|---|---|---|---|---|
request_body | object | Yes | Davinci 视频生成请求。 | — |
request_body.seed | integer | No | 随机种子,-1 表示随机。 | -1 |
request_body.audio | string | No | 可选,公开可访问的真实音频 URL,用于驱动人物口型与音频同步。提供 audio 后仍必须填写有效参考图片 image;音频不能代替图片。已有本地音频时可先通过 fileserver 上传,再使用返回的 downloadURL。不使用音频时省略此参数。 | — |
request_body.image | string | Yes | 必填,公开可访问的真实参考图片 URL,视频中的人物或主体以这张图片为基础。不得为空、传 null,或使用音频/视频文件链接代替;不能把 audio 的链接复制到 image。已有本地图片时,先调用 ae_file_service_get_upload_addr 获取 uploadURL/downloadURL,按 uploadURL 上传图片字节,再把 downloadURL 填入 image,不要填写 uploadURL 或本地路径。已有公开图片链接无需重复上传。没有参考图时,请先让用户提供或通过生图工具生成图片;仅有文字和音频不能替代本工具必需的图片输入。 | — |
request_body.prompt | string | Yes | 必填,描述参考图片中人物或主体的动作、场景、表情和镜头变化;文字提示不用于替代必需的 image。提供音频时,可在此说明口型与音频同步等要求。 | — |
request_body.duration | enum (5 | 6 | 7 | 8 | 9 | 10) | No | 生成视频的时长,单位为秒,可选 5、6、7、8、9、10,默认 5 秒。 | 5 |
request_body.resolution | enum (256p | 720p | 1080p) | No | 输出视频分辨率,可选 256p、720p、1080p,默认 720p。 | 720p |
request_body.aspect_ratio | enum (16:9 | 9:16) | No | 输出宽高比。 | 16:9 |
Invocation example
Copy the request and explore its response example.
/agent-api/v1/tool/executeX-Api-KeySubmit first, then query with the returned task_id until the task completes.
curl --request POST \
--url 'https://agentearth.ai/agent-api/v1/tool/execute?tool_name=ae_davinci_magihuman' \
--header "X-Api-Key: $AGENTEARTH_API_KEY" \
--header 'Content-Type: application/json' \
--data-raw '{
"params": {
"request_body": {
"aspect_ratio": "9:16",
"duration": 10,
"image": "https://agentearth.ai/fileserver/file/a57adf6b-0b59-43cb-aa98-51e99f247b86.png",
"prompt": "A professional Chinese male news anchor in a navy suit stands facing the camera in a TV studio, speaking directly to the audience with a serious, confident expression. Natural small head movements, occasional subtle hand gestures, steady upper body, professional presenter delivery, mouth moves naturally as if reporting urgent logistics news. Fixed medium shot, vertical composition.",
"resolution": "720p"
}
}
}'{
"description": "task accepted; poll the result via query_tool with this task_id",
"query_tool": "ae_get_result",
"status": "created",
"task_id": "d68a9d96-3488-4298-8660-1d6e3d377292"
}This example shows part of the response. Copied content is valid JSON.
Guides using this tool
See how this tool fits into real workflows


