kling-2.6 / motion-control
Commercial usePart of the Kling API family — compare all Kling models →

Developed by KuaiShou, Kling AI 2.6 Motion Control API is a performance-driven image-to-video model that transfers real human motion, gestures, and expressions from reference video to character images with stable timing and realism.

Pricing: 11 credits/s ($0.055) for 720P; 18 credits/s ($0.09) for 1080P High-tier top-ups (+10% bonus) bring effective pricing down to ~$0.050/s (720 p) and ~$0.081/s (1080 p).
Input

A text description of the desired output. Maximum length is 2500 characters.

0/2500

Click to upload or drag and drop

Supported formats: JPEG, PNG, JPG Maximum file size: 10MB; Maximum files: 1

Reference image. The characters, backgrounds, and other elements in the generated video are based on the reference image. Supports .jpg/.jpeg/.png, max 10MB, size needs to be greater than 300px, aspect ratio 2:5 to 5:2.

Click to upload or drag and drop

Supported formats: MP4, QUICKTIME Maximum file size: 100MB; Maximum files: 1

Reference video. The character actions in the generated video are consistent with the reference video. Supports .mp4/.mov, max 100MB, 3-30 seconds duration depending on character_orientation.

Generate the orientation of the characters in the video. 'image': same orientation as the person in the picture (max 10s video). 'video': consistent with the orientation of the characters in the video (max 30s video).

Output resolution mode. Use 'std' for 720p or 'pro' for 1080p.

Output
output typevideo
Examples

Explore different use cases and parameter configurations

README

Complete guide to using kling-2.6/motion-control

Kling 2.6 Motion Control API: Transfer Motion to a Character Image

Kling's kling-2.6/motion-control model applies the motion from one reference video to the subject in one reference image. It outputs 720p or 1080p video from a 3 to 30 second motion video, with character orientation from the image (videos up to 10 s) or from the video (up to 30 s).

Original image

Kling 2.6 Motion Control API Parameters

Send requests to POST /api/v1/jobs/createTask with the model ID and the parameters below. A request accepts at most one image and one video.

ParameterDetails
modelRequired. Set to kling-2.6/motion-control.
input_urlsRequired. Array containing one image URL (file URL after upload, not file content). Formats: JPEG, PNG, or JPG; up to 10MB; larger than 300px; aspect ratio 2:5 to 5:2; must clearly show the subject's head, shoulders, and torso.
video_urlsRequired. Array containing one video URL that defines the motion pattern to apply. Formats: MP4 or QuickTime; up to 100MB; 3 to 30 seconds; must clearly show the subject's head, shoulders, and torso.
character_orientationRequired. Set to "image" to match the orientation of the person in the picture (motion video up to 10 s), or "video" to stay consistent with the orientation of the characters in the video (motion video up to 30 s).
modeRequired. Output resolution: "720p" or "1080p".
promptOptional. Text description of the desired output, up to 2500 characters.
callBackUrlOptional. Webhook URL to receive task results (recommended for production). You can also poll the Get Task Details endpoint.

Kling 2.6 Motion Control Test Observations on Kie.ai

Each video was generated with Kling 2.6 Motion Control on Kie.ai in October 2026 from the same inputs: a 3D cartoon chef image we generated on Kie.ai with nano-banana-2-1, and the example motion video from the Kie.ai doc (a woman on a street making a heart with her hands, pointing, touching her hair and crossing her arms; 8.88 s). We checked that each output has an audio track but did not judge the audio.

Image Orientation, 720p, Kitchen Prompt

We set character_orientation to "image" and mode to "720p", with the prompt "The chef makes the same hand gestures in a bright restaurant kitchen." The API returned an 816x1104 MP4 of 8.53 s with an audio track. The chef follows the gestures, faces the camera, and stays in place. The first frame keeps the plain light grey studio background of the chef image, and from about 1 s the background becomes a restaurant kitchen as prompted; between about 2 s and 5 s a blurred figure appears in the kitchen background, which the prompt did not ask for.

Video Orientation, 1080p, No Prompt

We set character_orientation to "video" and mode to "1080p", with no prompt. The API returned a 1232x1680 MP4 of 8.73 s with an audio track. The plain light grey studio background of the reference image is kept, and the chef follows the same hand gestures (heart hands, pointing, a hand to the hair, crossed arms). The framing changes from a close shot at the start to a wider full-figure shot, and the chef's body turns slightly with the movement in the reference video.

Kling 2.6 Motion Control vs Kling 3.0 Motion Control on Kie.ai

This table compares documented parameters on Kie.ai only.

Kling 2.6 Motion ControlKling 3.0 Motion Control
model IDkling-2.6/motion-controlkling-3.0/motion-control
promptOptional, up to 2500 charactersOptional, 0 to 2500 characters
image1 URL (JPEG, PNG, JPG), up to 10MB, > 300px, aspect ratio 2:5 to 5:21 URL, > 340px, aspect ratio 2:5 to 5:2
motion video1 URL (MP4, QuickTime), up to 100MB, 3 to 30 s1 URL, up to 100MB, 3 to 30 s, > 340px, aspect ratio 2:5 to 5:2
character_orientationRequired: "image" (max 10 s video) or "video" (max 30 s video)Optional: "video" (default, recommended) or "image"
modeRequired: "720p" or "1080p"Optional: "720p" (default) or "1080p"

How to Use the Kling 2.6 Motion Control API on Kie.ai

Get started with our product in just a few simple steps...

  • 1. Create an API Key and Set the Authorization Header

    Create a key on the API Key Management page and send it as a Bearer token in the Authorization header of every request.

  • 2. Upload Your Image and Motion Video and Submit the Task

    Get URLs for one character image and one motion video; for local files, upload them first with the Kie.ai File Upload API to obtain file URLs. Then send a POST request to /api/v1/jobs/createTask with model set to kling-2.6/motion-control, input_urls, video_urls, character_orientation, mode, and an optional prompt or callBackUrl.

  • 3. Receive the Generated Video via Callback or Polling

    For production workflows, pass a callBackUrl in your request so Kie.ai sends the task result when generation finishes. You can also query the Get Task Details endpoint with the returned task ID to retrieve the output MP4 URL.

Tips Before You Ship

  • 01

    Match character_orientation to Documented Video Length

    Follow the documented duration rules when setting character_orientation: "image" supports motion videos up to 10 seconds, and "video" supports motion videos up to 30 seconds, so use "video" whenever your motion video is longer than 10 seconds.

  • 02

    Frame Head, Shoulders, and Torso in Both Inputs

    Both the reference image in input_urls and the motion video in video_urls must clearly show the subject's head, shoulders, and torso.

  • 03

    Send One Image URL and One Video URL Within Size Limits

    Each request accepts a maximum of one image and one video. Keep the image within 10MB (JPEG, PNG, or JPG, larger than 300px, with an aspect ratio between 2:5 and 5:2) and the video within 100MB (MP4 or QuickTime, 3 to 30 seconds).

  • 04

    Pass File URLs Using the File Upload API

    The input_urls and video_urls fields accept file URLs rather than raw file content. If your media is stored locally, upload the files with the Kie.ai File Upload API first and pass the returned URLs into your createTask payload.

    • 4.9/ 5
    45,673 ratings
    Pick a reaction

    Why Run Kling 2.6 Motion Control on Kie.ai

    What Kling Says VIDEO 2.6 Motion Control Does

    According to Kling, VIDEO 2.6 Motion Control takes a static character image and a 3 to 30 second movement video, supporting full-body movement sync with facial expressions, complex motion such as dance and martial arts, detailed hand movements and gestures, and text descriptions together with the motion.

    Browser Playground and REST API Access

    Kie.ai provides this model as an API with a playground; there is no app to download or install. You can try this model in the playground on this page without writing code, then move to POST /api/v1/jobs/createTask with webhook callbacks or task polling.

    Unified API Key and Credit Balance Across Models

    All models in the Kie.ai market use the same authentication method: an Authorization: Bearer YOUR_API_KEY header with the key from your Kie.ai API Key Management page. New Kie.ai users receive 80 free credits for testing. Your account has one credit balance that all generation services draw from; check it with the Get Remaining Credits API.

    Frequently Asked Questions about Kling AI 2.6 Motion Control API

    New Kie.ai users receive 80 free credits for testing. You can use them to try this model in the playground on this page without writing code, or to call the Kling 2.6 Motion Control API. Once they are used up, you can top up credits; the price is shown in the pricing card above.

    The price for 720p and 1080p output is shown in the pricing card above. Your account has one credit balance that all generation services draw from; check it with the Get Remaining Credits API.

    Yes. Kling motion control applies the motion pattern of a reference video to the subject in a static reference image. According to Kling, VIDEO 2.6 Motion Control takes a static character image and a video of a movement.

    Create a key on your Kie.ai API Key Management page. All models in the Kie.ai market use the same authentication method: an Authorization: Bearer YOUR_API_KEY header with the key from your Kie.ai API Key Management page.

    The required character_orientation parameter controls how the subject is oriented and how long the motion video can be. Setting it to "image" keeps the same orientation as the person in the picture and is documented for motion videos up to 10 seconds. Setting it to "video" stays consistent with the orientation of the characters in the video and supports motion videos up to 30 seconds; see the parameter table above.

    Each request accepts at most one image URL and one video URL; to get URLs for local files, upload them first with the Kie.ai File Upload API. The image must be JPEG, PNG, or JPG, up to 10MB, larger than 300px, with an aspect ratio of 2:5 to 5:2. The motion video must be MP4 or QuickTime, up to 100MB, and 3 to 30 seconds long. Both files must clearly show the subject's head, shoulders, and torso; see the parameter table above for full details.

    According to the Kie.ai documentation, the output may keep the audio track of the reference video. In all three of our October 2026 tests on Kie.ai, each returned MP4 contained an audio track. We only checked that the track existed; we did not listen to or evaluate the audio.

    According to Kling's Motion Control user guide, Kling VIDEO 3.0 Motion Control builds upon the Motion Control introduced in VIDEO 2.6 with improved facial consistency. In the Kie.ai documentation, kling-2.6/motion-control requires character_orientation and mode and needs an image larger than 300px. The kling-3.0/motion-control model makes mode (default 720p) and character_orientation (default video) optional, and needs both the image and the video to be larger than 340px with an aspect ratio of 2:5 to 5:2. See the comparison table on this page for the documented differences, and check each model's pricing card for its price.

    You can pass a callBackUrl parameter in your createTask request so Kie.ai sends the result to your webhook when the job finishes, which is recommended for production. Alternatively, you can poll the Get Task Details endpoint using the task ID returned when the job was created; see the parameter table above.