First & Last Frame to Video

Doc version：v1.0.0 | Last updated：2026-07-22
This platform fully supports the official Tongyi Wanxiang (Wan) video generation API. Requests and responses are transparently proxied; parameter semantics are identical to the official API.

Provide a first frame image and a last frame image, together with a text prompt, to generate a video that transitions smoothly between the two frames (kf2v series). The output video is fixed at 5 seconds, silent, MP4 (H.264 encoded), and exactly 1 video is produced. Tasks usually take 1–5 minutes; only asynchronous invocation is supported (create task → poll for results).

Supported Models

Model	Resolution Tiers	Default Resolution	Duration	Effect Templates	Description
`wan2.2-kf2v-flash`	480P / 720P / 1080P	720P	Fixed 5 seconds	—	Version 2.2, high cost-effectiveness; query response returns the resolution tier in `usage.SR`
`wanx2.1-kf2v-plus`	720P only	720P	Fixed 5 seconds	Supported (e.g. `mech1`, `hanfu-1`)	Version 2.1; query response `usage.video_ratio` is always `standard`

Resolution and duration directly affect cost: for the same model, 1080P > 720P > 480P; output is billed by the number of seconds of successfully generated video, and failed tasks are not billed.

Create Task

Note: The submission endpoint for first & last frame to video is image2video, which is different from the video-generation endpoint used for text-to-video / image-to-video.

POST https://platform.dataeyes.ai/ali/api/v1/services/aigc/image2video/video-synthesis

Request Headers

Parameter	Type	Required	Default	Description
`Content-Type`	string	Yes	application/json	Data exchange format
`Authorization`	string	Yes		Authentication, `Bearer {API_KEY}`

The X-DashScope-Async: enable header required by the official API is automatically added by the platform — you do not need to include it; including it is also compatible.

Request Body

Parameter	Type	Required	Default	Description
`model`	string	Yes		Model name. Options: `wan2.2-kf2v-flash`, `wanx2.1-kf2v-plus`
`input`	object	Yes		Basic input information
`input.first_frame_url`	string	Yes		First frame image. Supports: public URL (HTTP/HTTPS, recommended), Base64 (`data:{MIME_type};base64,{base64_data}`). The output video aspect ratio is based on this image; see the next section for image requirements
`input.last_frame_url`	string	No		Last frame image, same input format as above. In the regular first-and-last-frame scenario it is effectively required; ignored when using a `template` effect. Its resolution may differ from the first frame — no forced alignment is required
`input.prompt`	string	No		Text prompt, Chinese or English, ≤800 characters, automatically truncated if exceeded. When the subject/scene changes significantly between the first and last frames, it is recommended to describe the transition process (camera moves such as "the camera pans left", subject motion such as "a person running forward")
`input.negative_prompt`	string	No		Negative prompt describing content you do not want to appear, Chinese or English, ≤500 characters, automatically truncated if exceeded
`input.template`	string	No		Video effect template name (e.g. `mech1` Mech Awakening, `hanfu-1` Tang-Style Hanfu). Supported model: `wanx2.1-kf2v-plus`. When used, only `first_frame_url` is needed; `prompt` and `last_frame_url` are ignored. Passing only a last frame together with a template returns an error
`parameters`	object	No		Video processing parameters
`parameters.resolution`	string	No	720P	Resolution tier. flash: `480P` / `720P` / `1080P`; plus: `720P` only. Only adjusts total pixel clarity and does not change the aspect ratio (which follows the first frame). Directly affects cost
`parameters.duration`	integer	No	5	Video duration in seconds, fixed at 5, cannot be changed
`parameters.prompt_extend`	bool	No	true	Whether to enable intelligent prompt rewriting. Noticeably improves short prompts, but increases processing time
`parameters.seed`	integer	No	Random	Random seed, range [0, 2147483647]. Fixing the seed improves result reproducibility, but does not guarantee identical output
`parameters.watermark`	bool	No	false	Whether to add a watermark ("AI generated" mark in the bottom-right corner of the video)

Input Image Requirements

Item	Requirement
Count	1 first frame + 1 last frame (effect template mode uses the first frame only)
Format	JPEG, JPG, PNG (alpha channel not supported), BMP, WEBP
Resolution	Both width and height within [240, 8000] pixels
File size	≤10MB
First/last frame relationship	The last frame's resolution may differ from the first frame's — no forced alignment; the output video aspect ratio is based on the first frame

Output size rule: The model takes the first frame's aspect ratio as the baseline and, combined with the total pixel count of the resolution tier, computes the closest legal resolution; video width and height must be multiples of 16, so slight adjustments occur and a strict ratio cannot be guaranteed. Example: first frame 750×1000 (3:4) + 720P → actual output approximately 816×1104.

Request Example (wan2.2-kf2v-flash)

curl --request POST \
  --url 'https://platform.dataeyes.ai/ali/api/v1/services/aigc/image2video/video-synthesis' \
  --header 'Authorization: Bearer <token>' \
  --header 'Content-Type: application/json' \
  --data '{
    "model": "wan2.2-kf2v-flash",
    "input": {
      "first_frame_url": "https://wanx.alicdn.com/material/20250318/first_frame.png",
      "last_frame_url": "https://wanx.alicdn.com/material/20250318/last_frame.png",
      "prompt": "Realistic style, a small black cat curiously looks up at the sky, the camera gradually rises from eye level"
    },
    "parameters": {
      "resolution": "480P",
      "prompt_extend": true
    }
  }'

Request Example (wanx2.1-kf2v-plus)

curl --request POST \
  --url 'https://platform.dataeyes.ai/ali/api/v1/services/aigc/image2video/video-synthesis' \
  --header 'Authorization: Bearer <token>' \
  --header 'Content-Type: application/json' \
  --data '{
    "model": "wanx2.1-kf2v-plus",
    "input": {
      "first_frame_url": "https://wanx.alicdn.com/material/20250318/first_frame.png",
      "last_frame_url": "https://wanx.alicdn.com/material/20250318/last_frame.png",
      "prompt": "Petals drifting down in the wind, the camera pulls back from close-up to a wide shot"
    },
    "parameters": {
      "prompt_extend": true
    }
  }'

Response Example

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

Response Fields

Field	Type	Description
`output.task_id`	string	Task ID used for polling, valid for 24 hours
`output.task_status`	string	Initial task status, usually `PENDING`
`request_id`	string	Unique request identifier
`code` / `message`	string	Returned only when creation fails: error code and details (e.g. `InvalidParameter` for non-conforming image dimensions)

Query Task

Poll the task status using the task_id returned by the create-task call; once the status is SUCCEEDED, retrieve the video URL from output.video_url (valid for 24 hours).

GET https://platform.dataeyes.ai/ali/api/v1/tasks/{task_id}

See 09-Query-Task for details.

05-First-Last-Frame-to-Video

First & Last Frame to Video#

Supported Models#

Create Task#

Request Headers#

Request Body#

Input Image Requirements#

Request Example (wan2.2-kf2v-flash)#

Request Example (wanx2.1-kf2v-plus)#

Response Example#

Response Fields#

Query Task#