Wan 2.7 First-Last Frame to Video

Submit an asynchronous Ali wan2.7-i2v first-and-last-frame video generation task.

Use first_frame to specify the start frame, last_frame to specify the end frame, optionally paired with driving_audio for audio-driven generation.

Supported media combinations:

• first_frame + last_frame
• first_frame + last_frame + driving_audio

The prompt field is optional for this capability. size only supports 720P and 1080P. seconds only supports integer strings from 2 to 15. Also supports official fields negative_prompt, prompt_extend, watermark, and seed.

POST /v1/videos

cURL

first_last_frame

curl --request POST \
  --url https://baze-api.powerbuyin.top/v1/videos \
  --header 'Authorization: Bearer <token>' \
  --header 'Content-Type: application/json' \
  --data '
{
  "model": "wan2.7-i2v",
  "prompt": "A cat walking from one end of the hallway to the other.",
  "media": [
    {
      "type": "first_frame",
      "url": "https://example.com/assets/start.png"
    },
    {
      "type": "last_frame",
      "url": "https://example.com/assets/end.png"
    }
  ],
  "seconds": "5",
  "size": "1080P"
}
'

200

{
  "id": "<string>",
  "task_id": "<string>",
  "object": "video",
  "model": "<string>",
  "status": "queued",
  "progress": 123,
  "created_at": 123
}

Authorizations

Authorization · string · header · required

Pass Authorization: Bearer <token> in the request header.

Body

application/json

model · enum<string> · required

Model name. Available value: wan2.7-i2v.

Available options: wan2.7-i2v

media · object[] · required

List of media assets (images, audio) for video generation. Each element is a media object with type and url.

Supported combinations (invalid combinations will return an error):

• First & last frame: first_frame + last_frame
• First & last frame + audio: first_frame + last_frame + driving_audio

Each type can appear at most once in the array.

Show child attributes

prompt · string

Text prompt describing the desired video content and visual characteristics. Supports Chinese and English, up to 5000 characters (each character or letter counts as one). Excess characters will be automatically truncated.

size · enum<string> · default: 1080P

Specifies the output resolution tier for controlling video clarity (total pixels).

The model automatically scales to a similar total pixel count based on the selected resolution tier. The video aspect ratio will attempt to stay consistent with the input material (first frame or first clip).

Available options: 720P, 1080P

seconds · enum<string> · default: 5

Video duration as a string. Supports integer strings from 2 to 15.

Available options: 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15

negative_prompt · string

Negative prompt describing content you do not want in the video. Supports Chinese and English, up to 500 characters. Excess characters will be automatically truncated. Example: low resolution, worst quality, deformed, extra fingers, bad proportions.

prompt_extend · boolean

Whether to enable intelligent prompt rewriting. When enabled, a large language model rewrites the input prompt intelligently, significantly improving results for short prompts but increasing processing time. true (default): enable. false: disable.

watermark · boolean · default: false

Whether to add a watermark. The watermark is positioned at the bottom-right corner with the text 'AI生成'. false (default): no watermark. true: add watermark.

seed · integer

Random seed, range [0, 2147483647]. When unspecified, the system auto-generates a random seed. For better reproducibility, set a fixed seed. Note: due to the probabilistic nature of model generation, the same seed does not guarantee identical results every time.

Required range: 0 <= x <= 2147483647

Response

200

application/json

Submitted successfully. Returns an OpenAI-compatible video task object.

id · string

Public task ID.

task_id · string

Backward-compatible alias for the public task ID.

object · string

Example:

"video"

model · string

Model name.

status · enum<string>

Unified video task status.

Available options: queued, in_progress, completed, failed, unknown

progress · integer

Task progress percentage.

created_at · integer<int64>

Creation time as a Unix timestamp (seconds).