Video to Video (Motion Control)
AI-generatedSummary
Generate a motion-controlled video by applying the movements from a reference video to a character depicted in an input image using selected Kling model versions.
Inputs
- Model — Select the motion control model version to use for video generation (Kling 3.0 or Kling 2.6). Defaults to Kling 3.0.
- Character Image URL (required) — URL of the reference character image (jpg/jpeg/png; max 10MB; minimum 300px; aspect ratio between 2:5 and 5:2) that defines the characters appearing in the output video.
- Reference Video URL (required) — URL of the reference video (mp4/mov; max 100MB; duration 3–30 seconds) whose character actions will be emulated in the generated video.
- Prompt — Optional textual prompt (up to 2500 characters) to guide or influence the video generation.
- Character Orientation — Choose whether the character orientation in the output matches the orientation from the reference video or the character image; selecting 'image' limits output video length to 10 seconds.
- Resolution — Output video resolution, either 720p or 1080p, defaulting to 720p.
- Reply URL — Webhook URL to be called when the video generation task completes.
- Reply Ref — Custom reference string passed back in the webhook callback to correlate responses.
- Captcha Token — reCAPTCHA v3 Enterprise token; if omitted, kie.ai manages captcha automatically.
- Wait for Completion — Boolean flag to specify if the node should wait for video generation to complete before returning; defaults to true.
Output shape
a JSON array containing API response objects, either immediate task creation response or the completed task result with video URLs, one per input item.
If 'Wait for Completion' is true, the output includes the final video generation result after polling the task status; otherwise, it returns the initial task creation response with a task ID. Errors are returned per item if 'continueOnFail' is set.