Skip to main content
POST
/
generate
/
bailian
/
wan2-keyframe-to-video
Tongyi Wanxiang Wan2 Keyframe-to-Video
curl --request POST \
  --url https://api.sinancode.com/v1/generate/bailian/wan2-keyframe-to-video \
  --header 'Authorization: Bearer <token>' \
  --header 'Content-Type: application/json' \
  --data '
{
  "first_frame_url": "<string>",
  "last_frame_url": "<string>",
  "model": "wan2.2-kf2v-flash",
  "negative_prompt": "<string>",
  "prompt": "<string>",
  "prompt_extend": true,
  "resolution": "480P",
  "seed": 1073741823,
  "template": "<string>",
  "watermark": true
}
'
{
  "task_id": "01234567-89ab-cdef-0123-456789abcdef"
}

Documentation Index

Fetch the complete documentation index at: https://docs.sinancode.com/llms.txt

Use this file to discover all available pages before exploring further.

Generate smooth dynamic videos by providing just the first and last frame images using Alibaba Cloud’s Tongyi Wanxiang Keyframe-to-Video model. Key Features:
  • Keyframe Control - Precise control over video start and end frames
  • Smooth Transition - Automatically generate silky smooth intermediate frame animations
  • Multi-Resolution - Support 480P, 720P, 1080P output
  • Smart Rewrite - Support prompt intelligent rewriting for better results
  • Effect Templates - Support preset video effect templates

Authorizations

​
Authorization
string
header
required

Please use your API Token for authentication

Body

application/json

Task parameters

​
first_frame_url
string
required

First frame image URL (required) Supports public URLs (HTTP/HTTPS) or Base64 encoding Image format: JPEG, JPG, PNG (no transparency), BMP, WEBP Image resolution: width and height range [360, 2000] pixels File size: max 10MB Output video aspect ratio will be based on this image

​
last_frame_url
string

Last frame image URL (optional) Same requirements as first frame, resolution can differ from first frame Not required when using effect template

​
model
enum<string>

Model name (optional, recommended to leave blank) When left blank, the latest version will be used and future upgrades will be applied automatically Current default: wan2.2-kf2v-flash

Available options:
wan2.2-kf2v-flash
​
negative_prompt
string

Negative prompt (optional, max 500 characters) Describe content you don't want to see in the video

Maximum string length: 500
​
prompt
string

Text prompt (optional, max 800 characters) Describe the elements and visual features you want in the generated video When the subject/scene changes significantly between first and last frames, it's recommended to describe the transition process (e.g., camera movement, subject motion)

Maximum string length: 800
​
prompt_extend
boolean

Enable smart prompt rewriting (optional, default true) When enabled, uses large language model to intelligently rewrite input prompts

​
resolution
enum<string>

Resolution tier (optional, default 720P) wan2.2-kf2v-flash: available values 480P, 720P, 1080P

Available options:
480P,
720P,
1080P
​
seed
integer

Random seed (optional, range 0-2147483647)

Required range: 0 <= x <= 2147483647
​
template
string

Video effect template name (optional) When using this parameter, only need to provide first_frame_url Different models support different effect templates

​
watermark
boolean

Add watermark (optional, default false) Watermark is located at bottom right corner with fixed text "AI Generated"

Response

OK

​
task_id
string
required

Task ID

Example:

"01234567-89ab-cdef-0123-456789abcdef"