Wan 2.7 -- Generate a video from image

POST

/services/aigc/video-generation/video-synthesis

curl --location 'https://dashscope-intl.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \
  -H 'X-DashScope-Async: enable' \
  -H "Authorization: Bearer $DASHSCOPE_API_KEY" \
  -H 'Content-Type: application/json' \
  -d '{
  "model": "wan2.7-i2v",
  "input": {
    "prompt": "A scene of urban fantasy art. A dynamic graffiti art character. A boy made of spray paint comes to life on a concrete wall. He sings an English rap song at high speed while striking a classic, energetic rapper pose. The scene is set under an urban railway bridge at night. The light comes from a single street lamp, creating a cinematic atmosphere full of high energy and amazing detail. The audio of the video consists entirely of the rap, with no other dialogue or noise.",
    "media": [
      {
        "type": "first_frame",
        "url": "https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/wpimhv/rap.png"
      },
      {
        "type": "driving_audio",
        "url": "https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/ozwpvi/rap.mp3"
      }
    ]
  },
  "parameters": {
    "resolution": "720P",
    "duration": 10,
    "prompt_extend": true,
    "watermark": true
  }
}'

{
  "request_id": "4909100c-7b5a-9f92-bfe5-xxxxxx",
  "output": {
    "task_id": "0385dc79-5ff8-4d82-bcb6-xxxxxx",
    "task_status": "PENDING"
  }
}

Generate videos up to 15 seconds at 1080P from images, audio, and video clips, with optional audio sync and first-last frame control.

Changes from wan2.6

Unified API: First-frame, first-last-frame, and video continuation share one endpoint via the media array -- no separate APIs.
Audio-video sync: Provide a driving_audio file for lip-syncing. If omitted, the model auto-generates matching sound effects.
Resolution control: Set resolution (720P/1080P) instead of exact pixel size.
Longer prompts: Up to 5,000 characters (was 800).
Negative prompt moved: Now under input.negative_prompt instead of parameters.negative_prompt.
Watermark off by default: watermark defaults to false (was true).

Authorizations

string

header

required

DashScope API Key. Get one from the Qwen Cloud console.

Header Parameters

enum<string>

required

Must be set to enable for asynchronous task submission.

Available options:enable

Body

application/json

enum<string>

required

Model identifier. Fixed value: wan2.7-i2v.

Available options:wan2.7-i2v

Example:wan2.7-i2v

object

required

Input data for video generation.

Show child attributes

object[]

required

Reference materials for video generation. Each element specifies a type and url. Each type can appear at most once.

Supported combinations:

first_frame
first_frame + driving_audio
first_frame + last_frame
first_frame + last_frame + driving_audio
first_clip
first_clip + last_frame

Show child attributes

enum<string>

required

Asset type.

first_frame: First frame image. JPEG/JPG/PNG/BMP/WEBP, 240-8000 px per side, 1:8 to 8:1 ratio, max 20 MB.
last_frame: Last frame image. Same limits as first_frame.
driving_audio: Audio for lip-sync or speech-driven animation. WAV/MP3, 2-30 s, max 15 MB. If omitted, the model auto-generates matching sound effects.
first_clip: Input video for continuation. MP4/MOV, 2-10 s, 240-4096 px per side, 1:8 to 8:1 ratio, max 100 MB.

Available options:first_frame,last_frame,driving_audio,first_clip

string

required

Public URL (HTTP/HTTPS) of the media asset.

string

Describe the video you want. Supports Chinese and English, up to 5,000 characters (auto-truncated if exceeded).

Example:A kitten runs on the grass.

string

Describe what to exclude from the video (e.g., low quality, blurry, extra fingers). Supports Chinese and English, up to 500 characters (auto-truncated if exceeded).

Example:low resolution, error, worst quality, low quality, deformed, extra fingers, bad proportions

object

Video generation parameters.

Show child attributes

enum<string>

default"1080P"

Video clarity tier. The model automatically scales the output to match the selected tier, preserving the aspect ratio of the input material. Higher resolution costs more.

Available options:720P,1080P

integer

default5

Video duration in seconds (integer, 2-15). Longer videos cost more -- billing is per second. For video continuation (first_clip), this is the total output duration including the input clip. For example, if duration=15 and the input is 3 s, the model generates 12 s of new content.

Required range:2 <= x <= 15

boolean

defaulttrue

Rewrite the prompt using a Large Language Model before generation. Improves results for short or vague prompts but adds latency. Set to false to use your prompt verbatim.

boolean

defaultfalse

Add an "AI Generated" watermark in the lower-right corner.

integer

Seed for reproducible results. Same seed + same parameters produces similar (not identical) output.

Required range:0 <= x <= 2147483647

Response

200-application/json

string

Unique request identifier for tracing and troubleshooting.

Example:4909100c-7b5a-9f92-bfe5-xxxxxx

object

Show child attributes

string

Task ID for polling status. Use with GET /tasks/{task_id}.

Example:0385dc79-5ff8-4d82-bcb6-xxxxxx

enum<string>

Initial task status, typically PENDING.

Available options:PENDING,RUNNING,SUCCEEDED,FAILED

​Changes from wan2.6

Authorizations

Header Parameters

Body

Response

Changes from wan2.6